Significance of Tests and

Properties of Concrete
and Concrete-Making
Materials
STP 169D

Joseph F. Lamond and James H. Pielert, Editors

ASTM Stock No.: STP169D

ASTM International
100 Barr Harbor Drive
PO Box C-700
West Conshohocken, PA 19428-2959

Printed in the U.S.A.

Copyright © 2006 ASTM International, West Conshohocken, PA. All rights reserved. This material may
not be reproduced or copied, in whole or in part, in any printed, mechanical, electronic, film, or other
distribution and storage media, without the written consent of the publisher.

Photocopy Rights

Authorization to photocopy items for internal, personal, or educational classroom use, or the internal,
personal, or educational classroom use of specific clients, is granted by ASTM International (ASTM)
provided that the appropriate fee is paid to the Copyright Clearance Center, 222 Rosewood Drive,
Danvers, MA 01923; Tel: 978-750-8400; online: http://www.copyright.com/.

NOTE: The Society is not responsible, as a body, for the statements and opinions expressed in this
publication.

Printed in Bridgeport, NJ
April 2006
Foreword
THIS PUBLICATION is a revision and expansion of Significance of Tests and Properties of
Concrete and Concrete-Making Materials (STP 169C) published in 1994. That publication in
turn replaced editions published in 1956, 1966, and 1978. The present publication includes a
number of new materials and test methods that have been developed, or materials that have
increased in importance since the 1994 edition. Two most useful additions are the chapters on
slag as a cementitious material and self-consolidating concrete.
As in the previous editions, chapters have been authored by individuals selected on the ba-
sis of their knowledge of their subject areas, and in most cases because of their participation
in the development of pertinent specifications and test methods by ASTM Committee C09 on
Concrete and Concrete Aggregates and, in some cases, ASTM Committee C01 on Cement. The
authors developed their chapters in conformance with general guidelines only. Each chapter
has been reviewed and, where necessary, coordinated with chapters in which overlap of sub-
ject matter might occur.
This latest edition has been developed under the direction of the Executive Committee of
ASTM Committee C09 by coeditors Joseph F. Lamond, Consulting Engineer, and James H.
Pielert, Consultant, both members of Committee C09.
Contents
Chapter 1: Introduction—JOSEPH F. LAMOND AND JAMES H. PIELERT ..................................1

PART I
GENERAL
Chapter 2: The Nature of Concrete—RICHARD A. HELMUTH AND RACHEL J. DETWILER ......5
Chapter 3: Techniques, Procedures, and Practices of Sampling of Concrete
and Concrete Making Materials—TOY S. POOLE ...........................................16
Chapter 4: Statistical Considerations in Sampling and Testing—
GARLAND W. STEELE ................................................................................................22
Chapter 5: Uniformity of Concrete-Making Materials—ANTHONY E. FIORATO .......30
Chapter 6: Virtual Testing of Cement and Concrete—DALE P. BENTZ,
EDWARD J. GARBOCZI, JEFFREY W. BULLARD, CHIARA FERRARIS, NICOS MARTYS,
AND PAUL E. STUTZMAN ...........................................................................................38
Chapter 7: Quality Cement, Concrete, and Aggregates—The Role
of Testing Laboratories—JAMES H. PIELERT .....................................................51

PART II
FRESHLY MIXED CONCRETE
Chapter 8: Factors Influencing Concrete Workability—D. GENE DANIEL ...............59
Chapter 9: Air Content, Temperature, Density (Unit Weight), and
Yield—LAWRENCE R. ROBERTS ................................................................................73
Chapter 10: Making and Curing Concrete Specimens—JOSEPH F. LAMOND ...........80
Chapter 11: Time of Setting—BRUCE J. CHRISTENSEN .....................................................86
Chapter 12: Bleed Water—STEVEN H. KOSMATKA ...........................................................99

PART III
HARDENED CONCRETE
Chapter 13: Concrete Strength Testing—CELIK OZYILDIRIM
AND NICHOLAS J. CARINO........................................................................................125
Chapter 14: Prediction of Potential Concrete Strength at Later Ages—
NICHOLAS J. CARINO ..............................................................................................141
Chapter 15: Freezing and Thawing—CHARLES K. NMAI ............................................154
Chapter 16: Corrosion of Reinforcing Steel—NEAL S. BERKE .................................164
Chapter 17: Embedded Metals and Materials Other Than
Reinforcing Steel—BERNARD ERLIN ..................................................................174
Chapter 18: Abrasion Resistance—KARL J. BAKKE ....................................................184
Chapter 19: Elastic Properties, Creep, and Relaxation—JASON WEISS .................194
Chapter 20: Petrographic Examination—BERNARD ERLIN ........................................207
Chapter 21: Volume Change—FRED GOODWIN ...........................................................215
Chapter 22: Thermal Properties—STEPHEN B. TATRO .................................................226
Chapter 23: Pore Structure, Permeability, and Penetration Resistance
Characteristics of Concrete—NATALIYA HEARN, R. DOUGLAS HOOTON,
AND MICHELLE R. NOKKEN ......................................................................................238
Chapter 24: Chemical Resistance of Concrete—M. D. A. THOMAS
AND J. SKALNY .....................................................................................................253
Chapter 25: Resistance to Fire and High Temperatures—STEPHEN S. SZOKE ........274
Chapter 26: Air Content and Density of Hardened Concrete—
KENNETH C. HOVER ...............................................................................................288
Chapter 27: Analyses for Cement and Other Materials in Hardened
Concrete—WILLIAM G. HIME..............................................................................309
Chapter 28: Nondestructive Tests—V. MOHAN MALHOTRA .......................................314
vi CONTENTS

PART IV
CONCRETE AGGREGATES
Chapter 29: Grading, Shape, and Surface Texture—ROBIN E. GRAVES ..................337
Chapter 30: Bulk Density, Relative Density (Specific Gravity), Pore
Structure, Absorption, and Surface Moisture—JOHN J. YZENAS, JR. ........346
Chapter 31: Soundness, Deleterious Substances, and Coatings—
STEPHEN W. FORSTER ...................................................................................355
Chapter 32: Degradation Resistance, Strength, and Related Properties of
Aggregates—RICHARD C. MEININGER .................................................................365
Chapter 33: Petrographic Evaluation of Concrete Aggregates—
G. SAM WONG......................................................................................................377
Chapter 34: Alkali-Silica Reactions in Concrete—DAVID STARK.............................401
Chapter 35: Alkali-Carbonate Rock Reaction—MICHAEL A. OZOL ..........................410
Chapter 36: Thermal Properties of Aggregates—D. STEPHEN LANE .......................425

PART V
OTHER CONCRETE MAKING MATERIALS
Chapter 37: Hydraulic Cements—Physical Properties—LESLIE STRUBLE ................435
Chapter 38: Hydraulic Cement-Chemical Properties—SHARON M. DEHAYES
AND PAUL D. TENNIS ....................................................................................450
Chapter 39: Mixing and Curing Water for Concrete—JAMES S. PIERCE .................462
Chapter 40: Curing and Materials Applied to New Concrete
Surfaces—BEN E. EDWARDS ...............................................................................467
Chapter 41: Air-Entraining Admixtures—ARA A. JEKNAVORIAN ...............................474
Chapter 42: Chemical Admixtures—BRUCE J. CHRISTENSEN AND HAMID FARZAM ..........484
Chapter 43: Supplementary Cementitious Materials—SCOTT SCHLORHOLTZ .........495
Chapter 44: Slag as a Cementitious Material—JAN R. PRUSINSKI ...........................512

PART VI
SPECIALIZED CONCRETES
Chapter 45: Ready Mixed Concrete—COLIN L. LOBO AND RICHARD D. GAYNOR ..........533
Chapter 46: Lightweight Concrete and Aggregates—THOMAS A. HOLM
AND JOHN P. RIES ..................................................................................................548
Chapter 47: Cellular Concrete—FOUAD H. FOUAD .....................................................561
Chapter 48: Concrete for Radiation Shielding—DOUGLAS E. VOLKMAN .................570
Chapter 49: Fiber-Reinforced Concrete—PETER C. TATNALL .....................................578
Chapter 50: Preplaced Aggregate Concrete—EDWARD P. HOLUB ...........................591
Chapter 51: Roller-Compacted Concrete (RCC)—WAYNE S. ADASKA .....................595
Chapter 52: Polymer-Modified Concrete and Mortar—D. GERRY WALTERS .........605
Chapter 53: Shotcrete—JOHN H. PYE ..........................................................................616
Chapter 54: Organic Materials for Bonding, Patching, and Sealing
Concrete—RAYMOND J. SCHUTZ ..........................................................................625
Chapter 55: Packaged, Dry, Cementitious Mixtures—DENNISON FIALA ................631
Chapter 56: Self-Consolidating Concrete (SCC)—JOSEPH A. DACZKO
AND MARTIN VACHON ............................................................................................637

INDEXES
Index ..............................................................................................................................647
1
Introduction

Joseph F. Lamond1 and James H. Pielert 2

ASTM STP 169C, SIGNIFICANCE OF TESTS AND Part II deals with the properties of freshly mixed concrete.
Properties of Concrete and Concrete-Making Materials, was Part III concerns itself with the properties of hardened
published in 1994. ASTM Committee C9 on Concrete and Con- concrete.
crete Aggregates has once again decided the time was appro- Part IV deals with concrete aggregates. The order of the
priate to update and revise this useful publication to reflect chapters has been revised. They are now presented in the or-
changes in the technology of concrete and concrete-making der that most concerns concrete users: grading, density, sound-
materials that have taken place since that time. New materials ness, degradation resistance, petrographic examination, reac-
have appeared on the scene, along with a greater appreciation tivity, and thermal properties. Some of the chapter titles have
of the capabilities of concrete as a basic construction material. changed and the previous chapter on pore systems has been in-
Committee C9 and its subcommittees have made significant cluded in the chapter on density.
changes in many of its specifications and test methods to re- Part V includes materials other than aggregates. The title
flect these changes. New specifications and testing techniques of the chapter on curing materials was changed to reflect cur-
have been developed to provide for informed use of new ma- rent technology of materials applied to new concrete surfaces.
terials and new uses for concrete. The chapter on mineral admixtures has been separated into
Hydraulic cement concrete is a product composed of many two chapters, one on supplementary cementitious materials
materials and produced in many forms. The quality of concrete and the other on ground slag.
is dependent on the quality of the constituent materials and re- Part VI, on specialized concretes, contains one new chap-
lated manufacturing, testing, and installation processes. Since ter on self-consolidating concrete. The subcommittee structure
1914, ASTM Committee C9 has played a vital role in promoting of Committee C9 has been modified to accommodate this
the quality of concrete by developing specifications, testing need.
methods, and practices for concrete and concrete-making ma- The editors, along with ASTM Committee C9 on Concrete
terials. This has been possible through the dedication and com- and Concrete Aggregates, believe this new edition will serve
mitment of its volunteer members over the years. the concrete industry well. The editors selected authors and
Committee C9 first published Report on Significance of their chapters were reviewed in accordance with ASTM’s peer
Tests of Concrete and Concrete Aggregates, ASTM STP 22, in review procedures. C9 subcommittees having jurisdiction
1935, with an updated report published in 1943. ASTM STP over the subjects for pertinent chapters participated infor-
169 was published in 1956, followed by ASTM STP 169A in mally in the review process. The editors appreciate the help
1966, ASTM STP 169B in 1978, and ASTM STP 169C in 1994. and guidance of these people and the cooperation of ASTM
Following this brief introduction, this special publication Committee C1 on Cement in providing authors for the two
is organized into six parts: General, Freshly Mixed Concrete, chapters on cement. Some of the authors in ASTM 169C are
Hardened Concrete, Concrete Aggregates, Concrete-Making no longer active in Committee C9. The co-editors and Com-
Materials Other than Aggregates, and Specialized Concretes, mittee C9 members wish to dedicate this edition to those au-
with revised and new chapters. thors who have died since ASTM STP 169C was published.
In Part I, the chapters consist of general subjects on the They are Paul Klieger, Ed Abdur-Nur, Bill Dolch, Jack Scan-
nature of concrete, sampling, variability, and testing laborato- lon, Bob Philleo, Bill DePuy, Bryant Mather, Ron Mills, and
ries. A new chapter deals with modeling cement and concrete Owen Brown.
properties.

1
Consulting engineer, Jeffersonton, VA 22724.
2
Manager, Cement and Concrete Reference Laboratory, Gaithersburg, MD 20899.

1
PART I
General
2
The Nature of Concrete

Richard A. Helmuth1 and Rachel J. Detwiler 2

Preface materials for making concrete and their effects on concrete

properties are given in other chapters in this work.
T. C. POWERS AUTHORED THE FIRST VERSION OF Typical hydraulic-cement concretes have volume fractions
this chapter, which was published in ASTM STP 169A in of aggregate that range approximately from 0.7 to 0.8. The
1966. His chapter was reprinted without revision in ASTM STP remaining volume is occupied initially by a matrix of fresh
169B in 1978. In ASTM STP 169C (1994), Richard A. Helmuth cement paste consisting of water, cementitious materials, and
condensed some of that work and included more recent mate- chemical admixtures, that also encloses air voids. While the ag-
rial. The present version relies on the framework established by gregates occupy most of the volume, they are relatively inert
the earlier authors, while updating and adding to it. and intended to be stable. It is the cement paste matrix that
undergoes the remarkable transformation from nearly-fluid
Introduction paste to rock-hard solid, transforms plastic concrete into an
apparent monolith, and controls many important engineering
For thousands of years, mankind has explored the versatility of properties of hardened concretes.
materials that can be molded or cast while in a plastic state and
then hardened into strong, durable products [1]. As with Scope
ceramics and gypsum plasters, lime mortars and pozzolanic con- Hydraulic-cement concretes may be designed to provide prop-
cretes provided engineers with economical materials for pro- erties required for widely varying applications at low life-cycle
duction of diverse utilitarian and aesthetically pleasing struc- cost. If not properly designed or produced, or if exposed to
tures. Modern concretes preserve these ancient virtues while service conditions not understood or unanticipated, premature
greatly extending the range of technically achievable goals. failures may result. Successful use depends on understanding
the nature of concrete.
Concrete-Making Materials—Definitions The scope of this examination of the materials science of
Concrete is defined in ASTM Terminology Relating to Concrete concrete is mainly confined to concretes made with portland
and Concrete Aggregates (C 125) as “a composite material that cements, with or without supplementary cementitious materi-
consists essentially of a binding medium within which are als and chemical admixtures. The focus is mainly on how we
embedded particles or fragments of aggregate; in hydraulic- understand concrete performance in ordinary construction
cement concrete, the binder is formed from a mixture of practice. That understanding is based on knowledge of its
hydraulic cement and water.” Hydraulic-cement concretes are constituents, and their physical and chemical interactions in
those most widely used in the United States and worldwide. different environments.
Hydraulic cement is defined in ASTM Terminology Related to
Hydraulic Cement (C 219) as “a cement that sets and hardens Freshly-Mixed Cement Paste and Concrete
by chemical interaction with water and that is capable of doing
so under water.” Portland cement is the most important Water in Concrete
hydraulic cement. It is produced by pulverizing portland The properties of fresh cement paste and concrete depend on
cement clinker, consisting essentially of hydraulic calcium the structure and properties of ordinary water, which are
silicates, usually by intergrinding with small amounts of unusual for a substance of such low molecular weight. Each
calcium sulfate compounds to control reaction rates. It may be molecule has a permanent dipole moment, which contributes
used in combination with one or more supplementary to the strong forces of attraction between water molecules
cementitious materials, such as fly ash, ground granulated blast and results in unusually high melting and boiling points,
furnace slag (referred to as “slag” in the remainder of this chap- heats of fusion and vaporization, viscosity, and surface ten-
ter), silica fume, or calcined clay. sion [2].
Aggregate is defined in ASTM C 125 as “granular material, In addition to dipole interactions, hydrogen bonding
such as sand, gravel, crushed stone, or iron blast-furnace slag, between water molecules and thermal agitation affect the
used with a cementing medium to form hydraulic-cement structure of water and aqueous solutions. Hydrogen bonding
concrete or mortar.” Detailed descriptions of these and other favors formation of clusters of molecules, while thermal
1
Materials Research Consultant, Construction Technology Laboratories, Skokie, IL 60077-1030.
2
Senior Concrete Engineer, Braun Intertec, Minneapolis, MN 55438.
5
6 TESTS AND PROPERTIES OF CONCRETE

agitation, including translational, rotational, and vibrational that contain small percentages of ultrafine (submicron) parti-
motions, tends to disrupt the structure. cles may also aid in dispersing cement particles by adsorption
In the liquid state, the molecules are easily oriented in an of the ultrafine particles on the surfaces of the larger particles.
electric field so that water has a high dielectric constant (78.6 at This specific kind of fine-particle effect is responsible for the
25°C). This orientation, as well as molecular polarization, improved flow of many portland cement/fly ash mixtures [7,8].
means that the electric field strength and the forces between The average thickness of films of water separating dis-
charged particles, such as ions in solution, are reduced to 1/78.6 persed particles in the paste depends on the water-to-cement
relative to that in vacuum (or air). Because of its exceptionally ratio (w/c) and the cement fineness. A first approximation of
high dielectric constant, water is an excellent solvent for salts: the average thickness of these films is given by the hydraulic
the energy of separation of two ions in solution is an inverse radius: the volume of water divided by the specific surface. If
function of the dielectric constant of the solvent. Ions in it is assumed that the films are thin compared with the particle
solution are not separate entities but are surrounded by water sizes, the calculated thickness is 1.2
m for cement of specific
molecules attracted to them by ion-dipole forces. surface of 430 m2/kg, mixed at 0.5 w/c [9]. Since the assump-
A few minutes after mixing begins, about half of the tion is not valid for the finer fractions and much of the fine
cement alkalies are dissolved so that the concentration of the fraction in portland cement is composed of calcium sulfates
alkali and hydroxide ions may commonly be 0.1 to 0.4 mol/L, and other phases that dissolve within minutes after mixing be-
depending mainly on the water-to-cement ratio and the cement gins, the average film thickness for the larger particles in that
alkali content [3]. At 0.3 mol/L, each ion would be separated paste is probably about 2
m. For flocculated particles, the
from like ions, on the average, by about 1.7 nm, or about five films are much thinner between adjacent particles, so that
water molecules. much of the water is forced into relatively large cavities or cap-
illary-like channels.
Interparticle Forces
Atoms near the surface of solids are distorted and shifted rela- Cement Hydration and Structure Formation
tive to their positions in the interior because of the unsatisfied
atomic bonds at the surface. These distortions of the surface Early Hydration Reactions
produce net positive or negative surface charge, and elastic ex- It is convenient to divide the process of cement hydration into
cess surface free energy. In aqueous solutions, solid surfaces the early (within the first 3 h), middle, and late (after 24 h) pe-
may preferentially adsorb certain ions [4]. Particles with sur- riods. Soon after mixing cement with water, a gel layer forms
face charges of the same sign repel each other in suspensions on the surfaces of the cement grains. Taylor [10] characterized
and tend to remain dispersed. Particles of opposite sign attract this layer as “. . . probably amorphous, colloidal and rich in alu-
each other and flocculate [5]. mina and silica, but also containing significant amounts of cal-
In addition to these electrostatic forces, which can be at- cium and sulfate . . .” Within about ten minutes, stubby rods of
tractive as well as repulsive, there are forces among adjacent calcium aluminoferrite trisulfate hydrate (AFt) begin to form.
surfaces of solids, atoms, and molecules that are always attrac- They appear to nucleate in the solution and on the outer sur-
tive. These van der Waals, or dispersion, forces exist because face of the gel layer.
even neutral bodies constitute systems of oscillating charges During the middle period of hydration approximately 30 %
that induce polarization and oscillating dipole interactions [5]. of the cement reacts. The rapid formation of calcium silicate
The combined action of the different forces causes sorption of hydrate (C-S-H) and calcium hydroxide (CH) is accompanied by
water molecules and ions from solution, which can neutralize significant evolution of heat. The CH forms massive crystals in
surface charge and establish separation distances of minimum the originally water-filled space. The C-S-H forms a thickening
potential energy between solid particles [6]. The mechanical layer around the cement grains. As the shells grow outward,
properties of fresh and hardened cement pastes and concretes they begin to coalesce at about 12 h, a time coinciding with the
depend on these forces. maximum rate of heat evolution (Fig. 1) and corresponding ap-
proximately to the completion of setting. The shells are
Structure of Fresh Cement Paste apparently sufficiently porous to allow the passage of water in
Modern portland cements have mass median particle sizes that and dissolved cement minerals out. A gap begins to appear
are about 12 to 15
m (diameter of an equivalent sphere), al- between the hydration shell and the surface of the cement
most all particles being smaller than 45
m, and very little of the grain. Toward the end of the middle period the growth of AFt
cement being finer than 0.5
m. During grinding, calcium sul- crystals resumes; however, this time they are distinctly more
fates grind faster and usually become much finer than the acicular in shape. Their formation coincides with a shoulder on
clinker. After mixing with water, the solid surfaces become cov- the heat evolution curve [10].
ered by adsorbed ions and oriented water molecules forming a Like most chemical reactions, cement hydration proceeds
layer of solution of different composition and properties from more rapidly with increasing temperature. Verbeck and
those of the bulk aqueous phase; the layer extends out to a dis- Helmuth [11] postulated that because of the low solubility and
tance at least several times the diameter of a water molecule. low diffusivity, the ions forming the cement hydration products
These surface layers have the effects of simultaneously separat- would not have time to diffuse any significant distance from the
ing and weakly binding the particles into a flocculated structure. cement grain, thus forming a highly nonuniform distribution of
In fresh cement pastes and concretes made with high solid phases. They believed that the dense hydration shells
doses of water-reducing admixtures, cement particles may be- would serve as diffusion barriers, hindering further hydration.
come almost completely dispersed (deflocculated) because A consequence of the uneven distribution of the solid phases is
large organic molecules are adsorbed on their surfaces, dis- a coarser pore structure. Skalny and Odler [12] found that C3S
placing water films, and greatly reducing attractive forces be- pastes of a given w/c hydrated at temperatures of 50 to 100°C
tween cement particles. Supplementary cementitious materials had a coarser structure and greater volume of large pores than
 HELMUTH AND DETWILER ON THE NATURE OF CONCRETE 7

Before setting, two, and sometimes three, kinds of volume

changes occur. Sedimentation causes subsidence of the floc
structure and collection of bleeding water on the top surface,
if evaporation is not excessive. If the surface becomes partly
dried, capillary tension in the water can cause plastic shrinkage
and cracks. Chemical shrinkage is the volume change that
results from formation of hydrates that have less solid volume
than the volume of water and solids reacted. While the paste
is plastic, the entire volume of paste undergoes chemical
shrinkage. After setting, the external dimensions remain essen-
tially fixed and additional water must be imbibed to keep the
pores saturated with water. If sufficient water is not imbibed,
the paste undergoes self-desiccation.
Autogenous shrinkage is the volume change that results
when there is no moisture loss to the surrounding environ-
ment. It is most significant for concrete in which the water-to-
cement ratio is less than about 0.42. Before the concrete sets,
Fig. 1—Heat evolution of Type I/II portland cement paste autogenous shrinkage is equivalent to chemical shrinkage and
as measured by conduction calorimetry. The first heat peak is manifests itself as plastic settlement. Once a continuous struc-
associated with the initial hydrolysis of the C3S, the hydration
ture begins to form, the chemical shrinkage is restrained to
of the C3A in the presence of gypsum to form ettringite, and
some degree. As the internal structure of the cement paste ma-
rehydration of the hemihydrate to form gypsum. It is nor-
mally completed within 15 min. Deposition of ettringite on trix becomes more rigid, autogenous shrinkage is influenced
the surface hinders further hydration of the C3A. The first less by chemical shrinkage and more by self-desiccation [16].
peak is followed by a dormant period of 2 to 4 h, during Tazawa and Miyazawa [17] found that the amount of autoge-
which the paste remains in the plastic state as the C3S nous shrinkage increases with increasing C3A and C4AF con-
continues to dissolve slowly. The acceleration period begins tents of the cement and decreasing water-to-cement ratio.
with the renewed evolution of heat (beginning of the second Bentz and Geiker [18] found that the effects of self-desiccation,
peak) as the initial hydration products of the C3S begin to and thus the autogenous shrinkage, can be mitigated by the use
form. Initial set coincides with the beginning of the accel- of water-saturated low-density fine aggregates or of super-
eration period. The CH crystallizes from solution, while the
absorbent polymer particles to provide an internal source of
C-S-H deposits on the surface of the C3S, creating a diffusion
water, or by the use of shrinkage-reducing admixtures.
barrier. Eventually the rate of hydration slows due to the
difficulty of diffusion of water and ions through this barrier Gypsum and other sulfate-bearing materials are normally
[20]. Final set takes place just before the maximum point of interground with portland cement clinker in the production of
the second peak. The “shoulder” of the second peak, which in cement to control the hydration of the aluminates. If little or
this figure appears at about 9 h, is associated with the no gypsum is present, the cement experiences “flash set” or
renewed formation of ettringite [10]. Further hydration of “quick set,” in which the cement sets rapidly with much evolu-
the cement continues at a much slower rate, asymptotically tion of heat. Plasticity is not regained with further mixing, and
approaching 100 % [20]. (Image courtesy of E. Shkolnik.) the subsequent development of strength is poor. It is associated
with the rapid hydration of the aluminate and ferrite phases to
form plates of low-sulfate AFm phases such as C4AH13 and
those hydrated at 25°C. Kjellsen et al. [13] found that in cement C4AH19 throughout the paste. Some cements low in aluminate
paste hydrated at 50°C to a degree of hydration of approxi- phase do not flash set even without gypsum [10].
mately 30 % the hydration shells were already sufficiently dense Flash set and quick set may be caused by insufficient
to act as diffusion barriers to sulfate ions. The morphology of sulfates in the cement, or by the presence of the wrong form(s)
the CH crystals was dependent on the hydration temperature, of sulfate. During milling, gypsum can dehydrate to form
being more elongated in cement paste hydrated at 5°C and hemihydrate or so-called “soluble anhydrite” (-CaSO4). Some
more compact in cement paste hydrated at 50°C. byproducts from various industrial processes contain calcium
During the first hours of hydration, very fine supplemen- sulfite (CaSO3), known as hannebachite, rather than gypsum,
tary cementitious materials such as silica fume and fine fly ash and are much less soluble than gypsum. Byproduct gypsum
have some important physical effects on the development of the from flue gas desulfurization is often difficult to dispense
microstructure. According to Bache [14], the principal physical uniformly due to its high moisture content; thus the quantity
effect of silica fume in concrete is an enhancement of particle of gypsum in the cement can be variable. The form of sulfate
packing because the silica fume particles can fit into spaces has a direct bearing on the availability of sulfates in solution at
between cement grains in the same way that sand fills the the right time, since the solubility and rate of solution vary
interstices between particles of coarse aggregate and cement considerably from one form to another. Very finely ground
grains fill the interstices between sand grains. This analogy is cement requires more sulfate to control the reactivity of the
valid only when there is sufficient superplasticizer to counter- aluminates.
act the effects of attractive surface forces. As the cement begins Incompatibility among the various concrete ingredients
to hydrate, the fine particles of silica fume or fly ash serve as may also contribute to flash set or quick set. Lignosulfonates in
nucleation sites for the hydration products. Asaga et al. [15] water-reducing admixtures limit the solubility of sulfate and
found that fly ash, silica fume, and slag all increase the rate of calcium ions. Triethanolamine (TEA) in water-reducing admix-
early hydration of C3S, even though the supplementary cemen- tures makes the aluminate phases react faster while retarding
titious materials do not themselves react at this stage. the hydration of the calcium silicates. Some Class C fly ashes
8 TESTS AND PROPERTIES OF CONCRETE

Fig. 2—Computer-based model of the development of microstructure in hydrating

portland cement pastes: (upper left) portland cement paste before hydration begins; (upper
right) portland cement paste after 47 % hydration; (lower left) unhydrated portland cement
paste with fly ash; (lower right) partially hydrated portland cement paste with fly ash [images
courtesy of E. J. Garboczi]. All images 100
m  100
m.

contain sufficient quantities of reactive aluminate phases to form inside them. The availability of sulfate ions is very lim-
upset the balance between aluminates and sulfates unless ad- ited within the hydration shells, and calcium aluminoferrite
ditional gypsum is added to the system. If either the cement or monosulfate hydrate (AFm) forms, replacing AFt as the alu-
the fly ash has a high alkali content, it will increase the rate of minate phases continue to hydrate. After the spaces between
hydration. Hot weather accelerates the hydration of the alumi- the hydration shells and the cement grains fill with hydration
nates and reduces the solubility of the sulfates [19]. products, further hydration is slow and appears to occur by
Another type of rapid set is called “false set,” in which lit- a topochemical rather than a through-solution mechanism
tle heat is evolved. False set is associated with the rehydration [10].
of calcium sulfate hemihydrate to form secondary gypsum. Larger clinker particles in cement hydrate partly by disso-
The interlocking and intergrowth of the gypsum crystals re- lution and partly by in situ reactions so that a pseudomorph
sults in the stiffening observed. If the quantity of secondary of inner products is formed within the boundaries of the orig-
gypsum is not too great, it redissolves on further mixing, and inal grain. The depth of reaction increases with time, but at
the hydration reactions proceed normally. During milling, gyp- decreasing rates so that large particles may have unhydrated
sum can dehydrate to hemihydrate; hemihydrate is sometimes cores even after years of moist curing. The dissolved portion
added directly to the clinker when a more reactive form of forms outer products in the water-filled space near the grains.
sulfate is desired [10]. Grinding at a higher temperature or a The calcium silicates produce crystalline calcium hydroxide
lower relative humidity can often achieve the same result. and nearly amorphous calcium silicate hydrate (C-S-H gel) that
engulf crystalline phases formed by the early reactions. Capil-
Hardening Reactions and Microstructure lary pores remaining in mature cement paste increase in size
After the first 24 h, the hydration shells surrounding the ce- with water-to-cement ratio and have diameters ranging from 10
ment grains become less permeable, and C-S-H begins to nm to 10
m [20].
 HELMUTH AND DETWILER ON THE NATURE OF CONCRETE 9

Powers designated the hydration products of portland ce- monomolecular sheets, about 700 m2/g, as measured by small-
ments as “cement gel,” recognizing that they contained both C- angle X-ray scattering [26]. Because of the large internal surface
S-H gel and crystalline products, and micropores [21]. He area, the distances between solid surfaces of the pores in the gel
found that typical cement gels had minimum porosities of approach the size of water molecules; most of the gel water is
about 30 %, and specific surface areas of about 200 m2/g, as close to the surfaces. In such systems, it is not certain how much
calculated by Brunauer, Emmett, Teller (BET) theory [1] from of the volume change of chemical shrinkage should be attrib-
water-vapor adsorption data obtained after first drying to uted to the reaction itself, and how much to the possible change
remove all of the evaporable water. These studies also showed of density of water in pores as it is adsorbed on newly created
that at 0.38 water-to-cement ratio all of the capillary pore space surfaces. If it is assumed [27] that the adsorbed solution has the
was just filled by maximum density gel when all the cement same density as that in large pores, the apparent specific vol-
was hydrated. Mixtures made with water-to-cement ratios less umes of the nonevaporable (hydrate) water and solids were
than 0.38 cannot be completely hydrated; the amount of found to be 0.74 and 0.398 cm3/g, respectively, and the mini-
cement that can hydrate is less because hydration virtually mum porosity of the gel to be 30 %. The amount of chemical
stops when the capillary space is filled with gel of minimum shrinkage is expressed in terms of the change in the apparent
porosity. Saturated, fully hydrated cement pastes made at specific volume of the reacted water, from 0.99 to 0.74 cm3/g,
water-to-cement ratios above 0.38 have remaining capillary and the amount, wn, of nonevaporable water: 0.25 wn.
pore space (by definition) equal to the excess above 0.38. Hydration of each unit volume of cement produces about
Partially hydrated mixtures have proportionately less gel and 2.2 volumes of gel. This value does not depend on the assump-
more capillary space. Cements of different compositions tion concerning specific volumes. Although chemical shrink-
behave similarly, with similar values for the constants. age slightly reduces the space filling by solid hydrates, cement
Supplementary cementitious materials can alter the de- gel is an even more effective filler of the capillary space than
veloping microstructure of hydrating cement paste by means the solid hydrates because of the 30 % porosity of the gel.
of the pozzolanic reaction, in which silica reacts with CH and
water to form additional C-S-H. The composition of the C-S-H Effects of Drying
formed by pozzolanic reaction differs somewhat from that of Loss of moisture due to self-desiccation or evaporation par-
C-S-H formed by hydration of cement in that the former has a tially empties the largest capillaries at exposed surfaces. Ad-
lower Ca/Si ratio and a higher Al/Ca ratio [10]. In the hydration sorbed water remains on capillary walls as concave menisci
of cement pastes containing slag, Al substitutes for Si in the form and progress into smaller interconnected pores. Menis-
silicate chains [22]. With any supplementary cementitious cus curvature and capillary tension in the remaining water are
material, the pozzolanic reaction improves both the later-age increased as the relative humidity is decreased down to about
strength and the durability of concrete because it consumes 45 % RH, below which sorption effects prevail. Reductions in
the relatively weak, soluble CH and generates more C-S-H, relative humidity slow the hydration rate; at 80 % RH, hydra-
which further subdivides the larger pores and increases the tion is insignificant. Drying causes shrinkage of hardened ce-
strength of the cement paste. The microstructure of cement ment paste and major alterations of the gel microstructure.
pastes hydrated at elevated temperatures is marked by uneven Shrinkage and stabilization of cement paste by drying are com-
distribution of hydration products and, consequently, a coarser plex and partially irreversible processes involving capillary,
pore system. Goto and Roy found that the total porosities of sorption, and dehydration effects.
pastes hydrated at 60°C were greater than those of comparable Capillary tension in the pore water increases as the rela-
pastes hydrated at 27°C for the same length of time. They at- tive humidity decreases below the vapor pressure of the pore
tributed the difference in porosity largely to the difference in solution. For dilute solutions, tensions increase to about 97
volume of pores of radius 75–230 nm [23]. It is the larger pores MPa (14 000 psi) at 50 % RH. At lower relative humidities, 40
that contribute most to the permeability; Mehta and Manmo- to 45 % RH, the tension exceeds the cohesive strength of water
han proposed a pore radius of 50 nm as a somewhat arbitrary in capillaries and menisci can no longer exist [21]. Above 45 %
dividing point between “large” pores which contribute most to RH, capillary tension in the water must be balanced by
permeability and “small” pores which are much less significant compressive stresses in the solid structure, in which stress
[24]. For cement pastes hydrated at low temperatures, the hy- concentrations can produce irreversible effects. When the
dration products are more evenly distributed and the pores pores are nearly water-filled, the average stress is that
fine and discontinuous. Hydration shells around the cement produced by surface tension in the cross-sectional area that is
grains are not apparent. For cement pastes hydrated at ele- pore water; the resulting strain in the solid structure is the
vated temperatures, dense hydration shells are readily appar- beginning of the drying shrinkage. As the capillaries empty the
ent and the pores are coarse and interconnected [13]. cross sections of the remaining water-filled pores are smaller,
Supplementary cementitious materials mitigate the dele- but the capillary tension increases and causes local collapse of
terious effects of elevated-temperature curing. Cao and Det- less dense regions of the outer product, and enlargement of
wiler [25] found that both silica fume and slag were effective large pores. Desorption causes shrinkage both by permitting
in homogenizing the distribution of the hydrates of cement solids to come together, and by increasing solid surface
pastes cured at 70°C. While the total pore volume remained tension. Well-crystallized AFm and AFt hydrated phases also de-
essentially the same for pastes with and without silica fume or hydrate, decreasing lattice spacings, so that elastic restraint of
slag, the average pore size was significantly reduced. the shrinking C-S-H gel is reduced.
Drying and rewetting alter the microstructure, and differ- Cement paste cured for six months before drying at 47 %
ent adsorbates measure different surface areas. The sheet-like RH for the first time shows both reversible and irreversible wa-
crystallites are imperfectly stacked and separated by interlayer- ter loss and shrinkage [28]. Increased drying time causes in-
adsorbed water at relative humidities down to 11 %. Before creased water loss, shrinkage, and greatly reduced internal sur-
drying or aging, cement gels have specific surfaces of C-S-H face area. Rewetting causes sorption and swelling that only
10 TESTS AND PROPERTIES OF CONCRETE

trolled. When mixing is stopped, the floc structure reforms un-

til it becomes an essentially continuous floc. Cement pastes in
this condition are actually weak solids with measurable shear
stress yield values that depend on water-to-cement ratio, ce-
ment fineness, and other factors. Typical values for portland
cement pastes without admixtures range roughly from 10 to
100 Pa (0.0014 to 0.014 psi) for w/c from 0.6 to 0.35 [9]. Prior
to yield, the pastes are elastic and shear deformations can
reach about 20 deg, indicating a rubber-like elasticity.3 They
are also plastic solids with typical values for plastic viscosity
that range roughly from 10 to 100 mPa  s (centipoise). High
water-to-cement ratio pastes seem to be liquid and may be
poured easily because their yield stress values are so low. At
low water-to-cement ratios, the pastes are obviously plastic and
can be deformed by moderate forces. Standard test pastes
made at normal consistency have w/c about 0.25 and yield
stress values of about 2000 Pa (0.29 psi) [9].
At ordinary temperatures, portland cement hydration re-
actions cause progressive stiffening and setting during the first
few hours. Yield stress values increase to 2  104 Pa (2.9 psi)
at initial set and 1  105 Pa (14 psi) at final set [9].
Increasing the water-to-cement ratio of a cement paste
reduces both the yield stress and the plastic viscosity. Water
reducers and superplasticizers increase the dispersion of the ag-
glomerated particles of cementitious materials by adsorption on
the particle surfaces, thus reducing the amount of water needed
to produce a given flow. Taylor [10] discussed three mechanisms
by which these admixtures cause dispersion: (1) an increase in
the zeta-potential between the first adsorbed layer and the bulk
solution, so that like surface charges of sufficient magnitude
cause the particles to repel one another; (2) an increase in the
Fig. 3—Drying shrinkage, rewetting swelling, and
solid-liquid affinity, so that the particles are more attracted to the
evaporable water content of a 0.6 water-to-cement ratio
hardened portland cement paste. Numbers indicate sequence mix water than to one another; and (3) steric hindrance, the ori-
of measurements [28]. Reprinted with permission of the ented adsorption of a nonionic polymer that weakens the attrac-
Portland Cement Association. tion between particles. Thus the addition of water reducers and
superplasticizers reduces the yield stress of the fresh paste. The
addition of air-entraining admixture, on the other hand, stabilizes
partly reverses the first water loss and shrinkage (Fig. 3). The the air-water interface of the air voids [1] and reduces the surface
irreversible component of shrinkage increases with water- potential of the particles of cementitious material, reducing the
cement ratio (0.4 to 0.6) from 0.2 to 0.4 %, whereas the re- plastic viscosity of the paste [31]. In the presence of sufficient su-
versible component (after stabilization by drying) is only 0.2 % perplasticizer to overcome the effects of surface forces, small
and nearly independent of porosity. Even without drying, long- quantities of silica fume (less than approximately 5 % by mass of
term aging in moist conditions causes age stabilization at cement) reduce the plastic viscosity; larger dosages of silica fume
water-cement ratios above 0.4 so that even the irreversible increase both yield stress and plastic viscosity, making the paste
shrinkage tends to become porosity-independent at about sticky [31]. The effect at small dosages was attributed by Bache
0.2 %. The irreversible shrinkage volume is only about half of [14] to the displacement of water due to more efficient packing
the volume of the irreversibly lost water, if we assume its spe- of the particles when minute particles of silica fume can fill the
cific volume to be 0.99, which suggests that some pores emp- interstices between cement grains; at larger dosages, the particle
tied during drying become closed off and are not accessible packing is no longer efficient. Another possibility is that, like the
during rewetting [29], or have reduced capacity. Below 11 % finest fraction of fly ash particles, the silica fume particles adhere
RH, loss of interlayer water is accompanied by large partially to the surfaces of the cement grains, causing the cement grains
irreversible shrinkage and water loss effects [30]. to disperse [7].

Cement Paste Structure—Property Elasticity and Creep

Relationships Hardened cement pastes are not perfectly elastic, but are vis-
coelastic solids. Internal friction, creep, and stress relaxation are
Rheology of Fresh Cement Paste useful in dissipating vibrational energy, and preventing exces-
When cements are mixed with sufficient water and sufficient sive stress concentrations in concrete. They are a result of redis-
intensity, dry agglomerates of fine particles are first dispersed tribution of moisture, viscous flow of gel, and dissolution of
and then tend to form a floc structure, which is continuously solids under stress and recrystallization in pores. These
broken down by mixing if the early reactions are well con- processes, and slow growth of cracks, are thermally activated

3
Unpublished work done at Construction Technology Laboratories for the Portland Cement Association under Project HR 7190.
 HELMUTH AND DETWILER ON THE NATURE OF CONCRETE 11

processes in which random thermal motions provide sufficient each water-to-cement ratio when plotted against X3, and
energy, in addition to the applied stress, at sites of adsorbed wa- indicated intrinsic strengths of 134 and 97 MPa (19 400 and
ter molecules, or solid-solid bonds, to exceed the bond energy. 14 000 psi) at 0.36 and 0.54 w/c, respectively [35]. This result
Short-term loading tests of water-saturated cement paste show indicates that the intrinsic strength of the gel formed at
that creep and creep recovery versus time curves are bimodal water-to-cement ratios above about 0.38 decreases with
and consist of a component with retardation times ranging from increasing water-to-cement ratio, in contrast with Powers’s
0.2 to 2 s and a slower component that ranges over weeks. The mortar data.
short-time component was associated with redistribution of wa- If we consider fresh cement pastes to have strengths equal
ter in capillary pores [32]. Diffusion of strongly adsorbed and hy- to their yield stress values, typically 10 to 100 Pa, and ulti-
drate water, recrystallization, and other irreversible changes are mately harden to compressive strengths of 10 to 100 MPa, the
believed to contribute to the slower processes. Long-term creep increase is about six orders of magnitude.
of cement paste can be several times the elastic deformation.
Elastic moduli can be measured precisely by dynamic Permeability and Diffusivity
methods and are found to vary with porosity, , according to The transport properties of cement paste and concrete largely
determine its durability in most environments. Water itself
E  E0(1  )3 may be harmful because of its ability to leach CH from the
cement paste, and because ice crystals that grow on freezing
in which E0 is the modulus at zero porosity [33]. If the capil- generate tensile stresses that may cause cracking [36]. Water
lary porosity is used, E0 is the modulus of the cement gel, about may also carry harmful dissolved species such as chlorides or
34 GPa (5  106 psi) for Young’s modulus of water-saturated acids into the concrete. Cement and concrete are also
pastes. If the total porosity (including that of the gel) is used, permeable to gases such as oxygen and carbon dioxide, which
E0 is an average modulus for the solids, about 76 GPa (11  contributes to the corrosion of steel reinforcement. D’Arcy’s
106 psi). Equations of the same form apply for the shear and law describes the flow of water through saturated cement paste
bulk moduli. Drying significantly reduces Poisson’s ratio, from
about 0.3 to 0.18, and the bulk modulus; stresses are carried by dq/dt  K1  A  h/l
at least some of the water in pores.
Elastic moduli of saturated pastes increase moderately as where dq/dt is the flow rate, K1 is the permeability, A is the
the temperature is decreased to 0°C. At temperatures in the cross-sectional area, h is the hydraulic head, and l is the
freezing range down to about 60°C, ice formation in capil- thickness of the specimen [37]. Thus water-saturated flow is
laries increases the moduli as ice contents increase. At still proportional to hydraulic pressure differences, if corrections
lower temperatures, the moduli increase more rapidly; internal are made for osmotic effects; concentration gradients cause
friction reaches a peak at about 90°C as the gel water viscos- osmotic flow to higher concentrations, and diffusion of ions
ity increases as it approaches its glass transition temperature at to lower concentrations. When the cement paste is partly
119°C [34]. The gel water (adsorbed solution) does not freeze dried, relative humidity and moisture gradients cause flow be-
to ice, but becomes a glassy solid. cause of capillary tension and diffusion along surfaces and in
the vapor phase. The changes of microstructure that cause
great changes of elastic moduli and strength of cement pastes
Compressive Strength during hardening cause reductions of permeability and diffu-
The fraction, X, of the available space that is filled by cement sivity. The presence of supplementary cementitious materials
gel at any stage of hydration is called the gel/space ratio. It can can reduce the permeability by several orders of magnitude
be calculated from the water-to-cement ratio, the fraction of as the pozzolanic reaction reduces the continuity of the pore
the cement that has hydrated, and the volume of gel produced. system [10].
For fully hydrated cement pastes, it may be expressed as (1  Permeability coefficients of fresh portland cement pastes
), where  is the capillary porosity. Compressive strength, fc , of 0.5 and 0.7 water-to-cement ratio, calculated from bleeding
at different water-to-cement ratios and ages can be simply data, range from 5.7 to 20  105 m3/(s  m2  MPa/m) [6.1 to
expressed as 22  104 in.3/(s  in.2  psi/in.)], respectively. These coeffi-
cients for hardened pastes of the same water-cement ratios
fc  fcg X n after prolonged moist curing, determined with machined
samples, were reduced to ultimate values of 4.5 to 60  1012
in which fcg is the intrinsic strength of the gel (at X  1), and m3/(s  m2  MPa/m) [4.8 to 65  1011 in.3/(s  in.2  psi/in.)]
n has a value of about 3. Use of this equation by Powers indi- [37]. Permeability coefficients of fresh pastes are about ten
cated intrinsic strengths of cement gels ranging from 90 to 130 million times as great as when fully hydrated.
MPa (13 000 to 18 500 psi) in mortars made with five different The diffusion of ions in cement paste is described by Fick’s
cements [21]. However, mortars probably do not provide ac- second law once steady-state conditions have been reached:
curate measures of intrinsic strengths of pastes because of
transition zones at aggregate surfaces. Tests of cement pastes c/ t  Dc  2c/ x2
yielded higher strengths at gel/space ratios calculated to be
equal to those of mortars made with the same cement. where c is the concentration of the ion at distance x from the
Testing of pastes made with both normally ground and surface after time t and Dc is the diffusion coefficient. In prac-
controlled-particle-size-distribution portland cements has tice, the transport of ions may also be driven electrically or
shown that intrinsic strengths of the gel do not depend on by convection (in the case of partial saturation). Diffusion co-
cement particle size distributions over the range investigated, efficients for Na
are on the order of 1011 to 1013 m2/s and
although rates of strength development do. However, paste for Cl on the order of 1011 to 1012 m2/s [10]. Diffusion
strengths at several ages defined different straight lines for coefficients increase with increasing temperature and water-to-
12 TESTS AND PROPERTIES OF CONCRETE

cementitious materials ratio and decrease with degree of concrete so that the higher, or at least average, strengths may
hydration. be more significant. Ten different common types of rock tested
Specimens dried step-wise to 79 % RH and carefully resat- at the Bureau of Public Roads had average compressive
urated so as to avoid cracking had permeability coefficients strengths that ranged from 117 MPa (16 900 psi) for marble
about 70 times those of comparable specimens that were con- to 324 MPa (47 000 psi) for felsite [39]. A good average value
tinuously moist cured. Such changes indicate enlargement of for concrete aggregates is about 200 MPa (30 000 psi), but
large pores by partial drying. many excellent aggregates range in strength down to 80 MPa
(12 000 psi) [40]. These values are generally above strengths of
Thermal Expansion ordinary concretes.
Coefficients of thermal expansion of concretes are determined
mainly by those of their aggregates. However, thermal expan- Permeability
sions of cement paste depend strongly on their moisture Measurements of coefficients of permeability to water of
contents because retention of water by surface forces in the gel selected small (25-mm (1-in.) diameter) pieces of rock, free of
decreases as temperatures increase, and vice versa. When visible imperfections, yielded values several orders of magnitude
cooled without access to additional water, slightly dried ce- smaller than for larger specimens, which probably contained
ment gel has a linear coefficient of thermal expansion of about flaws [32]. Values for the small specimens ranged from 3.5 
27  106/°C. When cement gel is cooled in contact with suf- 1013 m3/(s  m2  MPa/m) [3.8  1012 in.3/(s  in.2  psi/in.)], for
ficient capillary water in cement paste or external water, mois- a dense trap rock, to 2.2  109 m3/(s  m2  MPa/m) [2.4  108
ture flows into the gel; the resulting expansion (during or after in.3/(s  in.2  psi/in.)] for a granite. These values are equal to
cooling) produces a net coefficient of thermal expansion of those measured for mature hardened portland cement pastes
about 11.6  106/°C [38]. Mature saturated cement pastes of made at water-to-cement ratios of 0.38 and 0.71, respectively, de-
low ( 0.45) water-to-cement ratio show transient effects spite the low (less than 1 %) porosities of these rocks.
caused by the relatively slow movement of moisture from cap-
illary to gel pores during cooling and vice versa during warm-
ing. At low relative humidities, coefficients of thermal expan- Concrete Proportioning, Structure, and
sion decrease to about the same value as for saturated pastes. Properties
Such differences between coefficients of thermal expansion of Proportioning and Consistency
pastes and aggregates may cause excessive local stresses in Two basically different kinds of concrete mixtures must be dis-
concretes unless relieved by creep. tinguished. Nonplastic mixtures made with relatively small
amounts of water show considerable bulking as water is added,
Concrete Aggregates and after compaction have sufficient strength to support their
own weight. The concrete block industry employs such non-
The major constituents of ordinary concretes are crushed plastic but cohesive mixtures. Void space in such mixtures is
rocks or gravels used as coarse aggregates and sands used as relatively high and filled mostly by air. The strength of the co-
fine aggregates. Materials used in concrete usually need to be hesive mixture results from capillary tension under menisci
processed to be of proper size grading and relatively free of bounding the water films on and between the solid particles,
such deleterious substances as clay, salts, and organic matter. and solid surface forces. Strength and bulking increase to a
maximum as water is added, and then decrease as the void
Specific Gravity and Porosity space nearly fills with water and capillary tension is dimin-
It is useful to classify aggregates by specific gravity and poros- ished. With sufficient water, the mixture is wetted so that sur-
ity into lightweight, ordinary, and heavyweight materials (ASTM face menisci and capillary tension disappear, void contents
C 125). Lightweight aggregates are used to reduce dead loads reach a minimum, and limited plastic deformation becomes
and stresses, especially in tall structures, and to provide thermal possible. The remaining cohesive force is a result of interparti-
insulation. Heavyweight aggregates are used mainly for radia- cle attraction between closely spaced fine particles. This mini-
tion shielding. Ordinary aggregates, such as sandstone, quartz, mum void space contains about 12 % air when such mixtures
granite, limestone, or dolomite, have specific gravities that are compacted by ordinary means, and the cement content is
range from about 2.2 to 3.0. Densities of ordinary concretes not below a certain limit. The water content at minimum voids
range from about 2.24 to 2.4 Mg/m3 (140 to 150 lb/ft3). content is called the “basic water content” [6].
Porosity reduces the weight, elastic modulus, thermal con- The consistency of cement paste at its basic water content
ductivity, and strength of aggregates and concretes, although is nearly the same as the normal consistency as defined in ASTM
the effect on strength may be significant only in high-strength standards. Normal consistency pastes, and mortar or concrete
concrete. Porosity increases permeability to fluids and diffu- mixtures made with different aggregates at their basic water
sivity to ions in pore solutions, especially if the pores are open contents, have slump values of about 42 mm (1.7 in.) in the stan-
(interconnected) rather than closed. Freezing of water in pores dard test. Such concretes are much stiffer than the plastic mix-
in aggregate particles in moist concrete can cause surface pop- tures commonly used in American practice that usually contain
outs or D-cracking in concrete pavements [20]. chemical admixtures and higher water contents. Further addi-
tions of water increase void volume, reduce interparticle forces,
Strength of Aggregate Particles and increase the capacity for plastic deformation.
Strength test results of individual samples of rock from any The main effect of adding increments of aggregate to paste
one source show wide variations that are caused by planes is to reduce the volume of voids and cement per unit volume.
of weakness, and their different orientations, in some of the The total effect is not just that of volume displacement, because
samples. Such weaknesses in the rock samples may not be sig- the cement paste void space is increased by the added aggregate
nificant once the rock has been crushed to the sizes used in surfaces, as described in the next section. Also, when aggregate
 HELMUTH AND DETWILER ON THE NATURE OF CONCRETE 13

is introduced, plastic strains in the paste during compaction are terms of the strength of the bond between paste and aggregate:
necessarily greater and the mixture is stiffer than the paste. If for concrete with a high water-to-cement ratio, failure is con-
such additions are begun using cement paste of the standard trolled by the properties of the paste; as the quality of the paste
normal consistency, and if the same compacting force or energy improves, the strength of the paste-aggregate bond controls the
is applied to the mixtures as to the paste, that consistency can strength of the concrete; with sufficient improvements in the
be maintained constant if increments of water are added with strength of the paste-aggregate bond, the strength of the ag-
each increment of aggregate. The ratio of the volume of water gregate becomes the controlling factor. Hadley [43] found that
plus air to the total volume of solids (voids ratio) decreases with the first hydration product to form on the aggregate surface is
added aggregate, but not as much as without the added water, a thin film of calcium hydroxide. While this film is still form-
until a minimum voids content is reached, and then increases. ing, calcium silicate hydrate gel begins to appear on the film.
Consistency of concrete depends on consistency of cement Small crystals of calcium hydroxide begin to form on the sur-
paste as well as on dispersion of aggregate by sufficient paste faces of the cement grains. These crystals also have a preferred
volume for each particular aggregate. Although concrete yield orientation parallel to the aggregate interface. With increasing
stress values can be calculated from slump values, there is as distance from the interface, the orientation of the crystals be-
yet no valid method of calculation of concrete slumps from come more random.
paste yield stress values for concretes made with different ag- Figure 4 [44] illustrates the microstructural characteristics
gregates and proportions. For fixed proportions, the stiffer the of the transition zone as compared to the bulk cement paste:
paste, the stiffer the concrete. In mixtures that are relatively higher void content, higher contents of CH and ettringite, re-
rich in cement and paste volume, adding increments of aggre- duced content of C-S-H, and larger crystals of CH strongly ori-
gate does not greatly increase water requirements for flow. In ented parallel to the aggregate surface. Factors contributing to
leaner mixtures, particularly those with aggregate contents the nature of the transition zone include bleeding, which cre-
above those at minimum voids ratios, but below those very lean ates pockets of water-filled space beneath aggregate particles;
mixtures that require excess amounts of entrapped air, the wa- less efficient packing of particles of cementitious materials in
ter requirement is proportional to the volume fraction of ag- the vicinity of a surface (the “wall effect”); and the “one-sided
gregate in the total solids [6]. This range comprises much of growth effect” of dissolved cementitious materials and hydra-
the concrete made for ordinary use. tion products diffusing in from the bulk cement paste (but not
from the aggregate) [45]. As the cementitious materials hydrate,
Structure the transition zone fills preferentially with hydration products
For concrete to possess plasticity, the aggregate must be dis- that form through solution (that is, CH and ettringite). Because
persed by a sufficient volume of cement paste to permit defor- of the relatively open space, the crystals can grow large.
mation under shear stress. For any aggregate size grading, the
minimum voids ratio indicates the volume required to fill the
voids in compacted (dry-rodded) aggregate. If the concrete is
plastic, it must contain a volume of paste and air above the min-
imum to disperse the aggregate, that is, to provide some sepa-
ration between particles that would otherwise be in contact.
Fine aggregate disperses coarse aggregate but also reduces av-
erage paste film thicknesses. For concretes made with nearly
the same voids ratios (about 0.20), at 75 to 100 mm (3 to 4 in.)
slump, and different aggregate finenesses, Powers calculated
the minimum separation distances between aggregate particles
from the excess paste volumes, by two different methods with
dissimilar results [6]. The average values by the two different
methods ranged from 26
m to 121
m for lean to rich mix-
tures, respectively, the latter having the highest percentage of
fine aggregate (43 %) and being close to the minimum voids ra-
tio. Such results indicate that many concretes, especially very
lean mixtures, suffer from poor workability because of particle
interference to flow by the larger ( 30
m) cement particles.
This indication has been confirmed by recent research. Al-
though cement pastes made with cements of 30-
m maximum
particle size were stiffer than those made with ordinary ce-
ments, improved flow was obtained using cements with con-
trolled particle size distributions in standard mortars and ordi-
nary (not lean) concretes [41]. Particle interference by large
particles is also one of the reasons that some fly ashes increase
water requirements of concretes [8].
The presence of the aggregates creates an anomaly in the
structure of hardened concrete known as the transition zone
between the cement paste and the aggregate. Cordon and Gille- Fig. 4—Representation of the transition zone at a paste/
spie [42] noted that for concrete with a water-to-cement ratio aggregate interface in concrete, showing a more coarsely
of 0.40, the larger the maximum size of the aggregate the lower crystalline and porous microstructure than that of the bulk
the strength of the concrete. They explained these results in cement paste [44]. Reprinted with permission of P. K. Mehta.
14 TESTS AND PROPERTIES OF CONCRETE

Mathematical modeling of the microstructure of concrete gates. Ice formation at frozen surfaces is propagated through
has improved our understanding of the mechanisms by which capillaries large enough for the contained water to freeze. Be-
the microstructure develops and by which it affects the prop- cause of ice/water interfacial tension, smaller capillaries re-
erties of the concrete. By developing models and verifying quire lower temperatures to be penetrated by the growing tips
them against experimental results, Bentz et al. [45] showed that of ice crystals. These crystals in frozen capillaries, and also
silica fume particles both reduce the initial thickness of the those in entrained air voids, grow by osmotic accretion of ice
transition zone and react to convert CH to C-S-H. Thus the tran- by diffusion of water from the gel pores. If entrained air voids
sition zone is thinner and more closely resembles the bulk are closely spaced, so that their void spacing factors are less
cement paste. Fly ash at the same dosage is less beneficial due than about 0.20 mm (0.008 in.), diffusion of moisture from
to its larger particle size and lower pozzolanic activity; thus it both capillary and gel pores to ice in the air voids dries the
is used at higher dosages. paste and prevents excessive expansions, otherwise caused by
Because of the greater porosity and the connectivity of the ice formation in capillary pores [46].
pores in the transition zone, the effect on the transport prop-
erties of the concrete is significant, particularly if the individ- Concluding Discussion
ual transition zones interconnect with one another to percolate
through the concrete [45]. The structure provides continuous The materials science of concrete provides a foundation for un-
pathways for the transport of fluids or ions. Similarly, the derstanding, predicting, and possibly controlling its behavior.
presence of the relatively weak transition zone affects the Our understanding of the fundamental relationships between
mechanical properties of concrete. The propagation of cracks structure and properties has been furthered by computer-
preferentially through transition zones reduces both the based models simulating the development of the microstruc-
strength and the modulus of elasticity of concrete. ture of hydrating cement paste and concrete [45]. These
models have advanced sufficiently to allow the prediction of
transport and mechanical properties and even provide insights
Properties of Hardened Concrete into the reasons for the behavior observed.
Other chapters in this publication give comprehensive treat- The increasing emphasis on high performance concrete
ments of many properties of hardened concrete. Some details provides an incentive for further development of performance-
are noted here to relate previous sections to specific properties based specifications. To make performance specifications truly
of the cement paste/aggregate composite. practical, and to allow us to use them with confidence, we need
In short-term loading tests for compressive strength, to continue to improve the test methods that measure the per-
stress/strain relationships for aggregates and cement paste can formance and correlate them with field performance. Ideally,
be sensibly linear up to near the compressive strength of the these correlations would not be merely statistical, but would be
paste, while those for concrete exhibit curvature with increas- based on the underlying physical and chemical mechanisms.
ing strain and pseudo-plasticity at stresses above about 0.4 times Some of the models used to predict field performance, for ex-
strength; microcracking at paste-aggregate interfaces develops ample for the purpose of comparing life-cycle costs of various
progressively with increasing strain [40]. For high strength con- alternative designs, are not solidly based on empirical data.
crete, however, there is little cracking at the interface until the Thus the use of such models is no better than extrapolation.
concrete reaches about 0.8 times its ultimate strength. Further work is needed to provide the necessary data.
Creep of many concretes, except possibly those loaded at Another implication of the use of high performance con-
early ages, is proportional to stress/strength ratios up to 0.3 to crete is that, as we seek to increase the strength or durability of
0.6; microcracking also begins in about this same range, de- the concrete, or the rate of construction, we observe seemingly
pending on the heterogeneity of the mixture. Mortars, for ex- new problems: a greater tendency to crack, for example.
ample, exhibit proportionality up to about 0.85. In concretes, Although the basics of good concrete practice had been long
stress/strength ratios near this value produce failure in time [40]. established when the first edition of this volume was published
Durability of concrete depends strongly on exposure and in 1966, these basics are often slighted in the interests of
service conditions, concrete properties—especially porosity and meeting the constraints of schedule and budget. Concrete is a
pore structure—and structural detailing. Exposure to acidic or remarkably forgiving material, performing amazingly well
neutral waters causes leaching of calcium hydroxide from the even under the less-than-ideal conditions that prevail in the
paste, increasing its porosity and permeability, and can even- field. However, high performance concrete is much less
tually soften even the much less soluble C-S-H. Sulfates in fresh “abuser-friendly”; it requires due attention to proper design
waters can penetrate into concrete to cause sulfate attack by re- and craftsmanship. Understanding the science behind the
action with aluminate phases in the cement paste to produce practice enables us to specify and enforce the right criteria to
ettringite, which can be destructively expansive if there is in- ensure the desired performance.
sufficient space to accommodate the volume expansion of that
highly hydrated reaction product. Calcium and sodium chlo- References
ride solutions react with aluminates to form Friedel’s salt and
[1] Lea, F. M., The Chemistry of Cement and Concrete, 3rd ed.,
other complex salts; some of these reactions are also expansive
Chemical Publishing Co., New York, 1971, pp. 1–10, 272, 596,
under some conditions. Seawater causes leaching and contains and 625.
sodium, chloride, magnesium, and sulfate ions in amounts suf- [2] “Water” in Van Nostrand’s Scientific Encyclopedia, 5th ed., Van
ficient to cause significant reactions, but the main effect is that Nostrand Reinhold, New York, 1976, pp. 2311–2312.
of erosion or loss of constituents [1]. Ion exchange, such as the [3] Gartner, E. M., Tang, F. J., and Weiss, S. J., Journal, American
replacement of calcium with alkalies or magnesium, can also Ceramic Society, Vol. 68, No. 12, Dec. 1985, pp. 667–673.
damage the concrete. [4] Weyl, W. A., in Structure and Properties of Solid Surfaces, R.
Moisture and freezing temperatures can cause damage by Gomer and C. S. Smith, Eds., University of Chicago Press,
ice formation in large pores in the paste and in some aggre- Chicago, 1953, pp. 147–180.
 HELMUTH AND DETWILER ON THE NATURE OF CONCRETE 15

[5] Adamson, A. W., Physical Chemistry of Surfaces, 2nd ed., Inter- [26] Winslow, D. N. and Diamond., S., Journal, American Ceramic
science Publishers, New York, 1960, pp. 209–245 and 317–379. Society, Vol. 57, No. 5, May 1974, pp. 193–197.
[6] Powers, T. C., The Properties of Fresh Concrete, Wiley, New [27] Copeland, L. E. and Verbeck, G. J., “Structure and Properties of
York, 1968, pp. 97–110, 121–125, 156, and 392–436. Hardened Cement Paste,” Proceedings, Sixth International
[7] Helmuth, R. in Fly Ash, Silica Fume, Slag, and Natural Pozzolans Congress on the Chemistry of Cement, English preprints,
in Concrete, SP-91, V. M. Malhotra, Ed., American Concrete Moscow, 1974, Paper II-5.
Institute, Detroit, MI, 1986, pp. 723–740. [28] Helmuth, R. A. and Turk, D. H., Journal, Portland Cement
[8] Helmuth, R. A., Fly Ash in Cement and Concrete, Portland Association, Vol. 9, May 1967, pp. 8–21.
Cement Association, Skokie, IL, 1987, pp. 14, 75, 76–79, [29] Parrott, L. J., Hansen, W., and Berger, R. L., Cement and Con-
112–117, and 156–163. crete Research, Vol. 10, No. 5, Sept. 1980, p. 647.
[9] Helmuth, R. A. in Principal Reports, Seventh International [30] Ramachandran, V. S., Feldman, R. F., and Beaudoin, J. J.,
Congress on the Chemistry of Cement, Paris, 1980, Paris Editions Concrete Science, Heyden, Philadelphia, 1981, pp. 77–88.
Septima, Vol. III, 1980, pp. VI-0/1–30. [31] Wallevik, O. H., Den Ferske Betongens Reologi og Anvendelse
[10] Taylor, H. F. W., Cement Chemistry, 2nd ed., Thomas Telford, på Betong med og uten Tilsetning av Silikastøv, [Rheology of
London, 1997, pp. 193–197, 218–219, 259, 280, 293, 332. Fresh Concrete and Application to Concrete With and Without
[11] Verbeck, G. J. and Helmuth, R. A., “Structures and Physical Prop- Addition of Silica Fume, in Norwegian with English summary],
erties of Cement Paste,” Proceedings of the Fifth International Doctoral dissertation 1990:45, Institutt for Bygningsmaterial-
Symposium on the Chemistry of Cement, Tokyo, 1968, pp. 1–32. lære, Norges Tekniske Høgskole, Trondheim, Norway, July
[12] Skalny, J. and Odler, I., “Pore Structure of Calcium Silicate 1990.
Hydrates,” Cement and Concrete Research, Vol. 2, No. 4, July- [32] Sellevold, E. J., “Inelastic Behavior of Hardened Portland
August 1972, pp. 387–400. Cement Paste,” Technical Report No. 113, Department of Civil
[13] Kjellsen, K. O., Detwiler, R. J., and Gjørv, O. E., “Development Engineering, Stanford University, Stanford, CA, Aug. 1969.
of Microstructures in Plain Cement Pastes Hydrated at Different [33] Helmuth, R. A. and Turk, D. H., Symposium on Structure of
Temperatures,” Cement and Concrete Research, Vol. 21, No. 1, Portland Cement Paste and Concrete, Special Report 90,
January 1991, pp. 179–189. Highway Research Board, Washington, DC, 1966, pp. 135–144.
[14] Bache, H. H., “Densified Cement/Ultra Fine Particle-Based [34] Helmuth, R. A., “Investigation of the Low-Temperature
Materials,” presented at the 2nd International Conference on Dynamic-Mechanical Response of Hardened Cement Paste,”
Superplasticizers in Concrete, Ottawa, June 1981. Technical Report No. 154, Department of Civil Engineering,
[15] Asaga, K., Kuga, H., Takahashi, S., Sakai, E., and Daimon, M., Stanford University, Stanford, CA, Jan. 1972.
“Effect of Pozzolanic Additives in the Portland Cement on the [35] Helmuth, R. A., “Phase I: Energy Conservation Potential of
Hydration Rate of Alite,” Proceedings of the 10th International Portland Cement Particle Size Distribution Control,” Construc-
Congress on the Chemistry of Cement, Gothenburg, Sweden, tion Technology Laboratories Final Report, CR7523-4330, to
2–6 June 1997, Vol. 3, paper 3ii107. U.S. Department of Energy, Washington, DC, March 1979.
[16] Holt, E. E., Early Age Autogenous Shrinkage of Concrete, VTT [36] Helmuth, R. A., Chemistry of Cement, Proceedings of the
Publication 446, Technical Research Centre of Finland, Espoo, Fourth International Symposium, Washington, D.C., 1960, NBS
2001. Also available through Portland Cement Association as Monograph 43, 1962, Vol. II, pp. 855–869.
LT257. [37] Powers, T. C., Copeland, L. E., Hayes, J. C., and Mann, H. M.,
[17] Tazawa, E. and Miyazawa, S., “Influence of Cement Composi- Journal, American Concrete Institute Proceedings, Vol. 51, Nov.
tion on Autogenous Shrinkage of Concrete,” Proceedings of 1954, pp. 285–298.
the 10th International Congress on the Chemistry of Cement, [38] Helmuth, R. A., Proceedings, Highway Research Board, Vol. 40,
Gothenburg, Sweden, 2–6 June 1997, Vol. 2, paper 2ii071. 1961, pp. 315–336.
[18] Bentz, D. P. and Geiker, M., “On the Mitigation of Early Age [39] Woolf, D. O., in Concrete and Concrete-Making Materials,
Cracking,” Self-Desiccation and Its Importance in Concrete ASTM STP 169A, ASTM International, West Conshohocken, PA,
Technology, Proceedings of the Third International Research 1966, pp. 462–475.
Seminar in Lund, 14–15 June 2002, pp. 195–203. [40] Neville, A. M., Properties of Concrete, Longman Scientific and
[19] Detwiler, R. J. and Shkolnik, E., “Avoiding Material Incompati- Technical, Harlow, Essex, UK, (Wiley, New York), 3rd ed., 1987,
bility Problems in Concrete,” Roads & Bridges, Vol. 42, No. 4, pp. 131, 364, and 401–402.
April 2004. [41] Helmuth, R. A., Whiting, D. A., and Gartner, E. M., “Energy
[20] Mindess, S. and Young, J. F., Concrete, Prentice-Hall, Engle- Conservation Potential of Portland Cement Particle Size
wood Cliffs, NJ, 1981, pp. 76–86, 95–99, and 566–567. Distribution Control Phase II,” Final Report to U.S. Department
[21] Powers, T. C., Proceedings, Fourth International Symposium on of Energy, Washington, DC, NTIS No. DE86-001926, 1984.
the Chemistry of Cement, Monograph 43, National Bureau of [42] Cordon, W. A. and Gillespie, H. A., “Variables in Concrete
Standards, U.S. Department of Commerce, Washington, DC, Aggregates and Portland Cement Paste which Influence the
Vol. II, 1962, pp. 577–608. Strength of Concrete,” Journal of the American Concrete
[22] Richardson, I. G., “The Structure of C-S-H in Hardened Slag Institute, Vol. 60, No. 8, August 1963, pp. 1029–1051.
Cement Pastes,” Proceedings of the 10th International Congress [43] Hadley, D. W., The Nature of the Paste-Aggregate Interface,
on the Chemistry of Cement, Gothenburg, Sweden, 2–6 June Ph.D. dissertation, Purdue University, November 1982.
1997, Vol. 2, paper 2ii068. [44] Mehta, P. K., Concrete Structure, Properties and Materials,
[23] Goto, S., and Roy, D. M., “The Effect of W/C Ratio and Curing Prentice-Hall, Englewood Cliffs, NJ, 1986, p. 38.
Temperature on the Permeability of Hardened Cement Paste,” [45] Bentz, D. P., Schlangen, E., and Garboczi, E. J., “Computer
Cement and Concrete Research, Vol. 11, No. 4, July-August Simulation of Interfacial Zone Microstructure and Its Effect on
1981, pp. 575–579. the Properties of Cement-Based Composites,” Materials Science
[24] Mehta, P. K. and Manmohan, D., “Pore Size Distribution and Per- of Concrete IV, Jan Skalny and Sidney Mindess, Eds., American
meability of Hardened Cement Pastes,” 7th International Con- Ceramic Society, Westerville, OH, 1995, pp. 155–199.
gress on the Chemistry of Cement, Paris, 1980, Vol. III, pp. VII-1–5. [46] Helmuth, R. A., Proceedings, Fourth International Symposium
[25] Cao, Y. and Detwiler, R. J., “Backscattered Electron Imaging of on the Chemistry of Cement, Monograph 43, National Bureau
Cement Pastes Cured at Elevated Temperatures,” Cement and of Standards, U.S. Department of Commerce, Washington, DC,
Concrete Research, Vol. 25, No. 3, 1995, pp. 627–638. Vol. II, 1962, pp. 829–833.
3
Techniques, Procedures, and Practices of
Sampling of Concrete and Concrete-Making
Materials

Toy S. Poole1

Preface Sampling Concepts

THE SUBJECT COVERED IN THIS CHAPTER WAS The two major problems sampling protocols seek to address
previously covered by C. E. Proudley in STP 169, by E. A. Ab- are variation in a material or concrete structure and bias in the
dun-Nur in STP 169A and 169B, and by Abdun-Nur and Poole in taking of the sample. If sampling is inadequate, and either or
STP 169C. The chapter in 169C was largely written by Ed Abdun- both of these problems exist in a material source or in hard-
Nur and contained an extensive bibliography on sampling. The ened concrete, then test results may poorly represent the ma-
chapter in this edition focuses more on the details of sampling terial under examination.
as they are represented in current standards. The reader is re- Variation is a normal part of the production stream of all
ferred to the bibliography in 169C as an excellent source for ref- concrete-making materials and of all finished concrete. The
erences on general sampling theory and practice. challenge for sampling is to either capture the nature of this
variation, or to smooth out this variation, depending on the
Introduction purposes for which the sampling and testing is being em-
ployed. In some instances the purpose of the work is to de-
The purpose of this chapter is to discuss general sampling con- scribe the variation. This information on uniformity is often
cepts as they apply to concrete and concrete-making materials, needed for monitoring material supplies to help control con-
to discuss details critical to specific sampling problems, and to crete uniformity and is often part of the description of the
outline the major features of sampling practices described in hardened concrete properties in a structure. In other cases,
current ASTM standards. the variation is not of particular interest, but rather the aver-
It is not difficult to make the argument that the details of age property of the material is the object of the testing. This
how, when, and where samples are taken can often be of criti- would plausibly be the case for some materials sources in
cal importance in determining the meaning of the test results which considerable blending of material will occur prior to
representing that sample. However, sampling is often a ne- making concrete. In this case, it is important for sampling to
glected part of the testing process. At least two things con- cause test results to represent the average property of the ma-
tribute to this. One is that details of sampling for various terial.
materials are seldom obvious, and therefore require knowl- Bias in the taking of a sample is particularly a problem in
edge and training of the individuals involved, who may already sampling heterogeneous materials, which include to some de-
be under severe time constraints. Another reason may be that gree most concrete-making materials. A major source of bias in
adhering to the details of standard guidance is time consuming sampling is segregation within the lot of material being sam-
and expensive to execute in some situations. It is at least im- pled. The intention of the sampling protocol is to avoid mis-
portant for people responsible for the quality of construction representing a material source due to the sampling of a segre-
to have some grasp of the important details of sampling so gated part of the material. An exception to this would be the
that, as a minimum, they can develop an understanding of the case where the sampling and testing is to verify whether or not
limits of interpretation of test results when sampling history is a stockpile is segregated to the point of causing a problem with
unknown. uniformity of concrete production.

1
U.S. Army Corps of Engineers, Engineer Research and Development Center, Vicksburg, MS.

16
 POOLE ON SAMPLING 17

So the importance of problems of variation and bias varies contain 1 to 3 metric tons (Mg) of cement, and fine and coarse
with the purposes of the testing program, which explains much aggregate amounts might each range from 2 to 10 metric tons of
of the differences in the details of sampling requirements that material. So it would be a reasonable practice to composite grab
are found in the various ASTM standards. samples over a total amount of material approximating this
Several terms and concepts consistently appear in sam- range with the expectation that tests would then better correlate
pling sections of ASTM standards. These are the lot, the sample with performance of a batch of concrete. However, for reason-
unit (or sometimes unit), the grab sample (or sometimes incre- ably uniform materials, this may be an unnecessary detail.
ment), the composite sample, and the number of samples. The The number of samples required to adequately represent
guidance for each sampling problem uses these concepts to a lot of material varies considerably among testing standards,
guide the user towards solving or avoiding the problems with usually depending on the amount of variation expected in a
uniformity and segregation problems typical of that material. material or structure being sampled and the confidence inter-
A lot is the fundamental unit of material subject to sam- val on the test. This can be a relatively complicated statistical
pling. Definitions vary among standards, but the phrase “. . . problem. ASTM standards that cover acceptance testing and
amount of material offered for sale. . .” features prominently in uniformity testing vary quite a bit as to the number of grab
most definitions. Some standards are more specific. In practice samples required or allowed to make a composite sample and
the definition of a lot is often determined by the user for his own as to the number of composite samples that need to be ana-
purposes. It is typically the amount of material in which a user lyzed to make a determination.
is interested in determining the properties by sampling and Mostly for purposes of economy, some standards vary the
testing. For acceptance purposes this is often the amount of amount of sampling or the amount of testing per sample de-
material for which a financial payment is agreed on by contract. pending on the purposes of the testing. For example C 183 uses
A lot might reasonably be the amount of product delivered to a a normal and reduced testing concept, depending on the knowl-
project over a day, a week, a month, etc., or it may be defined as edge of the quality history of a cement source. C 494 uses the con-
some discrete quantity, such as a barge load or a train load. It cept of quality and uniformity testing. The former is used when
may also be defined by the user as the total amount of aggregate the full properties of a lot of material are unknown. In the latter
to be used in a construction project. only those tests that are good indicators of uniformity are run.
A sample unit (sometimes called simply a unit) is a subdi-
vision of a lot. Again, its exact definition is often determined by Acceptance Testing of Concrete Materials
the user for his own purposes, but typically a lot will be subdi-
vided into a relatively large number of sample units, with a Aggregate
relatively small number of units being actually sampled and Practice D 75 covers sampling of aggregates. The major prob-
tested. A sampling unit may have some meaning with respect lem associated with sampling aggregate is segregation, which
to the production of concrete. For example, it may be a truck can lead to a biased sample. All concrete aggregate is a mixture
load of material, the amount used to produce a single batch of of a wide range of particle sizes, which tend to easily segregate
concrete, be based on some time interval of production, or the during handling. The practice covers sampling of aggregate
elements of a grid superimposed on a structure. Some production from flowing discharges, conveyor belts, stockpiles,
standards will define a sampling unit, while others leave it to transportation units, and from in-place roadway bases. Appen-
the user to define. It is normally required that the sample units dices are included that give guidance on practices for sampling
to be sampled and tested be selected at random. potential aggregate sources—quarries and gravel deposits.
The simplest type of sample is the grab sample. Grab sam- The practice strongly favors sampling aggregate from a
ples are defined in various ways, but all definitions contain the flowing discharge or from a conveyor belt because of the rela-
concept that they are samples taken as a single effort, and that tive absence of segregation at these locations. The practice
each grab sample represents a relatively small fraction of the discourages sampling coarse-aggregate and mixed coarse-
material being sampled. Terms such as “one pass” or “one aggregate and fine-aggregate from stockpiles and transporta-
scoop,” etc., are often used. The term increment is synonymous tion units, particularly if grading is one of the critical proper-
with grab sample (as in D 75). It is usually required that grab ties to be tested, because of the ease with which segregation
samples be taken at random locations within a sampling unit. can occur and the difficulty in dealing with it in these cases. An
An important concept about grab sampling is that test results appendix to the practice does give some recommendations on
among grab samples will tend to show the maximum variation practices to use if sampling these is required.
in the material being tested, since there is essentially no aver- The problem with stockpiles is the tendency for the larger
aging in the process. particles to become segregated at the bottom of the pile. Rec-
The composite sample is put together by blending two or ommended sampling involves using power equipment to dig
more grab samples. Depending on the amount of such com- into the stockpile and taking samples from many locations in
positing directed in a sampling program, composite samples hopes of overcoming the possible segregation effect.
may represent quite a bit of averaging of properties. Standards The problem with sampling conveyances is that it is diffi-
covering specific materials are usually rather specific as to the cult to get to various locations in the shipment unless power
amount of compositing allowed or required. equipment is available, so that if some segregation has oc-
As mentioned above, grab samples show the maximum vari- curred in the loading of the conveyance, it may be difficult to
ation in a lot or production stream of material. This is often un- capture this with a series of grab samples.
realistic because the material is rarely or never used in the small The definition of what constitutes a lot and what constitutes
amounts represented by a grab sample. In the case of concrete a sample unit (called a unit in D 75) is left to the user. Units are
production, the smallest amount of material typically used is the selected at random. A test sample is defined as a composite of
amount that goes into a single batch of concrete. For example, in three increment samples (grab samples). The number of samples
commercial production a batch of concrete might reasonably needed to test a lot is left undefined and must be determined by
18 TESTS AND PROPERTIES OF CONCRETE

the user. The number of samples is intentionally left undefined, The practice distinguishes between sources of cement that
as explained in the practice, because the number needed to char- have a documented testing history and undocumented sources.
acterize a lot depends on the variability of the material in the lot Procedures are given for calculating control limits on docu-
and the criticality of the test result. If it is relatively uniform or mented sources. These procedures cause the sampling and test-
the criticality of test results is low, then a small number of sam- ing frequency to be adjusted according to the uniformity of the
ples is adequate, but if the material is quite variable or the criti- source and the proximity of a material’s properties to the spec-
cality is high, then more samples must be taken to capture this ification limit. A reduced sampling and testing requirement is
property. Practice E 122 is designed to assist in determining the allowed for these sources relative to an undocumented source.
number of samples required to develop the required resolution. Hydraulic cement sources typically issue mill certificates
listing test data for the cements offered for sale. The user should
Cementitious Materials be aware that the sampling and testing frequency described in
The major problem sampling schemes for cementitious mate- C 183 is probably different from the sampling procedures used
rials must overcome is manufacturing variation. Standards by manufacturers for quality control and for generating mill
covering these materials typically give rather specific instruc- certificate data. Cement plants typically take grab samples every
tions on when and where grab samples can be taken, on how hour during production, then make a 24 h composite sample
many grab samples can be composited for a test sample, and for full testing. The mill certificate typically reflects values from
also relatively specific instructions on the number of test sam- this composite sample. Therefore, mill certificate data probably
ples required to adequately represent a lot of production. Seg- represent at least a 24-h average. Some may represent a longer
regation is not heavily covered, but many standards do caution period of averaging. Sampling and testing for acceptance typi-
about sampling from the surface of a storage or transportation cally involves considerably less compositing, so test values are
unit because of the layer of fine cementitious dust that can set- more likely to reflect more of the manufacturing variation in
tle there as a result of the loading or transfer process. The pro- the cement. Therefore these test data may differ from the mill
cedure is typically to remove several inches of material before certificate data, depending on the amount of this variation.
taking a grab sample. Lack of agreement between a user’s test data and mill cer-
Cementitious materials are sometimes carried in trucks or tificate values can be a serious source of contention for prop-
rail cars used for hauling other materials. Small residues of erties that typically fall near the specification limit, such as al-
these other materials may reside in the bottom of these con- kali (Na2Oe) and sulfate content. Manufacturing variation will
tainers. This may constitute a relatively minor contamination sometimes cause grab samples to slightly exceed the specifica-
considering the amount of material in the load, but this rela- tion limit, while larger composites will show the average to
tively minor contamination may show up as a significant con- comply with the limit. Recognizing this problem, some cement
tamination if a grab sample is taken from the first material plants will sample and test for these properties at a higher fre-
taken from the container. A common result of such a sample quency, such as every hour during production.
(due to, for example, fly ash and cement being transported in There are also considerable differences among cement
the same trucks or stored in close proximity such that con- companies in the frequency with which they produce mill cer-
tamination can occur) is that the insoluble residue of the ce- tificates for distribution to customers. In some cases, the cer-
ment will exceed normal levels or fail to meet requirements. tificate actually represents reasonably closely the cement being
Sampling of hydraulic cement is covered by Practice C 183. sold. In other cases the mill certificate is only a representation
This the oldest (1944) standard covering the sampling of cemen- of typical values one can expect in the product. In these cases,
titious materials and the most detailed. This practice contains it is not uncommon that the same mill certificate will be issued
considerable detail on the sampling of hydraulic cement for pur- with shipments for quite a long time.
poses of determining conformance with purchase specifications. Sampling fly ash and natural pozzolans for acceptance
A number of configurations exist by which hydraulic cement is testing is covered by Method C 311. Information is given on
stored, transferred among storage units, and shipped to a site; types of samples, size of samples, and on frequency of samples
therefore, this amount of detail is required to cover all scenarios. required for acceptance testing.
Two standard methods of sampling for acceptance testing Like hydraulic cement, pozzolan is stored and shipped in
purposes are allowed under all circumstances. The methods are: a number of configurations. Three standard sampling proce-
1. from a conveyor (from process stream) to bulk storage; dures are allowed:
2. during transfer between storage bins. 1. from bulk storage at point of discharge, from rail cars, and
Each method directs that a specific number of grab sam- from trucks (road tankers);
ples be taken. The standard gives instructions on compositing 2. from bags;
these into a test sample. 3. from conveyor delivering to bulk storage.
Four optional methods of sampling for acceptance testing This practice prescribes different frequencies and differ-
purposes are allowed contingent upon approval by the pur- ent levels of compositing for testing of different properties.
chaser. These methods are: Moisture content, loss on ignition, and fineness are believed to
1. from bulk storage at point of discharge; require daily (or at least every 360 Mg) testing, while other
2. from bulk storage and bulk shipment by means of a slot- physical tests and chemical analyses are recommended only on
ted tube sampler; monthly composites. The standard also distinguishes between
3. from bags or packages; new and established sources, the latter requiring six months of
4. from bulk shipment of rail car or truck. quality assurance records.
As with the two standard methods, a specific number of grab Sampling of ground-granulated-blast-furnace slag is cov-
samples are prescribed for each of these optional methods, ered in Specification C 989, though not in extensive detail. Sam-
with additional instructions in the standard on compositing pling is allowed to be either by grab sample or by a composite
into a test sample. sample. Each sample is represented by no more that 115 Mg of
 POOLE ON SAMPLING 19

material. This detail is currently under consideration for revi- Sampling for Determination of Uniformity
sion. The standard is not exactly clear on this, but apparently this
mass of material represents the maximum size of a sampling Uniformity testing takes two forms. One is testing to determine
unit. Sampling units are chosen for testing such that, on the av- mixer performance as measured by uniformity within a batch
erage, one sample is collected for each 2300 Mg of material pur- of concrete. The other is testing to determine production
chased. stream uniformity, as in uniformity in freshly made concrete
Sampling of silica fume is covered in Specification C 1240. or in concrete-making materials.
As with slag, sampling of silica fume is not covered in great de- Mixer uniformity is covered in Specification C 94 on
tail. The standard relies heavily on Practice C 183 for guidance, ready-mixed concrete. Mixer-uniformity tests are normally run
although sampling frequencies are somewhat higher than for to determine the minimum mixing time required per batch.
hydraulic cements. There are two important requirements in sampling for mixer
uniformity. One is that samples be taken from either the ex-
Chemical Admixtures treme ends of a mixer, if concrete is sampled directly from the
Sampling of chemical admixtures is covered in Specifications mixer, or from the first and last 15 % of the batch, when con-
C 494 and C 1017. The specification recognizes two purposes crete is sampled on discharge from the mixer. These are ap-
for sampling. One is for determining properties for acceptance parently the locations in a mixer most likely to show the effects
testing and the other is for determining uniformity. of poor mixing. The other requirement is that, when samples
A major concern in sampling liquid products is segrega- are taken from the discharge stream, an entire cross section of
tion on storage due to settling or floating of one or more of the the discharge stream be collected and that nothing be done to
components. This is particularly a problem if the liquid is not impede the flow of concrete from the mixer, such as partially
a true solution, but rather a suspension or an emulsion of solid closing the gate. The purpose of these precautions is to prevent
particles in a liquid carrier. The solids may either settle or float getting a segregated sample due to separation of coarse aggre-
on prolonged storage. True solutions do not segregate on gate from mortar.
standing. Determining uniformity of concrete-making materials is
The specification directs that containers be agitated or covered by Practice C 1451. Method C 917 is a similar standard
stirred prior to sampling. In the case of large storage tanks that specific to hydraulic cement. C 917 predates C 1451 and is the
cannot be stirred, sampling at different levels of the tank using model on which the latter was developed. Current specifica-
a special sampling bottle is required. tions on concrete-making materials do not have uniformity lim-
Solid-phase admixtures (called non-liquid admixtures in its. The practice is based on analysis of grab samples, so results
the specification) are sampled using the same grab-sample and show the maximum variation in a material. Since it is a prac-
composite-sample concepts as for liquid admixtures, although tice and not a specification, the exact structure of the sampling
the physical acquisition of the samples more resembles the scheme is not defined, but left to the discretion of the user. A
techniques used for cementitious materials. materials supplier might determine uniformity for its own pur-
Acceptance testing is performed on composite samples, poses using a practice of regular sampling, while a user might
each required to be made up as a composite of at least three want to determine uniformity of the materials according to the
grab samples. There is no specific guidance as to the location schedue on which they are received, or for other specific uses.
of these grab samples, but the general guidance is offered that These two sampling schemes could easily result in different
they should be distributed to insure that the composite sample calculated levels of uniformity.
is representative of the lot under test. No definition of lot is Analysis of uniformity among concrete batches is covered
offered. by ACI 214 and 318. These standards use compressive strength
Uniformity may be determined using either grab or com- as the metric for the analysis and determination of compliance.
posite sampling. If the uniformity within a lot or sampling unit ACI 214 gives sampling and analysis procedures for determining
is required, the analysis should be on grab samples. If unifor- a concrete producer’s performance criteria. ACI 318 gives mini-
mity among lots or sampling units is required, then the analy- mum sampling frequencies for quality assurance purposes.
sis should be on composite samples taken from each.
Sampling of air-entraining admixtures is covered in Speci- Sampling Fresh Concrete
fication C 260. The sampling guidance is adapted from Speci-
fication C 494, and does not differ substantially. Sampling from a single batch of fresh concrete is covered by
Practice C 172. The principal precautions are to avoid sam-
Curing Compounds pling the very first or the very last material delivered from a
The sampling of curing compounds is covered in Specifica- mixer and to avoid practices that will cause the concrete to seg-
tions C 309 and C 1315, which contain identical guidance. Seg- regate during sampling. The first precaution is the exact oppo-
regation is the major source of problems in sampling curing site of the guidance given in C 94 on mixer uniformity, which
compound. On standing, the solid component of many curing is to intentionally sample the first and last part of the batch.
compounds will either float to the surface or settle to the bot- The apparent purpose in C 172 is to capture the average prop-
tom of the container. Therefore, agitating and sampling from erties of the batch of concrete and not let irregularities in the
the top and bottom of containers is required. These standards fringes bias that determination. Segregation of the concrete
contain specific guidance for determining the number of con- can occur when the gate on the delivery stream is partially
tainers to be sampled from a lot. The number is the cube root closed or when only part of the delivery stream is collected, so
of the total number of containers in the lot, rounded to the guidance is given to avoid these conditions.
next largest whole number. For example, if there are 100 con- Specific instructions are given for sampling from station-
tainers in a lot, then five (4.6 rounded to 5) would be sampled ary mixers, paving mixers, revolving-drum or agitating trucks,
for testing. and open-top, non-agitating trucks. Instructions are also given
20 TESTS AND PROPERTIES OF CONCRETE

for sampling concrete containing large nominal maximum- termine whether the number of events in one category occur
sized aggregate by wet screening. significantly more or less frequently than the number of events
in one or several other categories. Such a test might be applied
Sampling Hardened Concrete in the analysis of the data obtained from samples of a concrete
of mixed types or conditions to determine the statistical signifi-
Sampling concrete from structures for purposes of determin- cance of the frequencies of each type of concrete found (if
ing in-place properties is covered by Practice C 823. The prac- samples were taken at random). No other guidance is given
tice distinguishes two types of problems, each requiring differ- specific to Situation 2. This section of the practice probably
ent approaches to sampling. One is sampling for purposes of needs to be revised.
identifying the cause of some kind of concrete problem, and
the other is sampling for purposes of describing the average Other Guidance
and distribution of properties of concrete in a structure.
Sampling for the purpose of analysis of a concrete prob- Practice D 3665
lem is relatively simple if the location of the problem is known. This practice was developed to assist in random sampling of
Samples of problem concrete and similar concrete that does road and paving materials. The practice gives examples of how
not appear to have a problem, if such exists, are usually taken to use a random number table to determine which sublots
using judgment as to number and location to insure that the (sampling units) should be sampled from a lot of material. The
full expression of the problem or problems is represented. practice also gives information on how to determine the num-
Sampling to determine the range and distribution of ber and size of sublots. The practice contains a large random-
properties, particularly when property variation is not obvious number table.
on visual inspection, is more complicated. If sampling is not
done properly, an incorrect description of the state of a struc- Practice E 105
ture may result. The practice recognizes two conditions. One is This practice describes recommended rules for setting up a
when all of the concrete appears to be of similar condition random sampling plan, but is not specific to concrete and con-
(called Situation 1). The other is when the concrete appears to crete-making materials. It also discusses some problems en-
be in two or more categories with respect to composition or countered in executing probability-sampling plans.
quality (called Situation 2).
For Situation 1, any random or systematic sampling Practice E 122
scheme is allowed, as long as there is no bias in the selection This practice gives guidance on how to determine the number
of sampling sites. As shown in a later paragraph, the practice of samples needed to give an estimate of a material property to
actually recommends using random sampling methods, which the desired level of precision. It is not specific to concrete or
are preferable since systematic sampling can sometimes cause concrete-making materials. The user must determine the de-
the analyst to miss or to overrepresent regularly occurring fea- sired level of precision. This type of calculation also requires
tures in a structure. some knowledge of the variation in the material and testing
The term Probability Sampling is sometimes used in this process. The practice gives information on how to estimate this
context. The practice recommends dividing the structure or from other sources if there is no such information in the actual
parts of the structure under investigation into a number of testing program.
equal sections. The individual sections are sample units. De-
termining the number of sample units requires some experi- Practice E 141
ence. The practice recommends at least ten, “. . . but not less This practice gives recommendations for rules for setting up a
than the number of separate areas that are included in the in- sampling plan that will be suitable for evidentiary purposes. It
vestigation.” The meaning of this phrase is unclear. is not specific to concrete or concrete-making materials. The
Sampling units are chosen by some random selection practice is particularly designed to assist in one-of-a-kind stud-
method. The practice comments on the issue of the number of ies, where there is no previous experience to rely on. It uses ter-
samples taken, but no guidance is offered on how to determine minology that is unconventional relative to that commonly
this number. Practices E 105 and E 122 are cited for this pur- found in concrete standards, and so requires some study to fol-
pose (see below). Determining an adequate number of samples low. It gives rules for accepting or rejecting evidence based on
requires some experience and judgment as to the purposes and a sample and for data collection procedures for legal purposes.
economic limitations of the analysis. Clearly, the more samples The practice covers topics on auditing the sampling plan, re-
taken, the more detail will be developed in the description of porting formats, sample sizes, probability sampling rules, and
the properties of the structure. A useful approach, if possible, procedures for determining unbiased estimates of experimen-
is to take a preliminary round of samples to determine whether tal or measurement errors.
there is cause to believe that there are problems in the con-
crete, then, if such are found, design a more comprehensive ACI 214
plan to determine the extent of the problem in more detail. This standard gives guidance on sampling, calculating, and us-
Practice E 141 gives guidance on how to properly sample for ing among-batch variation in a concrete production for deter-
cases likely to be part of a legal dispute. mining project specifications on strength.
Paragraph 11.3, on sampling for Situation 2, appears to be
very confusing. The paragraph reads, “In Situation 2, samples Building Code ACI 318
may be taken for comparison with respect to several categories This is the Building Code. It gives minimum sampling fre-
of performance by using the Chi-Square test (5).” The (5) is a quencies for concrete for quality assurance and acceptance
reference to a standard statistics text. The Chi-Square test is purposes, as well as sampling guidance for hardened concrete
usually applied to the analysis of count or frequency data to de- in cases of disputes over strength.
 POOLE ON SAMPLING 21

Referenced Documents ASTM C 917-98. Test Method for Evaluation of Cement

Strength Uniformity from a Single Source.
ACI 214-77 (Reapproved 1997). Practice for Evaluating ASTM C 989-99. Specification for Ground Granulated Blast-
Strength Results of Concrete. Furnace Slage for Use in Concrete and Mortars.
ACI 318-99. Building Code Requirements for Structural ASTM C 1017-03. Specification for Chemical Admixtures for
Concrete. Use in Producing Flowing Concrete.
ASTM C 42-03. Test Method for Obtaining and Testing Drilled ASTM C 1240-03a. Specification for Use of Silica Fume
Cores and Sawed Beams of Concrete. for Use as a Mineral Admixture in Hydraulic Cement Concrete.
ASTM C 94-03a. Specification for Ready-Mixed Concrete. ASTM C 1315-03. Specification for Liquid Membrane-forming
ASTM C 172-99. Practice for Sampling Freshly Mixed Concrete. Compounds Having Special Properties for Curing and Sealing
ASTM C 183-02. Practice for Sampling and the Amount of Test- Concrete.
ing of Hydraulic Cement. ASTM C 1451-99. Practice for Determining Uniformity of
ASTM C 260-01. Specification for Air-Entraining Admixtures Ingredients of Concrete from a Single Source.
for Concrete. ASTM D 75-03. Practice for Sampling Aggregates.
ASTM C 309-03. Specification for Liquid Membrane-Forming ASTM D 3665-02. Practice for Random Sampling of Construc-
Compounds for Curing Concrete. tion Materials.
ASTM C 311-02. Test Methods for Sampling and Testing Fly ASTM E 105-58 (Reapproved 1996). Practice for Probability
Ash or Natural Pozzolans for Use in Portland-Cement Con- Sampling of Materials.
crete. ASTM E 122-00. Practice for Calculating Sample Size to Esti-
ASTM C 494-99a. Specification for Chemical Admixtures for mate, with a Specified Tolerable Error, the Average for a Char-
Concrete. acteristic of a Lot or Process.
ASTM C 823-00. Practice for Examination and Sampling of ASTM E 141-91 (Reapproved 2003). Practice for Accep-
Hardened Concrete in Constructions. tance of Evidence Based on the Results of Probability Sampling.
4
Statistical Considerations in
Sampling and Testing

Garland W. Steele1

Preface concrete and concrete-making materials. If needed, detailed

texts on statistical methods and procedures are available from
THE APPLICATION OF STATISTICAL CONSIDERA- many sources. A few are listed in the references.
tions to the sampling and testing of concrete and concrete-
making materials has been addressed by chapters in each of General Considerations
the four previous editions of ASTM STP 169. The first edition,
published in 1956, contained the chapter entitled “Size and Statistical Parameters [2–5]
Number of Samples and Statistical Considerations in Sam- If the characteristics of concrete or of a material used in the con-
pling,” by W. A. Cordon. The second edition, published in 1966, crete are to be determined with a known probability of meeting
contained the chapter entitled “Evaluation of Data,” by J. F. the user’s specifications, a plan is required. Such plans are com-
McLaughlin and S. J. Hanna. The second edition also con- monly called acceptance plans or other similar names that ap-
tained the chapter by W. A. Cordon that first appeared in the peal to the designer of the plan. Two types of risks that are often
1956 edition. The third edition, published in 1978, contained determined for acceptance plans are the seller’s risk (
), or
the chapter entitled “Statistical Considerations in Sampling supplier’s/contractor’s risk, and the buyers risk (), or owner’s/
and Testing,” by H. T. Arni. The fourth edition, published in specifying authority’s risk. When a material is exactly at the ac-
1994, contained the chapter entitled “Statistical Considerations ceptable quality level (AQL) as set forth in the acceptance plan
in Sampling and Testing,” by the author of this chapter. The for a contract, the probability of rejecting that material is the
contributions of previous authors regarding the practical ap- seller’s risk. When a material is exactly at the rejectable quality
plication of statistical probability to the field of concrete and level (RQL) as set forth in the acceptance plan for a contract, the
concrete-making materials are hereby acknowledged. This edi- probability of accepting that material is the buyer’s risk. Since it
tion will review and update topics as addressed by the previous is possible for the quality level to vary over the entire range be-
authors, introduce new clarification of the concepts presented, tween the AQL and the RQL, an operating characteristics (OC)
and include appropriate references. curve can be constructed to show the probability of acceptance
(or rejection) at any quality level for the acceptance plan, if the
Introduction plan specifies accept/reject decisions based on specified num-
bers of tests. If the acceptance plan incorporates pay adjustment
The use of statistical methods to assess test data derived for the factors for various quality levels, determination of the
and 
purpose of determining the characteristics of concrete and risks is a more complex task. Construction of OC curves for each
concrete-making materials is now established industry prac- of the various pay factors is one requirement necessary for fur-
tice. The use of probability-based acceptance criteria founded ther evaluation of the risks. Other tools that are helpful include
upon statistically assessed test data applicable to concrete and computer simulation programs to be used in the development
concrete-making materials is now an accepted industry prac- of percent within limit (or percent defective) acceptance plans
tice in many areas. Subsequent to the pioneering work by tailored to provide a predetermined buyer’s (or seller’s) risk, and
Walker [23], much research and development work was done expected pay (EP) curves that are developed to show the ex-
in and around the decade of the 1960s that aided in establish- pected average pay for given levels of quality.
ing the rational foundation for current practices. Many papers When a plan is designed to obtain the desired information
documenting the work were published and are still available in through a process called inspection by attributes, each item or
various records. One of the most succinct papers, by Abdun- group of items will usually be classified only as satisfactory or
Nur, explaining the need for probability-based concrete unsatisfactory so that the relevant parameter is percent satis-
strength specifications was published in 1962 [1]. The purpose factory or percent unsatisfactory. The number of each is
of this chapter is to provide suggestions regarding the use of recorded for use in decisions concerning the use or other dis-
statistical applications that are practical, valuable, and appro- position of the item(s). Attributes inspection is occasionally
priate when used in the sampling, testing, and evaluation of used on precast concrete items.

1
Consulting Engineer, Steele Engineering, Box 173, Tornado, WV 25202; or R. D. Zande & Associates, Inc., 424 2nd St., St. Albans, WV 25177.

22
 STEELE ON STATISTICAL CONSIDERATIONS 23

The majority of plans designed for concrete and concrete- not to use the range for this purpose. The standard deviation is
making materials use a process called inspection by variables. the preferred statistic, particularly when it is constant over the
Inspection by variables is done by measuring the selected char- range of values collected for the measured characteristic.
acteristic of a material or product and recording the measured When the standard deviation is not reasonably constant for the
value. The values are used to calculate fundamental statistical range of values encountered, the coefficient of variation
parameters needed to describe the characteristic. Such de- should be considered. If the coefficient of variation is constant,
scriptions are used in decisions concerning the use or other it would normally be used. If neither statistic is constant over
disposition of the material or product. the range of values obtained, either statistic may be used. How-
The fundamental statistics derived from variables inspec- ever, the standard deviation is preferred in such cases unless
tion that are most useful in making decisions concerning con- standard practice dictates otherwise. Note that it is acceptable
crete and concrete-making materials are the mean and the practice to apply either statistic to subset ranges of the overall
standard deviation. The mean (arithmetic), when derived from set of values when said statistic is constant for the subset.
data gathered by a properly designed plan, conveys, in one When the behavior of one variable is to be compared to the
value, the average value (central tendency) of the measured behavior of another variable, a correlation coefficient may be
characteristic of the material or product. It also indicates that determined. A good, or significant, positive correlation {A} im-
approximately one half of the measured values of the charac- plies that as one of the variables being correlated increases, the
teristic will have values greater than, and approximately one other tends to increase also.2 Alternatively, the second variable
half will have values smaller than the average value, when may tend to decrease as the first increases (negative correlation).
measurements are made. Although the mean is a very impor- All that is indicated by a good correlation is that the trend ex-
tant statistic, it does not show how far or in what way the hibited by the data when one variable is plotted against the other
greater or lesser values may be distributed from the mean. This is sufficiently well defined to permit rejection of the hypothesis
information is conveyed by the standard deviation, a powerful that there is no relationship between the two variables—in other
statistic that measures variation about the mean, or dispersion. words, that the relationship is completely random.
The standard deviation indicates, in one figure, how far above Other statistical terms will be encountered when an
or below the mean other values will be and how many values owner/specifying authority requires comparison of test data
will likely be found at any distance from the mean. derived by said authority’s representative with test data derived
Other statistics occasionally derived from variables inspec- by the supplier/contractor’s representative. The purpose of
tion of concrete and concrete-making materials include skew- such comparisons will usually be twofold. First, to determine
ness and kurtosis. Skewness indicates whether the distribution whether the two datasets are from the same population, and
of values tend to be grouped unequally above or below the second, to determine whether the material and completed
mean (nonsymmetrical) rather than equally occurring on each work comply with the governing specifications for the con-
side (symmetrical). Kurtosis indicates whether the frequency of tract. The additional terms encountered may include some or
occurrence of values at any distance from the mean is greater all of the following: difference two sigma limit (d2S limit), dif-
or less than the frequency expected from a normal distribution. ference two sigma percent limit (d2S % limit), t-test, F-test.
Other statistics representing central tendency, such as geo- The d2S limit is the maximum acceptable difference be-
metric mean, mode, etc., are seldom calculated for concrete tween two test results obtained on a split sample of the same
and concrete-making materials characteristics. Likewise, other material using the same process followed in deriving the limit
statistics concerning dispersion, such as the average deviation, value. (Additional discussion of this limit and the d2S % limit is
are usually not calculated. However, in addition to the stan- in the Precision and Bias section of this chapter.) Since d2S lim-
dard deviation, two statistics concerning dispersion are fre- its appear in many ASTM standards, the comparison process is
quently calculated. These are the range and the coefficient of easily followed. When more than one pair of split sample test
variation (CV). The range conveys the difference between the results are to be compared, a paired t-test may be used. The cal-
largest and smallest value of the measured characteristic. It is culated t-value, based on differences within pairs, is compared
not uncommon to see the range used in decisions concerning to a critical t-value at the selected significance level. Both of
the use of materials or products. The coefficient of variation is these comparisons provide information only on test procedure
the ratio of the standard deviation to the mean of a group of and equipment variability, as they are designed to do.
values, usually expressed as a percentage. This statistic is fre- A more powerful comparison process involves independent
quently calculated for measured characteristics of concrete random samples obtained by representatives of the owner/
and concrete-making materials. specifying authority and representatives of the supplier/con-
Generally, as previously noted, the most useful statistic tractor. First, the F-test is used to compare the variances of the
concerning the central tendency of data collected during the two independent sets of data. The ratio of the variances is cal-
measurement of concrete and concrete-making materials char- culated and compared to a critical F value at the selected signif-
acteristics is the arithmetic mean. The most useful statistic con- icance level. If the null hypothesis (i.e., the hypothesis that the
cerning the dispersion of said data is less straightforward. The two independent sets of data are from the same population) is
simplest of the three statistics that are commonly encountered upheld for the variances, a t-test is then used to compare the
is the range. The range is also the least powerful and its use is means of the two independent datasets. The calculated t-value is
normally limited to specific applications (for example, control compared to a critical t-value at the selected significance level. If
charts). The range is also used occasionally to estimate the stan- the null hypothesis is upheld for the means, the two sets of data
dard deviation. However, since many pocket calculators and are generally considered to be from the same population. This
spread sheet software used in construction activities have keys comparison provides information on test procedure and equip-
for direct calculation of the standard deviation, it is preferable ment variability, materials variability, and process variability.

2
Letters in braces refer to the notes attached to the end of this chapter.
24 TESTS AND PROPERTIES OF CONCRETE

Regression Lines
The standard method of using paired data as a source in de-
veloping a process for predicting one variable from another is
by calculating a linear regression line using the method of least
squares [6–10]. This method is very useful in situations that in-
volve actual calibration of a measuring device and in which the
plotted points approximate the calculated regression line very
closely with small scatter {B}. When a coefficient of determi-
nation, r2 (i.e., the square of the coefficient of correlation, r), is
calculated for an assumed linear relationship, consideration
should be given to selecting a minimum acceptable r2. How-
ever, note that an r2 value that is closer to the value of 0 than
the selected minimum only means that there is little linear cor-
relation between the variables. There may or may not be a high
nonlinear correlation between the variables.
When two separate measurements, each obtained from dif-
ferent test methods, are statistically related to each other and
one of the test methods is used to obtain measurements that are
then used to predict measurements of the other type, there may
be a large variation in the prediction obtained from the meas-
ured value. Note C and the references contained therein illus-
trate this point. While both sets of measurements are affected
by changes in the property of interest, each is actually measur-
ing two different quantities and each is affected by different sets
of influences extraneous to the property of interest. This is usu-
ally accounted for by recognizing that many measurements are
needed to derive a reliable line.
If regression lines showing the relationship between data
from two measurement systems are to be calculated and used
for purposes other than illustrating the relationship existing in
the data, several points should be considered. Fig. 1—Regression curve and confidence limits for
compressive strength versus rebound numbers [12].
• Derivation of a linear regression line involves determina-
tion of two parameters, the slope of the line, b1, and the Y
intercept of the line, b0. Due to the departures of the data regression equation to convert measurements obtained in one
points from the line, each of these parameters has an esti- system to measurements that might have been obtained by
mated variance and corresponding standard error (the some other system introduces an additional degree of uncer-
square root of the variance): sb12 for the slope and sb02 for tainty into the process. Extrapolation beyond the limits of the
the intercept. In addition, there is an estimated variance, data from which the equation was derived increases the degree
sy2, and corresponding standard deviation, sy, calculated of uncertainty in the derivation. Further, the use of relation-
from the sum of the squares of the deviations of the meas- ships other than linear equations will, in appropriate situa-
ured Ys above and below the fitted line. The three meas- tions, provide more reliable data {H}.
ures of variation (variances or their corresponding square
roots) should always be given whenever a regression line Sampling
is reported. The standard errors of the slope and intercept
indicate the significance of the relationship {D}. In addi- The goal of sampling concrete and concrete-making materials
tion, the number of pairs of data from which the regres- is to obtain samples from which an unbiased estimate of the
sion was calculated and the upper and lower limits of the characteristic of interest of a lot of material or product can be
data of both X and Y values should be reported along with obtained. Therefore, when a sample is used to evaluate a char-
the other parameters. acteristic of a lot (quantity) of material, all material in the lot
• A confidence interval for the fitted line should be shown should have an equal chance of being included in the sample.
on the graph of the line {E} (Fig. 1). Alternatively, if the probability of inclusion is not the same for
• There are three kinds of confidence intervals that can be all parts of the lot, this fact should be considered when evalu-
calculated for a fitted regression line: the line as a whole, ating data derived from the sample.
a point on the line, or a future value of Y corresponding Given the preceding goal, the simplistic answer may seem
to a given value of X {F}. to be a requirement that all material in the lot should have
• When appropriate confidence limits are correctly calcu- an equal chance of being included in any sample obtained
lated and reported, proper use requires a practical under- from the lot. However, in many situations this may not be
standing of the confidence interval given {G}. practical. Consider, for example, a 100-yd3 (76-m3) concrete
It should be reiterated when using regression lines that placement.
new measurements are probably subject to the same degree of • Using an appropriate plan, the fresh concrete could be
variation that characterized the original data. sampled during the placement process in a manner that
In general, then, if any system of measurements is valid, would assure that all concrete had an equal chance of be-
“the system must have validity within itself.” The use of a ing included in the sampling operation.
 STEELE ON STATISTICAL CONSIDERATIONS 25

• In contrast, using an appropriate plan for sampling hard- easily implemented by dividing the lot or quantity of material
ened concrete, it is likely that in most situations the hard- into a number of sublots or subquantities. The most conven-
ened concrete in different segments of the placement ient number is usually equal to the number of samples to be
would have an unequal chance of being included in the obtained. Each sample is then randomly obtained from one
sampling operation (or even no chance). Final geometry, sublot or subquantity. The purpose of stratified random sam-
critical stresses, reinforcement, and lack of accessibility all pling is to prevent the possibility that several samples would be
contribute to the situation. obtained from one segment of the lot or quantity of material
It is, therefore, very important that the circumstances in being sampled.
each sampling situation be recognized and receive competent An adequate plan for evaluating specific characteristics of
consideration. a material (or process) will provide sampling direction. This
The term “representative sample” has been used in many should include the type, frequency, method, and location of
different ways in the construction industry. The most useful sampling. Furthermore, the quantity and handling of samples
definition would paraphrase the goal of sampling previously subsequent to selection should be clearly detailed [19].
noted; that is, a representative sample is a sample that is ob-
tained from a lot or quantity of material, using procedures that Testing
will allow an unbiased estimate of a particular characteristic of
the material to be derived from the sample. If the stated defi- The reason for testing concrete and concrete materials is to
nition is to be operant, the sample must be selected randomly. produce data from which unbiased estimates of certain char-
ASTM Practice for Random Sampling of Construction Materi- acteristics of the material can be derived. The reliability of
als (D 3665) contains a random number table that can assist these estimates improves as the number of test results in-
the sampler in obtaining random samples. creases. Also, although depending somewhat on the purpose
The definition should also assist in a better understanding for which the estimates will be used, the reliability tends to
of composite samples. If, for example, a particular characteris- increase as the quantity of material undergoing test increases.
tic of several randomly selected 2000-lb (900-kg) units of aggre- To illustrate, consider the 2000-lb (900-kg) unit of aggregate
gate is to be used to estimate that characteristic in 10 000 tons example set forth in the previous section on sampling. With-
(9000 metric tons) of material, the characteristic can be deter- out resorting to mathematical proof (which could be done), it
mined from testing each of the 2000-lb (900-kg) units in total. is intuitively seen that the performance of a test on the entire
Alternatively, five (or some other suitably determined number) 2000-lb (900-kg) unit will likely yield a better estimate of the
randomly selected subsamples may be selected from each of characteristic being determined than would the performance
the 2000-lb (900-kg) units for compositing. The five randomly of a test on a 10-lb (4 1/2-kg) subsample obtained from the
selected subsamples from each of the 2000-lb (900-kg) units 2000-lb (900-kg) unit. Likewise, estimates based on ten 10-lb
may then be combined to form a single sample obtained from (4 1/2-kg) subsamples obtained from the 2000-lb (900-kg) unit
each 2000-lb unit randomly selected for sampling. This is called will be more reliable than the estimate derived from one 10-
a composite sample. Results from the single composited lb (4 1/2-kg) subsample. Then, as the number of subsamples
samples may then be used to estimate the characteristic of the is increased, the reliability of estimates based thereon will
10 000-ton (9000-metric ton) lot of aggregate. Obviously, this approach that of the 2000-lb (900-kg) unit. Finally, when the
alternative provides a possible advantage in that much smaller number of subsamples obtained from the 2000-lb (900-kg)
quantities of aggregate are handled in the final sample. How- unit equals 200, the reliability of the estimates will be equal.
ever, a possible disadvantage to be considered is that variations However, as noted in the original example, the information
in the characteristic of interest that exist in the 2000-lb (900-kg) concerning the characteristic that can be derived from the
unit of aggregate will remain unknown. A second alternative for 200 subsamples will be far greater than that which can be
compositing is available that provides the information needed derived from treating the 2000-lb (900-kg) unit as a single
for estimating the variations within the 2000-lb (900-kg) units of sample.
aggregate for the characteristic of interest. The second alterna- It is also intuitively evident from the previous example that
tive requires that each subsample be individually retained and testing additional samples or samples of greater size increases
used for testing. The data derived from each subsample in a the cost of testing. It is necessary, therefore, to establish the
2000-lb (900-kg) unit may then be averaged and used in the reliability required in each case commensurate with the re-
same manner as the data derived from the composited sample sources that are to be made available and with how much in-
in the first alternative. formation it is necessary to obtain. These decisions should also
Similar types of examples could be presented for cement, be based on a determination of the consequences of inadver-
other concrete-making materials, and concrete, either fresh or tently accepting defective material due to a low level of relia-
hardened. bility in the sampling and testing process.
The number of samples needed to provide the information Similar examples could be presented for cement, other
necessary for estimating a particular characteristic of a lot or concrete-making materials, and concrete, either fresh or hard-
quantity of material with a given confidence in the results can ened.
be easily determined. ASTM Practice for Calculating Sample Although it is seldom articulated, it may be better to make
Size to Estimate with a Specified Tolerable Error, the Average no tests than to make tests with poor samples that do not por-
for Characteristic of a Lot or Process (E 122) contains equa- tray the actual properties of the materials. An engineer, repre-
tions for calculating sample size. Note that “sample size” as senting the owner or specifying authority, who must rely on
used in ASTM E 122 is equivalent to the “number of samples” samples or tests that do not provide reasonably unbiased esti-
as commonly used in the construction industry. mates of the properties of interest for materials or structures
A useful concept when sampling concrete and concrete could probably make more appropriate decisions if there were
materials is “stratified random sampling.” This concept can be no samples or tests available. Such decisions would likely be
26 TESTS AND PROPERTIES OF CONCRETE

based on conservative assumptions, with large safety factors, should be fully compatible with the precision and bias infor-
rather than reliance on fallacious information with consequent mation contained in the test standard used for generating the
unknown risks. While neither of these situations will generally data. Reference [27] provides additional information concern-
yield optimum cost effectiveness, the greater risks inherent in ing this point. Test standards that contain no information con-
using faulty data are obvious. Optimum cost effectiveness mini- cerning precision are of limited value in making acceptance or
mizes the risk that the owner will accept faulty material and rejection decisions.
the risk that the supplier will have acceptable material rejected. Standards {L} best suited for the determination of preci-
These risks were previously defined and discussed in the Sta- sion and bias of tests for concrete and concrete-making
tistical Parameters section of this chapter. materials are ASTM Practice for Preparing Precision and Bias
Statements for Test Methods for Construction Materials (C 670)
Evaluation and ASTM Practice for Conducting an Interlaboratory Test Pro-
gram to Determine the Precision of Test Methods for Con-
Test data, to be useful after they have been obtained, must be struction Materials (C 802). The indices of precision used in
evaluated with respect to a standard or potential standard such these standards are, in order of preference, the “difference two-
as contract documents, specifications, design criteria, or other sigma limit (d2S) and the difference two-sigma limit in percent
previously determined concepts. Any evaluation of concrete or (d2S %). The indices are determined by multiplying the factor
concrete-making materials will be more practical and informa- 2(2)1/2  2.828 by the standard deviation of test results or the
tive when commonly used statistical procedures are applied. coefficient of variation of test results, respectively. This is the
Typical guidance can be obtained in numerous publications. standard deviation or coefficient of variation obtained from a
ASTM Practice for Sampling and the Amount of Testing of Hy- properly planned interlaboratory study. The d2S or d2S % limit
draulic Cement (C 183) utilizes control charts {I} and quality published in a test standard will generally have descriptive
history to determine the frequency of testing and the evalua- terms preceding the value listed that will define the process fol-
tion of hydraulic cement. The versatile concept of moving av- lowed in deriving the value. Typical descriptive terms include
erages is demonstrated in the evaluation of cement strength the following: “single laboratory,” “within laboratory,” “multi-
uniformity as set forth in ASTM Test Method for Evaluation of laboratory,” “between laboratory,” “single operator,” “within
Cement Strength Uniformity from a Single Source (C 917). ACI test,” etc. When one of these indices is provided in a test stan-
214, Recommended Practice for Evaluation of Strength Test dard, the user of the standard can presume that only about one
Results of Concrete {J}, provides detailed procedures for the time in 20 will the difference between two test results {M} on
evaluation of concrete strength tests. split samples, obtained under approximately the same condi-
The principles contained in the three cited standards tions noted in the precision information, exceed the value
above can be extended to many other concrete and concrete- indicated by the index. If more than two results are to be
making materials tests. However, for a few nonparametric {K} compared, different multipliers are required to determine the
tests this is not the case. Examples are ASTM Test Method for index. ASTM C 670 contains a table of multipliers for
Scaling Resistance of Concrete Surfaces Exposed to Deicing determining indices for use in comparing up to ten test results.
Chemicals (C 672) that uses an ordinal scale of measurement When test values from independently obtained samples
and ASTM Test Method for Organic Impurities in Fine Aggre- are to be compared, the F-test and t-test should be used as pre-
gates for Concrete (C 40) that uses a nominal or classification viously discussed in the Statistical Parameters section of this
scale. The median can be used to indicate central tendency for chapter. Test values derived from independently obtained sam-
such tests. Reference [26] should be consulted for other pro- ples contain additional components of variance not present in
cedures applicable to these types of test data. test values derived from split samples. The following items are
Another consideration that may be encountered when very important in the maintenance of valid precision informa-
evaluating values derived from test data is whether retesting tion in standard test methods:
is appropriate. Generally, when the level of quality of a ma- • Requirements for the designated number of test results
terial or process is estimated from the evaluation of one set (one or more) that constitute a valid test should be stated
of test results that have been obtained in accordance with in a test standard.
appropriate standards, said estimate will be improved by in- • Criteria used in deriving the precision information should
cluding another set of test results, obtained in accordance be outlined in the standard.
with appropriate standards, in the evaluation procedure. The • When a standard test is revised, other than editorially, the
practice of retesting is fully acceptable if all test results, both precision information should be reviewed to determine
original and retests, are 1) obtained and tested in accordance whether a new precision index should be derived.
with appropriate standards, 2) included in the final evalua- Although other indices based on different criteria can be
tion, and 3) the final evaluation recognizes the increased derived and used, those noted earlier have been recommended
number of test results existing for that particular process or as the most appropriate for test standards developed by ASTM
lot of material. Committees C1, C9, D4, and D18.

Precision and Bias Statements Other Useful Standards

ASTM Practice for Conducting A Ruggedness or Screening
One of the most important factors concerning a test method Program for Test Methods for Construction Materials (C 1067)
that is used to determine acceptance and rejection of materials provides an economical procedure for the detection and re-
and construction in a buyer-seller relationship is the informa- duction of variations in test methods before initiating a com-
tion contained in the precision and bias statement. plete interlaboratory study.
Numerical limits, based on standard test data, which are Procedures for determination of precision based on values
included in specifications to govern acceptance decisions, from other related tests are presented in ASTM Practice for
 STEELE ON STATISTICAL CONSIDERATIONS 27

Calculating Precision Limits Where Values are Calculated from For a hypothetical rebound number of 25, this figure in-
Other Test Methods (D 4460). dicates a calculated average compression strength of 31.2 MPa
Procedures for deriving the variation of properties of (4530 psi), with the 95 % confidence interval extending from
concrete-making materials from a single source and appropri- 30.0 to 32.5 MPa (4350 to 4710 psi). Since the rebound num-
ate comparisons of such derivations to precision statement bers themselves have a distribution with a characteristic scat-
data are presented in ASTM Practice for Determining Unifor- ter, illustrated in this case by a standard deviation of 0.50 for
mity of Ingredients of Concrete from a Single Source (C 1451). averages of 20 rebound numbers, the 95 % confidence interval
ASTM Practice for Inclusion of Precision Statement Varia- for the average rebound measurement is from 24 to 26 (i.e.,
tion in Specification Limits (D 6607) presents guidance con- 25  [2(0.50)]}. These figures combined with the confidence
cerning the appropriate inclusion of test method precision interval for the line give an approximately 90 % confidence in-
values in specifications containing limits that are based on data terval for the predicted compressive strength of 27.6 to 35.1
obtained by use of said test method. MPa (4010 to 5070 psi).
Unfortunately, this is not the limit of the final uncertainty
Notes of the predicted result. A 95 % confidence limit is often inter-
preted as meaning that 95 % of future results will be within the
{A} For example, many in-place tests on concrete tend to limits given. What it actually means, however, is that if the ex-
show increases in the measurements obtained as concrete periment is repeated a large number of times, each with the
strength increases. A good correlation (or high correlation same materials and conditions and with the same number of
coefficient) between the results of any test that exhibits such determinations, and each time the 95 % confidence interval is
behavior and the results of standard strength tests does not in calculated, then 95 % of the intervals so calculated will include
itself constitute evidence that the relationship is sufficiently the true average. This does not mean that any particular one of
close to permit the use of one type of test as a means of pre- the intervals will actually contain the true average in its exact
dicting what will happen in another type of test. In fact, the cor- center. For any given determination of the line and its confi-
relation coefficient has limited use in the field of analyzing en- dence interval, it is highly unlikely that the calculated line will
gineering data. coincide exactly with the true line.
{B} An example is the calibration of proving rings that are {H} One example that illustrates the successful use of non-
used for calibrating testing machines. This use is discussed by linear equations for determining the relationship between cer-
Hockersmith and Ku [11]. tain variables in concrete is contained in ASTM Test Method
{C} Even when the existence of this uncertainty is recog- for Developing Early-Age Compressive Strength and Projecting
nized, its size and affect on the derived measurements may be Later-Age Strength (C 918). This standard involves the applica-
underestimated. This point is illustrated in the case concerning tion of the maturity concept. The concept goes back a long way,
prediction of compressive strength measurements from pene- but detailed work was done on its application to evaluation of
tration and rebound tests in Refs [12] and [13]. concrete strengths in the 1970s [15–18].
{D} In general, if the slope is not at least twice as large as The standard involves establishing a relationship between
its standard error, this is insufficient evidence to conclude that the strength and the logarithm of a quantity called maturity of
the true slope is other than zero (that is, the relationship is ran- the concrete that is defined as the product of temperature at
dom). which the curing is taking place and the curing time in hours
{E} Calculation of the quantities needed to plot the confi- (degree-hours).
dence interval is described elsewhere [6]. It should be noted {I} Monitoring Production-Continuous Evaluation (adapted
that the upper and lower confidence limits are represented by from Ref [20])—One of the most effective means of maintaining
two branches of the hyperbola that are closest together at the the quality of a manufactured product is by continuously moni-
point where X  X
, the average of all the X values used in cal- toring the quality by means of regularly performed tests through-
culating the relationship. Thus uncertainty of the predicted Y out the process of production. The best tool for doing this is by
increases as the X value departs from X
in either direction. Oc- means of control charts.
casionally, confidence intervals have been plotted by multiply- The control chart became a well-established technique in
ing the SY (standard error of the Y estimate) by the t value for production quality control during the World War II era. The
the number of points used and drawing parallel straight lines control chart is a combination of both graphical and analytical
above and below the regression line. This is not correct and procedures. The basis of the theory arises from the fact that the
gives an optimistic picture of the uncertainty of estimated Y variation of a process may be divided into two general cate-
values. gories. One portion of the variation can be described as ran-
{F} Reference 6 describes how to calculate these three dom or chance variation of the process and the other as the
confidence intervals. It is important to note that only the con- variation due to assignable causes. A process that is operating
fidence interval for the location of the line as a whole is ap- with only chance variation should result in some distribution
propriate to use if the calculated line is to be used repeatedly of the measured characteristics, and one should be able to pre-
for predicting future values of Y from future observed values dict a range within which a certain percentage of the data
of X. should fall. If some assignable cause (such as an increased
{G} Figure 1 (taken from Ref [12]) illustrates the point. The water-cement ratio) results in a change in the distribution, then
regression line shown is based on 16 plotted points relating the values of the measured characteristic could fall outside the
strengths of 28-day cylinders to the average of 20 Swiss hammer predicted range.
rebound numbers obtained on slabs made from the same Three types of control charts that are frequently used are
batches of concrete as the cylinders. The figure also shows the control charts for averages (or moving averages), control
hyperbolic curves representing the upper and lower 95 % con- charts for standard deviations (or moving standard deviations),
fidence limits for location of the line referred to earlier. and control charts for ranges (or moving ranges). Also, control
28 TESTS AND PROPERTIES OF CONCRETE

charts for other measures, such as percent defective (or per- other end and the stages in between represent progression
cent within limits), may be useful. Detailed treatment of this from one level to another.
subject and tables of control chart constants for determining {L} As a result of concern about problems connected with
upper and lower control limits are presented in texts on statis- precision statements and how to develop and use them, a joint
tical quality control [14,19,21]. Table 27 in Ref [22] contains task group of ASTM Committee C1 on Cement, C9, D4 on Road
the control chart constants for averages, standard deviations, and Paving Materials, and D18 on Soil and Rock for Engineer-
and ranges. ing Purposes developed two practices: ASTM Practice for
{J} Evaluation of Strength Tests—One of the earliest and Preparing Precision and Bias Statements for Test Methods for
most widely used applications of statistics in the concrete field Construction Materials (C 670) and ASTM Practice for Con-
has been in the area of evaluation of strength tests both of ducting an Interlaboratory Test Program to Determine the Pre-
mortar cubes for the testing of cement strength and, more ex- cision of Test Methods for Construction Materials (C 802).
tensively, for the analysis of strengths of concrete specimens, ASTM C 670 gives direction and a recommended form for writ-
usually, in the United States, in the form of 6 by 12-in. cylin- ing precision statements when the necessary estimates (usually
ders. The chief pioneer in this effort was Walker who published standard deviations) for precision and/or bias are in hand.
his study in 1944 [23]. In 1946, largely at the instigation of ASTM C 802 describes a recommended method for conducting
Walker, the American Concrete Institute (ACI) began work on an interlaboratory study and analyzing the results in order to
statistical evaluation of compression tests that eventually obtain the necessary estimates. Both of these standards should
resulted in the publication of ACI Standard Recommended be studied and followed closely by any task group that is
Practice for Evaluation of Strength Test Results of Concrete charged with writing a precision and bias statement for con-
(ACI 214-03 ). First published as a standard in 1957, this docu- struction materials.
ment has undergone a number of revisions. {M} If two results differ by more than the (d2S) limit, a
In 1971 a symposium was conducted at the ACI Fall Con- number of interpretations are possible. Which interpretation is
vention on the subject “Realism in the Application of ACI Stan- most appropriate depends on various circumstances con-
dard 214-65.” This symposium presented valuable information nected with the situation, and in most cases a degree of judg-
on the meaning and use of ACI 214 and resulted in a sympo- ment is involved.
sium volume that included seven papers presented at the The limit in a precision statement is to provide a crite-
symposium, a reprint of ACI 214-65, and reprints of two earlier rion for judging when something is wrong with the results.
papers dealing with evaluation of concrete strengths [24]. Thus the failure of a pair of results to meet the (d2S) crite-
{K} Nonparametric Tests—There are some test methods rion causes concern that the conditions surrounding the two
that do not provide numbers for which the customary tests may not be the same as those existing when the preci-
processes of calculating means, standard deviation, (d2S) lim- sion index was derived or that the samples used in the two
its, and other so called parametric statistics are applicable. tests are not unbiased samples from the same type of mate-
Such tests measure on a nominal or classification scale, or on rial. The appropriate action to take depends on how serious
an ordinal or ranking scale [25]. Test methods of the latter type the consequences of failure are. In most cases, a single iso-
sometimes cause problems because of the fact that numbers lated failure to meet the criterion is not cause for alarm, but
are assigned to the different levels of quality of performance in an indication that the process under consideration should be
the method, and then the numbers are treated as though they watched to see if the failure persists. If appropriate, the tests
represented measurement on an interval scale, which is the can be repeated, and usually the procedures of the labora-
type of measurement scale appropriate to most concrete or tory(s) involved should be examined to make sure that the
concrete-making materials test methods. Because of the num- test is performed in accordance with the standard from
bers derived, there is a temptation to average results of several which the precision statement was developed.
specimens and even to calculate standard deviations to indi- Failure to meet a multilaboratory precision limit may en-
cate scatter. Such calculations are inappropriate when the tail more serious consequences than those connected with fail-
magnitudes of the numbers indicate only order or rank and ure to meet a single-operator criterion. The latter is sometimes
not measurements of quantities. When lengths are measured, used to check the results and procedures of a single operator
for instance, the difference between an object that measures 5 in a laboratory, and failure to meet the criterion leads to reex-
cm and one that measures 6 cm is a length of 1 cm. The same amination of the materials and procedures. If the test is being
difference applies to two objects that measure 9 and 10 cm, re- used to determine compliance with a specification, the single-
spectively. The increment of one between scaling ratings of one operator (d2S) limit should be used to check whether or not
and two, however, is not necessarily the same increment as that the results obtained are a valid test for the purpose. The former
between three and four. Adding ranking numbers of this type may occur in situations where there is a dispute about accept-
and dividing by the number of measurements may have little ance of materials. In these cases, both laboratories should
significance. Central tendency and scatter can be indicated by obtain two results by the same operator who was used in the
giving the median and the range. multilaboratory tests, and use the single-operator difference as
A test that provides measurement on a nominal or a check on proper performance of the test method within the
classification scale is one in which results merely fall into dif- laboratories.
ferent categories without any judgment being made that one Also note that conditions, materials, apparatus, operators,
category is higher or lower than another; for example, ASTM etc., change with time. In many cases, the subcommittee re-
C 40. In one procedure, a solution from the test sample is sponsible for the test method can obtain proficiency sample
compared to a reference solution and judged to be lighter, data from the Cement and Concrete Reference Laboratory or
darker, or the same. In another procedure, five color stan- the AASHTO Materials Reference Laboratory from which ap-
dards may be used. The latter may be treated as an ordinal propriate revisions to update a precision statement can be
scale if one end of the scales is judged to be better than the drafted as shown in the ASTM C 670 appendix.
 STEELE ON STATISTICAL CONSIDERATIONS 29

[15] Hudson, S. B. and Steele, G. W., “Prediction of Potential

References Strength of Concrete from the Results of Early Tests,” High-
[1] Abdun-Nur, E. A., “How Good is Good Enough,” Proceedings, way Research Record No. 370, Highway Research Board,
American Concrete Institute, Vol. 59, No.1, Jan.1962. Washington, DC, 1971, pp. 25–28.
[2] AASHTO (American Association of State Highway and Trans- [16] Hudson, S. B. and Steele, G. W., “Developments in the Predic-
portation Officials), “Implementation Manual for Quality As- tion of Potential Strength of Concrete from Results of Early
surance,” Publication Code IMQA-1, February 1966. Tests,” Transportation Research Record No. 558, Transportation
[3] AASHTO, “Quality Assurance Guide Specification,” Publication Research Board, Washington, DC, 1975, pp. 1–12.
Code QA-1, February 1966. [17] Lew, H. S. and Reichard, T. W., “Prediction of Strength of Con-
[4] AASHTO Standard Recommended Practice R9, “Acceptance crete from Maturity,” SP-56, American Concrete Institute, 1978,
Sampling Plans for Highway Construction,” AASHTO Standard pp. 229–248.
Specifications for Transportation Materials and Methods of [18] Hudson, S. B., Bowery, F. J., and Higgins, F. T., “Research Study
Sampling and Testing, Part 1B. to Refine Methods and Procedures for Implementing the
[5] AASHTO Standard Recommended Practice R10, “Definition of Method of Early Prediction of Potential Strength of Portland
Terms for Specifications and Procedures,” AASHTO Standard Cement Concrete,” West Virginia Department of Highways Re-
Specifications for Transportation Materials and Methods of search Project 47, Final Report, Woodward-Clyde Consultants,
Sampling and Testing, Part 1B. Rockville, MD, 1976.
[6] Natrella, N. G., Experimental Statistics, Handbook 91, National [19] Bennett, C. A. and Franklin, N. L., Statistical Analysis in Chem-
Bureau of Standards, Gaithersburg, MD, 1963. istry and the Chemical Industry, Wiley, New York, 1954.
[7] Dixon, W. J. and Massey, F., Jr., Introduction to Statistical Analy- [20] McLaughlin, J. F. and Hanna, S. J., “Evaluation of Data,” Signif-
sis, McGraw-Hill, New York, 1957. icance of Tests and Properties of Concrete and Concrete-Mak-
[8] Daniel, C. and Wood, F. S., Fitting Equations to Data, Wiley, ing Materials, ASTM STP 169A, ASTM International, West Con-
New York, 1971. shohocken, PA, 1966, p. 36.
[9] Acton, F. S., The Analysis of Straight Line Data, Wiley, [21] Duncan, A. J., Quality Control and Industrial Statistics, rev. ed.,
New York, 1959. Richard D. Irwin, Inc., Homewood, IL, 1959.
[10] Mandell, J., The Statistical Analysis of Experimental Data, Wiley, [22] Manual on Presentation of Data and Control Chart Analysis,
New York, 1964. ASTM STP 15D, ASTM International, West Conshohocken, PA,
[11] Hockersmith, T. E. and Ku, H. H., “Uncertainty Associated with 1976.
Proving Ring Calibration,” Reprint Number 12.3-2-64, Instru- [23] Walker, S., “Application of Theory of Probability to Design of
ment Society of America, 1964, (reprinted in Precision Mea- Concrete for Strength,” Concrete, Vol. 52, No. 5, Part 1, May
surement and Calibration, National Bureau of Standards, Spe- 1944, pp. 3–5.
cial Technical Publication 300, Vol. 1, 1969.) [24] “Realism in the Application of ACI Standard 164–65,” SP-37,
[12] Arni, H. T., “Impact and Penetration Tests of Portland Cement American Concrete Institute, Detroit, MI, 1973.
Concrete,” Report No. FHWA-RD-73-5, Federal Highway Ad- [25] Siegel, S. in Nonparametric Statistics for the Behavioral
ministration, Washington, DC, Feb. 1972. Sciences, McGraw-Hill, New York, 1956, pp. 16–28.
[13] Arni, H. T., “Impact and Penetration Tests of Portland Cement [26] Siegel, S. in Nonparametric Statistics for the Behavioral
Concrete,” Highway Research Record No. 378, Highway Re- Sciences, McGraw-Hill, New York, 1956, pp. 26–30.
search Board, Washington, DC, 1972, pp. 55–67. [27] Philleo, R. E., “Establishing Specification Limits for Materials,”
[14] Burr, I. W., Engineering Statistics and Quality Control, McGraw- ASTM Cement, Concrete, and Aggregates Journal, Vol. 1, No. 2,
Hill, New York, 1953. 1979, pp. 83–87.
5
Uniformity of Concrete-Making Materials

Anthony E. Fiorato1

Preface placing, and curing. Uniformity of properties may be as impor-

tant to the concrete supplier as the individual properties them-
THE SUBJECT OF VARIABILITY OF CONCRETE- selves. Within relatively broad limits, the supplier can adapt
making materials was first covered in ASTM STP 169C. It was concrete mixtures to accommodate individual properties of
derived from the work of the joint ASTM C01/C09 Task constituent materials. However, once that is done, it is essential
Group on Variability of Concrete-Making Materials. The Task to maintain uniformity to assure consistent concrete properties
Group provided the impetus for development of guidelines and performance. For example, once a mixture has been devel-
on determining uniformity of concrete-making materials, oped for a specific cement, water reducer, retarder, air-entrain-
which eventually became ASTM Standard Practice for Deter- ing admixture, aggregates, and batch water, unanticipated
mining Uniformity of Ingredients of Concrete from a Single changes in critical properties of individual components can
Source (C 1451). cause problems with fresh or hardened concrete performance.
The ASTM C01/C09 Uniformity Task Group identified im-
Introduction portant properties of constituent materials, how much varia-
tion is acceptable, and how uniformity from a single source of
The goal of the concrete supplier is to provide a material that these materials can be controlled.
consistently meets requirements set out by the buyer, whether
these are defined in the form of prescriptive or performance Properties of Constituent Materials that
specifications. The question is, “How do we define and assure Affect Concrete Performance
uniformity of concrete?” To answer this question, it is neces-
sary to consider those factors that affect concrete properties In 1988, members of ASTM Committee C1 on Cement and
and performance. Committee C9 on Concrete and Concrete Aggregates were
The steps to obtaining concrete performance are concep- surveyed to obtain their impressions on the relative impor-
tually illustrated in Fig. 1. The process starts with a mix design tance of concrete-making materials. Members were asked to
and specification developed for the particular application. It is rank major constituent materials in their order of importance
followed by selection and acquisition of constituent materials relative to variability, and also in the context of field practices
and processing of those materials in accordance with the speci- for three categories of construction (residential, low-rise com-
fications. Presumably, if the design, selection, and implemen- mercial, and high tech/high strength). In addition, each con-
tation steps are properly conducted, the concrete properties stituent material was rated relative to its own properties and
and performance will meet job requirements. However, it is attributes.
naive to assume that the steps to obtaining properties and per- Twenty-eight members responded to the survey. Since no
formance can be achieved without accommodating variations. attempt was made to scientifically select the sample popula-
But what level of variation can be accepted without detrimen- tion, no claim can be made for statistical significance.
tal impact on performance? This chapter will address the vari- However, the respondents are among the world’s most knowl-
ability of concrete-making (constituent) materials and their edgeable and experienced individuals in concrete materials
effects on performance. technology. Therefore, the survey can be considered a valid
With improvements in concrete technology, concrete has representation of industry experience and perceptions regard-
become more versatile, but also more complex in that the num- ing those materials’ characteristics that affect concrete per-
ber of mixture constituents has increased. It is rare to encounter formance. And anecdotal evidence indicates these perceptions
concrete that consists only of cement, fine aggregate, coarse ag- remain unchanged today.
gregate, and water. Today, most mixtures also contain chemical The survey provides guidance on specific materials’ prop-
admixtures or mineral admixtures (supplementary cementi- erties and performance attributes that impact concrete
tious materials) or both. To minimize the variability of concrete, properties and performance. This information is valuable in
it is necessary to control the uniformity of constituent materi- identifying properties that must be controlled to achieve uni-
als as well as the uniformity of batching, mixing, transporting, formity of performance.

1
President and CEO, CTLGroup, Skokie, IL 60077.

30
 FIORATO ON UNIFORMITY OF CONCRETE-MAKING MATERIALS 31

Figure 2 is a summary of responses (27 for this part) to a

question that required the respondent to rank ten major con-
stituent materials in order of importance from one being most
important to ten being least important. In the context of the
question, “important” relates to what impact variations in
the constituent material would have on concrete performance.
No distinction was made as to what performance aspect—
constructibility, strength, or durability—might be affected, but
it is likely that strength was the most commonly considered at-
tribute. Variations in cement are identified as the most impor-
tant by a significant margin. Variations in batch water are iden-
tified as least important. The relatively “unimportant” rankings
given to slag and silica fume may be related to a belief that
these materials have little variability, or to the fact that they are
less frequently used than the other constituents.
Figure 3 shows results when constituent materials were
rated within different construction types (residential, low-rise
commercial, and high tech/high strength). The intent of the
question was to determine the overall importance of potential
variability in the constituent material for selected types of con-
struction. Answers were to reflect whether the variability of the
constituent material can be considered to produce few or nu-
merous field problems. In this part of the survey, the materials
were rated (not ranked) on a scale of one (important) to ten
(not important). Not surprisingly, the overall importance
(lower rating numbers) increased from residential to commer-
cial to high tech. Cement was considered the most important
for all construction categories. For residential and low-rise
commercial construction, silica fume was considered least im-
portant (not likely to be used), while for high-tech/high-
strength concrete, batch water was considered least important.
While the results in Figs. 2 and 3 provide a rather general
picture of perceptions about the relative importance of con-
Fig. 1—Uniformity of concrete is a function of the entire stituent materials, another valuable part of the survey is sum-
design and construction process. marized in Table 1. For this part, major constituent materials
were evaluated independently of each other to identify those
characteristics that are important to performance. Respon-
dents were asked to rate each material property or perform-
ance attribute on a scale of one to three with one being most

Fig. 2—A 1988 survey of ASTM C1 and C9 committee members revealed their
perceptions about the impact of variability of constituent materials on concrete
performance.
32 TESTS AND PROPERTIES OF CONCRETE

example, Johansen and Taylor’s summary of the effect of ce-

ment characteristics on concrete performance [1]. Table 1
provides a comprehensive list of properties and attributes for
the major constituents in concrete. It also provides an indica-
tion of their perceived level of importance relative to defining
potential variations in concrete performance. This identifies
critical characteristics that should be considered in uniformity
standards for constituent materials. Such was the case for the
only existing uniformity standard for concrete-making materi-
als, ASTM Test Method for Evaluation of Cement Strength Uni-
formity from a Single Source (C 917).

Evaluation of Uniformity
An Example: ASTM C 917
The development of ASTM C 917 took place over a number
of years, starting in the 1960s and culminating in its first edi-
tion in 1979 [2]. It is worth reviewing the development of
ASTM C 917 because it is representative of efforts that are
needed to implement uniformity standards for properties or
attributes of other concrete-making materials. Therefore, the
following discussion is presented not to focus on cement
strength issues, but to illustrate the process of developing a
uniformity standard.
The fact that cement strength was selected as the first at-
tribute to be standardized is not too surprising given the ear-
lier discussion of Figs. 2 and 3 and Table 1 from the 1988
ASTM survey. In fact, the initial impetus for development of
ASTM C 917 can be traced to work by Walker [3] and Walker
and Bloem [4]. The key point is that a specific attribute of a
constituent material for concrete was identified as important
to the uniformity of concrete. This led to the establishment of
a joint committee of the Portland Cement Association and the
National Ready Mixed Concrete Association to address
strength uniformity [5]. The joint committee planned a pro-
gram to develop data on uniformity of cement strengths from
individual cement plants [6].
The joint committee selected 7- and 28-day strengths of
mortar cubes that conform to ASTM Test Method for Com-
pressive Strength of Hydraulic Cement Mortars (Using 2-in.
or 50-mm Cube Specimens) (C 109) as the reference for ce-
ment strength. After a pilot program in 1975, a one-year vol-
untary sampling and testing program was initiated in 1976.
Forty-six cement companies, representing over 100 plants in
the United States and Canada, participated. Testing was con-
ducted on grab samples representing 25 ton (23 Mg) lots of
Fig. 3—ASTM C01/C09 1988 survey of perceptions about cement at the rate of 30 samples per calendar quarter
the impact of variability of constituent materials on concrete (preferably ten per month and not more than one per day).
performance based on field practice for different construction Mortar cubes were prepared in each plant’s laboratory with
types. ten duplicate sets of cubes prepared each quarter to evaluate
within-laboratory test error. Data were submitted quarterly
for statistical analysis.
important and three being least important. Performance at- Data from the program that includes information on stan-
tributes reflect the behavior of the constituent material when dard deviations, coefficients of variation, 7- and 28-day average
incorporated in paste, mortar, or concrete. strengths, and ratios of 28-day to 7-day strengths were pub-
With further quantification, data such as listed in Table 1 lished in the appendix to ASTM C 917.2 They were arranged in
serve as the basis for uniformity standards. Before such stan- terms of cumulative percentage of plants falling below the
dards can be developed, specific characteristics to be con- value indicated for the statistic of interest, and provide a refer-
trolled must be identified and then quantified with respect to ence point for comparing strength uniformity results from a
their impact on concrete performance variations. See, for particular source.

2
Data developed in 1991 were used to update the standard [7].
 FIORATO ON UNIFORMITY OF CONCRETE-MAKING MATERIALS 33

TABLE 1—Perceived Relative Importance of Materials Characteristics to Concrete Quality (ASTM

C01/C09 1988 Survey)
Number of Respons Sulfate expansion 1.79 7 20 1
Volume changes 1.80 8 14 3 Number of Responsesa
Bleeding characteristics 1.86 10 12 6
Air entrainment dosage 1.88 8 13 5 Average 1 2 3
FinishingUNIFORMITY
CEMENT characteristics 1.96 8 13 7
Strengthproperty
Material gain beyond 28 days 2.07 8 10 10
Autoclave
Sulfate form expansion
and content2.07 5 15 7 1.36 19 8 1
Expansion
Fineness in moist storage2.12 5 12 8 1.39 17 11 0
FINEC3A,
AGGREGATE
C3S, C2S UNIFORMITY 1.43 18 8 2
Material property
Alkalies 1.50 17 8 3
Grading including
Solubility fineness modulus 1.18
of alkalies 23 5 0 1.68 12 13 3
Deleterious particles
Chemical composition 1.43 18 8 2 1.78 9 15 3
Particles
200 sieve
Microscopically (amount and
determined type)
composition 1.50 15 12 1 1.93 8 14 6
Moisture
Heat of content
hydration 1.68 16 5 7 2.04 6 15 7
Particle shape
Air content 1.82 8 17 3 2.04 6 15 7
Absorption
Loss on ignition2.07 (L.O.I.) 3 20 5 2.11 8 9 11
Specific
SiO2, Algravity
2O3, MgO 2.18 4 15 9 2.15 3 17 7
Attrition
Specific (grinding
gravity during mixing) 2.29 4 12 12 2.46 3 9 16
Performance
Performanceattribute
attribute(in(inpaste,
paste,mortar,
mortar,concrete)
concrete)
Water requirement
Strength 1.21 22 6 0 1.14 25 2 1
Air entertainment
Strength gain from 7–28 days 1.36 20 6 2 1.21 22 6 0
Concrete strength
Setting time 1.57 14 12 2 1.25 22 5 1
Water
Early reducer
age strengtheffectiveness
gain 1.96 6 17 5 1.25 21 7 0
COARSE
Slump AGGREGATE UNIFORMITY
loss with admixtures 1.43 17 10 1
Material
Slumpproperty
loss with temperature 1.71 12 12 4
Grading
Drying shrinkage 1.18 23 5 0 1.72 10 12 3
Deleterious particles (amount and type)
Sulfate expansion 1.46 17 9 2 1.79 7 20 1
Particle
Volume shape
changes1.68 10 17 1 1.80 8 14 3
Particles
Bleeding
200 sieve (amount and type)
characteristics 1.75 11 13 4 1.86 10 12 6
Moisture content dosage 1.82
Air entrainment 14 5 9 1.88 8 13 5
Absorption 2.04
Finishing characteristics 6 15 7 1.96 8 13 7
Attrition
Strength(grinding
gain beyondduring 28mixing)
days 2.07 7 12 9 2.07 8 10 10
Specific gravity
Autoclave 2.14
expansion 5 14 9 2.07 5 15 7
Temperature
Expansion in moist 2.29storage 3 14 11 2.12 5 12 8
Chemical composition 2.32 3 13 12
FINE AGGREGATE
Performance UNIFORMITY
attribute (in paste, mortar, concrete)
Material
Concreteproperty
strength 1.39 17 11 0
Grading
Water including fineness
requirement modulus17
1.43 10 1 1.18 23 5 0
Deleteriousdurability
Freeze-thaw particles (D-cracking) 1.63 13 11 3 1.43 18 8 2
Particles
Drying
200 sieve (amount
shrinkage1.82 7 and type)
19 2 1.50 15 12 1
Moisture
Thermal vol.content
changes (cracking, etc.) 2.00 5 18 5 1.68 16 5 7
FLY Particle shape
ASH UNIFORMITY 1.82 8 17 3
Absorption
Material property 2.07 3 20 5
Specific
Loss gravity(L. O. I.) 1.11
on ignition 25 1 1 2.18 4 15 9
Attrition (grinding
Fineness 1.37during17 mixing) 10 0 2.29 4 12 12
Performance
Variations in attribute
CaO (in paste,
1.65 mortar,
13 concrete)
9 4
Water requirement
Variations in SO3 1.76 11 9 5 1.21 22 6 0
Air entertainment
Alkalies 1.78 10 13 4 1.36 20 6 2
Concrete in
Variations strength
SiO2 2.15 7 8 11 1.57 14 12 2
Water reducer
Variations in Al2Oeffectiveness 2.27 4 11 11 1.96 6 17 5
3
Variations
COARSE in Fe2O3 UNIFORMITY
AGGREGATE 2.48 2 9 14
Specific gravity
Material property2.26 3 13 10
Performance
Grading attribute (in paste, mortar, concrete) 1.18 23 5 0
Required
Deleteriousair-entrainment dosage
particles (amount 1.26
and type) 21 5 1 1.46 17 9 2
Reactivity with different cements
Particle shape 1.54 14 7 3 1.68 10 17 1
Time of set
200 sieve
Particles 1.58 (amount 13 and type)
11 2 1.75 11 13 4
Reactivity
Moistureat different temperatures 1.71
content 12 7 5 1.82 14 5 9
Response
Absorption to admixtures 1.81 9 13 4 2.04 6 15 7
Pozzolanic activity index
Attrition (grinding during1.81mixing) 10 12 5 2.07 7 12 9
Shrinkage
Specific gravity 2.00 3 18 3 2.14 5 14 9
WATER REDUCERS, HRWR, RETARDERS UNIFORMITY
Temperature 2.29 3 14 11
Material property
Chemical composition 2.32 3 13 12
Sensitivity to cement composition 1.30 19 8 0
Sensitivity to time of addition 1.48 15 11 1

(continues)
34 TESTS AND PROPERTIES OF CONCRETE

TABLE 1—Perceived Relative Importance of Materials Characteristics to Concrete Quality (ASTM

C01/C09 1988 Survey) (Continued)
Number of Responsesa

Average 1 2 3
Performance attribute (in paste, mortar, concrete)
Concrete strength 1.39 17 11 0
Water requirement 1.43 17 10 1
Freeze-thaw durability (D-cracking) 1.63 13 11 3
Drying shrinkage 1.82 7 19 2
Thermal vol. changes (cracking, etc.) 2.00 5 18 5
FLY ASH UNIFORMITY
Material property
Loss on ignition (L.O.I.) 1.11 25 1 1
Fineness 1.37 17 10 0
Variations in CaO 1.65 13 9 4
Variations in SO3 1.76 11 9 5
Alkalies 1.78 10 13 4
Variations in SiO2 2.15 7 8 11
Variations in Al2O3 2.27 4 11 11
Variations in Fe2O3 2.48 2 9 14
Specific gravity 2.26 3 13 10
Performance attribute (in paste, mortar, concrete)
Required air-entrainment dosage 1.26 21 5 1
Reactivity with different cements 1.54 14 7 3
Time of set 1.58 13 11 2
Reactivity at different temperatures 1.71 12 7 5
Response to admixtures 1.81 9 13 4
Pozzolanic activity index 1.81 10 12 5
Shrinkage 2.00 3 18 3
WATER REDUCERS, HRWR, RETARDERS UNIFORMITY
Material property
Sensitivity to cement composition 1.30 19 8 0
Sensitivity to time of addition 1.48 15 11 1
Compatibility with other admixtures 1.48 16 9 2
Percent solids 1.59 15 8 4
Composition and concentration 1.59 15 8 4
Sensitivity to temperature 1.62 15 6 5
Variations in chlorides 1.78 9 15 3
Temperature stability (freezing, etc.) 1.81 11 10 6
Variation in alkalies (HRWR) 1.85 8 14 4
Stability in storage 1.93 8 13 6
Performance attribute (in paste, mortar, concrete)
Time of set 1.11 24 3 0
Rapid stiffening 1.15 23 4 0
Early-age strength 1.41 17 9 1
Later-age strength 1.85 8 15 4
Finishing characteristics 1.93 10 9 8
AIR-ENTRAINING ADMIXTURE UNIFORMITY
Material property
Percent solids (specific gravity) 1.54 14 7 3
Composition (infrared spectra) 1.91 7 11 5
pH 2.00 7 10 7
pH in deionized water 2.17 3 13 7
Performance attribute (in paste, mortar, concrete)
Stability of air with fly ash 1.11 24 3 0
Air void system characteristics 1.27 20 5 1
Sensitivity to cement composition 1.37 19 6 2
Sensitivity to temperature 1.44 16 10 1
Generation of air voids 1.50 16 7 3
Compatibility with other admixtures 1.63 14 9 4
Sensitivity to aggregate grading 1.78 10 13 4
Sensitivity to mix water composition 2.15 7 9 11

(continues)
 FIORATO ON UNIFORMITY OF CONCRETE-MAKING MATERIALS 35

TABLE 1—Perceived Relative Importance of Materials Characteristics to Concrete Quality (ASTM

C01/C09 1988 Survey) (Continued)
Number of Responsesa

Average 1 2 3
LIGHTWEIGHT AGGREGATE UNIFORMITY
Material property
Unit weight 1.23 20 6 0
Absorption 1.27 19 7 0
Grading 1.38 16 10 0
Moisture content 1.38 19 4 3
Specific gravity 1.60 14 7 4
Particle shape 1.77 8 16 2
Attrition (grinding during mixing) 1.81 8 15 3
Performance attribute (in paste, mortar, concrete)
Concrete strength 1.31 20 4 2
Shrinkage and volume changes 1.58 12 13 1
Air entrainment 1.85 10 10 6
Absorption of admixtures 1.96 8 10 7
SLAG UNIFORMITY
Material property
Fineness 1.24 19 6 0
Glass content 1.48 14 10 1
Variation in chemical composition 1.74 10 9 4
Specific gravity 2.12 5 12 8
Performance attribute (in paste, mortar, concrete)
Activity index 1.25 19 4 1
Temperature 1.63 13 7 4
Required air-entrainment dosage 1.83 7 14 3
Required water reducer, HRWR dosage 2.00 7 10 7
Shrinkage 2.00 5 14 5
SILICA FUME UNIFORMITY
Material property
Composition 1.40 16 8 1
Percent solids 1.68 11 7 4
Stability in storage 1.73 11 6 5
Specific gravity 2.23 6 5 11
Performance attribute (in paste, mortar, concrete)
Air entertainment and air void system 1.44 16 7 2
MIX WATER UNIFORMITY
Material property
Chloride content 1.52 16 5 4
Organics content 1.62 14 8 4
Alkali content 1.83 9 10 5
Sulfate content 1.92 8 11 6
Hardness 2.25 4 10 10
pH 2.29 3 11 10
Solids content 2.38 2 11 11
Performance attribute (in paste, mortar, concrete)
Air entertainment 1.46 14 9 1
Time of set (Cl, Na2CO3) 1.72 10 12 3
Temperature 1.80 10 10 5
Durability (ASR, sulfate resist., etc.) 2.00 7 10 7

a
1  important to 3  unimportant.
36 TESTS AND PROPERTIES OF CONCRETE

In addition to comprehensive data that quantified poten-

tial strength uniformity, the PCA/NRMCA program provided
extensive information on sampling and testing procedures,
and correction factors for testing variations. Compilation of
this information into a draft recommended practice docu-
ment greatly facilitated the ASTM development process for
the new standard. Thus, the first ASTM standard for unifor-
mity of a concrete-making material, ASTM C 917, was ap-
proved in 1979, approximately two years after completion of
the test program.
Even with approval of a document such as ASTM C 917,
there is a continuing development and educational process
that must take place to foster appropriate use. A uniformity
standard should provide a communication tool between manu-
facturer and customer that will improve overall concrete qual-
ity and performance [2,8,9]. Education of both those providing
and those using uniformity data is an important step once a
standard is introduced [10,11]. A final test of the standard is
whether it is used. Widespread adoption of ASTM C 917 has
been slow [12,13]. However, with the increasing trend toward
total quality management, use has been growing.
The early work by Walker recognized that concrete
strength uniformity was not solely a function of cement
strength uniformity and included discussion of such factors as
sampling and testing variations, temperature effects, and age
effects [3]. Batching, mixing, transporting, placing, and curing
also have important implications. Walker also noted that
strength was not the only concrete performance attribute of
importance; he recognized constructibility and durability as
other critical attributes [3].
Given that today’s concretes are more sophisticated, it is
appropriate to consider standardization of uniformity provi-
sions for other constituent material properties and attributes
that affect concrete constructibility, strength, and durability.
The following section discusses a standard for determining
uniformity of other characteristics of concrete-making materi-
als that was developed by the ASTM C01/C09 Task Group and
is now designated as ASTM C 1451.

Standard for Determining Uniformity of

Concrete-Making Materials
Figure 4 illustrates the process for developing a uniformity
standard. First, the critical property or attribute of the particu-
lar constituent material must be identified. Potential properties
and attributes are listed in Table 1. It would be prohibitive
and unnecessary to develop uniformity requirements for each
specific property or attribute in Table 1, so those that are most Fig. 4—Process for developing uniformity standards for
critical must be selected. This is an appropriate responsibility concrete-making materials.
for ASTM committees that govern standards for materials
listed in Table 1. the effect of inherent variability of testing on the property or
Once the attribute has been identified for evaluation, the attribute being measured.
test method must be selected. In most cases, an appropriate Testing errors can be significant, particularly for between-
ASTM method exists, particularly for material properties. How- laboratory tests, and must be accommodated. Generally, single-
ever, for some attributes, standard procedures may need to be laboratory testing is preferred for uniformity work. For
developed or existing procedures modified. example, as pointed out by Gaynor [12], the precision state-
After test methods are selected, a comprehensive sampling ment in ASTM C 109 implies that duplicate batches of mortar
and testing program to quantify uniformity potential and test- made in a single laboratory should give strengths that do not
ing variability for the particular property should be conducted. differ by more than 10.7 % 1 time in 20. However, single
This will provide data for a specific standard’s provisions on batches mixed by two different laboratories should not differ
sampling frequency and protocol, testing criteria, statistical by more than 20.7 %.
corrections for testing variations, and reporting requirements. ASTM C 1451 should speed the standards development
The development of comprehensive test data is considered an process because the test program data for any property or
essential part of the process because it is necessary to define attribute can be “plugged in” to a standard format. It addresses
 FIORATO ON UNIFORMITY OF CONCRETE-MAKING MATERIALS 37

the following components for determining the uniformity of and a protocol for developing uniformity standards is dis-
properties of a material from a single source: cussed. With the growing sophistication of concrete mixtures,
1. Sampling and the importance of total quality to concrete users, it is es-
(a) Sampling is performed by trained personnel. sential that the industry be prepared to provide uniform prod-
(b) Grab samples are taken at a frequency defined by uct performance. This will lead to increasing reliance on
maximum lot sizes. uniformity standards by suppliers of constituent materials and
(c) Sampling protocols are defined by existing ASTM concrete producers.
standards.
(d ) A minimum rate of sampling of ten per month or two References
per week is required.
2. Evaluation Procedure [1] Johansen, V. and Taylor, P., “Effect of Cement Characteristics
on Concrete Properties,” EB226, Portland Cement Association,
(a) Samples are tested in accordance with standard
Skokie, IL, 2003.
ASTM procedures.
[2] Peters, D. J., “Evaluating Cement Variability—The First Step,”
(b) Variations from a single source are corrected for
NRMCA Publication 161, National Ready Mixed Concrete Asso-
variations inherent in the test procedures. ciation, Silver Spring, MD, 1 April 1980.
(c) Within- and between-laboratory testing variations are [3] Walker, S., “Uniformity of Concrete Strength as Affected by Ce-
considered as required. ment,” NRMCA Publication No. 77, National Ready Mixed Con-
(d ) Single-laboratory test variations are established by du- crete Association, Silver Spring, MD, March 1958.
plicate testing if no history is established. [4] Walker, S. and Bloem, D. L., “Variations in Portland Cement,”
(e) Between-laboratory variations are quantified by sam- Proceedings, ASTM International, West Conshohocken, PA, Vol.
ple exchange or standard reference samples. 58, 1958; also reprinted as NRMCA Publication No. 76.
3. Statistical Calculations [5] Davis, R. E., “Uniformity of Portland Cements: Facts and Fan-
(a) Equations are defined for average and total standard tasies,” Modern Concrete, April 1975, p. 61.
deviations of the measured values. [6] Klieger, P., “The PCA/NRMCA Strength Uniformity Study,” Re-
(b) Equations are defined for testing standard deviation search and Engineering Standards Committee Meeting, Na-
and coefficient of variation. tional Ready Mixed Concrete Association, Houston, TX, 2 Feb.
1976.
(c) Equations are defined for standard deviation cor-
[7] Sykora, V., “Amemdment of Appendix X2 in ASTM C 917 Evalu-
rected for testing variations.
ation of Cement Uniformity from Single Source,” Cement, Con-
4. Report Requirements
crete, and Aggregates, Vol. 17, No. 2, Dec. 1995, pp. 190–192.
(a) Reports identify materials tested.
[8] Davis, R. E., “How Uniform is a Portland Cement From a Single
(b) Reports cover a minimum of three and a maximum of Source,” Concrete Construction, Feb. 1977, p. 87.
twelve months. [9] Davis, R. E., “The Cement Producer’s Role in Providing Informa-
(c) Reports include duplicate test results. tion for an ASTM C 917 Method for Evaluation of Cement
(d ) Reports include specific statistical results for time Strength From a Single Source,” Research and Engineering
period covered. Standards Committee Meeting, National Ready Mixed Concrete
This format follows that of ASTM C 917 and provides Association, St. Louis, MO, 22 July 1980. (Available as NRMCA
a “generic” approach to accommodating any property or Back Engineering Report No. 80–10.)
attribute. [10] Gaynor, R. D., “Use of Cement Strength Uniformity Informa-
As discussed here, even after the development of the con- tion,” National Ready Mixed Concrete Association, Silver
sensus standard it will be necessary to educate users. This will Spring, MD, Feb. 1978.
be a continuing process. In addition, uniformity of concrete- [11] Gaynor, R. D., “Has There Been a Change in Strength Level? (Or
How to Look at C 917 Data When It Comes In),” NRMCA Semi-
making materials is necessary, but not sufficient, for obtaining
nar on Cement Strength Uniformity, St. Louis, MO, 22 July 1980.
uniform concrete performance. Field practices must also be
(Available as NRMCA Back Engineering Report No. 80–11.)
addressed.
[12] Gaynor, R. D., “Cement Strength Variability and Trends in
Cement Specifications in the U.S.,” European Ready Mixed Con-
Summary crete Organization (ERMCO) Congress, Rome, May 1986. (Avail-
able as NRMCA Back Engineering Report No. 86–3.)
This chapter has addressed uniformity of concrete-making ma- [13] Uniformity of Cement Strength, ASTM STP 961, E. Farkas and
terials. Those properties and attributes of constituent materials P. Klieger, Eds., ASTM International, West Conshohocken, PA,
that are considered to affect concrete uniformity are identified 1987.
6
Virtual Testing of Cement and Concrete

Dale P. Bentz,1 Edward J. Garboczi,1 Jeffrey W. Bullard,1

Chiara Ferraris,1 Nicos Martys,1 and Paul E. Stutzman1

Introduction accompanying standards, the truth is that a functional VCCTL

places a greater burden on testing and standards than ever be-
CONCRETE IS A UNIQUE MULTI-PHASE, RANDOM, fore. First, model property predictions can only be as good as
complex, and composite material that provides both strength the characterization of the starting materials. Thus, standardized
and impermeability to engineered structures. Unlike most con- methods for preparing and analyzing these materials will be crit-
struction materials, the properties of concrete continue to de- ical. And second, measurements of fundamental properties are
velop over time and in place—an aid in processing, but a com- needed in order to empower and validate the prediction models.
plication in testing and quality assurance. This is clearly The VCCTL will drive the standard tests to move from empiri-
exemplified by the bread-and-butter test of concrete quality cism towards a firm materials science basis, not eliminate them.
compliance, the 28 day compressive strength test. Having to Standards and testing are critical for validating and extending
wait 28 days to assure performance compliance is another the computer models. While current models can predict several
unique feature of concrete, and not necessarily an properties that are already covered by existing ASTM C01 (Ce-
attractive one. This holds true both for the field engineer, who ment) and C09 (Concrete) standards, they can also quantitatively
would prefer to directly proceed with further aspects of the predict other properties for which no standard test method cur-
construction, and the industrial or academic researcher, who rently exists. Even though much work remains to be done on re-
would prefer to avoid waiting 28 days (or more) before truly fining and extending these models, it is just as critical that the
knowing the effects of a new chemical additive or an alterna- necessary standards work be pursued and completed. The dual
tive processing methodology. Thus, many efforts have been objectives of this chapter, then, are (1) to show what current
made to predict (and ensure) concrete performance based on models can do, and (2) to highlight needs relative to existing and
waiting periods of less than 28 days. future standards for characterizing cement and concrete mate-
One area of effort has been the development of computer rials and for evaluating their properties.
models to predict cement and concrete properties via virtual
testing. In these models, typically, starting materials are char- What Really is Virtual Testing?
acterized in some fashion and this information is input into
one or more computer models to predict a variety of proper- “Virtual testing” is an exciting name, given the ubiquity of “vir-
ties of the fresh and hardening cement or concrete. If proper- tual” everything around us these days, but what does it really
ties can be successfully predicted, the time and cost savings to mean for cement and concrete materials? In the field of con-
the cement and concrete industry can be tremendous. Many densed matter physics, material properties are measured at a
companies spend millions of dollars per year on the testing of fundamental level. These measurements are then compared to
concrete, including material costs, storage, labor, and disposal. quantitative predictions from condensed matter theory that is
An additional benefit of virtual testing would be the capability based on valid mathematical principles and atomic and molec-
to rapidly perform a large number of “what-if” type computa- ular arrangements. “Simple” materials are usually studied. In
tions to explore new material systems and optimize existing materials science, more complex materials are studied, like ran-
ones. With this goal in mind, in January of 2001 a joint indus- dom composites and biological materials. For many of these ma-
try/government consortium was initiated to develop a web- terials, it is not possible to carry out analytical calculations and
based Virtual Cement and Concrete Testing Laboratory so the field of computational materials science has been devel-
(VCCTL). An earlier version of the VCCTL, 1.1, is available for oped. All this entails is the usual condensed matter theory, but
the general public at http://vcctl.cbt.nist.gov. The progress of with the mathematics solved on a computer, since the problems
the consortium can be seen in three annual reports [1–3], and are analytically intractable. Cement and concrete are complex,
in numerous examples of properties that can be successfully multiscale materials. The chemistry of cement is well developed
predicted using the VCCTL [4] and other existing models. and quite fundamental and analytically based, but analyzing and
While at first glance, it may appear that virtual testing has predicting the materials science properties requires a level of
the potential to eventually eliminate physical testing and its theory involving computational materials science. So the

1
Chemical engineer, physicist, materials scientist, physicist, physicist, and geologist, respectively, Materials and Construction Research Division, National Institute of
Standards and Technology, Gaithersburg, MD 20899.

38
 BENTZ ET AL. ON VIRTUAL TESTING 39

definition of virtual testing of cement and concrete is just the of a cement’s hydration properties. There is work in ASTM
computational materials science of cement and concrete. subcommittee C01.23 at present on replacing the inaccurate
The ideal model for concrete would start from the known Bogue calculations with more accurate X-ray diffraction
chemical composition of the material. Starting with the correct Rietveld measurements [7]. But the hydration of a cement
proportions and arrangement of atoms, it would build up the takes place at the individual particle level, so to model the hy-
needed molecules, the nanostructure and microstructure dration, one must go to the particle level. Correct particle level
needed, and eventually predict properties at the macroscale. information is needed as an input into the virtual hydration
Even with modern-day computers, however, this ideal model is model, CEMHYD3D [8–10].
a long way away. It is still not possible to systematically build Particle level information includes the detailed breakdown,
up multiscale models starting from the atomic scale. The per particle, of the various clinker phases. In a cement, some
science-based virtual models that do exist need good data as particles are purely C3S, for example, while others are a complex
input, data from careful characterization of materials. The mixture of clinker phases. To acquire this information, a dry ce-
models can predict physical properties of interest to actual ma- ment powder is mixed into an epoxy and cured. A polished sec-
terials users, but are based on fundamental parameters, not tion is prepared, and a combination of back-scattered electron
empirical tests. For example, measuring the Blaine fineness scanning electron microscopy (SEM) and X-ray microprobe
(using ASTM Standard Test Method for Fineness of Hydraulic analysis are used to identify the chemical phase belonging to
Cement by Air Permeability Apparatus (C 204)) as a character- each pixel in the image. This procedure is being investigated for
ization of a cement does not give useful data for a model. On standardization within ASTM subcommittee C01.23. Figure 1(a)
the other hand, careful measurement of the cement particle shows the ordinary gray-scale back-scattered electron SEM im-
size distribution (PSD) does give information vital to the suc- age for a portland cement, while Fig. 1(b) shows the image after
cessful modeling of how the cement hydrates and develops all of the major chemical phases have been identified. More de-
microstructure. So the rise of virtual testing in this instance tails on this procedure are available elsewhere [8,11,12]. This
should drive the testing of Blaine fineness toward the meas- procedure can identify calcium sulfate, but cannot distinguish
urement of cement particle size distribution. Analogous exam- between its different hydrated forms. Quantitative X-ray diffrac-
ples, in which virtual testing could push existing test methods tion or thermogravimetric analysis [13] can give this informa-
toward more fundamental measurements, could be given in tion in bulk, but not at the particle level. In addition to the meas-
the areas of rheological properties and durability. urement of bulk volume fractions, the final segmented SEM
images can also be analyzed to determine phase surface frac-
The Importance of Materials Characterization tions and autocorrelation functions. Along with a measured par-
ticle size distribution, these three characteristics can be utilized
The heart of concrete, as well as the heart of its virtual testing, to reconstruct an initial three-dimensional microstructure of ce-
lies in the cement. The VCCTL models start at the level of the ment particles in water that is a very realistic representation of
cement particles. Specific needs are what cement particles look the specific cement in question [8,10]. This is a critical step in
like, how the various chemical phases of the cement are the successful simulation of cement paste microstructure and
distributed in the cement particles, what their particle size dis- performance properties. Mineral admixtures need to be charac-
tribution is like, and what is their shape. By saying “cement,” terized in the same way. A few forms of fly ash and slag have
gypsum is also included since it is almost always interground been so characterized [14,15], though further research on better
with portland cement. If there are other mineral admixtures, adapting the above characterization techniques to these materi-
such as fly ash, ground granulated blast furnace slag, als will prove fruitful [16].
metakaolin, silica fume, or something else, the same kind of Cement particle shape is important, too. The characteriza-
information is needed about them as well. tion technique for cement particle shape is very similar to that
To characterize a cement so that one has a hope of predict- for aggregates, so aggregate shape is considered first. Aggre-
ing its hydration and general performance, one must first meas- gates used in concrete, both fine and coarse, are of many dif-
ure the particle size distribution using some method. By far the ferent mineralogical types, and are either found naturally in
most popular method among cement companies and other lab- the desired gradation, or are crushed to the desired sizes. Their
oratories is the laser diffraction method [5,6]. This technique shape, which is measured in the ASTM Standard Test Method
uses optical laser light diffracting from the various cement for Flat Particles, Elongated Particles, or Flat and Elongated
grains to measure/compute an equivalent spherical diameter for Particles in Coarse Aggregate (D 4971), can vary greatly. The
each particle. This is critical for virtual testing, since smaller amount of internal porosity and hence water absorption can
cement grains hydrate faster and more completely than larger vary as well. Their different mineralogical types imply that
cement grains, having more exposed surface area, a phenome- their elastic moduli will also vary. To make a virtual concrete,
non that must be captured accurately for any model to have a one must in general use realistic aggregates. For some proper-
hope of being accurate. A measurement such as Blaine fineness, ties, like chloride diffusivity, the shape of model aggregates
while useful in conveying information about the relative fine- used does not seem to matter much. For other properties, like
ness of the cement, does not give the distribution of sizes. The fresh concrete rheology and mechanical properties, especially
hydration performance of a cement couples more to the distri- at early ages, aggregate shape means a lot. There are few stan-
bution of sizes, not just some measure of the average size. In dard tests that address aggregate shape (ASTM D 4971 and
modeling the hydration of a cement, a 3-D model of the particles ASTM Standard Guide for Petrographic Examination of
themselves is built up, so an accurate size distribution is needed. Aggregates for Concrete (C 295)), and none that attempt to
Accurate bulk measures of cement chemical phases using characterize the full 3-D aspects of shape, which must be used
Rietveld analysis of X-ray diffraction data, combined with par- to understand and predict the effect of shape on properties.
ticle size distribution information, can probably capture most To be able to build models using real-shape aggregates,
of the information needed for experimental characterization one must be able to treat a given aggregate particle as a
40 TESTS AND PROPERTIES OF CONCRETE

Fig. 2—Reconstructed images of aggregates sampled from

AASHTO Materials Reference Laboratory (AMRL) proficiency
sample #39. The left image is of a coarse limestone aggregate
with an equivalent spherical diameter of 13.80 mm, and the
image on the right is of a fine limestone aggregate with an
equivalent spherical diameter of 1.46 mm.

tained, then real-shaped particles can be handled in a model

with the same ease as spherical particles. Figure 2 shows a
VRML (virtual reality modeling language) picture of a fine and
a coarse limestone aggregate, from an American Association of
State Highway and Traffic Officials (AASHTO) Materials Ref-
erence Laboratory (AMRL) proficiency sample. Aggregate data-
bases are currently being built up for various aggregates and
incorporated into the VCCTL.
To characterize cement particle shape requires X-ray
microtomography, which can achieve the length scales of
approximately 1
m/voxel necessary to capture the shape of
particles the average size of which are usually about 10
m
to 20
m [18]. A voxel is a small cubic or rectangular paral-
lelepiped-shaped element of a 3-D digital image. After such
an image is obtained, the same methods used to character-
ize the shape of aggregates can be used for the cement
particles [19].

Virtual Testing of the Rheology of Fresh

Cement and Concrete
Rheology of concrete is the study of how concrete flows,
before the setting point is reached. This is important because
Fig. 1—(a) Backscattered electron SEM image of a cut and the concrete must be placed by some kind of pouring or pump-
polished section of cement 140 from the Cement and Concrete ing into the prepared forms. The slump test specified in the
Reference Laboratory proficiency sample program. (b) False
ASTM Standard Test Method for Slump of Hydraulic-Cement
gray scale image of the same section, based on X-ray micro-
Concrete (C 143/143M) is an empirical measurement of how
probe analysis. The meaning of each gray scale is indicated in
the accompanying legend. Images are 256
m  200
m. concrete flows in a given situation. However, work in the last
few decades has shown clearly that concrete rheology is char-
acterized by at least two parameters, yield stress and plastic vis-
cosity. In a similar way, an elastic solid is characterized by two
parameters, Young’s modulus and Poisson’s ratio. Typically,
only the Young’s modulus is considered, but really both are
mathematical object. Spheres are easy to so use, since there is needed to completely understand and predict the elastic per-
a simple equation that defines their surfaces: the distance from formance of an elastic solid. In the same way, the slump test
the center of mass to the surface is a constant. Real-shaped par- only measures one parameter, the yield stress, but the plastic
ticles can also be characterized in terms of mathematical func- viscosity is needed as well to be able to fully understand and
tions, although this is a little more complicated. A recent paper predict the rheology of concrete, which in turn determines the
[17] showed how a combination of X-ray computed tomogra- workability and flowability of the concrete into a form or
phy (CT) and spherical harmonic functional analysis could pro- through a pumping process.
duce an analytical function for the surface of an arbitrary Because concrete is a multi-scale material, investigating the
aggregate particle. Once this mathematical function is ob- rheology involves a multi-scale approach. Rheology of the
 BENTZ ET AL. ON VIRTUAL TESTING 41

some ways to the SCC tests being considered for standardiza-

tion. The coarse aggregate shapes were taken from the shapes
of real aggregates, characterized by the process described in
the previous section.

Virtual Testing of the Properties of Hardening

Cement Paste and Concrete
Hydration and Degree of Hydration
To model the development of cement and concrete properties
over time, a proper understanding (and model) of the hydra-
tion process is essential. While a complete understanding of
cement hydration is still lacking, a significant knowledge base
[23] has been accumulated in over 100 years of experimenta-
tion. The two most influential parameters for the properties of
cement and concrete are water-to-cement mass ratio (w/c) and
degree of hydration. If degree of hydration (of both cement
and pozzolans) can be accurately predicted, many properties
Fig. 3—Comparison of DPD model to experimental data can be conveniently computed, as will be demonstrated in the
on the dependence of relative viscosity of fresh concrete on subsections that follow. To model the increase in degree of
the volume fraction of coarse aggregate. The first three sets of
hydration with time, two approaches are possible, depending
data in the legend are the simulations, and the remaining
on whether or not the underlying (blended) cement paste
datasets are experimental data obtained using various
rheometer types. See Ref [21] for detailed discussions of these microstructure is explicitly considered.
concrete rheometer types. The first approach consists of utilizing some functional
form to describe the relationship between degree of hydration
and time. The dispersion models of Knudsen [24,25] or the use
cement paste greatly influences the time-dependent rheology of of nuclei-growth models [26] are examples of this approach. In
the concrete, and is itself non-Newtonian and complicated. this approach, the parameters obtained from fitting the experi-
However, having such a large volume fraction of aggregates, mental data may or may not have physical significance. In ad-
60 % or more, also has a very large effect on concrete rheology. dition, prediction of the performance of other systems (change
Simultaneous modeling of the hydration process and in cement composition, PSD, w/c, etc.) based on the fitting of
cement paste rheology is beyond current computational one (or more) set(s) of experimental data may or may not be
capabilities. Therefore, we have taken a combined theoretical- possible. Still, these equations can be useful, as they usually do
experimental approach. Cement paste and mortar rheology is provide an excellent fit to an individual experimental dataset,
measured in a custom rheometer [20]. The effect of coarse ag- and thus could be used to predict later-age degree of hydration
gregates on concrete rheology is modeled using a dissipative from early-age measurements.
particle dynamics (DPD) approach. This is similar to a molec- The second approach attempts to make greater use of the
ular dynamics approach for the movement of atoms and detailed characterization of the starting materials described
molecules, but adapted for coarse aggregate-size particles. The above by directly modeling the microstructure development of
rheological properties of the matrix of the suspension come the cement paste. In this approach, cement PSD, phase com-
from the cement paste and mortar measurements. Figure 3 position and distribution, and w/c can all be explicitly consid-
shows experimental and DPD simulated plots of the plastic ered. In the last 15 years or so, significant developments have
viscosity of a concrete plotted versus the volume fraction of ag- been made in this approach, concurrent with the vast increases
gregate. Adding more aggregate clearly increases the apparent
viscosity.
Experimentally, research is being done on measuring
concrete rheology using various concrete rheometers. Several
different designs are currently available. An intercomparison
effort is underway [21,22], and it may be possible to extract
fundamental rheological parameters (plastic viscosity and
yield stress) from these measurements. This would both enable
the rheological models to be validated and, more importantly,
begin to allow fundamental rheological parameters to be used
in analyzing and predicting concrete flow in field conditions.
The DPD simulations can be used to analyze flow in different
rheometer designs and extract fundamental parameters from
empirical test results. There has also been DPD modeling work
of concrete flow in self-compacting concrete (SCC) empirical
tests, thus offering the potential for putting these tests on a Fig. 4—Static image from simulation of vertical flow of a
more fundamental materials science basis. Some of these tests concrete with real particles through a grid of four steel rein-
are currently being considered for standardization in ASTM forcing bars which are separated by a distance of 200 mm. The
subcommittee C09.47. Figure 4 shows a DPD simulation of particles are dropping down, due to gravity, through the re-
coarse aggregate falling through four parallel rebars, similar in bars.
42 TESTS AND PROPERTIES OF CONCRETE

in computer speed and available memory. A pioneering effort

in this field was the work of Jennings and Johnson [27], who
modeled the cement powder as a collection of spherical parti-
cles of tricalcium silicate that resulted in the formation of cal-
cium hydroxide and calcium silicate hydrate reaction prod-
ucts, while explicitly considering the PSD of the powder.
Similar models continue to be developed and utilized to this
day [28,29]. In the early to mid 1990s, three other research
groups, all active at the present date, initiated research on this
topic. In Japan, Maekawa et al. developed the DuCOM (dura-
bility model of concrete) model, including modules for hydra-
tion, microstructure formation, and heat and mass transport
[30]. The underlying hydration model, while focused on heat
generation and water consumption, also considers the stereo-
logical aspects of the overlapping hydrating cement particles.
A demonstration of the model is accessible over the Internet at Fig. 5—Typical 3-D reconstructed microstructure used as
http://concrete.t. u-tokyo.ac.jp/en/demos/ducom (accessed input for the CEMHYD3D cement hydration model. The phase
March 2005). In the Netherlands, van Breugel published his designated by each gray scale is shown in the legend accom-
Ph.D. thesis [31] on the HYMOSTRUC model for the hydration panying Fig. 6.
and microstructure development of cement paste. This model
considers the cement particles to be spherical and explicitly an-
alytically accounts for the overlap of hydration products as the cent version of CEMHYD3D also includes reactions for a vari-
individual cement particles expand during the hydration ety of mineral admixtures such as silica fume, fly ash, slag,
process. The model has been applied by van Breugel and his and limestone [1–3]. Figure 6 provides an example two-
subsequent Ph.D. students to predict a variety of cement and dimensional image of a portion of a hydrated CEMHYD3D mi-
concrete properties, including heat development, strength de- crostructure.
velopment, and autogenous shrinkage [31–33]. Experimentally, no ASTM standard method exists for eval-
During the same time period, Bentz and Garboczi initiated uating the degree of hydration of either portland or blended
a cement hydration modeling effort [34] at the National cements. For portland cements, non-evaporable water content
Institute of Standards and Technology (NIST) that ultimately [42] is one generally accepted method for estimating the de-
culminated in the CEMHYD3D three-dimensional cement gree of hydration of the hardened paste. In this test, the mass
hydration model [8–10]. Unlike the other models mentioned loss of the hydrated cement paste occurring between 105°C
previously, CEMHYD3D considers the cement particles at the and 1000°C (or another similar range of temperatures de-
sub-particle level, applying a digital-image-based approach to pending on the preference of the researcher) relative to the
the modeling of cement hydration and microstructure devel- mass of the ignited paste (corrected for the loss on ignition of
opment. Each digitized spherical cement particle is composed the original cement powder) provides a measure of the non-
of one or more digital elements (voxels) that can be assigned evaporable water content. Degree of hydration (0.0–1.0) is then
to be any of the possible phases of the starting cement powder. obtained by dividing the non-evaporable water content of the
Using computational algorithms, the user is able to create an specimen by that measured for a fully hydrated cement paste
initial three-dimensional cement (in water) microstructure that specimen. The value for a fully hydrated specimen can also be
matches the following measured features of the real cement estimated based on the potential Bogue composition of the ce-
powder, obtained via the characterization methods outlined ment using coefficients provided by Molina and given in Table
earlier in this chapter: PSD, bulk phase composition, and intra- 1 [43] (the effect of the more correct composition as deter-
particle chemical phase distribution. An image of a portion of mined by Rietveld X-ray diffraction analysis is not known at
a typical starting 3-D microstructure is provided in Fig. 5.
In CEMHYD3D, a cellular automaton-like computer algo-
rithm is used to simulate the hydration process. Individual
cement phase voxels can dissolve, diffuse within the available
capillary porosity, and react to form solid hydration products.
The model proceeds as a series of dissolution/diffusion/
reaction cycles, and a single user-provided parameter is used to
approximately convert between hydration cycles and real time.
The algorithms are applied so as to maintain the correct
volume stoichiometry for the generally accepted reactions of
cement hydration [8,10,23], including the chemical shrinkage
first highlighted by Le Chatelier over 100 years ago [35].
Degree of hydration, of the cement powder or any combina-
tion of the original cement phases, can be determined by a
simple voxel counting algorithm, either on a mass or a volume Fig. 6—2-D cross section of a 3-D model cement paste with
basis. CEMHYD3D has been utilized, worldwide, by cement re- w/c  0.4, hydrated for 2000 computational cycles using the
searchers to predict performance properties and to generate CEMHYD3D hydration model. The phase designated by each
microstructures for the subsequent evaluation of degradation gray scale is indicated in the accompanying legend. The image
[36–41]. In addition to cement hydration reactions, the most re- is 100
m  100
m.
 BENTZ ET AL. ON VIRTUAL TESTING 43

Setting Time
TABLE 1—Non-evaporable Water Contents Setting time is one of the most important properties of a ce-
for Major Phases of Cement ment, as it will determine how much time is available to place
Cement Phase Coefficient (g of water per g of phase) and finish the concrete. Two ASTM standards exist for the eval-
uation of setting time: ASTM Standard Test Method for Time
C3S 0.24 of Setting of Hydraulic Cement by Vicat Needle (C 191) and
C2S 0.21 ASTM Standard Test Method for Time of Setting of Hydraulic-
C3A 0.40 Cement Paste by Gillmore Needles (C 266). The two usually
C4AF 0.37 provide slightly different values, with the Gillmore needle
Free lime (CaO) 0.33 times for initial and final set generally being slightly greater
than those obtained using the Vicat needle. Both methods ba-
sically measure when the hydrating cement paste develops
some finite value of resistance to penetration.
present). For Type I portland cements, a typical value is on the To predict setting times using a computer model, one first
order of 0.23 g of water per gram of cement. For blended ce- needs to decide what constitutes the physical process of set-
ments, non-evaporable water content is no longer an accurate ting. Fundamental research has indicated that setting is nor-
measure of degree of hydration due to the confounding influ- mally controlled by the formation of calcium silicate hydrates
ence of the pozzolanic reactions, which consume calcium hy- (bridges) that link together the original cement particles
droxide (and its accompanying non-evaporable water) in pro- [49,50]. In systems that undergo a rapid reaction of the alumi-
ducing other cement hydration products [44,45]. The use of nate phases and flash set, it is likely the aluminate hydrates that
SEM/image analysis (e.g., point counting) to estimate the form these linking bridges. Within the VCCTL software, setting
degree of hydration of both portland and blended (with fly ash is assessed using a specialized percolation algorithm [10]. The
or slag) cements appears to be a promising new option [46]. algorithm measures the fraction of total solids (mainly cement
Clearly, this is an area where further standardization efforts particles at this point) that are linked together by calcium sili-
are needed. cate hydrate gel, ettringite, and calcium aluminate hydrate hy-
Figure 7 provides a comparison of VCCTL-predicted and dration products. Thus, two touching cement particles are not
experimentally measured (via non-evaporable water content) considered to be connected, for determining setting, unless
degrees of hydration for Cement and Concrete Reference Lab- some hydration product bridges them. When such a structure
oratory (CCRL) cement 135 with a w/c  0.40, hydrated under spans the entire system, one says that the structure is perco-
both saturated and sealed curing conditions [47,48]. Under lated or connected or, in this case, set. In this way, the setting
sealed curing conditions, the paste self-desiccates due to the behavior of both well-dispersed and flocculated cement pastes
chemical shrinkage that occurs during hydration, which also can be consistently evaluated. The VCCTL program returns the
decreases the achieved degree of hydration at later ages. It can percolated (connected) fraction [0,1] for the total solids as a
be observed that the computer model provides an excellent fit function of either hydration time or degree of hydration.
to the experimental data for both curing conditions, using Figure 8 shows these percolation plots versus time for
a parameter of 0.0003 h/cycle2 to convert between cycles and CCRL cements 135 and 141 [48,51], both hydrated at a w/c
time. determined by the ASTM Test Method for Normal Consis-
Degree of hydration

Fig. 7—Degree of hydration as a function of time for CCRL cement 135

with w/c  0.4, cured isothermally at 25°C. The solid curve is the prediction by
CEMHYD3D using saturated curing conditions, and the dashed line is the pre-
diction using sealed curing conditions. The symbols are experimental meas-
urements, based on nonevaporable water content, using saturated and sealed
curing conditions, respectively. The error bars indicate  one standard devia-
tion from the mean value, and are about the same size as the data symbols.
44 TESTS AND PROPERTIES OF CONCRETE

Fig. 8—Fraction of solids connected as a function of time for (a) CCRL

cement #135, w/c  0.25, and (b) CCRL cement #141, w/c  0.267. The w/c
ratio of both cements was determined by the ASTM Normal Consistency test
(ASTM C 187). The fluctuations in the model curves is an indication of the
randomness of the model due to its relatively small size. The vertical lines in
each plot mark the measured initial and final setting times using the Vicat
needle test and the Gilmore needle test. The sharp rise in the curves gives
an indication of the setting times predicted by CEMHYD3D.

tency of Hydraulic Cement (C 187). The initial and final set- Chemical Shrinkage
ting times as determined experimentally by both the Vicat In addition to being identified by Le Chatelier over 100 years
and Gillmore needle tests are shown as vertical lines on the ago, the measurement of chemical shrinkage was also one of
graphs, as noted in the caption. For the Vicat needle, the ini- the first subjects investigated by Powers early in his career [53].
tial and final setting times are shown to approximately cor- Because the cement hydration products occupy less volume
respond to percolated fractions of 0.4 and 0.75, respectively. than the starting materials (cement and water), a hydrating
The Gillmore setting times are slightly longer, with the initial cement paste will imbibe water in direct proportion to its on-
and final setting times corresponding to percolated fractions going hydration [53,54]. This holds true except for low w/c
of 0.6 and 0.8, respectively. Setting time will be strongly in- pastes (less than w/c  0.4) after a few days of curing, in which
fluenced by the w/c of the cement paste and the PSD of the the depercolation of the capillary porosity may dramatically
cement powder. As would be expected, coarser cements re- reduce the permeability of the cement paste and limit its imbi-
quire more hydration time to achieve set, due to their slower bition rate below that required to maintain saturation during
hydration rate. But, these coarser systems actually achieve set the continuing hydration [8,54]. Chemical shrinkage has been
at a lower degree of hydration, as fewer interparticle bridges shown to be in direct proportion to other measures of degree
are needed to percolate the microstructure in a coarser par- of hydration such as non-evaporable water content and heat re-
ticle system [52]. lease [8,55]. It appears to provide a rapid method for assessing
 BENTZ ET AL. ON VIRTUAL TESTING 45

Chemical shrinkage (ml / g cement)

Fig. 9—Chemical shrinkage of CCRL cement 135 with w/c  0.3, hy-
drated isothermally at 25°C under saturated curing conditions. The solid
curve shows the prediction using CEMHYD3D, and the circles are experi-
mental measurements.

early hydration rates and may provide a means for evaluating Heat Release and Adiabatic Temperature Rise
cement cracking susceptibility [56]. While no standard method Another convenient measure of degree of hydration of port-
currently exists for the measurement of chemical shrinkage, land cements is heat of hydration. Heat of hydration is con-
ASTM C01.31 subcommittee on Volume Change is currently ventionally measured using the ASTM Standard Test Method
balloting a draft standard for this test. for Heat of Hydration of Hydraulic Cement (C 186). Within
Knowing the volume stoichiometry (e.g., molar volumes) the CEMHYD3D computer model, heat of hydration is com-
of all ongoing hydration reactions, it is straightforward to puted based on the individual heats of hydration of the vari-
compute chemical shrinkage in the VCCTL CEMHYD3D ous cement clinker (and pozzolan) phases. Knowing the de-
model. Figure 9 provides a comparison of model and experi- gree of hydration of each individual cement phase, the
mental results for CCRL cement 135 with w/c  0.3. The devi- overall heat of hydration is easily computed. For adiabatic
ation between model and experimental results at later ages boundary conditions, this heat of hydration can be readily
( 40 h) is likely due to the depercolation of the capillary converted to an adiabatic temperature rise, if the heat ca-
porosity mentioned above, as excellent agreement is ob- pacity, specific gravity, and mixture proportions of the con-
served between model and experiment up until this time. crete are known [36]. Figure 10 provides an example of the
Temperature ( oC )

Fig. 10—Comparison of predicted (solid line) and experimental measured (data

points) adiabatic heat signature curves for a w/c  0.65 OPC concrete with 50 % fly
ash replacement of cement, by mass.
46 TESTS AND PROPERTIES OF CONCRETE

E, G (GPa)

Fig. 11—Comparison of elastic moduli predictions to experimental data, versus

w/c ratio, for 28-day and 56-day specimens. At each w/c ratio, the upper point is the
56-day value and the lower point is the 28-day value.

generally observed agreement between model and experi- concrete cylinders is assessed according to ASTM Standard Test
mental data for the adiabatic temperature rise of a blended Method for Compressive Strength of Cylindrical Concrete Spec-
cement concrete. imens (C 39/C 39M). Computationally, several approaches are
possible for the prediction of compressive strength. The most
Virtual Testing of the Properties of Hardened widely used to date has been Powers’s gel-space ratio theory
Cement Paste and Concrete [61]. In the CEMHYD3D model, voxel counting is utilized to cal-
culate the volume fractions of “gel” and “space,” and compres-
Elastic Moduli sive strength is estimated as a function of the ratio of the two.
As stated above, the output of the CEMHYD3D model is a dig- Since a pre-factor is present in the strength prediction equation,
ital hydrated 3-D cement paste microstructure, in which each at least one experimental measurement of compressive
voxel is occupied by a unique cement paste phase. By treating strength is required to estimate the pre-factor. Usually, either a
each voxel as a tri-linear finite element, the overall elastic 3-day or a 7-day strength measurement is performed and the re-
moduli of the cement paste model can be computed directly sulting pre-factor is used to predict the 28-day strength. While
using finite element techniques. A typical model is 1003 voxels originally developed for portland cement systems, it has been
in size, so that a conjugate gradient relaxation solver must be recently verified experimentally that the theory can also be ap-
used [57–59]. This size system can routinely be done on a mod- plied to blended cement (specifically fly ash) systems [62]. The
ern desktop in an hour or less. theory has also been used in commercially available software
Recently, a careful test of this algorithm applied to the packages that predict compressive strength development based
microstructures resulting from operation of CEMHYD3D was on the measurement of heat signature curves (as a measure of
carried out [60]. Cement paste samples were prepared from a degree of hydration) [63]. Figure 12 provides an example of the
German cement, for w/c ranging from 0.25 to 0.6. After peri- agreement between model predictions and measured compres-
ods of 28 days and 56 days of curing, the degree of hydration sive strengths for CCRL cement 133 [10]. While it would be
was measured using loss on ignition (a method described preferable to directly predict compressive strength develop-
above), and the elastic moduli were measured using a version ment without requiring an early-age measurement, the above
of the ASTM Standard Test Method for Fundamental Trans- procedure could still result in considerable time and cost sav-
verse, Longitudinal, and Torsional Resonant Frequencies of ings to the industry, as it reduces the 28-day evaluation window
Concrete Specimens (C 215). Figure 11 shows the comparison down to either 3 days or 7 days.
between model and experimental results for 28-day- and 56- The previous subsection has illustrated that elastic moduli
day-old samples, graphed as a function of w/c. There is excel- development can be accurately predicted using the VCCTL
lent agreement between the virtual and experimental results. software. Another approach to predicting compressive
strength, currently being pursued within the VCCTL consor-
Compressive Strength tium, is to first compute elastic moduli, and then estimate
Actually, compressive strength is a more highly utilized param- compressive strength based on some functional relationship
eter in the cement and concrete industries than are elastic mod- between compressive strength and elastic modulus. A conven-
uli. Therefore, prediction of the compressive strength of mor- ient empirical equation relating Young’s modulus and com-
tar cubes and concrete cylinders is an obvious application for pressive strength, based on many experimental results, is
computer modeling. The compressive strength of mortar cubes found in the ACI standards [64]. It is possible that a multiscale
is generally assessed based on the ASTM Standard Test Method strength of materials theory can be formulated to give accu-
for Compressive Strength of Hydraulic Cement Mortars (using rate, microstructure-based predictions of compressive strength
2-in. [50-mm] cube specimens) (C 109/ C 109M), while that of via similar kinds of equations that are microstructurally based.
 BENTZ ET AL. ON VIRTUAL TESTING 47

Compressive strength (MPa)

Fig. 12—Experimentally measured (circles) and model-predicted compressive

strengths for ASTM C 109 mortar cubes prepared from CCRL cement 133 at 25°C.
The error bars indicate  one standard deviation from the mean, as determined in
the CCRL testing program.

Diffusivity [68,69]. Thus, equations have been developed to predict

As more and more concrete structures are designed for concrete diffusivity for limited ranges of these four input pa-
durability as well as strength, transport properties such as rameters. It must be recognized that these equations ignore a
diffusivity are becoming more important. In the past, ASTM number of field concrete realities such as (micro)cracking, par-
Standard Test Method for Electrical Indication of Concrete’s tial saturation (drying), differential curing, and leaching (of
Ability to Resist Chloride Ion Penetration (C 1202) has been calcium hydroxide) from the surface layer. While they may
used extensively to measure the “rapid chloride permeability” provide adequate predictions of the diffusivities of laboratory
of a concrete cylinder. Even though the results of this test are concretes produced under carefully controlled conditions,
known to be confounded by the conductivity of the pore solu- they should be used with caution in field applications.
tion and various temperature effects [65], it is still widely speci-
fied within the industry. While for a set of similar materials, the Modeling the Degradation and Service Life of
test method may produce the correct performance ranking, us- Hardened Cement and Concrete
ing the test as a performance criterion for conformance to spec-
ifications is a dangerous but ever-growing practice. The newly An important area for virtual testing is durability, as durability-
approved ASTM Test Method for Determining the Apparent type tests are usually time consuming. Degradation mecha-
Chloride Diffusion Coefficient of Cementitious Mixtures by nisms at the microstructural level have been considered, such
Bulk Diffusion (C 1556) is a welcome addition that will provide as leaching of calcium [72], how leaching affects diffusion
diffusion coefficients that can be subsequently used to obtain rates via changes in the amount and connectedness of the cap-
service life estimates for concrete structures. However, one has illary pore space [73–76], and how leaching affects elastic
to be careful in evaluating chloride diffusion through a strong moduli via the dissolution of solid phases [41,77]. Virtual dura-
ionic electrolyte such as cement paste pore solution, and so it is bility testing is still in its infancy, however, although it will
possible that more refinements will have to be made [66,67]. become a major area of focus as time goes on.
For the case of diffusivity, successful predictions at the ce-
ment paste level have been also successfully extended to mor- Prospectus and Future Directions
tar and concrete [68,69]. For this application, a hard core/soft
shell (HCSS) microstructural model of concrete has been de- The previous sections have illustrated the wide variety of
veloped [70]. The concrete is modeled as a three-dimensional cement and concrete properties that can be predicted and
continuum volume of spherical aggregates, in which each ag- studied using computer modeling. As computer modeling
gregate particle is surrounded by an interfacial transition zone slowly moves into the industrial mainstream, new and di-
(ITZ). The diffusivity of the diffusing species is different (gen- verse applications surely will be discovered. In addition, it
erally higher) in the ITZ regions than in the bulk cement paste. has been clearly shown that standards and computer model-
Random walker techniques are then employed to determine ing can and should function in a synergistic relationship. The
the diffusivity of the overall 3-D model concrete microstruc- computer modeling efforts of recent years have clearly
ture [71]. The HCSS model is currently being extended to use pointed out the need for new materials characterization stan-
real-shape aggregates instead of model spheres. dards and test methods. The models’ performance depends
Using this approach, it has been determined that the ma- critically on high quality input concerning the materials in
jor variables influencing concrete diffusivity are w/c, degree of question. Many of the needed new standards and test meth-
hydration, aggregate volume fraction, and silica fume addition ods are already being worked on by the appropriate ASTM
rate (other mineral admixtures have not been considered) subcommittees.
48 TESTS AND PROPERTIES OF CONCRETE

High quality quantitative standards are also needed for the virtual testing. We would also like to thank Mr. James Pielert,
experimental validation of the computer models. Experimental former Manager of the ASTM Cement and Concrete Reference
validation is critical both during model development and Laboratory (CCRL) and the AASHTO Materials Reference Lab-
during their extension to new systems/environments. The oratory (AMRL), Mr. Ray Kolos and Mr. Robin Haupt of CCRL,
models can only be proved, disproved, and improved based on and Mr. Ron Holsinger of AMRL, for their support of this work
comparison to carefully measured quality experimental results. via material samples, test data, and useful conversations. Fi-
As the above sections indicate, computer modeling has exhib- nally, the authors would like to thank Dr. Claus Haecker, for-
ited the potential to predict a wide variety of cement and con- merly of Dyckerhoff Zement GmbH, for collaborating on the
crete properties, several of which are lacking a standard test elastic work by supplying elastic and degree of hydration data.
method for obtaining the necessary experimental data. While
degree of hydration, chemical shrinkage, and heat of hydration References
are all fundamental measures of the reactivity of a (blended) ce- [1] Bentz, D. P., Ed., “The Virtual Cement and Concrete Testing
ment system, only heat of hydration is covered by an existing Laboratory Consortium Annual Report 2001,” NISTIR 6840, U.S.
standard (ASTM C 186). Even this standard only provides a Department of Commerce, December 2001.
measure of heat of hydration at limited fixed times such as 7 [2] Bullard, J. W., Ed., “The Virtual Cement and Concrete Testing
days and 28 days. Both chemical shrinkage and isothermal Laboratory Consortium Annual Report 2002,” NISTIR 6962, U.S.
calorimetry can provide nearly continuous monitoring of the Department of Commerce, January 2003.
early hydration rate of cement-based systems. For both of these [3] Bullard, J. W., Ed., “The Virtual Cement and Concrete Testing
measures, ASTM standard test methods would be a welcome ad- Laboratory Consortium Annual Report 2003,” NISTIR 7096, U.S.
dition to the research, production, and end-user communities. Department of Commerce, 2004.
While degree of hydration by non-evaporable water content (for [4] Haecker, C. J., Bentz, D. P., Feng, X. P., and Stutzman, P. E.,
“Prediction of Cement Physical Properties by Virtual Testing,”
portland cement systems) or SEM point counting (for portland
Cement International, Vol. 1, No. 3, 2003, pp. 86–92.
cements and those blended with slag and fly ash) typically pro-
[5] Ferraris, C. F., Hackley, V. A., Aviles, A. I., and Buchanan Jr.,
vide measures of reactivity at limited points in time, standardi-
C. E., “Analysis of the ASTM Round-Robin Test on Particle Size
zation efforts for these two measures are also critically needed. Distribution of Portland Cement: Phase II,” NISTIR 6931, U.S.
Standard test methods are lacking in a number of other ar- Department of Commerce, 2002.
eas. The results of virtual testing clearly point this out, by com- [6] Hackley, V. A., Ferraris, C. F., and Aviles, A. I., “Measurement of
puting values of important parameters for which no standard Particle Size Distribution in Portland Cement Powder: Analysis
measurement is available. Examples include measurement of of ASTM Round-Robin Studies,” ASTM Cement, Concrete, and
internal relative humidity in hardened concretes, measurement Aggregates, Vol. 26, 2004, pp. 71–81.
of elastic moduli and creep of concrete at early ages, and mech- [7] Stutzman, P. E. and Leigh, S., “Phase Composition Analysis of
anistic-based tests for common degradation problems such as the NIST Reference Clinkers by Optical Microscopy and X-ray
sulfate attack and freeze-thaw degradation. In addition to its us- Powder Diffraction,” NIST Technical Note 1441, U.S. Depart-
age to predict performance properties, computer modeling ment of Commerce, September 2002.
should also prove invaluable in providing guidance and direc- [8] Bentz, D. P., “Three-Dimensional Computer Simulation of
Cement Hydration and Microstructure Development,” Journal
tion for these new standards development efforts.
of the American Ceramic Society, Vol. 80, No. 1, 1997, pp. 3–21.
[9] Bentz, D. P., “Modeling Cement Microstructure: Pixels, Particles
Conclusion and Property Prediction,” Materials and Structures, Vol. 32,
1999, pp. 187–195.
Advances in ceramics, metals, and polymers have been driven [10] Bentz, D. P., “CEMHYD3D: A Three-Dimensional Cement
by the development of sufficient theory to help explain puz- Hydration and Microstructure Development Modeling Package
zling experimental results and point the way to new experi- Version 2.0,” NISTIR 6485, U.S. Department of Commerce, April
ments that had not been thought of previously. Theoretical (or 2000.
virtual) understanding of the material gave material designers [11] Bentz, D. P. and Stutzman, P. E., “Petrography of Cementitious
a whole new level of control over their material, leading to im- Materials,” ASTM STP 1215, Sharon M. DeHayes and David
Stark, Eds., ASTM International, West Conshohocken, PA, 1994,
provements that opened up new opportunities. Virtual testing
pp. 60–73.
for cement and concrete can do the same thing for these ma-
[12] Bentz, D. P., Stutzman, P. E., Haecker, C. J., and Remond, S.,
terials. Virtual testing will drive the development of empirical “SEM/X-ray Imaging of Cement-Based Materials,” Proceedings
physical testing towards materials science-based measure- of the 7th Euroseminar on Microscopy Applied to Building
ments that will in turn empower the material models to have Materials, H.S. Pietersen, J.A. Larbi, and H.H.A. Janssen, Eds.,
even greater predictive power and usefulness. Rightly viewed, Delft University of Technology, 1999, pp. 457–466.
physical testing and virtual testing are complementary, not [13] Blaine, R., “Determination of Calcium Sulfate Hydrates in Build-
competitive. But the development of virtual testing will drive ing Materials Using Thermal Analysis,” American Laboratory,
changes in physical testing—there is no doubt about that. Vol. 27, No. 14, September 1995, pp. 24–28.
[14] Bentz, D. P. and Remond, S., “Incorporation of Fly Ash into a 3-D
Acknowledgments Cement Hydration Microstructure Model,” NISTIR 6050, U.S.
Department of Commerce, 1997.
[15] Remond, S., Pimienta, P., and Bentz, D. P., “Effects of the
The authors would like to thank Dr. Geoffrey Frohnsdorff, for-
Incorporation of Municipal Solid Waste Incineration Fly Ash in
mer Chief of the Building Materials Division at NIST and cur- Cement Pastes and Mortars I. Experimental Study,” Cement and
rent vice-chairman of ASTM C01, for his continuing enthusias- Concrete Research, Vol. 32, No. 2, 2002, pp. 303–311.
tic support for the computer modeling efforts described in this [16] Williams, P. J., Biernacki, J. J., Rawn, C. J., Walker, L., and Bai, J.,
chapter, and the members of the VCCTL consortium for their “Microanalytical and Computational Analysis of a Class F Fly
financial, technical, and moral support for the development of Ash,” ACI Mater. J., Vol, 102, No. 5, 2005. pp. 330–337.
 BENTZ ET AL. ON VIRTUAL TESTING 49

[17] Garboczi, E. J., “Three-dimensional Mathematical Analysis of [37] van Eijk, R. J, and Brouwers, H. J. H., “Prediction of Hydroxyl
Particle Shape using X-ray Tomography and Spherical Harmon- Concentrations in Cement Pore Water Using a Numerical Ce-
ics: Application to Aggregates used in Concrete,” Cement and ment Hydration Model,” Cement and Concrete Research, Vol.
Concrete Research, Vol. 32, No. 10, 2002, pp. 1621–1638. 30, No. 11, 2000, pp. 1801–1806.
[18] Bentz, D. P., Mizell, S., Satterfield, S., Devaney, J., George, W., [38] Matte, V., Moranville, M., Adenot, F., Richet, C., and Torrenti,
Ketcham, P., Graham, J., Porterfield, J., Quenard, D., Vallee, F., J. M., “Simulated Microstructure and Transport Properties of
Sallee, H., Boller, E., and Baruchel, J., “The Visible Cement Data Ultra-High Performance Cement-Based Materials,” Cement and
Set,” Journal of Research of the National Institute of Standards Concrete Research, Vol. 30, No. 12, 2000, pp. 1947–1954.
and Technology, Vol. 107, No. 2, 2002, pp. 137–148. [39] Remond, S., Bentz, D. P., and Pimienta, P., “Effects of the Incor-
[19] Garboczi, E. J. and Bullard, J. W., “Shape Analysis of a Reference poration of Municipal Solid Waste Incineration Fly Ash in Cement
Cement,” Cem. Conc. Res., Vol. 34, 2004, pp. 1933–1937. Pastes and Mortars. II- Computer Modelling Results,” Cement
[20] Ferraris, C. F., “Measurement of the Rheological Properties of and Concrete Research, Vol. 32, No. 4, 2002, pp. 565–576.
Cement Paste: A New Approach,” RILEM International Sympo- [40] Brouwers, H. J. H. and van Eijk, R. J., “Alkali Concentrations of
sium on The Role of Admixtures in High Performance Concrete, Pore Solution in Hydrating OPC,” Cement and Concrete
Monterrey, Mexico, 1999. Research, Vol. 33, No. 2, 2003, pp. 191–196.
[21] Ferraris, C. F. and Brower, L. E., Eds., “Comparison of Concrete [41] Kamali, S., Mouranville, M., Garboczi, E. J., Prené, S., and Gérard,
Rheometers: International Tests at LCPC (Nantes, France) in B., “Hydrate Dissolution Influence on the Young’s Modulus of
October, 2000,” NISTIR 6819, U.S. Department of Commerce, Cement Pastes,” in Proceedings of the 5th International Confer-
2001. ence on Fracture Mechanics of Concrete and Concrete Structures
[22] Ferraris, C. F. and Martys, N. S., “Relating Fresh Concrete (FraMCos 5), Vail, Colorado, Victor C. Li, C. K. Y. Leung, K. J.
Willam, and S. L. Billington, Eds., 2004, pp.631–638.
Viscosity Measurements from Different Rheometers,” Journal
of Research of the National Institute of Standards and Technol- [42] Copeland, L. E. and Hayes, J. C., “The Determination of Non-
ogy, Vol. 108, 2003, pp. 229–234. evaporable Water in Hardened Portland Cement Pastes,” PCA
Bulletin 47, also ASTM Bulletin 194, Dec. 1953.
[23] Taylor, H. F. W., Cement Chemistry, 2nd ed., Thomas Telford,
London, 1997. [43] Molina, L., “On Predicting the Influence of Curing Conditions
on the Degree of Hydration,” CBI report 5:92, Swedish Cement
[24] Knudsen, T., “The Dispersion Model for Hydration of Portland
and Concrete Research Institute, Stockholm, 1992.
Cement, I. General Concepts,” Cement and Concrete Research,
[44] Yogendran, V., Langan, B. W., and Ward, M. A., “Hydration of
Vol. 14, 1984, pp. 622–630.
Cement and Silica Fume Paste,” Cement and Concrete Research,
[25] Bezjak, A., “An Extension of the Dispersion Model for the
Vol. 21, No. 5, 1991, pp. 691–708.
Hydration of Portland Cement,” Cement and Concrete
[45] Zhang, M.-H. and Gjorv, O. E., “Effect of Silica Fume on Cement
Research, Vol. 16, 1986, pp. 260–264.
Hydration in Low Porosity Cement Pastes,” Cement and
[26] Bezjak, A., “Nuclei Growth Model in Kinetic Analysis of Cement
Concrete Research, Vol. 21, No. 5, 1991, pp. 800–808.
Hydration,” Cement and Concrete Research, Vol. 16, 1986, pp.
[46] Feng, X. P., Garboczi, E. J., Bentz, D. P., Stutzman, P. E., and
605–609.
Mason, T. O., “Estimations of the Degree of Hydration of
[27] Jennings, H. M. and Johnson, S. K., “Simulation of Microstruc- Blended Cement Pastes by a Microscopical Point-Count
ture Development During the Hydration of a Cement Com- Procedure,” Cem. Conc. Res., Vol. 34, 2004, pp. 1787–1793.
pound,” Journal of the American Ceramic Society, Vol. 69,
[47] Bentz, D. P., Feng, X., Haecker, C. J., and Stutzman, P. E.,
1986, pp. 790–795.
“Analysis of CCRL Proficiency Cements 135 and 136 Using
[28] Navi, P. and Pignat, C., “Simulation of Cement Hydration and CEMHYD3D,” NISTIR 6545, U.S. Department of Commerce, May
the Connectivity of the Capillary Pore Space,” Advanced 2000.
Cement-Based Materials, Vol. 4, 1996, pp. 58–67. [48] Cement and Concrete Reference Laboratory, “Cement and
[29] Navi, P. and Pignat, C., “Three-Dimensional Characterization of Concrete Reference Laboratory Proficiency Sample Program:
the Pore Structure of a Simulated Cement Paste,” Cement and Final Report on Portland Cement Proficiency Samples Number
Concrete Research, Vol. 29, No. 4, 1999, pp. 507–514. 135 and 136,” Gaithersburg, MD, March 2000.
[30] Maekawa, K., Chaube, R., and Kishi, T., Modelling of Concrete [49] Chen, Y. and Odler, I., “On the Origin of Portland Cement
Performance: Hydration, Microstructure Formation, and Mass Setting,” Cement and Concrete Research, Vol. 22, 1992,
Transport, E&FN Spon, London, 1999. pp. 1130–1140.
[31] van Breugel, K., “Simulation of Hydration and Formation of [50] Nonat, A., “Interactions Between Chemical Evolution (Hydra-
Structure in Hardening Cement-Based Materials,” Ph.D thesis, tion) and Physical Evolution (Setting) in the Case of Tricalcium
Delft University of Technology, Delft, The Netherlands, 1991. Silicate,” Materials and Structures, Vol. 27, 1994, pp. 187–195.
[32] Koenders, E. A. B., “Simulation of Volume Changes in Harden- [51] Cement and Concrete Reference Laboratory, “Cement and
ing Cement-Based Materials,” Ph.D. thesis, Delft University of Concrete Reference Laboratory Proficiency Sample Program:
Technology, Delft, The Netherlands, 1997. Final Report on Portland Cement Proficiency Samples Number
[33] Ye, G., van Breugel, K., Fraaij, A. L. A., “Three-Dimensional Mi- 141 and 142,” Gaithersburg, MD, Sept. 2001.
crostructure Analysis of Numerically Simulated Cementitious [52] Bentz, D. P., Garboczi, E. J., Haecker, C. J., and Jensen, O. M.,
Materials,” Cement and Concrete Research, Vol. 33, No. 2, 2003, “Effects of Cement Particle Size Distribution on Performance
pp. 215–222. Properties of Cement-Based Materials,” Cement and Concrete
[34] Bentz, D. P. and Garboczi, E. J., “A Digitized Simulation Model Research, Vol. 29, 1999, pp. 1663–1671.
for Microstructural Development,” in Advances in Cementitious [53] Powers, T. C., “Adsorption of Water by Portland Cement Paste
Materials, Ceramics Transactions, Vol. 16, 1991, pp. 211–226. during the Hardening Process,” Industrial and. Engineering
[35] Le Chatelier, H., “Sur Les Changements de Volume Qui Chemistry, Vol. 27, 1935, pp. 790–794.
Accompagnent le Durcissement des Ciments,” Bulletin de la [54] Geiker, M., “Studies of Portland Cement Hydration: Measure-
Societe d’Encoragement pour l’Industrie Nationale, 1900, ment of Chemical Shrinkage and a Systematic Evaluation of
pp. 54–57. Hydration Curves by Means of the Dispersion Model,” Ph.D.
[36] Bentz, D. P., Waller, V., and De Larrard, F., “Prediction of Adia- thesis, Technical University of Denmark, Lyngby, Denmark, 1983.
batic Temperature Rise in Conventional and High-Performance [55] Parrott, L. J., Geiker, M., Gutteridge, W. A., and Killoh, D., “Mon-
Concretes Using a 3-D Microstructural Model,” Cement and itoring Portland Cement Hydration: Comparison of Methods,”
Concrete Research, Vol. 28, No. 2, 1998, pp. 285–297. Cement and Concrete Research, Vol. 20, 1990, pp. 919–926.
50 TESTS AND PROPERTIES OF CONCRETE

[56] Burrows, R. W., Kepler, W. F., Hurcomb, D., Schaffer, J., and Sell- [68] Bentz, D. P., Garboczi, E. J., and Lagergren, E. S., “Multi-Scale
ers, G., “Three Simple Tests for Selecting Low-Crack Cement,” Microstructural Modelling of Concrete Diffusivity: Identifica-
Cement and Concrete Composites, Vol. 26, 2004, pp. 509–519. tion of Significant Variables,” Cement, Concrete and Aggre-
[57] Garboczi, E. J. and Day, A. R., “An Algorithm for Computing the gates, Vol. 20, No. 1, 1998, pp. 129–139.
Effective Linear Elastic Properties of Heterogeneous Materials: [69] Bentz, D. P., “Influence of Silica Fume on Diffusivity in Cement-
3-D Results for Composites with Equal Phase Poisson Ratios,” Based Materials. II. Multi-Scale Modeling of Concrete Diffusiv-
Journal of the Physics and Mechanics of Solids, Vol. 43, 1995, ity,” Cement and Concrete Research, Vol. 30, 2000, pp.
pp. 1349–1362. 1121–1129.
[58] Garboczi, E. J., “Finite Element and Finite Difference Programs [70] Bentz, D. P., Garboczi, E. J., and Snyder, K. A., “A Hard Core/Soft
for Computing the Linear Electric and Elastic Properties of Shell Microstructural Model for Studying Percolation and
Digital Images of Random Materials,” NISTIR 6269, U.S. Depart- Transport in Three-Dimensional Composite Media,” NISTIR
ment of Commerce, 1998. 6265, U.S. Department of Commerce, 1999.
[59] Bohn, R. B. and Garboczi, E. J., “User Manual for Finite Element [71] Garboczi, E. J., Schwartz, L. M., and Bentz, D. P., “Modelling the
and Finite Difference Programs: A Parallel Version of NISTIR Influence of the Interfacial Zone on the D.C. Electrical Conduc-
6269,” NISTIR 6997, U.S. Department of Commerce, 2003. tivity of Mortar,” Advanced Cement-Based Materials, Vol. 2,
[60] Haecker, C. J., et al., “Modeling the Linear Elastic Properties of 1995, pp. 169–181.
Portland Cement Paste,” Cem. Conc. Res., 2005, in press. [72] Marchand, J., Beaudoin, J. J., and Pigeon, M., “Influence
[61] Powers, T. C., “Physical Properties of Cement Paste,” Proceed- of Ca(OH)2 Dissolution on the Properties of Cement Systems,” in
ings of the 4th International Symposium on the Chemistry of Materials Science of Concrete- Sulfate Attack Mechanisms,
Cement, Washington, D.C., National Bureau of Standards American Ceramic Society, Westerville, OH, 1999, pp. 283–293.
Monograph 43, Vol. 2, 1962, pp. 577–613. [73] Bentz, D. P. and Garboczi, E. J., “Modeling the Leaching of
[62] Lam, L., Wong, Y. L., and Poon, C. S., “Degree of Hydration and Calcium Hydroxide from Cement Paste: Effects on Pore Space
Gel/Space Ratio of High-Volume Fly Ash/Cement Systems,” Percolation and Diffusivity,” Materials and Structures, Vol. 25,
Cement and Concrete Research, Vol. 30, 2000, pp. 747–756. 1992, pp. 523–533.
[63] Radjy, F. F. and Vunic, D. W., “Heat Signature Testing of [74] Revertegat, E., Richet, C., and Gegout, P., “Effect of pH on the
Concrete,” Proceedings of Structural Materials Technology- An Durability of Cement Pastes,” Cement and Concrete Research,
NDT Conference, Techomic Publishing, Lancaster, PA, 1994, Vol. 22, 1992, pp. 259–272.
pp. 8–15. [75] van Eijk, R. J. and Brouwers, H. J. H., “Study of the Relation
[64] Neville, A. M., Properties of Concrete, 4th ed., Prentice-Hall, Between Hydrated Portland Cement Composition and Leaching
Harlow, England, 1995, p. 418. Resistance,” Cement and Concrete Research, Vol. 28, No. 6,
[65] Snyder, K. A., Ferraris, C., Martys, N. S., and Garboczi, E. J., “Us- 1998, pp. 815–828.
ing Impedance Spectroscopy to Assess the Viability of the Rapid [76] Marchand, J., Bentz, D., Samson, E., and Maltais, Y., “Influence
Chloride Test for Determining Concrete Conductivity,” Journal of Calcium Hydroxide Dissolution on the Transport Properties
of Research of the National Institute of Standards and Technol- of Hydrated Cement Systems,” in Materials Science of Concrete:
ogy, Vol. 105, 2000, pp. 497–509. Calcium Hydroxide in Concrete, J.P. Skalny, J. Gebauer, and I.
[66] Snyder, K. A., “The Relationship between the Formation Factor Odler, Eds., American Ceramic Society, Westerville, OH, 2001,
and the Diffusion Coefficient of Porous Materials Saturated pp. 113–129.
with Concentrated Electrolytes: Theoretical and Experimental [77] Carde, C. and Francois, R., “Effect of the Leaching of Calcium
Considerations,” Concrete Science and Engineering, Vol. 3, Hydroxide from Cement Paste on Mechanical and Physical Prop-
2001, pp. 216–224. erties,” Cement and Concrete Research, Vol. 27, No. 4, 1997, pp.
[67] Snyder, K. A. and Marchand, J., “Effect of Speciation on the 539–550.
Apparent Diffusion Coefficient in Nonreactive Porous Systems,”
Cement and Concrete Research, Vol. 31, No. 12, 2001,
pp. 1837–1845.
7
Quality Cement, Concrete, and Aggregates—
The Role of Testing Laboratories

James H. Pielert1

Preface process—poor specimens and faulty sampling techniques will

defeat the purposes for which the tests are made. A laboratory
cannot produce satisfactory information if the samples it
THE CONTRIBUTIONS OF J. J. WADDELL AND J. R. DISE, receives do not represent the material under consideration
who prepared the chapters on Laboratory testing in ASTM STP either because they have not been taken with appropriate care,
169A and ASTM STP 169B, respectively, are acknowledged. or have been altered by mistreatment in initial storage, curing,
or shipment. Suitable detailed sampling procedures are
Introduction described in standards for concrete and concrete-making
materials prepared by the American Society for Testing and
Concrete construction, which is vital to the nation’s economic Materials (ASTM). Comparable national standards are also
health, makes up a significant portion of the more than $900B prepared by the American Association of State Highway and
spent annually on construction. Field testing and inspection Transportation Officials (AASHTO) for highways and bridge
(including sampling) and laboratory testing of concrete and construction. Among other requirements, adequate training
concrete-making materials are key activities in the construction and instruction of sampling personnel is essential. Supervisors
process. The results of these activities are important to build- at all levels must also be fully acquainted with sampling plans
ing officials, contractors, concrete producers, owners, and and procedures and recognize their responsibility for ensuring
architectural and engineering firms in determining whether that the plans and procedures are rigorously followed.
the qualities of the materials in the construction comply with Concrete specifications enumerated in construction con-
contract documents. The role of field and laboratory testing in tracts are often based on the results of test methods devel-
promoting the quality of construction and protecting the pub- oped by ASTM. For example, the Building Code Require-
lic safety is receiving added consideration, both nationally and ments for Structural Concrete of the American Concrete
internationally. The evaluation of testing laboratories by Institute (ACI 318), which is frequently referenced, specifies
outside organizations plays a vital role in providing the con- that tests of concrete materials and of concrete must be made
struction community with confidence in the quality of testing. in accordance with ASTM standards [1]. No matter who the
This chapter discusses the role of laboratory testing in originator of the chosen standards may be, it is imperative
assuring the quality of in-place construction through testing and that every effort be made to avoid the use of any but the lat-
inspection activities. Specific areas discussed are the role of stan- est testing techniques. Employment of an unsatisfactory pro-
dards-developing organizations in the United States in preparing cedure is potentially dangerous because it can lead to erro-
standards which promote quality testing; the implementation of neous conclusions about the characteristics of the concrete
quality system concepts in testing laboratories; the role of evalu- and the safety of the structure.
ating authorities, including accreditation bodies, in recognizing In studying quality assurance for highway construction
the competency of testing laboratories; and the existing programs materials, variance in quality of the product can be divided
used to recognize the competency of laboratory technicians. into material or process variance, sampling variance, and test-
ing variance. In a study conducted by the Bureau of Public
Concerns in Testing Roads in the late 1960s, it was found that 50 % or more of the
overall variance could be attributed to two of these factors:
The testing of concrete and its component materials consists of sampling and testing [2]. These two processes must be con-
sampling, performing specified tests, and reporting test results. stantly monitored if reliable results are to be obtained.
The procurement of representative specimens through stan- On completion of testing, a laboratory is usually required
dard sampling procedures is a critical step in the testing to submit a written report to its client. The report should be

1
Manager, Cement and Concrete Reference Laboratory, National Institute of Standards and Technology, Gaithersburg, MD 20899.

51
52 TESTS AND PROPERTIES OF CONCRETE

complete and factual. According to ASTM Standard Practice • ASTM Standard Guide for Development of a Directory of
for Laboratories Testing Concrete and Concrete Aggregate for Accredited Laboratories by an Accrediting Body (E 1738)—
Use in Construction and Criteria for Laboratory Evaluation Provides guidance on criteria to be used by laboratory ac-
(ASTM C 1077), the report should cite the name and address of crediting bodies in the development of directories of ac-
the laboratory, report identification and date issued, name of credited laboratories.
the client and project identification, sample identification, • ASTM Standard Guide for Selection, Assignment, and
identification of the standard test method(s) and known devia- the Monitoring of Persons to be Utilized as Assessors/
tions from the standard, test results, and other information Auditing Technical Experts (E 2159)—Provides guidance
required by the appropriate standard [3]. This may include to organizations that use independent auditors or asses-
comments about any unusual aspects of the appearance or sors to assure the soundness of their quality systems
behavior of specimens that might in any way be relevant to the and practices.
interpretation of the results obtained. The key to competent testing is the laboratory’s use of an
When a final report is available, it can be determined by effective and comprehensive quality system involving both
the laboratory or another representative of the client whether quality assurance and quality control activities. ASTM E 1187
a material complies, or fails to comply, with specification re- defines quality assurance as “all the planned and systematic
quirements [4]. If there is a question about the sampling or activities implemented with the quality system, and demon-
testing methods used to generate results recorded in the re- strated as needed, to provide adequate confidence that an
port, then the reliability of the entire report should be ques- entity will fulfill requirements for quality”; and quality con-
tioned. trol as “operational techniques and activities that are used to
fulfill requirements for quality.” In recent years, there has
Trends in Improving and Promoting Quality been increasing emphasis, both nationally and internation-
of Testing ally, on the need for laboratories to establish and maintain
quality systems [6]. ASTM E 1187 defines a quality system as
ASTM Committee E36 on Laboratory and Inspection Agency “the organizational structure, procedures, processes and re-
Evaluation and Accreditation has responsibility for preparing sources needed to implement quality management.” A quality
standards related to testing laboratory quality. In some techni- system must be tailored to the unique characteristics and ca-
cal areas, the work of the committee stimulated the drafting of pabilities of each laboratory. The International Organization
similar international standards, while in other areas, cog- for Standardization/International Electrotechnical Commis-
nizance of international work has had a beneficial influence on sion (ISO/IEC) 17025 General Requirements for the Compe-
ASTM standards. Relevant standards prepared by Committee tence of Testing and Calibration Laboratories [7], and the
E36 have done much to raise all aspects of the quality of AASHTO R18 Standard Recommended Practice for Establish-
testing. They are as follows [5]: ing a Quality System for Construction Materials Testing Lab-
• ASTM Specification for Agencies Engaged in the Testing oratories [8] provide information similar to that contained in
and/or Inspection of Materials Used in Construction (E ASTM E 1187.
329)—Provides requirements for laboratories testing and Each of these standards concerning quality systems require
inspecting construction materials; it references other that the policies, procedures, and practices comprising a labo-
ASTM standards for specific materials. ratory’s quality system be documented in a quality manual.
• ASTM Guide for Calibration and Testing Accreditation ASTM E 1187 defines a quality manual as “a document stating
Systems—General Requirements for Operation and Recog- the quality policy, quality system, and quality practices of an or-
nition (E 994)—Provides requirements for systems that ganization.” Such a manual provides the staff with an under-
accredit organizations involved in testing, measuring, standing of the laboratory’s quality policies and operating pro-
inspecting, and calibrating activities. cedures and the extent of their duties and responsibilities.
• ASTM Standard Terminology Relating to Conformity Assessment of a laboratory by an evaluating authority is based
Assessment (E 1187)—Provides definitions of terms under on the existence of a comprehensive quality manual and docu-
the jurisdiction of ASTM Committee E36. mentation confirming that the laboratory operates as stated in
• ASTM Standard Guide for Proficiency Testing by Use of the manual.
Interlaboratory Comparisons (E 1301)—Provides guidance A quality manual is essential since it provides the basic ref-
for the development and operation of proficiency sample erence to a laboratory’s quality system. Typical topics covered
programs including management structure, specimen in the quality manual include: quality policy; description of the
preparation and distribution, and the analysis and report- laboratory structure; relations between management, technical
ing of results. operations, support services, and the quality system; job
• ASTM Standard Guide for Evaluating Laboratory Mea- descriptions of staff; identification of the laboratory’s ap-
surement Practices and the Statistical Analysis of the proved signatories; procedures for achieving traceability of
Resulting Data (E 1323)—Provides guidance for assessors measurements; scope of testing and references to test proce-
to evaluate measurement practices of laboratories and dures; arrangements for ensuring that the laboratory reviews
ensures that appropriate protocols are provided for statis- new work to ensure that it has appropriate facilities and re-
tically analyzing data. sources; procedures for handling test items; reference to major
• ASTM Standard Guide for Surveillance of Accredited equipment and reference standards used; reference to proce-
Laboratories (E 1580)—Presents procedures which a labo- dures for calibration, verification, and maintenance of equip-
ratory accreditation body may use to provide assurance ment; reference to interlaboratory comparisons; procedure for
that accredited laboratories continue to satisfy criteria and feedback and corrective action; procedures for dealing with
conditions under which they were accredited. complaints; procedures for protecting confidentiality and
 PIELERT ON THE ROLE OF TESTING LABORATORIES 53

proprietary rights; and procedures for audit and review. The Evaluation Authorities
actual quality system requirements depend on the standard
that a laboratory is trying to meet. ASTM C 1077 defines an evaluation authority as “an independ-
ent entity, apart from the organization being evaluated, that
Continuing Improvements in the Quality of can provide an unbiased evaluation of that organization.” The
Concrete Testing standard lists the Cement and Concrete Reference Laboratory
(CCRL) sponsored by ASTM Committees C01 and C09 as an
Among other things, ASTM Committee C09 on Concrete and evaluation authority that provides laboratory inspection and
Concrete Aggregates, and ASTM Committee C01 on Cement proficiency sample services. Accrediting bodies listed include
prepare documents for use by inspection/accreditation agen- the “National Voluntary Laboratory Accreditation Program
cies or other parties on evaluating concrete and cement testing (NVLAP), the American Association for Laboratory Accredita-
laboratories. ASTM C 1077 prepared by Committee C09 pro- tion (A2LA), Construction Materials Engineering Council
vides criteria for the evaluation of the capability of testing lab- (CMEC), AASHTO Accreditation Program (AAP), and other
oratories to perform designated ASTM tests on concrete and recognized agencies as may be established.”
concrete aggregates [3]. The standard establishes minimum CCRL and the AASHTO Materials Reference Laboratory
requirements for the testing laboratory’s personnel, equip- (AMRL) comprise the Construction Materials Reference Labo-
ment, and quality system. The 2003 version of C 1077 lists ratories (CMRL) [10,11] which are located at the National In-
seven mandatory ASTM test methods in which competence is stitute of Standards and Technology (NIST). CCRL and AMRL
required for laboratories testing concrete (C 172, C 143/ are NIST Research Associate Programs that operate at NIST un-
C 143M, C 138/C 138M, C 173/C 173M, C 1064/C 1064M, der Memoranda of Agreement between the sponsoring organi-
C 31/C 31M, and C 39/C 39M), and five mandatory ASTM stan- zations and NIST. ASTM is the sponsor of CCRL and AASHTO
dards for laboratories testing concrete aggregates (C 136, is the sponsor of AMRL. ASTM provides programmatic and
C 117, C 127, C 128, and C 40). Further, it lists optional ASTM technical oversight of CCRL through a Joint C01/C09 Subcom-
concrete and aggregate test methods for which a laboratory mittee on the CCRL, while AASHTO provides oversight to
may request evaluation. A laboratory complying with ASTM C AMRL through the AASHTO Subcommittee on Materials.
1077 must establish and maintain a quality system that in- CMRL promotes the quality of testing by assessing the
cludes procedures for personnel evaluation and training, par- performance of construction materials testing laboratories;
ticipation in a proficiency sample program, procedures for providing support to the sponsoring standards committees in
record keeping, procedures for equipment calibration and the preparation of test methods; operating a research program
maintenance, an inventory of test equipment, procedures for which complements and benefits from interaction with the NIST
handling technical complaints, and procedures for assuring Building and Fire Research Laboratory (BFRL) program; and
the quality of external technical services. ASTM C 1077 also re- through use of its programs by accrediting bodies, governmen-
quires that testing services offered by the laboratory be under tal agencies, and other organizations involved in quality assess-
the full-time technical direction of a professional engineer with ment. The primary functions of the CMRL are the assessment of
at least five years experience in construction materials testing, testing laboratories and the distribution of proficiency test sam-
and that the laboratory be periodically assessed by an ples. Laboratory Assessors from the CMRL visit laboratories to
independent evaluating authority. evaluate equipment, procedures, and quality systems according
Turning to cement, ASTM Standard Practice for Evalua- to the requirements of the test methods, and provide a report of
tion of Laboratories Testing Hydraulic Cement (C 1222), which findings to the laboratory. Concrete materials included in the
was first published by Committee C01 in 1993, identifies mini- CCRL Laboratory Inspection Program are hydraulic cements,
mum training and experience requirements for personnel, and concrete, reinforcing steel, aggregates, masonry materials, and
equipment requirements for cement testing laboratories [9]. pozzolans. Concrete materials routinely distributed to laborato-
ASTM C 1222 does not identify mandatory test methods that a ries by the CCRL Proficiency Sample Program include portland
laboratory must be able to perform, but does require that it cement, blended cement, masonry cement, portland-cement
have the capability of performing all laboratory testing associ- concrete, and pozzolan materials; while the AMRL Proficiency
ated with its intended functions. Standard chemical and physi- Sample Program distributes fine and coarse aggregate profi-
cal requirements for various types of cements are listed in the ciency samples as well as samples of other highway materials.
ASTM Specification for Masonry Cement (C 91), ASTM Stan- Data from these programs are provided to standards commit-
dard Specification for Portland Cement (C 150), ASTM Stan- tees of ASTM and AASHTO for assessing the adequacy of test
dard Specification for Blended Hydraulic Cements (C 595), methods, determining the impact of revisions to standards, and
ASTM Standard Specification for Expansive Hydraulic Cement for use in preparing precision statements. Over 1500 different
(C 845), and ASTM Standard Performance Specification for laboratories in the United States and 25 other countries
Hydraulic Cement (C 1157). The scope of a testing laboratory currently participate in the CMRL Laboratory Assessment
may be chemical testing, physical testing, or both. A laboratory and Proficiency Sample Programs. Utilization of the CMRL
complying with ASTM C 1222 is required to establish and programs is voluntary and laboratories are not rated, certified,
maintain a quality system for cement analogous to that for con- or accredited by the CMRL itself, though because of its reputa-
crete, as described in ASTM C 1077. The manager of the labo- tion for integrity, results from its programs are used by three of
ratory should be a chemist, materials analyst, or an engineer the four accrediting authorities referenced in ASTM C 1077.
with at least three years of supervisory experience in the test- The American Association for Laboratory Accreditation
ing of hydraulic cement; or a person with equivalent science- (A2LA) was formed in 1978 as a nonprofit scientific member-
oriented education or experience. A periodic assessment by an ship organization dedicated to the formal recognition of test-
evaluation authority is also required. ing organizations that have been shown to be competent [12].
54 TESTS AND PROPERTIES OF CONCRETE

A2LA grants accreditation in the following fields of testing: Technician Competency

acoustics and vibration, biology, calibration, chemistry, con-
struction materials, environmental, geotechnical, electrical, The competency of laboratory technicians conducting the tests
mechanical, nondestructive testing, and thermal. The con- plays a vital role in assuring the quality of testing. Laboratories
struction materials field of testing which includes cement, con- have a responsibility to have a means for training their
crete, soils, asphalt, and aggregates currently has about 100 personnel, assessing their competency, and documenting their
laboratories accredited. A2LA requires laboratories to partici- qualifications to perform assigned tasks. As a means of ensur-
pate in the applicable proficiency sample programs of CCRL ing this competency, technician training and certification pro-
and AMRL, and generally uses contract assessors who are grams are offered by independent organizations including the
experts in their field for on-site assessment of laboratories. American Concrete Institute (ACI) and the National Institute
The AASHTO Accreditation Program (AAP) was started by for Certification in Engineering Technologies (NICET). ASTM
AASHTO in 1988 for construction materials testing laborato- C 1077 references compliance with programs of these organi-
ries [11,13]. AAP certifies the competency of testing laborato- zations, or of an equivalent certification program, as a means
ries in carrying out specific tests on soils, asphalt cements, cut- of demonstrating the competency of a technician in perform-
back asphalts, emulsified asphalts, hot-mixed asphalt, ing tests on concrete materials.
aggregates, hydraulic cement, and portland cement concrete. ACI currently operates eight certification programs for
The Laboratory Inspection and Proficiency Sample Programs portland cement concrete and aggregates: Field Testing
of CCRL and AMRL are used by AAP to evaluate the perform- Certification Program, Craftsman Certification Program,
ance of laboratories that test these materials. As of June 2004, Strength Technician Certification Program, Laboratory
AASHTO had accredited 939 laboratories in the United States Technician Certification Program, Aggregate Technician
and Canada of which 599 were portland cement concrete Certification Program, Inspector Certification Program, Tilt-
laboratories and 68 were hydraulic cement laboratories. AAP Up Certification Program, and Shotcrete Nozzleman Program
accreditation may also be obtained for compliance with ASTM [19]. NICET offers four levels of certification for concrete
C 1077, ASTM C 1222, and ISO/IEC 17025. technicians [20].
The Construction Materials Engineering Council (CMEC)
was founded in 1983 for the purpose of improving the quality
of production, inspection, and testing of construction materials Conclusions
through accreditation, education, and certification programs
[14]. CMEC operates laboratory accreditation and proficiency High-quality field inspection and laboratory testing services are
testing programs for cement, concrete, aggregates, soils, as- important to achieving safe, efficient, and cost-effective
phalt, masonry, and chemical testing, and is a sponsoring concrete structures. The mechanisms discussed in this chapter
group for the American Concrete Institute (ACI) Technician which are being developed, standardized, and implemented
Certification Program. CMEC provides inspection and profi- will continue to promote the quality of these services. The
ciency sample programs to over 400 different laboratories an- importance of concrete sampling and accurate testing will
nually, throughout 15 states (from coast to coast) and three increase as structural design and modeling by computers
foreign countries. Because CMEC provides its own proficiency require that material properties be determined more precisely.
samples, CMEC does not use CMRL programs in accrediting Additionally, the development of high-performance concretes
laboratories. with enhanced mechanical, volume stability, durability, and
The National Voluntary Laboratory Accreditation placement properties require more accurate test results.
Program (NVLAP), which is administered by NIST, accredits
public and private testing laboratories based on the evalua- References
tion of their quality systems, equipment, test procedures, and
[1] Building Code Requirements for Structural Concrete and Com-
the technical qualifications and competence of their staffs for mentary, ACI 318, American Concrete Institute, Farmington
conducting specific tests [15]. NVLAP accreditation is based Hills, MI, 2002.
on the requirements of ISO/IEC 17025 and the relevant [2] McMahon, T. F. and Halstead, W. J. “Quality Assurance in High-
requirements of ISO 9002. NVLAP accreditation in the way Construction Part 1—Introduction and Concepts,” Public
construction materials testing field is available for selected Roads, Vol. 35, No. 6, Feb. 1969, p. 1929.
methods of testing for concrete, aggregates, cement, admix- [3] Annual Book of ASTM Standards, Volume 04.02 Concrete and
tures, geotextiles, soil and rock, bituminous materials, and Aggregates, ASTM International, West Conshohocken, PA,
steel materials. Participation in the CCRL and AMRL profi- 2003.
ciency sample programs may be used by laboratories to meet [4] Waddell, J. J., “Quality of Testing,” Significance of Tests and
the NVLAP proficiency testing requirements for aggregates, Properties of Concrete and Concrete-Making Materials, ASTM
cement, concrete, soil, and bituminous material. NVLAP STP 169A, ASTM International, West Conshohocken, PA, 1975,
generally uses contract assessors who are experts in their p. 32.
field for on-site assessments of the laboratories. There are [5] Annual Book of ASTM Standards, Volume 14.02 General Test
Methods, Forensic Sciences, Terminology, Conformity Assess-
currently around 20 laboratories accredited by NVLAP in the
ment, Statistical Techniques; ASTM International, West Con-
construction materials testing field. shohocken, PA, 2003.
Other organizations which accredit construction materials [6] Locke, J. W., “Quality Assurance in the Construction Materials
testing laboratories include the International Accreditation Laboratory,” Proceedings, Workshop on Evaluation of Cement
Service [16], the Concrete Advisory Board of Georgia [17], and and Concrete Laboratory Performance, NIST Special Publication
the Washington Area Council of Engineering Laboratories 788, National Institute of Standards and Technology, Gaithers-
[18]. burg, MD, July 1991, p. 105.
 PIELERT ON THE ROLE OF TESTING LABORATORIES 55

[7] General Requirements for the Competence of Testing and Cali- [13] Information on the AASHTO Accreditation Program is available
bration Laboratories, ISO/IEC 17025, International Organization at http://www.amrl.net
for Standardization, Geneva, Switzerland, 1999. [14] Information on the Construction Materials Engineering Council
[8] Standard Specifications for Transportation Materials and Meth- is available at http://www.cmec.org
ods of Sampling and Testing, Part 1B: Specifications, AASHTO, [15] Information on the National Voluntary Laboratory Accredita-
Washington, DC, 2003. tion Program is available at http://ts.nist.gov/ts/htdocs/210/214/
[9] Annual Book of ASTM Standards, Volume 04.01 Cement, Lime 214.htm
and Gypsum, ASTM International, West Conshohocken, PA, 2003.
[16] Information on the International Accreditation Service is
[10] Pielert, J. H., “The Cement and Concrete Reference Labora- available at http://www.iasonline.org
tory—Promoting Quality in Laboratory Testing,” ASTM Stan-
[17] Information on Concrete Advisory Board of Georgia is available
dardization News, ASTM International, West Conshohocken,
at http://www.cabofgeorgia.org
PA, November 2002, pp. 34–37.
[11] Pielert, J. H. and Spellerberg, P. A., “AASHTO Materials Refer- [18] Information on the Washington Area Council of Engineering
ence Laboratory—Thirty Years of Service to the Transportation Laboratories is available at http://www.wacel.org
Community,” TR News, Number 183, Transportation Research [19] ACI Certification Guide, American Concrete Institute, Farming-
Board, Washington, DC, March-April 1996, pp. 22–28. ton Hills, MI, 2002.
[12] Information on the American Association for Laboratory [20] Information on the National Institute for Certification in
Accreditation is available at http://www.a2la.org Engineering Technologies is available at http://nicet.org
 PART II
Freshly Mixed Concrete
8
Factors Influencing Concrete Workability

D. Gene Daniel 1

Preface quantity of concrete with minimum loss of homogeneity.” The

American Concrete Institute (ACI) defines workability in ACI
I. L. TYLER AND FRED HUBBARD PREPARED CHAP- 116R [6] as “that property of freshly mixed concrete or mortar
ters on “Uniformity, Segregation and Bleeding” and “Workabil- that determines the ease with which it can be mixed, placed,
ity and Plasticity,” in ASTM STP 169 [1,2]. C. A. Vollick com- consolidated, and finished to a homogeneous condition.” Work-
bined, revised, and updated these two chapters into “Uniformity ability includes such items as compactibility, consistency, fin-
and Workability” for ASTM STP 169A [3]. D. T. Smith revised ishability, harshness, and pumpability. Project specifications
and updated this chapter in ASTM STP 169B [4]. John M. Scan- will normally provide guidance on maximum workability as as-
lon revised, updated, and changed the title to “Factors Influenc- sociated with water content of the concrete via maximum wa-
ing Concrete Workability” in ASTM STP 169C [5]. The current ter-cement ratio, range of slump as measured in accordance
edition reviews, revises, and updates the topics as addressed by with ASTM Test Method for Slump of Hydraulic-Cement Con-
the previous authors. Mr. Tyler was with the Portland Cement crete (C 143/C 143M), and at times a minimum cement content.
Association, Mr. Hubbard with the National Slag Association, These specified properties of the fresh concrete may not fully
Mr. Vollick with Sika Chemical Corporation, Mr. Smith with describe the degree of workability desired by the contractor in
Marquette Cement Manufacturing, and Mr. Scanlon was a the fulfillment of the contractor’s responsibilities. Concrete
senior consultant with Wiss, Janney, Elstner Associates, Inc. mixture proportioning to provide greater workability then be-
comes an issue between the contractor and the concrete pro-
Introduction ducer to determine a method to provide a greater workability
and maintain concrete that in the eyes of the design engineer
The workability of concrete is partially determined in the eyes meets the specified criteria. Four primary terms have now be-
of the beholder, who is influenced by the space to be filled with come important in describing some basic concrete properties.
the concrete and the equipment available to assist in the place- They are workability, which was previously defined, rheology,
ment and finishing process. Each of the parties having a consistency, and self-consolidating concrete (SCC). Rheology is
responsibility for the completed concrete structure views work- defined as “the science dealing with the deformation and flow
ability somewhat differently depending upon his or her of matter”and “the ability to flow or be deformed.” As related to
particular responsibilities. The involved parties may include concrete there is a direct relationship between shear stress and
the owner, engineer, contractor, concrete materials suppliers, the rate of shear plus a shear stress when the rate of shear stress
concrete producer, testing agency, inspectors, and others. The is zero. Another approach is to characterize the rheological
concrete construction portion of a project cannot be success- properties of a concrete mixture by the yield stress and plastic
ful without reasonable guidelines pertaining to the workability viscosity. These properties are currently measured by rather
of the concrete for each portion of the work. There are a num- large laboratory rheometers of several designs. The rheology of
ber of factors that influence the workability of concrete. How- concrete, like slump, is time dependent as measured from the
ever, there is no widely accepted test method to measure the time of initial mixing to the time of testing. ACI 116R [6] defines
workability of concrete. rheology as “the science dealing with flow of materials, includ-
It is important to note that by using today’s technology, con- ing studies of deformation of hardened concrete, the handling
crete mixtures can be proportioned with practically any worka- and placing of freshly mixed concrete, and the behavior of slur-
bility and retain the capabilities to develop a wide range of hard- ries, pastes, and the like.”
ened properties needed by the structure. Such technology may The importance of rheology technology and measure-
require the use of specialty products such as nonlocal aggre- ments is that it advances the industry past a single descriptive
gates, supplementary cementitious materials, and admixtures. measurement for concrete workability. Information on behav-
ior of fresh concrete during vibration and rheology on consoli-
Terminology dation can be found in ACI 309.1R [7]. Unfortunately a reliable
field test has not been developed to measure the rheological
ASTM C 125 Terminology Relating to Concrete and Concrete properties. What has been shown is a reasonably definite rela-
Aggregates defines workability of concrete as “that property de- tionship between slump and a rheometer-determined yield
termining the effort required to manipulate a freshly mixed stress [8,9]. Both a field test and a greater understanding of the

1
Consulting Engineer, D. Gene Daniel, Inc., Rogers, AZ 72758.
59
60 TESTS AND PROPERTIES OF CONCRETE

information provided by rheology test results and the relation- with an appropriate ASTM specification. To cover the wide
ship to the concrete composition [8] are current research range of raw materials being dealt with these specifications are
goals. predominately composed of minimum or maximum require-
Consistency is a term related to the freshly mixed con- ments and rarely have an allowable tolerance range for a pre-
crete. ASTM and ACI definitions of consistency are virtually scribed property. The uniformity of product achieved is actu-
identical. ACI 116R [6] defines it as “the relative mobility or ally the result of the desire for customer satisfaction rather
ability of freshly mixed concrete or mortar to flow; the usual than any specified tolerances. Chemical admixture specifica-
measurements are slump for concrete, flow for mortar or tions contain minimum and maximum limits for properties
grout, and penetration for neat cement paste.” The word con- compared to control mixtures that do not contain the particu-
sistency has multiple usages and two of these are pertinent to lar admixture being tested. Chemical admixture specifications
concrete. The flow of the mixture, as measured by the slump also contain maximum variances between lots and thus are de-
test, and the variability from batch to batch are each important signed to maintain consistency of the product. The tests in-
properties for successful projects. Once an acceptable level of volved typically involve residue content, infrared analysis, and
workability has been established, the concrete quality relies on specific gravity. Control must be maintained at the raw mate-
control of variability by maintaining uniform consistency. rial sources and at the batching and mixing plants and related
components.
Uniformity of Concrete Aggregates should be tested for grading, density, absorp-
tion, and moisture content, and mixtures should be adjusted to
Concrete mixture proportions should always be developed in correct for changes in these properties. ASTM Specification for
such a way that the finished hardened concrete will attain the Concrete Aggregates (C 33) covers the requirements for both
required physical properties and be able to withstand exposure coarse and fine normal weight aggregates. Tests for aggregate
to the anticipated environmental conditions. Equally impor- properties are all defined as maximum allowables except for
tant is that the freshly mixed concrete should possess the work- prescribed grading ranges for each size aggregate. Only the fine
ability and other characteristics that permit it to be mixed, aggregate with a specified maximum variation of the fineness
transported, placed, consolidated, and finished in a timely fash- modulus of 0.20 from the base fineness modulus addresses con-
ion without hardship under prevailing conditions. After mix- sistency of the product. Consistency of aggregate products is
tures have been developed and the necessary characteristics driven by the need for customer satisfaction balanced against
that affect workability are determined, it is essential that the economic realities and not by specification requirements. As the
quantities of ingredients be kept relatively uniform. demands for more consistent concrete increase the need for
Nonuniformity may be evident in freshly mixed and hard- consistency requirements in aggregate properties will also in-
ened concrete. There are two distinct measurements of nonuni- crease. Items that may be subject to future variation restrictions
formity: within-batch variations and batch-to-batch variations. include coarse aggregate gradings, moisture contents, absorp-
Within-batch variations refer to variations relating to concrete tion, relative density, and bulk density.
sampled from the front, middle, and end of the batch as dis- Control of aggregate moisture has steadily progressed
charged from the mixer. These variations may be attributed to with computerized batching and the development of moisture
a buildup of hardened concrete in the drum, excessive wear of probes for aggregate feed bins, particularly for fine aggregates.
the mixer blades, inadequate or excessive mixing time and Van Alstine [10] attributed the uniformity obtained at Denver
speed, improper loading sequence, or possibly overloading of Reservoir No. 22 Dam, where sand bins were filled 17–20 times
the mixer. each day with sand of widely varying moisture content, to the
Batch-to-batch variations in concrete may be attributed to use of an electrical resistance meter. Even batch plants that do
variations in aggregate moisture conditions, aggregate grad- not use moisture determining probes have increased the moni-
ings, inaccurate weighing or volumetric dispensing equipment, toring of fine aggregate moisture by the use of rapid moisture
and, of course, all of the variations related to within-batch vari- content determinations. The available methods include using
ations. The concrete mixture proportions may result in segre- hot plates, ovens, or microwave ovens and ASTM Test Method
gation of the mixture due to a relatively high water-cement for Total Evaporable Moisture Content of Aggregate by Drying
ratio or poor mixture proportions. The use of improper mate- (C 566), a Chapman flask with ASTM Test Method for Surface
rial-handling equipment may also cause a marginal mixture to Moisture in Fine Aggregate (C 70), or a Speedy Moisture Tester
separate. The uniformity of concrete production and delivery as described in ASTM Test Method for Field Determination of
should always be evaluated, because the better the uniformity Water (Moisture) Content of Soil by the Calcium Carbide Pres-
during production, transportation, and placing, the greater the sure Tester (D 4944). The chapter in this volume by Yzenas con-
opportunity to obtain desired hardened concrete properties. tains information on absorption and surface moisture.
Mixed concrete should be tested for consistency, air con-
Control of Concrete Production tent, temperature, and density. Product tolerances are pro-
vided by ASTM Specification for Ready-Mixed Concrete (C
Uniform concrete can be obtained only through proper quality 94/C 94M) for slump and air content. Slump tolerances are in
control of all operations from selection and production of ma- two different formats: a “maximum” or “not to exceed” value
terials through batching, mixing, transporting, conveying, plac- or a specified target value. The tolerance varies by level of
ing, consolidation, finishing, and curing. All materials must be slump. The stated tolerance for air content is
1.5 % of the
kept relatively uniform. There is great reliance on the manu- specified value. Temperature and concrete density tests are
facturers or producers of the raw materials for relatively useful production control tests, but do not have specific toler-
uniform products. Hydraulic cements and supplementary ce- ances. Concrete specimens must be properly fabricated, cured,
mentitious materials, such as fly ashes, GGBF slag, silica fume, and tested for evaluating hardened concrete properties. The
and natural pozzolans are each specified to be in compliance criteria for these specimens is provided in ASTM Practice for
 DANIEL ON CONCRETE WORKABILITY 61

Making and Curing Concrete Test Specimens in the Field (C place two samples and water content replaces concrete density
31/C 31M) and ASTM Test Method for Compressive Strength of as a sixth measurement and potential criterion. Tests are to be
Cylindrical Concrete Specimens (C 39/C 39M). performed on samples of concrete representing each of the
Routine instructions for measuring, mixing, and placing three thirds of the batch, but not the very first and very last por-
concrete are given by the American Concrete Institute (ACI) tions of the batch to be discharged. The tests are performed
Manual of Concrete Inspection [11]. Additional information on also to determine the feasibility of altering the mixing time.
practices that lead to better uniformity are found in the ACI The standard guide specifications that are used to prepare proj-
304R [12]. Some variation must be accepted, but consistent ect specifications include limits for these various tests.
concrete of satisfactory quality can be obtained if proper con-
trol is maintained. ASTM C 94/C 94M provides specific in- U.S. Bureau of Reclamation Test of Mixer
structions for critical segments of batching operations plus Performance
maximum acceptable batching tolerances for each material. The U.S. Bureau of Reclamation’s mixer performance test [14]
The tolerances for measuring the materials vary with the ma- is used to evaluate the ability of a mixer to mix concrete that
terial, the size of the batch, and the type of weigh batcher sys- will be within prescribed limits of uniformity. The uniformity
tem for aggregates. For batches that use more than 30 % of of freshly mixed concrete is evaluated by comparing variations
scale capacity cement and cement plus other cementitious ma- in quantity of coarse aggregate, air content, and the density of
terials have a tolerance of
1 %. Water added to the batch air-free mortar of two samples, one taken from each of the first
must be measured to an accuracy of 1 % of the total required and last portions of the batch.
mixing water. Total water includes any wash water and free wa- Large variations in the density of air-free mortar may indi-
ter on aggregates and is measured to an accuracy of
3 % of cate that the batching procedure is incorrect or mixer blades are
the specified total water quantity. Chemical admixtures are worn. Additional mixing time may be required if the unit weight
measured to a tolerance of
3 %. of air-free mortar varies more than 24 kg/m3 (1.5 lb/ft3) [15].
Concrete uniformity generally has been measured in
terms of compressive strength, slump, density, air content, and Dunagan Test
content of coarse aggregate and cementitious materials. Uni- Dunagan [16] proposed a method for measuring the propor-
formity tests have been used to establish required mixing time, tions of cement, water, sand, and coarse aggregate in fresh con-
mixing speed, mixer batch capacity, and to verify efficient crete by a series of wash separations and weighing in air and
batching procedures. Tests by a number of investigators have water. This method has been used by Slater [17], Hollister [18],
been considered in the preparation of ASTM C 94/C 94M. Cook [19], and others to study the effects of different rates of
ASTM C 94/C 94M also contains tolerances of test results rotation of truck mixers, effect of time of haul, and effect of
that are requirements of uniformity of a single batch of mixing time on uniformity of concrete. The Dunagan test has
concrete. limited usefulness because of sampling errors and difficulties
in distinguishing between cement and very fine sand.
Methods of Measuring Uniformity
Air-Free Unit Weight Test
Tests for Mixer Uniformity A study designed to establish test methods and limits for varia-
tions in truck-mixed concrete was reported by Bloem et al. [15].
ASTM C 94/C 94M, Annex Al. Concrete Variations in slump, air content, percent of coarse aggregate,
Uniformity Requirements air-free density of mortar, water content by oven drying, and
Measuring the uniformity of concrete mixers has been a part compressive strength of concrete obtained after approximately
of ASTM Specification for Ready-Mixed Concrete (C 94/C 94M) 1/6, 1/2, and 5/6 of discharge from a truck mixer were deter-
since the 1940s. The requirements of a mixer uniformity test mined. They concluded that the air-free density test was an
have expanded and changed but the premise of checking for improvement over the density test because the number of vari-
uniformity of concrete within a batch has remained the same. ables was reduced and excessive changes in this property re-
The test method requires two separate samples from each end flected changes in water or in proportions of cement and sand.
of the middle portion of the mixer and involves determining According to their data, a difference in air-free density of mor-
the air content of each sample as well as the quantity of coarse tar of 17.6 kg/m3 (1.1 lb/ft3) corresponds to a change in water
aggregate, concrete density, density of air-free mortar, calcu- of about 9.9 L/m3 (2 gal/yd3) when the proportions of sand to
lated density of air-free concrete, slump determinations, and cement were kept constant and the water alone was varied.
seven-day compressive strength. The uniformity requirements They suggested that a variation of more than 16.0 kg/m3
involve six comparison values, assigning a maximum variation (1 lb/ft3) in this test indicates real differences in the propor-
to each tested or calculated property. The ASTM C 94/C 94M tions of the mortar ingredients, and differences of more than
requirement is that an acceptable mixer must meet not less 32.0 kg/m3 (2 lb/ft3) should be considered evidence of unsatis-
than five of the six measured properties. factory uniformity.

U.S. Army Corp of Engineers Test Method for Tests for Quality Control Uniformity
Within-Batch Uniformity of Freshly Mixed
Concrete (CRD-C 55-92) [13] Slump
This method of evaluation tests three samples of concrete for The slump test, ASTM Test Method for Slump of Hydraulic-
water content, density of air-free mortar, coarse aggregate con- Cement Concrete (C 143/C 143M), is essentially an indication of
tent, air content, slump, and seven-day compressive strength. the consistency of an individual batch of concrete. Large within-
The difference between this test method and the ASTM C 94/C batch variations in slump indicate incomplete mixing and
94M Annex Al mixer performance test is that three samples re- nonuniform distribution of water or other ingredients through-
62 TESTS AND PROPERTIES OF CONCRETE

out the batch. Batch-to-batch variations may result from batch- times until all mortar has been removed from the cup. With the
ing tolerances or errors, uncorrected changes in moisture indicator in a vertical position, the finger is carefully removed
content, grading of the aggregate, or variations in temperature. from the stem, and the number of graduations from the top to
In reasonably uniform concrete, the slump measurement the new liquid level gives an indication of the air content in the
should not vary more than about 2.5 cm (1 in.) within a batch. mortar sample. A correction factor, based on the mixture pro-
portion, must be applied to convert to percent air in concrete.
Air Content Meticulous care must be used in the selection of the mortar sam-
Air content has an important influence on concrete workabil- ple, method of inserting the stopper, agitation of the sample,
ity. Air entrainment increases slump with each 1 % of addi- and removal of the finger from the tube. The test can only pro-
tional air being approximately equivalent to 2.5 cm (1 in.) of vide an indication of relative air content and cannot be consid-
slump. Less air than desired will detrimentally affect workabil- ered as reliable as the pressure meter or the volumetric method
ity. Sudden loss of workability may indicate a major change in and should not be used to accept or reject concrete. It is most
air content. A sudden stiffening of the mixture may indicate useful as an indicator of the batch-to-batch consistency of the
loss of air or a lower air content than desired while a sticky air content of the concrete.
mixture or reduced bleeding is an indicator of increased air Another non-ASTM standard air indicator is a mini-
content. It is important that the concrete contain a uniform volumetric air meter developed by K. Nasser in Canada [21,22].
quantity of air. Within-batch variations should not exceed 1 %. This instrument is approximately 50 % of the height of an
Several methods have been developed to directly deter- ASTM C 173/C 173M volumetric meter. Aggregates larger than
mine the air content of fresh concrete. The principally ac- 25 mm are removed by sieving before the test procedure com-
cepted methods include the pressure method, ASTM Test mences. The concrete is placed in an inner bowl using two lay-
Method for Air Content of Freshly Mixed Concrete by the Pres- ers of material that are each rodded 15 times with a 9.4-mm di-
sure Method (C 231), and the volumetric method, ASTM Test ameter rod before tapping the sides to remove large air
Method for Air Content of Freshly Mixed Concrete by the Vol- bubbles and striking off the top to produce a level surface. Wa-
umetric Method (C 173/C 173M). ter is placed in an outer bowl before the concrete-containing
The pressure meter method, ASTM C 231, consists of a inner bowl is inserted. This water covers the concrete upon in-
special pressure-tight container and accessories designed to sertion of the inner bowl. The long-necked top section is then
hold a precalibrated volume of concrete. This method is used connected to the outer bowl and more water is added through
more than any other and is considered satisfactory for all types the open neck of the top section until the apparatus is
of concrete and mortar except those made with highly porous completely full. The apparatus is inverted and shaken vigor-
lightweight aggregate. This apparatus must be calibrated peri- ously a minimum of 15 times. After shaking, the apparatus is
odically to guard against changes caused by rough usage, and, set upright and jarred to release trapped foam. The top is
if the elevation of the place at which the apparatus is used opened and a measured quantity of antifoaming agent that
changes by more than 183 m (600 ft), it should be recalibrated. may be an isopropyl alcohol is added. A calibrated rod is in-
An aggregate correction factor should be determined with the serted through the top of the neck and the air content is read
materials used and subtracted from the apparent reading to de- to the nearest 0.25 %.
termine the actual air content. The aggregate correction factor This test method can be used to measure air contents of
varies only slightly for the same type of aggregate and need concrete containing any type of aggregate including light-
only be checked when there is a definite change in materials. weight. The potential problem with either of these non-ASTM
Due to the advent of new air-entraining admixtures and test methods is greater variability of test results than with ap-
their ability to entrain much smaller air bubbles, it is recom- proved test methods.
mended that the results of the pressure meter tests be periodi- The chapter in this publication by Roberts contains addi-
cally verified by the density test. tional information on air content, temperature, density and
The volumetric method, ASTM C 173/C 173M, consists of yield.
removing air from a concrete sample by mixing it with water
and isopropyl alcohol in a long-necked, closed-top special con- Density (Unit Weight)
tainer. The volume of air is determined from the difference in Air content of normal weight concrete may be computed by
volume of the sample containing entrained air and the volume comparing the actual density of concrete with the theoretical
of the sample after it has been agitated to permit the air to es- mass based on the density of the materials used, as outlined in
cape. This method is recommended particularly for light- ASTM Test Method for Density (Unit Weight), Yield, and Air
weight concrete, but it may be applied to other types of Content (Gravimetric) of Concrete (C 138/C 138M). Results ob-
concrete as well. tained by this method are influenced by variations in mixture
A non-ASTM standard air indicator is a miniature device proportions, density of ingredients, and changes in moisture
called the Chace Indicator, which uses the volumetric principle. content of aggregates. Consequently, variations in concrete
This test method is described in detail in AASHTO T 199 Air Con- density are difficult to evaluate as to cause or significance. The
tent of Freshly Mixed Concrete by the Chace Indicator [20]. A test is more definitive when the mass and solid volume of
small sample of carefully selected mortar is obtained from the coarse aggregate and volume of air are eliminated as in the
concrete and placed in a brass cup measuring 1.9 cm (3/4 in.) density of air-free mortar test.
in diameter by 1.3 cm (1/2 in.) high and compacted with a wire The density test is recommended as a job control measure,
or knife blade. The glass tube that comes with the device is filled especially for lightweight aggregate concrete in conjunction
to the top line with isopropyl alcohol, the brass tube is inserted with air determinations and slump. If the slump and air con-
in the tube, and the liquid level is adjusted to the top line. The tent are kept constant, a change in density indicates a change
finger is placed over the stem to prevent alcohol from escaping, in mass of aggregate. If the mass of aggregate per cubic meter
and the indicator is rolled from vertical to horizontal several of lightweight concrete changes, it may be the result of a
 DANIEL ON CONCRETE WORKABILITY 63

change in moisture content, grading, or density of the aggre- strength overdesign needed to ensure strength requirements
gate. Additional tests, including density, moisture content, and are met. Not to be overlooked in an evaluation process are any
grading of the aggregate are required in order to determine the improper laboratory procedures that may be a source of
cause of the variation. strength variations. ASTM C 94/C 94M lists requirements for
The chapter in this publication by Roberts contains added within-batch uniformity of concrete, and these are expressed as
information on air content uniformity as related to density. the maximum permissible difference in results of samples
taken from two locations in a concrete batch. The permissible
Temperature difference for the average compressive strength at seven days
Workability is also effected by the temperature. As the tempera- for each sample (not less than three cylinders), based on the
ture increases the time period for a given workability decreases. average strength of all comparative specimens, is 7.5 %. In a
The heat increases both the hydration rate and the rate of water well-controlled laboratory, compressive strength of cylinders
evaporation. Measurement of the freshly mixed concrete tem- fabricated from the same concrete sample may vary from 3 to
perature is accomplished using a calibrated temperature meas- 5 %. Variations in excess of this amount must be attributed
uring devise accurate to
1°F (
0.5°C). The measurement and to mixing variations within the batch of concrete being tested
reading are performed in accordance with ASTM Test Method for uniformity. An excessive variation for within batch com-
for Temperature of Freshly Mixed Hydraulic Cement Concrete pressive strengths may indicate a mixer containing fins with ex-
(C 1064/C 1064M). The primary precaution in determining the cessive wear or excessive hardened buildup.
temperature of fresh concrete is to properly and adequately Another ASTM strength test of concrete available for
cover the temperature sensor with concrete. strength comparisons is ASTM Test Method for Splitting Ten-
sile Strength of Cylindrical Concrete Specimens (C 496/C
Flow Test 496M). This test method also uses cylindrical specimens, but
The measurement of concrete flow by ASTM Test Method for the load is applied to the edge of the cylinder producing a ten-
Flow of Freshly Mixed Hydraulic Cement Concrete (C 1362) is sile failure. The variation in test results is greater than with the
a quick indicator of batch-to-batch consistency of the concrete ASTM C 39/C 39M compressive strength test, thus limiting the
or within-batch consistency. It is applicable for concrete with practical usefulness of this test to projects using the splitting
coarse aggregate up to 37.5 mm (112⁄ in.). The dimensions of this tensile strength as a pass/fail criterion for the concrete.
relatively small and compact apparatus are provided in ASTM Flexural strength testing is normally limited to paving ap-
C 1362. The flow testing device was originally known as a K- plications. The available test procedures are ASTM Test Method
slump tester [23]. ASTM C 1362 contains a precautionary note for Flexural Strength of Concrete (Using Simple Beam with
that this test method may not be appropriate for use with gap- Third-Point Loading) (C 78) or ASTM Test Method for Flexural
graded aggregate concrete. Strength of Concrete (Using Simple Beam With Center-Point
Loading) (C 293). Flexural specimens are sensitive to small varia-
Vebe Apparatus tions in molding, curing, and handling. The result is wider varia-
A modified Vebe apparatus is now used to determine the con- tions in test results than with testing for compressive strength.
sistency of very stiff concrete. The current test method, prima- When strength is to be used as a measure ASTM C 39/C 39M is
rily used for roller-compacted concrete, is applicable for any the test method of choice unless one of the other methods is re-
low-slump concrete. The ASTM Test Methods for Determining quired as a pass/fail test for strength. The chapter in this volume
Consistency and Density of Roller-Compacted Concrete Using by Ozyildirim and Carino on concrete strength testing contains
a Vibrating Table (C 1170) were approved in May 1991 and are information on quality control uniformity.
under the jurisdiction of Subcommittee C09.45 on Roller-
Compacted Concrete. This is a laboratory test method, thus Tests for Cement Content Uniformity
limiting its usefulness as a quality control method. This test
method includes a vibrating table with a surcharge mass (Test Centrifuge Test
Method A) or without a surcharge mass (Test Method B). ASTM This test, also known as the Willis-Hime method, is described
C 1170 also prescribes a method to determine the density of in detail elsewhere [25]. It provides a basis for within-batch
the consolidated concrete specimen. comparisons of cement content of concrete and employs a liq-
uid with a density greater than sand but less than cement. This
Strength Testing liquid is used to separate the components of a carefully pre-
Measured concrete strength is used widely as a criterion of pared mortar sample extracted from the concrete.
concrete quality. Other factors such as durability, abrasion re- The centrifuge test was used in the study by Bloem et al.
sistance, thermal properties, dimensional stability, placeability, [15]. They concluded that the test is quite involved and, in most
and compactibility may be more critical, but strength tests are cases, the information gained is not commensurate with the
easily made and variations in strength are assumed to be in- time and labor required. This test for cement content was an
dicative of variations in other properties. alternative to the air-free mortar density in mixer uniformity
Compressive strength as a control test is particularly use- testing during the 1960s but was then dropped from ASTM C
ful in determining the degree of uniformity of concrete. ASTM 94/C 94M.
Test Method for Compressive Strength of Cylindrical Concrete
Specimens (C 39/C 39M) is the most-used strength test for this Rapid Analysis for Determining the Cement
purpose. ACI Committee 214 [24] has developed ACI 214R-02, Content of Freshly Mixed Concrete
Evaluation of Compressive Test Results of Concrete as a statis- This test method was ASTM C 1078 and was discontinued in
tical evaluation method for control of compressive strength August 1998. This test method is discussed in detail in Chapter
based on the coefficient of variation. The larger the variation 13, “Cement and Water Content of Fresh Concrete,” in ASTM
in strengths of production concrete the larger the compressive STP 169C [5].
64 TESTS AND PROPERTIES OF CONCRETE

Cement Content Determined by Nuclear The 2-min cycles continue until the change in mass is less than
Methods 1 g. The water content is calculated by dividing the total change
An instrument small enough to be used in a field laboratory in mass by the mass of the fresh specimen.
uses nuclear technology to measure the amount of cement in The American Association of State Highway and Trans-
a sample of freshly mixed concrete [26]. portation Officials (AASHTO) has adopted this principal in Test
The test apparatus contains radioisotopic materials. Ce- Method T-318 [28]. ASTM has not adopted this procedure due
ment, primarily composed of calcium, is measured by the ab- to its poor precision. Potential modifications being studied to
sorption of low-energy photons emitted by the apparatus. The improve the test precision include the use of specimens larger
calcium atoms absorb the photons in proportion to the quantity than the current 1500 g minimum plus the use of more pow-
of calcium present in the concrete. Calcareous aggregates and erful ovens.
Class C fly ash can create potential measurement errors. Several
calibration tests are necessary with each concrete mixture and Water Content Determined by Nuclear
with each change in source of materials. A calibration curve Methods
must be developed for each mixture to be tested. The sensitivity An instrument small enough to be used in a field laboratory uses
of the apparatus to the chemical make-up of other materials has nuclear technology to measure the amount of water in a sample
produced variations in reported cement contents in excess of of freshly mixed concrete. The test apparatus contains radioiso-
permitted batching tolerances. The use of nuclear methods may topic materials. Water is actually measured as hydrogen because
be satisfactory for project averages but the accuracy is not suffi- there is little hydrogen in the other materials making up the con-
cient for measuring batch-to-batch cement quantities. As recy- crete. The apparatus contains a fast neutron source that emits
cling of mixer wash water and process water increases this water these neutrons into a concrete sample by a probe extended into
becomes another source of non-cement calcium that varies and the sample. Upon striking hydrogen atoms the source neutrons
must be accounted for in the measurements by this apparatus. lose energy. The apparatus measures the quantity of these neu-
trons, converting the results into pounds of water [26,29]. Sev-
Cement Content of Fresh Concrete eral calibration batches are required for each different concrete
Constant neutralization by 3 N hydrochloric acid (HCl) for a mixture. The nuclear apparatus has not been incorporated in an
fixed period of time (1 h in this procedure) has been employed ASTM test method due to precision variations larger than the al-
by the California Department of Transportation [27]. Accuracy lowable batching tolerances of ASTM C 94/C 94M.
of this procedure was reported to be within about 14 kg of
cement/m3 (24 lb/yd3). Field tests indicate the procedure is Workability
most useful for evaluating the performance of a concrete
mixer. Relative cement contents of various portions of a batch Workability is an everyday concern in concrete construction,
can be determined in about 1 h. and it is a factor easily appreciated in practice. Workability
This test procedure was apparently used on concrete that means different things to different people and for different
did not contain calcareous aggregates; at least, there was no placing conditions. Various nonstandard methods have been
discussion of the reaction of HCl on the aggregates, and this is developed for its measurement. None of these tests evaluate all
a phenomenon that should have a profound effect on the end characteristics that are involved in this property.
results of the test. The test procedure does require a new cali- Granville [30] defined workability as “that property of the
bration curve if there is any change in cement or aggregate concrete which determines the amount of useful internal work
source and ordinarily a new curve each day. necessary to produce full compaction.” Powers [31] defined it
The authors of the report state in their conclusions that as “that property of the plastic concrete mixture which deter-
the test was not proven to be of sufficient accuracy for routine mines the ease with which it can be placed and the degree to
control of cement content during normal concrete production which it resists segregation.” Both relate to the physical char-
and, at present, the most applicable use of the procedure is in acteristics of the concrete alone, being independent of the
evaluating mixer efficiency. methods of placing and compacting.
In actual practice, workability is related directly to the type
Tests for Water Content Uniformity of construction and methods of placing, mixing, and transport-
ing. Concrete that can be placed readily without segregation or
Rapid Analysis for Determining Water separation in a mass dam could be entirely unworkable in a thin
Content of Freshly Mixed Concrete structural member. Workable concrete compacted by means of
This test method was ASTM C 1079 and was discontinued in high-frequency vibrators would be unworkable if vibrators
August 1998. This test method is discussed in detail in Chapter could not be used and hand tamping and spading were required.
13, “Cement and Water Content of Fresh Concrete,” in ASTM Concrete having suitable workability for a pavement might be
STP 169C [5]. unsuitable for use in a thin, heavily reinforced section.
Properties involved in workability include finishing char-
Water Content Determined by Microwave acteristics, consistency or fluidity, pumpability, mobility, seg-
Oven Drying regation, and bleeding. None of the test methods proposed or
A sample of the concrete is wrapped in a fiberglass cloth and in use today simultaneously measures all of these properties.
dried in a microwave oven. The particulars are the use of not Consequently, measurement of workability is determined to a
less than three drying cycles with crushing of lumps and sepa- large extent by judgment, based on experience.
rating coarse aggregate after the first cycle. After each subse- Workability is dependent upon the physical and chemical
quent drying cycle the test specimen is stirred and its mass properties of the individual components and the proportions
determined. Each of the first three drying cycles is approxi- of each in the concrete. The degree of workability required for
mately 5 min and subsequent cycles are approximately 2 min. proper placement and consolidation of concrete is governed
 DANIEL ON CONCRETE WORKABILITY 65

by the type of mixing equipment, size and type of placing The chapters in this volume by Graves, Yzenas, Forster,
equipment, method of consolidation, and type of concrete. and Meininger have additional information on factors affect-
ing workability.
Factors Affecting Workability
Some of the factors that affect the workability [30] of concrete Coarse Aggregate
are quantity of cementitious materials, characteristics of these The particle size distribution of coarse aggregate influences the
materials, consistency, grading of fine aggregate, shape of sand water requirements and workability of concrete. Coarse aggre-
grains, grading and shape of coarse aggregate, proportion of fine gates meeting standard grading requirements, such as ASTM
to coarse aggregate, percentage of air entrained, type and quan- Specification for Concrete Aggregates (C 33), should be used.
tity of pozzolan or other supplementary cementitious material, After the grading is established, it should be maintained within
quantity of water, mixture and ambient temperatures, amount rather close tolerances to avoid sudden changes in workability
and characteristics of admixtures used, and time in transit. and other concrete properties. Segregation is reduced and uni-
formity improved by separating the aggregate into several size
Cement fractions and recombining these fractions when concrete is
Very lean mixes tend to produce harsh concrete having poor
manufactured. This recombined grading should then be bene-
workability. Rich mixes are more workable than lean mixes,
ficial in establishing the fine aggregate (sand) content.
but concrete containing a very high proportion of cementitious
Breakage, separation, and contamination of aggregate can
materials may be sticky and difficult to finish. An increase in
occur during handling and stockpiling. Introduction into the
the fineness of cement increases the cohesiveness of the con-
mixer of a large quantity of undersize material that may have
crete mix as well as the rate at which the cement hydrates and
accumulated will result in a sudden change in workability re-
the early strength development.
sulting in a demand for additional water. Plant layouts to mini-
Consistency mize aggregate handling are desirable. Training of front-end
Consistency (according to ASTM C 125) and plasticity are loader operators on the use of techniques to minimize aggre-
terms often used to indicate workability. Consistency generally gate segregation will assist in avoidance of sudden changes in
denotes the wetness of the concrete, which is commonly meas- workability.
ured by the slump test. It must not be assumed that the wetter Production of workable concrete with sharp, angular, or
the mix the more workable the concrete. If a mix is too wet, seg- crushed aggregates generally requires more sand than similar
regation may occur with resulting honeycomb or sand streak- concrete made with rounded aggregates. The water demand
ing on the exposed surface; finishing properties will be may be increased 9–15 kg/m3 (15–25 lb/yd3). If the water-
impaired because of the accumulation of laitance on the sur- cement ratio is held constant, more cement is required. Flat or
face. If a mix is too dry, it may be difficult to place and com- elongated particles, defined as particles having a ratio of width
pact, and separation may occur because of the tendency for to thickness or length to width, respectively, greater than 3:1,
larger particles to roll towards the outer edge of the heap are detrimental to concrete workability and finishability. More
formed when it is deposited. It is agreed generally that con- sand, cement, and water are required when the coarse aggre-
crete should have the driest consistency that is practicable for gate contains flat and elongated particles.
placement with available consolidation equipment. The consis- The maximum size of aggregate that can be used to pro-
tency necessary for full compaction varies with the type of duce workable concrete is limited by practical considerations
structure, type and size of aggregate, and type of compaction including type and size of structure, amount and spacing of
equipment available. reinforcing bars, method of placing, and availability of mate-
rials. Generally, aggregate should not be larger than three-
Sand fourths of the maximum clear spacing between reinforcing
Concrete containing fine sand requires more water for the same bars nor larger than one-fifth of the wall thickness or nar-
consistency, as measured by the slump test, than an equivalent rowest dimension between sides of forms. ACI Practice for Se-
amount of coarse sand. Very coarse sand can have an undesir- lecting Proportions for Normal, Heavyweight, and Mass Con-
able effect on finishing quality. Neither very fine nor very coarse crete (ACI 211.1) [32] provides recommendations on the
sand is desirable but both have been used satisfactorily. maximum sizes of aggregate for various types of construc-
Rounded river sand gives greater workability than crushed sand tion. The U.S. Bureau of Reclamation’s [33] experience in
composed of sharply angular pieces with rough surfaces. Angu- pumping concrete indicates that concrete containing 6.4 cm
lar sand particles have an interlocking effect and less freedom (2 12⁄ in.) maximum size aggregate can readily be pumped
of movement in the freshly mixed concrete than smooth through a 20.3 cm (8 in.) pipe, but aggregate larger than 6.4
rounded particles. Natural sand may give satisfactory results cm (2 12⁄ in.) may cause difficulty. ACI 304R [12] recommends
with a coarser grading than would be permitted with crushed limiting the maximum size of angular or crushed coarse ag-
manufactured sand. In addition, concrete must contain 2–3 % gregate to one-third of the smallest inside diameter of the
more sand by absolute volume of total aggregate and 6–9 kg pump or pumpline.
more water/m3 (10–15 lb/yd3) when crushed sand is used. The chapters in this volume by Graves, Yzenas, Forster,
Manufactured fine aggregate (MFA) processed from and Meininger have additional information on factors affect-
crushed stone is gaining wider use as natural sands become de- ing workability.
pleted in some geographic areas. The MFA generally contain a
greater quantity of fines than natural sands and often mask Air Entrainment
their good workability with low slump test results. The aggre- Entrained air increases the paste volume, acts as a lubricant,
gate particle shape and subsequent finishability characteristics and improves the workability of concrete. It reduces bleeding
are influenced by the type of crusher used in the MFA produc- and segregation during handling and placing of concrete and
tion process. increases the cohesiveness or “fattiness” of the concrete.
66 TESTS AND PROPERTIES OF CONCRETE

Improvement in workability resulting from air entrainment is coarse aggregate plus a sufficient amount to permit the con-
more pronounced in lean mixes that are harsh and unwork- crete to be placed readily in forms and vibrated around rein-
able because of poor aggregate grading or type of aggregate forcement as necessary. An excess of mortar increases worka-
used. The chapter in this volume by Roberts has additional in- bility, but excess workability is inefficient. Too much mortar
formation on factors affecting workability. can result in a sticky mixture. It should not be more than is re-
quired for consolidation by available equipment. The quantity
Finely Divided Material of mortar required to produce the desired workability with a
The addition of finely divided material, including inert or ce- given coarse aggregate can be determined more effectively by
mentitious materials or pozzolans, generally improves the laboratory tests. ACI 211.1 provides a basis for estimating the
workability of the concrete. Improvement is more noticeable in proportions of coarse aggregate to be used in trial mixes.
lean mixes than in rich mixes. These materials have been used
to improve the grading of sands deficient in fines. Cementitious Methods of Measuring Normal Consistency
and pozzolanic materials are usually substituted volumetrically Concrete
for 10–70 % of the cement. Workability will be improved if these
materials are added as a replacement for part of the sand, in- Slump Test
stead of substituted for part of the cement. The slump test (Fig. 1) is the most commonly used method of
measuring the consistency of concrete. It is not suitable for
Chemical Admixtures very wet concrete, very dry concrete, or SCC. It does not meas-
Water-reducing admixtures, when added to concrete, permit a ure all factors contributing to workability, nor is it always rep-
reduction in mixing water with no loss in slump, or, if the wa- resentative of the placeability of the concrete. However, it is
ter content is held constant, produce an increase in slump. Set- used conveniently as a control test and gives an indication of
retarding admixtures reduce the early rate of hardening and the uniformity of concrete consistency from batch to batch. Re-
permit concrete to be handled and vibrated for a longer period peated batches of the same mix, brought to the same slump,
of time. will have the same water content and water-cement ratio pro-
It has been reported that there is a decrease in the fre- vided weights of aggregate, cement, and admixtures are uni-
quency of plugged pump lines when water-reducing retarders form and aggregate gradings are within acceptable tolerances.
are used in the concrete. This is partially due to the increased Additional information on the mobility of the concrete can
slump and typically an increased sand content with the use of be obtained if, after removing the slump cone and measuring
a water-reducing chemical admixture. The use of high-range- the slump, the concrete is tapped on the side with the tamping
water-reducing (HRWR) admixtures have greatly increased the rod. Two concretes with the same slump may behave differ-
placeability capabilities for high-strength, low-water-cement ra- ently, that is, one may fall apart after tapping and be harsh with
tio concretes. HRWR admixtures have been used with silica a minimum of fines, and the other may be very cohesive with
fume providing highly workable concrete mixtures. The self- surplus workability. The first concrete may have sufficient
consolidating concretes require maximum utilization of chem- workability for placement in pavements or mass concrete, but
ical admixtures. A late generation HRWR is necessary to the other concrete may be required for more difficult place-
produce good flowability and a viscosity-modifying admixture ment conditions.
(VMA) is usually needed to increase the consistency (viscosity) The slump test should be performed in strict accordance
and prevent segregation. with the requirements of ASTM C 143/C 143M. Tests are often
made at the point of placement and should be made whenever
Mixture Proportions specimens are molded for strength testing. Slump tests may be
Concrete workability can be controlled by proper proportion- made at the batch plant in order to check the uniformity of
ing of the constituent materials. As the proportion of mortar, batching operations.
including sand, cement, water, and air, is increased, the grad- Popovics [34] has presented data indicating that the rela-
ing and angularity of the coarse aggregate become less impor- tionship between consistency values, as measured by the slump
tant. There should be sufficient mortar to fill the voids in the test and the water content of concrete, is parabolic, that is, the
percentage change in water content required to increase the
slump 25 mm (1 in.) may vary from 2.0 % when the initial
slump is 127 mm (5 in.) to approximately 4.5 % when the ini-
tial slump is 51 mm (2 in.). An average change in water content
of 3 % generally is considered necessary for a 25 mm (1 in.)
slump change.
As the temperature of the concrete increases, the slump
decreases. Concrete placed at a slump of 100 mm (4 in.) at
21°C (70°F) may only have a 76 mm (3 in.) slump when placed
at 32°C (90°F), or the same concrete may have a slump of 140
mm (5.5 in.) when placed at 10°C (50°F).
Air-entrainment and water-reducing admixtures will in-
crease the slump of concrete if all other conditions remain the
same. Each 1 % increase or decrease in air content will pro-
duce approximately the same influence as a change in water
content of 3 %.
The slump generally is reported to the nearest 5 mm (1/4
Fig. 1—Slump test. in.). Slump reported by different operators on the same batch
 DANIEL ON CONCRETE WORKABILITY 67

of concrete may vary by as much as 13 mm (1/2 in.). The most

unsatisfactory form of slump is the shear slump, that is, a
falling away or shearing off of a portion of the concrete from
the mass. If this condition exists, the concrete probably lacks
the necessary plasticity for the slump test.
Slump cones manufactured using a stiff, durable plastic
material have been approved by ASTM C 143/C 143M after a
series of comparative tests at three different slump ranges in-
dicated no measurable difference in test results. Both the clas-
sic metal cones and plastic cones can now be used for the
slump test.

Flow Tester
The test apparatus is used to measure a concrete flow value
with ASTM Test Method for Flow of Freshly Mixed Hydraulic-
Cement Concrete (C 1362). The insertion of the instrument
into the concrete is limited to 40 s before the measuring rod is
lowered and the flow measurement read.
The flow tester (K-slump tester) (Fig. 2) is reported to
measure slump directly 1 min after the tester is inserted in the
concrete [23]. It measures an index that is related to workabil-
ity after the device is removed from the concrete. The first
reading is taken after the tester has been in place (the disk rest-
ing on the concrete surface) for 60 s. This reading, in centime- Fig. 3—Ball penetration apparatus.
tres, is referred to as the K-slump. The device is removed from
the concrete and the measuring rod is again lowered to rest on
the surface of the concrete remaining in the tube; this reading,
in centimetres, represents the workability of the mix. crete from one definite shape to another by means of jigging.”
Coarse aggregate up to 37.5 mm (1 12⁄ in.) is permitted in The equipment consists of a metal cylinder mounted inside a
the concrete. The minimum depth of concrete at testing is 175 larger cylinder and a suspended plate that fits inside the smaller
mm (7 in.) and the minimum distance from the tube to the cylinder. A slump cone is placed inside the smaller cylinder so
nearest edge of the level surface of concrete to be tested shall that the bottom rests on the base. It is filled with concrete, the
be 75 mm (3 in.). Concrete in the forms can be tested or con- slump cone is removed, and the plate is placed on top of the
crete may be in a pail, wheelbarrow, or other container. The concrete. The flow table on which the apparatus is mounted is
concrete may be retained in the original test position and then operated. The number of 6 mm (1/4 in.) drops required to
retested later if desired. The test was developed to provide a mold the concrete into a cylindrical form is a measure of the
method of testing using a minimal concrete area and volume. workability of the concrete. This method has not found wide-
Studies have been made and reported on 420 concrete batches spread use and no ASTM standard has been written about it.
by five laboratories. Statistical determinations and equations
are reported elsewhere [23]. Ball Penetration Test
The Kelly ball test [35] was developed principally as a conven-
Remolding Test ient method of measuring and controlling consistency in the
The remolding test apparatus was developed by Powers [31] to field. The ball test can be performed on concrete in the forms,
measure “the relative effort required to change a mass of con- space permitting, and it is claimed that tests can be performed
faster and precision is greater than with the slump test. One
disadvantage of this test is that it requires a large sample of
concrete.
The apparatus (Fig. 3) weighs 13.6 kg (30 lb) and consists
of a 15.2-cm (6-in.) diameter ball and stem that can slide
through the center of a stirrup, the legs of which rest on the
concrete to be tested. The depth of concrete must be at least 20
cm (8 in.), and the minimum distance from the center of the
ball to the nearest edge of the concrete is 23 cm (9 in.).
The surface of the concrete is struck off level, avoiding ex-
cess working. The ball is lowered gradually onto the surface of
the concrete, released, and the depth of penetration read im-
mediately on the stem to the nearest 6 mm (1/4 in.). The ratio
of slump to the penetration of the ball is between 1.5 and 2 and
is fairly constant for a given mix but varies according to the
mix. This ASTM Test Method for Ball Penetration in Fresh Port-
land Cement Concrete (C 360) was adopted in 1955 and dis-
continued by ASTM in February 1999. Information on this test
Fig. 2—Flow test apparatus. method was reported in Ref 35 and ASTM STP 169C [5].
68 TESTS AND PROPERTIES OF CONCRETE

Truck Slump Meter

Truck mixers now use the hydraulic pressure needed to turn
mixing drums to measure the consistency (slump) of the con-
crete contained in the drum. Tests indicate that higher
pressures for mixing indicate lower slump values. After several
calibration comparisons of hydraulic pressure readings versus
measured slump test results, the truck mixer gage can be used
to closely estimate the slump of the concrete. Load size is a fac-
tor, but equipment manufacturers claim it is not a major fac-
tor. These pressure readings will vary for each truck and mixer
necessitating calibration for each mixing unit not of identical
design, age, and condition.
Daczko [36] discusses the use of truck mixer slump meters
as a method of measuring rheology properties of production
concrete. Experimenting with two high-fluidity concrete mix-
tures having comparable slump values, slump meter pressure
readings were taken at various drum speeds and graphically
plotted as drum speed versus hydraulic pressure. The results
produced the same pattern as that of a Bingham material and
relative results comparable to the tests of these mixtures in an
IBB rheometer [37]. The latter apparatus is not a true rheome- Fig. 5—Thaulow concrete tester.
ter, but is the instrumented mixer described in the Two-Point
Workability Tests.

Wigmore Consistometer pacted by turning the handle attached to the cam at the rate of
The Wigmore consistometer is described by Orchard [38]. This about 1 rps. The table drops 0.6 cm (7/32 in.) four times per
apparatus consists of a galvanized container and a hand- revolution of the cam, and the number of drops required to
operated compaction table. A 5.1-cm (2-in.) diameter ball that lower the ball and stem 19.7 cm (734⁄ in.) into the concrete is
is fastened to a sliding stem is mounted in the lid of the con- considered a measure of the consistency of the concrete.
tainer. The container is filled with concrete that is compacted The number of drops required varies by a factor of ten as
on the table by eight drops. The container is again filled with the concrete changes from a 150-mm (6 in.) slump to very stiff
concrete and leveled off; then the lid and the ball are placed in no-slump concrete. It is claimed that the Wigmore consistome-
position with the ball resting on the surface of the concrete. ter is an improvement over the slump test because work is ac-
The apparatus is placed on the table and the concrete is com- tually done on the concrete in a way that resembles field con-
ditions. Variations in results may be expected if the ball comes
in contact with large aggregate.

Methods of Measuring Consistency of No-Slump

Concrete

Vebe Apparatus
The modified Vebe apparatus (Fig. 4) is used to measure the
consistency of Roller-Compacted Concrete. ASTM Test Meth-
ods for Determining Consistency and Density of Roller-
Compacted Concrete Using a Vibrating Table (C 1170) is
discussed in the chapter by Adaska in this volume.
The Vebe consistometer [39] includes a vibrating table, a
sheet metal pan, slump cone, and plastic plate attached to
a graduated, free-moving rod that serves as a reference end
point. The cone is placed in the pan, filled with concrete, and
removed. The plastic disk is brought into position on top of the
concrete, and the vibrating table is set in motion. The number
of seconds required to remold the cone of truncated concrete
to the shape of the cylinder is the measure of consistency and
is reported as the number of Vebe seconds or degrees. This
method is very suitable for very dry concrete, but the vibration
is too vigorous for concrete with a slump greater than about
5.1 cm (2 in.). For example, 0–3 s are required for concrete
with a slump of 7.6–10.2 cm (3–4 in.), and 10–32 s [40] may be
required for concrete with less than zero slump. The Vebe con-
sistometer differs from the modified Vebe Apparatus of ASTM
C 1170 and the results should not be interchanged. In the ACI
Fig. 4—Modified vebe apparatus. Guide for Selecting Proportions for No-Slump Concrete (ACI
 DANIEL ON CONCRETE WORKABILITY 69

211.3R) [40], several tables are included referencing Vebe test the discharge tube is closed by placing the finger over the end,
results. These results are based upon the Vebe consistometer and 1725 mL of mixed grout is poured into the cone. The fin-
and not the modified Vebe apparatus of ASTM C 1170. ger is removed, and the number of seconds until the first break
in the continuous flow of grout is the efflux time. When com-
Other Methods paring grouts, the speed of mixing and the mixing time have an
ACI 211.3R [40] provides a comparison of consistency meas- influence on efflux time and should be kept constant.
urements by three methods. These methods are the conven-
tional Vebe consistometer [39], the Thaulow Concrete Tester Grout Consistency Meter
developed in Norway [41] (Fig. 5), and the Compacting Factor A meter for measuring the consistency of grout has been de-
test developed in Great Britain [30]. Differences in consistency veloped at the University of California and is described else-
of very dry mixes cannot be measured with the slump cone, but where [42]. The grout consistency meter is essentially a torque
the Thaulow drop table is considered to have merit for this ap- meter. The sample of grout is placed in a metal pan mounted
plication. Concrete with a slump of 0–2.5 cm (0–1 in.) requires on a platform that can be rotated at a constant speed of 60
14–28 revolutions of the drop table, and concrete with a slump rpm. Suspended from a music wire is a 7.3 kg (16 lb) paddle
of 7.6–10.2 cm (3–4 in.) requires less than seven revolutions. assembly to which a torque is applied as the sample of grout is
The Compacting Factor test is considered only marginal for rotated. The angle of twist or consistency factor is read by an
very dry concrete. index pointer attached to a cross strut.

Methods of Determining Consistency of Grout Measuring Workability of Self-Consolidating

Concrete (SCC)
Flow Cone The general characteristics of SCC make the usual workability
ASTM Test Method for Flow of Grout for Preplaced-Aggregate tests ineffective. The high flowability and variable viscosities of
Concrete (Flow Cone Method) (C 939) is a method used in the SCC have caused the development of a series of new test meth-
laboratory and in the field for determining the consistency of ods. Some of these tests will be developed into ASTM test
grout mixtures by measuring the time of efflux of a specified methods while others will not. All of these test methods have a
volume of grout from a standardized flow cone or funnel. The goal of being relatively inexpensive and providing immediate
flow cone (Fig. 6) is mounted firmly with the top surface level, results, with good repeatability (single operator) and repro-

Fig. 6—Flow cone.

70 TESTS AND PROPERTIES OF CONCRETE

ducibility (multilaboratory). The chapter in this volume by

Daczko and Vachon presents detailed information on self-
consolidating concrete and the testing methods associated with
its workability.

Two-Point Workability Tests

Tattersall [43] discusses the principles of measurements of the
workability of fresh concrete and a simple two-point test. In
this paper, the author points out that an understanding of
workability of fresh concrete is important for the following rea-
sons: (a) to make possible the design of mixtures for particular
purposes; (b) to provide a method of control in the manufac-
ture of the mixture; and (c) to contribute to the efficient use of
manufacturing processes such as vibration, pumping extru-
sion, and finishing. Fig. 7—Rheometer test results.
Tattersall contends that in spite of all the efforts (papers
that have been written and the many proposed tests) over the
past 30 plus years, there is no test that is fully satisfactory and Rheometers
the property of workability cannot even be defined except in The basis of the two-point test to measure workability is a vari-
the most general terms. Each of the test methods is capable of able constant-speed rheometer. In the last ten years five rota-
classifying as identical concretes that can be shown to be tional rheometers have been designed to measure the flow
dissimilar. properties of concrete [46]. Three different types of configura-
Tattersall [37] suggests the following summary of pro- tions are used in these instruments. Two use coaxial cylinders,
posed terminology as an effort toward standardization. one uses parallel plates, and the other two use rotating vanes.
Unfortunately each rheometer produces numerical test values
I. Qualitative that are not directly comparable to values from a different
Workability rheometer design. Brower and Ferraris [46] found when test-
Flowability ing several different concrete mixtures in each of the five styles
Compactibility of rheometers that comparisons of test results were reasonably
Stability good. It is nevertheless advisable, if the properties of different
Finishability concrete mixture proportions are to be compared, that all tests
Pumpability be performed in one style of rheometer.
II. Quantitative empirical
Slump
IBB Rheometer
Compacting factor
The IBB Rheometer, the principals of which were developed by
Vebe time
Tattersall [37,47], is not like other rheometers. The original de-
Flow table spread
sign of the IBB rheometer was a converted food-mixer with a
III. Quantitative fundamental
planetary motion and ultimately a pressure gage and an H-
Viscosity Mobility Fluidity
shaped impeller. The test procedure consists of measuring the
Yield value
power required at three separate speeds to operate an 18.9 L
The objection to the several workability tests are that, al- (20 qt) food mixer when empty, then repeating the power
most without exception, they are single-point tests, whereby measurement at all three speeds when the bowl contains a pre-
only one measurement is made at one specific rate of shear or scribed quantity of concrete. Values for yield and plastic vis-
set of shearing conditions. Tattersall [44] states that such a pro- cosity are obtained by plotting (P  PE) / w against w, where w
cedure is valid only for a simple Newtonian liquid whose flow is speed, P is power under load, and PE is power when the bowl
properties are completely defined by the constant ratio of is empty. Tattersall [37] claims thousands of tests have proven
stress to shear rate, and that ordinary observation shows that beyond doubt that the flow properties of fresh concrete closely
fresh concrete is not a Newtonian liquid and that, conse- approximate the Bingham model and that there is a simple
quently, any test based explicitly or implicitly on the assump- straight line relationship between torque and speed with an in-
tion that it is will be inadequate. tercept on the torque axis. The two-point test yields more in-
Tattersall also states that there is evidence to indicate that, formation concerning the performance of a concrete mixture
in practice, it may be sufficient to treat freshly-mixed concrete (Fig. 7) and appears to have good potential for determining
as conforming with the Bingham model, which describes a ma- and consequently controlling concrete uniformity.
terial with a yield value and a plastic viscosity, which together
describe the shear stress of the material. The yield value and Conclusions
plastic viscosity are constants, and it follows that measure-
ments at two shear rates are required to determine them. The Concrete knowledge and technology have advanced slowly
balance of this paper by Tattersall [44] discusses test proce- during the years, but additional knowledge will be required if
dures, results, and conclusions, and further modifications to al- concrete is to maintain the position it has established as the
low application of vibration. universal building material. New products and new technolo-
Tattersall [37,43,45] further explains the rationale of a two- gies are being developed in all phases of the concrete industry
point workability test and the relationships between slump, at a rapid rate as evidenced by SCC, truck mixer slump meters,
compacting factor, Vebe time, and the two-point test. and an increasing use of rheology in concrete research. Im-
 DANIEL ON CONCRETE WORKABILITY 71

provements in concrete production, control, delivery, and plac- [8] Ferraris, C. F. and De Larrard, F., “Testing and Modeling of Fresh
ing techniques are also being developed. At the same time, Concrete Rheology,” NISTIR 6094, National Institute of Stan-
more widespread use of available knowledge for controlling dards and Technology, Gaithersburg, MD, 1998, pp. 43–44.
uniformity, quality, and workability will improve the competi- [9] Domone, P. L. J., Yongmo, Xu, and Banfill, P. F. G., “Develop-
tive position of concrete. ments of the Two-Point Workability Test for High-Performance
Concrete,” Magazine of Concrete Research, Thomas Telford
New methods of mixing, placing, consolidating, and fin-
Ltd., London, Vol. 51, No. 3, 1999, pp. 171–179.
ishing concrete may permit the use of less water and improve
[10] Van Alstine, C. B., “Water Control by Use of a Moisture Meter,”
concrete quality. New tests and methods of quality control and Journal, American Concrete Institute; Proceedings, Nov. 1955,
measurement of workability must be developed concurrently pp. 341–347.
with methods of mixing and placing concrete. The slump test [11] ACI Manual of Concrete Inspection, SP-2, 9th ed., American
and other tests used to measure uniformity may be replaced by Concrete Institute, Farmington Hills, MI, 1999.
more efficient test methods in the future. [12] Guide for Measuring, Mixing, Transporting, and Placing Con-
Meters have been developed to measure moisture content crete (ACI 304R-00), ACI Manual of Concrete Practice 2004,
of sand and coarse aggregate in the bins at concrete plants and Part 2, American Concrete Institute, Farmington Hills, MI,
promote better control of mixing water. The widespread use of 2004.
aggregate moisture meters, particularly in sand, has greatly im- [13] U.S. Army Corps of Engineers Handbook for Concrete and Ce-
proved the quality control of concrete throughout the industry. ment, updated December 2002, http://www.wes.army.mil/SL/
The development of slump meters for equipment, both MTC/handbook/handbook.htm. Waterways Experiment Sta-
hydraulic and electrical resistance, has improved the consis- tion, Vicksburg, MS.
tency of concrete while being mixed and transported, but this [14] Bureau of Reclamation, “Variability of Constituents in Concrete
(A Test of Mixer Performance), (Designation 26),” Concrete
technology is expected to become even more accurate in the
Manual, 8th ed. Revised, U.S. Dept. of the Interior, Denver, CO,
future. The slump test and other tests used to indicate con- 1981, pp. 558–562.
sistency can be made only after the concrete is discharged, [15] Bloem, D. L., Gaynor, R. D., and Wilson, J. R., “Testing Unifor-
and corrections can be applied only to subsequent batches. mity of Large Batches of Concrete,” Proceedings, ASTM Inter-
Perhaps future research will reach into such areas as detect- national, Vol. 61, 1961.
ing changes in aggregate absorption and gradings during the [16] Dunagan, W. M., “A Method of Determining the Constituents
batching process. of Fresh Concrete,” Journal, American Concrete Institute; Pro-
The slump test is an old friend and has served its purpose ceedings, Vol. 26, 1930.
well. Some say that the slump test has already outlasted its [17] Slater, W. A., “Tests of Concrete Conveyed from a Central Mix-
time, but several of the new tests for SCC are being developed ing Plant,” Proceedings, ASTM International, Vol. 31, Part II,
around the slump mold. As antiquated as it sometimes seems 1931.
the slump test continues to be in demand because of its sim- [18] Hollister, S. C., “Tests of Concrete From a Transit Mixer,” Pro-
plicity, economy, and friendliness in the field. The develop- ceedings, American Concrete Institute, Vol. 28, 1952, pp.
405–417.
ment of improved testing procedures for freshly mixed con-
[19] Cook, G. C., “Effect of Time of Haul on Strength and Consis-
crete in both the laboratory and the field remains a
tency of Ready-Mixed Concrete,” Journal, American Concrete
requirement of the concrete industry. Institute; Proceedings, Vol. 39, 1952, pp. 413–428.
[20] AASHTO, “Method of Test for Air Content of Freshly Mixed
Concrete by the Chace Indicator,” AASHTO T 199, Standard
References Specifications for Transportation Materials and Methods of
Sampling and Testing, 24th Edition, Part II, American Associa-
[1] Tyler, I. L., “Uniformity, Segregation, and Bleeding,” Signifi- tion of State Highway and Transportation Officials, Washing-
cance of Tests and Properties of Concrete and Concrete Aggre- ton, DC, 2004.
gates, ASTM STP 169, ASTM International, West Conshohocken, [21] Nasser, K. W. and Whaley, B. D., “Mini Air Meter for Fresh Con-
PA, 1956, p. 37. crete,” Concrete International: Design & Construction, Vol. 10,
[2] Hubbard, F., “Workability and Plasticity,” Significance of Tests No. 3, Mar. 1988, pp. 43–46.
and Properties of Concrete and Concrete Aggregates, ASTM STP [22] Khayat, K. H., Nasser, K. W., and Hester, W. T., “Alternative
169, ASTM International, West Conshohocken, PA, 1956, p. 47. Method to Measure Air Content in Fresh Concrete,” ACI Mate-
[3] Vollick, C. A., “Uniformity and Workability,” Significance of rials Journal, Vol. 87, No. 4, July-August 1990, pp. 355–362.
Tests and Properties of Concrete and Concrete-Making Materi- [23] Nasser, K. W., “New and Simple Tester for Slump of Concrete,”
als, ASTM STP 169A, ASTM International, West Conshohocken, ACI Journal, Proceedings, Vol. 73, No. 10, Oct. 1976, pp.
PA, 1966, pp. 73–89. 561–565.
[4] Smith, D. T., “Uniformity and Workability,” Significance of [24] Evaluation of Strength Test Results of Concrete (ACI 214R-02),
Tests and Properties of Concrete and Concrete-Making Materi- ACI Manual of Concrete Practice, Part 1, American Concrete In-
als, ASTM STP 169B, ASTM International, West Conshohocken, stitute, Farmington Hills, MI, 2004.
PA, 1978, pp. 74–101. [25] “Proposed Tentative Method of Test for Cement Content of
[5] Scanlon, J. M., “Factors Influencing Concrete Workability,” Sig- Freshly Mixed Concrete,” Bulletin, No. 239, ASTM International,
nificance of Tests and Properties of Concrete and Concrete- West Conshohocken, PA, July 1959, p. 48.
Making Materials, ASTM STP 169C, ASTM International, West [26] Highway Innovative Technology Evaluation Center, “Evalua-
Conshohocken, PA, 1994, pp. 49–64. tion of the Troxler Model 4430 Water-Cement Gauge,” CERF
[6] Cement and Concrete Terminology (ACI 116R-00), ACI Manual Report: HITEC 96-03-F, Civil Engineering Research Foundation,
of Concrete Practice 2004, Part 1, American Concrete Institute, Reston, VA, August 1996, 41 pp.
Farmington Hills, MI, 2004. [27] Report No. CA-DOT-TL-5149-1-76-65, California Department of
[7] Behavior of Fresh Concrete During Vibration (ACI 309.1R-93), Transportation, Sacramento, CA, Sept. 1976.
ACI Manual of Concrete Practice, Part 2, American Concrete In- [28] Test Method for Water Content of Freshly Mixed Concrete Us-
stitute, Farmington Hills, MI, 2004. ing Microwave Oven Drying (T 318-02), American Association of
72 TESTS AND PROPERTIES OF CONCRETE

State Highway and Transportation Officials, Washington, DC, [38] Orchard, D. F., Concrete Technology, Vol. 2, Wiley, New York,
2003. 1962.
[29] Monfore, G. E., “A Review of Methods for Measuring Water [39] Bahrner, V., “Report on Consistency Tests on Concrete Made by
Content of Highway Components in Place,” Highway Research Means of the Vebe Consistometer,” Report No. 1, Joint Re-
Record Number 342, Environmental Effects on Concrete, High- search Group on Vibration of Concrete, Svenska Cement-
way Research Board, Washington, DC, 1970, pp. 17–26. foreningen, March 1940.
[30] Granville, W. R., Collins, A. R., and Mathews, D. D., “The Grad- [40] Guide for Selecting Proportions for No-Slump Concrete
ing of Aggregate and Workability of Concrete,” Road Research (ACI211.3R-02), ACI Manual of Concrete Practice, Part 1,
Technical Paper No. 5, Her Majesty’s Stationary Office, London, American Concrete Institute, Farmington Hills, MI, 2002, 26
1947. pp.
[31] Powers, T. C., “Studies of Workability of Concrete,” Journal, [41] Thaulow, Sven, Field Testing of Concrete, Norsk Cementforen-
American Concrete Institute, Vol. 28, Feb. 1932, p. 419. ing, Oslo, 1952. Also, Resume in ACI Journal, Proceedings, Vol.
[32] Practice for Selecting Proportions for Normal, Heavyweight, 50, No. 7, Mar. 1954, pp. 10–11, 24–26.
and Mass Concrete (ACI 211.1-91) (Reapproved 1997), ACI Man- [42] Davis, R. E., Jansen, E. C., and Neelands, W. T., “Restoration of
ual of Concrete Practice, Part 1, American Concrete Institute, Barker Dam,” Journal, American Concrete Institute; Proceed-
Farmington Hills, MI, 2004. ings, Vol. 44, April 1948.
[33] Bureau of Reclamation, Concrete Manual, 8th ed. Revised, U.S. [43] Tattersall, G. H., “The Rationale of a Two-Point Workability
Dept. of the Interior, Denver, CO, 1981, p. 63. Test,” Magazine of Concrete Research, Vol. 25, No. 84, Sept.
[34] Popovics, S., “Relations Between the Change of Water Content 1973.
and the Consistence of Fresh Concrete,” Magazine of Concrete [44] Tattersall, G. H., “Principles of Measurement of the Workability
Research, London, July 1962. of Fresh Concrete and a Proposed Simple Two-Point Test,” Fresh
[35] Kelley, J. W. and Polivka, M., “Ball Test for Field Control of Con- Concrete, Rilem Seminar Proceedings, Vol. 1, 1973.
crete Consistency,” Journal, American Concrete Institute; Pro- [45] Tattersall, G. H., “The Relationships Between the British Stan-
ceedings, Vol. 51, May 1955, p. 881. dard Tests for Workability and the Two-Point Test,” Magazine
[36] Daczko, A. D., “A Proposal for Measuring Rheology of Produc- of Concrete Research, Vol. 28, No. 96, Sept. 1976.
tion Concrete,” Concrete International, Vol. 22, No. 5, May [46] Brower, L. E. and Ferraris, C. F., “Comparison of Concrete
2000, pp. 47–49. Rheometers,” Concrete International, Vol. 25, No. 8, American
[37] Tattersall, G. H., “Workability and Quality Control of Concrete Institute, Farmington Hills, MI, Aug. 2003, pp. 41–47.
Concrete”, E & FN Spon, an imprint of Chapman & Hall, London, [47] Tattersall, G. H. and Banfill, P. F. G., The Rheology of Fresh
1991. Concrete, Pitman Books Limited, London, 1983.
9
Air Content, Temperature, Density
(Unit Weight), and Yield

Lawrence R. Roberts1

Preface concrete, and to determine the volume of concrete being pro-

duced from a given batch. The yield data so obtained are then
THIS CHAPTER CONTINUES THE TRADITION OF available for the calculation of unit cement and aggregate con-
previous editions of ASTM STP 169. Due to the fundamental tents, which are essential in mixture development and may be
nature of these topics, much of what is here has appeared be- required in some specifications. Tests for unit weight are also
fore in ASTM STP 169B, authored by F. F. Bartel, and ASTM performed to control concrete weight per se of both lightweight
STP 169C, authored by the present author. This chapter and heavyweight concretes.
updates the language and expands on the utility of density While air content is most commonly determined to ensure
measurement in conjunction with air content measurement to the presence of air entrainment for freeze-thaw durability,
improve test reliability. The author would also like to acknowl- knowledge of air content of non-air-entrained concrete is also
edge the support and encouragement of his company and the important, due to the strong negative impact that unexpected
many colleagues who have provided useful insight. increases in air content can have on compressive and flexural
strengths. The author is aware of a number of job problems in
Introduction non-air-entrained concrete that could have been prevented by
air measurement or even simple density measurements,
Measurements of air content, temperature, and density (unit thereby significantly reducing the economic impact caused by
weight)2 of freshly mixed concrete are the backbone of field excess air, brought about by contaminations, changes in mate-
quality control of concrete construction. Without these, pro- rials, etc. Thus, checking air content and density is also a nec-
portioning of mixtures, control of yield, and assurance of essary safety precaution for these concretes.
placeability, strength development, and durability of concrete Air contents of air-entrained concretes can vary for a large
exposed to freezing conditions are not possible. The tests in- number of reasons. Among these are changes in air-entraining
volved are relatively straightforward, use simple equipment, admixture type or dosage, changes in cement alkali content
and are thoroughly described in the applicable ASTM test [1,2], fine aggregate grading, slump, concrete temperature, mix-
methods. Despite this, job problems, ranging from simple dis- ing intensity and duration, and many others. While a discussion
putes over yield to complete performance failures, frequently of the impact of each of these is beyond the scope of this work
can be attributed directly to faulty application of these meth- and is well-covered elsewhere [3–5], suffice it to say that the
ods. The purpose of this chapter is to help the interested con- possibility of a change due to any one factor is great enough to
crete technician understand some of the key issues controlling warrant close control of the air content. Further, nominally non-
the correct application and interpretation of the methods, and air-entrained concretes can also develop significant air contents,
hopefully to encourage more careful and frequent application. due to errors in batching or contamination of materials. The in-
terrelationship between air content and density should be obvi-
Overview of Significance and Use ous: an increase in volume of air results in a lower density, all
other material contents being unchanged. The calculation of ma-
Temperature measurement of fresh concrete is vital to ensure terial quantities per unit volume from the density depends not
adherence to maximum temperature specifications, usually only on the correct determination of the concrete density, but
imposed to control thermal gradients and possible resultant also on the accurate knowledge of the weights actually batched.
thermal cracking as the cement hydrates and then cools to am- If the density test is to be used to estimate air content directly
bient conditions, and to minimum temperature specifications, from proportions, then the specific gravities of the aggregates
imposed in cold weather to ensure adequate setting and and cement (usually taken as 3.15 when portland cement is
strength performance. used) must be accurately known. Slight errors can result in
Tests for air content and density are performed on fresh significant error in estimating the air content. For this reason,
concrete to provide a control of these properties in the hardened the gravimetric determination of air is much more useful to

1
Key Accounts Technical Manager, Grace Construction Products, Cambridge, MA 02140.
2
The older term unit weight is now deprecated, but is included here as reference. Density will be used in the remainder of this document.

73
74 TESTS AND PROPERTIES OF CONCRETE

monitor variation in air content of mixtures, the reference air air is found to be within specifications, but the slump is too low,
content of which is determined by other methods. Also, when air and the specifications allow addition of retempering water. This
contents vary in concretes being controlled to a fixed slump, the water is added, the slump re-run and found to be within speci-
density changes may be less than predicted by air content fications, but the air content of the retempered concrete is not
changes alone, due to the reduction of required water for equal determined. The addition of water and further mixing alter the
slump in many concrete mixtures as air content increases, and air content, and subsequent determinations of the hardened air
the gravimetric method can be used only if all the water weights contents may not agree with the results recorded for the fresh
are accurately known. concrete properties. Questions about the accuracy of the air
It must be emphasized that all methods to be discussed here content measurements then arise, solely because the concrete
measure the total air in the concrete, subject to the limitations tested in the fresh state was altered after testing but before be-
of each method. While the adjectives “entrapped” and “en- ing allowed to harden. The job specifications should make clear
trained” are sometimes applied in technical discussion, and un- what steps are to be taken in this situation.
fortunately sometimes in contract documents, to distinguish be-
tween large and small air voids, respectively, these methods Test Methods
cannot in any way make such a distinction. Furthermore, the
fresh concrete air contents specified by the American Concrete Temperature
Institute (ACI) and others for concrete durability under freeze- Concrete temperature measurement is determined in accor-
thaw are uniformly the total measured air contents, yet are re- dance with ASTM Test Methods for Temperature of Freshly-
ferred to as entrained air. While some conclusions may be made Mixed Portland-Cement Concrete (C 1064). This test method
for research purposes about the expected amount of small air describes the types and precision (0.5°C) required of the tem-
voids by comparing the air contents of similar air-entrained and perature measuring devices to be used. Although liquid in glass
non-air-entrained concretes, there is no clear dividing line even thermometers may be used, the conditions of field concrete
by microscopic examination, and such distinctions are purely testing in many cases make metal dial thermometers more
arbitrary. In some cases, contract language for “entrained air practical, and these are most frequently employed. Yearly cali-
content” has been interpreted to mean that the specified level bration against 0.2°C precision reference temperature meas-
should be in addition to the base air content of non-air-entrained uring devices is required, using two temperatures at least 15°C
concrete. This is an inappropriate application of these fresh con- apart; but due to the ease with which some metal dial ther-
crete methods, since no such distinction is possible. mometers can lose calibration, it is recommended that a single
temperature comparison against a reference thermometer be
Sampling performed daily as an equipment check, with a full recalibra-
These methods are regularly applied under a variety of condi- tion being run if deviation is noted. Formerly, precision liquid
tions—in the laboratory during mixture proportion development, in glass reference thermometers was required by the test
at the plant for control of production, at the job site discharge method, but now direct-reading resistance thermometers are
point—to ensure compliance with specifications, and, at the acceptable, so long as their calibration is done at least yearly,
point of placement, to best estimate the resulting hardened con- and is traceable to NIST standards.
crete properties. Due to the limitations of these various loca- Another key provision of ASTM C 1064 relates to sample
tions, a clear understanding of proper sampling procedures and size. Clearly, if the sample is small enough to gain or lose sig-
acceptable deviations from test specifications is needed. nificant heat to its surroundings during the time of testing, the
For example, all the methods to be discussed nominally re- result will not be representative of the mass of the concrete. Ac-
quire that concrete be sampled according to ASTM Practice for cordingly, ASTM C 1064 calls for the sample to be large enough
Sampling Freshly Mixed Concrete (C 172), which requires that for a minimum of 75 mm of concrete to surround the temper-
the concrete come from two or more portions of the load taken ature-measuring device.
at regularly spaced intervals during discharge of the middle of While the test method does not now explicitly indicate this,
the batch. This is clearly not possible if a truck is being sam- it is obvious that the measuring device should remain embed-
pled prior to discharge for compliance with specifications, ded in the concrete while the reading is being taken, as rapid
which is perhaps the most common application of these meth- evaporative cooling could reduce the temperature read if the
ods. This represents an inconsistency that needs resolution; device were withdrawn prior to reading. Language to clarify
therefore it is best if the specification documents clearly out- this is currently in the balloting process.
line the sampling requirements. In practice, samples must be Finally, aggregate of 75 mm or greater in size may cause
taken from the initial discharge, but enough concrete must the concrete to take as much as 20 min to temperature equili-
be allowed to discharge, usually about 10 %, to obtain concrete brate, especially if there are large differences between the ag-
representative of the load, and all the other directions of care gregate and other material temperatures. Thus in large aggre-
described in ASTM C 172 must be adhered to. If, after testing, gate concretes sampled early, replicate readings over time are
obvious changes in the concrete being discharged are noticed, warranted to ensure that the equilibrium is reached. No guid-
this must be noted and good practice indicates re-testing. Re- ance is given in the method as to how far apart these should
solving this sampling time issue takes a clear eye for concrete, be, but, for practical purposes, if after a 5 min delay the tem-
good judgment, and is sometimes a source of contention on perature remains within 1°C, effective equilibrium is reached.
job sites.
One source of error in sampling during application of these Air Content
methods deserves special mention. Frequently, one or more of The three tests for air content of fresh concrete, using pres-
the air content methods is applied to a load of concrete arriving sure, volumetric, and gravimetric methods, each have their
at a job site at the same time the slump test, ASTM Test Method own advantages and limitations. We will discuss each in order,
for Slump of Hydraulic Cement Concrete (C 143), is run. The with focus on the proper selection of method.
 ROBERTS ON AIR CONTENT, TEMPERATURE, DENSITY, AND YIELD 75

Pressure Air Measurement true air content. This is corrected for by application of the re-
The pressure method for determining air content of fresh con- quired aggregate correction factor, the lack of which applica-
crete is based on Boyle’s law, which states that the volume of a tion is a common flaw in observed field testing procedures.
gas is inversely related to the pressure. By applying pressure to While this correction factor will compensate for the air in the
a known volume of concrete containing air voids, the voids are aggregate, it is not appropriate to apply the pressure test to
compressed, the concrete is reduced in volume, and the vol- concrete containing lightweight aggregates in which the
ume change can be measured and related to the initial volume. aggregate correction factor exceeds about 0.5 %, because with
Knowledge of the pressure difference allows calculation of the aggregate of this high porosity, relatively minor changes in ag-
total volume of air. This principle was first applied by Klein gregate porosity will lead to significant variation in measured
and Walker in 1946 [6], while Menzel [7] refined the apparatus air content.
and proposed a standard test procedure. This method has the Due to the use of water, the concrete must be discarded at
advantage that no knowledge of specific gravities or batch the end of the test.
weights is required to obtain the required answer. A degree of controversy has arisen regarding the ability of
Two types of meters are defined in ASTM Test Method for the pressure meter to measure air content when the air voids
Air Content of Freshly Mixed Concrete by the Pressure Method are very small. This is usually perceived to occur when the air
(C 231). The Type A meter relies on direct measurement of the contents are high and may be to some extent dependent on the
volume change, by means of a column of water above the type of air-entraining agent used. Several authors have re-
known volume of concrete. A calibrated sight tube allows ported significantly higher air contents in hardened concrete
measurement of the volume reduction directly as pressure is measured according to ASTM Practice for Microscopical De-
applied. This method is very straightforward, but has the termination of Air-Void Content and Parameters of the Air-Void
drawback that recalibration is necessary if barometric pressure System in Hardened Concrete (C 457) than was obtained using
or elevation changes exceeding 183 m occur. A change of 183 the pressure method [8–12], as shown in Fig. 1. Meilenz et al.
m is approximately equal to a 2 % change in barometric discussed the theoretical possibility of air in the smaller, higher
pressure at sea level. pressure air voids dissolving, then coming out of solution in
In the Type B air meter, a known volume of air at an es- lower pressure, larger voids, leading to a net increase in air vol-
tablished higher pressure is allowed to equilibrate with the ume [13]. Hover [14] calculated the effect of the incompressi-
known volume of concrete in a sealed container. The drop in bility of very small air voids. Other authors have failed to find
pressure measured in the high-pressure air chamber can be re- such an increase [15].
lated to the amount of air within the concrete. This method Closer examination of the data indicates the problem may
does not have the ambient air pressure recalibration require- largely be a manifestation of the sampling problem mentioned
ment of the Type A meter, but the complexity of valves and earlier. While Hover’s calculations [14] show that significant
seals make the apparatus prone to leakage; operators should underestimation could occur if the very small air voids com-
be prepared with tools and replacement parts for repair, and prised a significant portion of the air volume, this is not the
be sensitive to instability in dial readings that may signify leak- case, even in concretes of high air content [15]. Thus, this un-
age and therefore incorrect results. A common mistake is to derestimation is unlikely to exceed around 1 % air content, not
close the petcocks prior to pressurizing the high-pressure the 3–6 % reported by some.
chamber. If this is done, it is possible to not observe a leak in Ozyldirim [16] followed up on field reports of higher air in
the needle valve that would normally be noticed due to air bub- the hardened concrete with a thorough test program in which
bles escaping the open petcocks. This would result in incor- all types of fresh concrete air content tests were compared with
rectly low air content readings. hardened air results from ASTM C 457. In no case did the dif-
This method requires complete consolidation of the con- ferences display the extreme variation previously reported.
crete in the bowl; any large air voids due to lack of consolida- Only when the concrete was retempered and the hardened air
tion will be measured as air content of the concrete. To ensure contents compared to the fresh concrete results prior to retem-
proper consolidation, rodding is required above a slump of 75 pering were significant differences seen. Re-examination of the
mm, internal vibration below 25 mm, and either in between. data included in reports of such underestimation problems [10]
Vibration should cease when all the coarse aggregate is sub-
merged and the surface takes on a smooth, glistening appear-
ance. Extreme care must be taken to avoid removal of the
intentionally entrained air by over-vibration, and for this
reason vibration is rarely used in slumps above 25 mm.
Concretes containing aggregates larger than 50 mm must be
screened using the 37.5 mm sieve prior to testing, since repre-
sentative sampling becomes difficult with larger aggregate.
Strike-off of the concrete is possible with either a bar or
a strike-off plate. In the case where only a pressure air test is
being run, the precise volume of concrete is less critical, and
the bar is acceptable. When the air meter base is being used
for determination of density, a strike-off plate must be used,
as described later.
The pressure method is limited to use with concretes con-
taining relatively dense aggregates. Air in the interconnected
porosity within the aggregate particles will be compressed just Fig. 1—Pressure versus microscopic air content (data
as air within the cement paste, thus indicating a higher than from Ref 10).
76 TESTS AND PROPERTIES OF CONCRETE

level from its original mark provides a measure of the air con-
tent. During agitation, it should be possible to hear the aggre-
gate rolling around within the chamber. To ensure that all the
concrete is dislodged from the base, it is frequently necessary
to tip the apparatus upside down, but this should be done
briefly, as aggregate particles can lodge in the neck of the me-
ter. If this occurs, they usually may be dislodged by sloshing
the apparatus from side to side. The agitation must be repeated
until there is no further drop in the water level. As defined in
ASTM C 173/C 173M, this is considered to be when there is less
than 0.25 % change in the measured air content.
This method has the advantage that air trapped in the
aggregates has no impact on the test results, and thus is the
method of choice for lightweight aggregate concretes. How-
ever, it has two significant disadvantages. First, the effort in-
Fig. 2—Pressure versus gravimetric air content (data
from Ref 10). volved in agitating the filled apparatus is significant and can
lead to severe operator fatigue over a work day. The fatigue fac-
tor can be reduced by using smaller apparatus with containers
shows that the pressure test corresponds well with the gravi- of 2.0 L capacity as are now allowed. Second, it is absolutely vi-
metric method results (see Fig. 2). This would seem to preclude tal for accurate measurements that the agitation indeed be re-
a specific problem with the pressure test, although some change peated until no significant change in air content occurs. These
in the actual concrete from fresh to hardened state could be two factors work in opposition to each other in practice, espe-
invoked. cially in high cement factor, sticky concretes having high air
Further consideration makes this unlikely. The air differ- contents.
ences reported are up to 6 %. Accepting that the fresh concrete The operator should be alert to any free water on or
air content measurements are accurate, a net growth of the around the apparatus, indicative of a failure to properly seal
concrete would be required to cause this measured difference. the unit. Such a leak will result in a higher air content being
This would result in up to an 18 mm increase in height of a measured than actually is present in the concrete. Fine parti-
typical 300 mm cylinder. Such growth is not reported. The an- cles adhering to gaskets and the cap frequently cause such
swer may well lie in the difficulties associated with the ASTM leaks.
C 457 test, which are covered elsewhere [3]. Some preliminary A further constraint is that with high-air-content concretes
data showing this effect have been presented [17], as seen in it may be very difficult to dispel residual bubbles of air in the
Table 1. Poor sample surface preparation tends to erode the neck of the apparatus. It is for this reason that the addition of
void edges, effectively making them seem larger, and thus con- isopropyl alcohol, previously an option, is now required in the
tributing higher measured air contents. test method. Generally the higher the air content and the stick-
ier the concrete, the more isopropyl alcohol will be found
Volumetric Method necessary to dispel the bubbles during the first rocking and
The ASTM Test Method for Air Content of Freshly Mixed Con- rolling, as described in notes 2 and 3. When needed, further
crete by the Volumetric Method (C 173/C 173M) measures air increments of isopropyl alcohol may be added using the small
content by washing the air out of the concrete through agita- cup provided, the volume of which is equal to 1 % air content
tion of the concrete with an excess of water containing some as measured in the calibrated neck. Care must then be exer-
level of isopropyl alcohol, primarily by a rolling action. A cised to note the volumes added and add to the air content fi-
known volume of concrete is covered by water, an amount of nally read as described in the calculation section of ASTM C
isopropyl alcohol appropriate to the type of concrete added, 173/C 173M. For a consistent class of concrete, generally an
and the water level adjusted to a zero mark in a calibrated clear ideal level of isopropyl alcohol will be found and it should be
neck on the described apparatus. The apparatus is sealed; then added to each test. When the test operator is monitoring more
the concrete, alcohol, and water are agitated by rocking and than one class of concrete on a job site, very different levels of
rolling until the air in the original concrete is displaced and isopropyl alcohol may be needed, and it will require extreme
rises to the top of the neck of the apparatus. The drop in water care to remember to add the different amounts to different
concrete classes.
Recent reports of problems with volumetric measure-
ment of high-air-content concretes have paralleled the prob-
TABLE 1—Effect of Surface Preparation on lems attributed to the pressure method as described earlier,
Measured Air Content (Data from Ref 17) and may be related to the wider application of air-entrained
high cement factor concrete in recent years. In most cases,
Air Content, % the air contents measured by pressure and volumetric
methods agree with each other reasonably well, and with the
ASTM C 457 gravimetric method, leading to the conclusion that the
ASTM C 231 Good Polish Bad Polish problem is not in these test methods. Figure 3, based on data
8.0 7.7 11.0 from one paper in which a lack of correspondence with hard-
8.8 8.4 11.2 ened air contents is shown [10], confirms the good cross-
7.5 7.23 8.69 correlation of fresh concrete methods. This is in line with
previous experience [18,19].
 ROBERTS ON AIR CONTENT, TEMPERATURE, DENSITY, AND YIELD 77

The top surface of the concrete must be obtained with a

strike-off plate. This is best accomplished by covering about
two-thirds of the surface with the plate, pressing down, and
withdrawing with a sawing motion. Then replace the plate on
the smoothed two-thirds, press down, and advance it over the
unsmoothed portion with a sawing motion. Striking off with
a bar is specifically not permitted, as it leads to a less precise
filling. Do not overfill the measure then push the coarse ag-
gregate down into the measure with the strike-off plate, as
this results in mortar being squeezed out, giving a nonrepre-
sentative sample.
Subsequent calculation of volume of concrete produced
per batch, the relative yield, actual cement content per unit vol-
Fig. 3—Fresh versus microscopic air content (data from ume, and air content are outlined clearly in ASTM C 138/C
Ref 10). 138M, and will not be repeated here. Each calculation requires
precise knowledge of batch weights and material properties
not readily available in the field. When the gravimetric proce-
However, in some instances [20], it does appear that with dure for determining air content of fresh concrete is used, the
extremely sticky concretes it can be so difficult to wash out all density of concrete as determined is compared with the theo-
the air that at the very least the volumetric method takes an im- retical density of air-free concrete. This is calculated from the
practically long time to complete (this information was previ- weight and specific gravity of each ingredient of the concrete
ous to the inclusion of isopropyl alcohol). Therefore, care mixture, and requires knowledge of the saturated, surface dry
should be exercised in the use of this method when such con- specific gravities and their moisture contents as batched. This
cretes are encountered, with special attention paid to using the can lead to serious errors. For example, when there is a 2 % er-
proper amount of isopropyl alcohol, to continuation of agita- ror in the moisture content of the fine aggregate or an error of
tion until no air increases are registered, and to the checking 0.04 in the aggregate specific gravities, an approximate 1 % er-
of the bowl after completion to ensure that all the concrete was ror in the air content computed according to ASTM C 138/C
loosened from the base. If such attention results in unaccept- 138M will result. Thus, ASTM C 138/C 138M is not appropriate
ably long test times, alternate methods should be employed. for determining air contents of lightweight aggregate concrete,
Such situations may require two test technicians to ensure that since aggregate-specific gravities and moisture contents are
fatigue does not unduly influence the willingness to ensure both subject to wide variation.
that thorough separation of the concrete has taken place. Further, when air contents increase, required water con-
tents will be reduced in most concretes. The amount of reduc-
Gravimetric Method tion is greatest in the lower cement content mixtures [21], due
In the gravimetric procedure, the density of concrete is deter- to the air substituting for the water needed to fill in the voids
mined in accordance with ASTM Test Method for Density (Unit between the fine aggregate particles. Accordingly, air content
Weight), Yield, and Air Content (Gravimetric) of Concrete (C increases measured by ASTM C 138/C 138M would be reduced
138/C 138M). The concrete is placed in a container of known by approximately one-third unless completely accurate water
volumetric capacity and weighed. Using the known volume, the content information were available.
density of the concrete is calculated and, from this, the air con- Due to these uncertainties associated with direct calcula-
tent and yield can be determined provided the requisite mix- tion of air by the gravimetric method, it is the author’s experi-
ture proportion information is available. This dependence on ence that the greatest value is realized when density is used as
extraneous information is a major limitation of this method a cross-check to other methods of air measurement, and that
when used for other than density. in that regard it is extremely valuable. Testing of concrete air
Aggregate size determines the minimum capacity of the in the field is subject to a number of errors, many of which are
measures used, since as the aggregate increases in size, edge enumerated above. The ability to have a simple cross-check by
effects tend to bias the sampling toward lesser aggregate con-
tents. The minima given in ASTM C 138/C 138M are repro-
duced in Table 2, from which it is important to note the very
high volumes required when aggregate size exceeds 25 mm.
Safety considerations require that means be provided for TABLE 2—Minimum Capacity of Measures
movement of the buckets on and off the scales without undue for Use in ASTM C 138
strain on the individuals conducting the test.
As with the other air content methods discussed here, the Nominal Maximum Size Minimum Measure
results of this test are completely dependent on proper filling of Coarse Aggregate, Capacity,
mm (in.) L (ft3)
and consolidation of the measure. Rodding may be used above
25-mm slump; vibration is required for lower slumps. Vibra-
25.0 (1) 6 (0.2)
tion may not be used if the slump exceeds 75 mm. ASTM C 138 37.5 (1 1/2) 11 (0.4)
requires that measures smaller than 11 L be consolidated by 50 (2) 14 (0.5)
rodding, due to the possibility of excessive loss of entrained air, 75 (3) 28 (1.0)
while ASTM C 231 permits vibration at this size. Thus, the 112 (4) 71 (2.5)
smallest measures cannot be used in accordance with ASTM C 150 (6) 99 (3.5)
138/C 138M for concretes of less than 25-mm slump.
78 TESTS AND PROPERTIES OF CONCRETE

Fig. 4—Air versus density data from field testing of different trucks during one day of
field testing.

which problems with the apparatus or method can be found in Fig. 3. Figure 4 shows data from a typical well-run field test
quickly can prevent incorrect rejection of concrete, and also in- of a number of different admixtures yielding different air con-
correct adjustment of air-entraining agent doses. Figure 2 is an tents, with the same base concrete mixture. The dotted lines
illustration of such a use. represent the expected relationships, while the fitted regres-
The author is aware of a number of cases in which con- sion line shows by its slope that the test is well controlled. On
crete ultimately had to be removed because faulty pressure close inspection, one mix, at 2.9 % air, is somewhat question-
method air measurement at the job site indicated low air and able and would bear repeating; Fig. 5 shows the improvement
plant dosage adjustment was made. Later, when strength prob- when that mix is eliminated from the dataset. Having prepared
lems led to determination of high air in the hardened concrete, blank graphs at the job site on which the data can be plotted
the faulty testing became manifest. This could have been easily makes this approach easier.
precluded in the case of pressure meter testing by the simple Since no water comes in contact with concrete tested ac-
expedient of using the air pot base to determine the density, cording to ASTM C 138/C 138M, the concrete may be used in
and keeping a running simple air cross-check graph as shown other testing.

Fig. 5—Air versus density data from field testing of different trucks during one day of
field testing, after elimination of one data point.
 ROBERTS ON AIR CONTENT, TEMPERATURE, DENSITY, AND YIELD 79

Summary and Future Trends [7] Menzel, C. A., “Procedures for Determining the Air Content of
Freshly-Mixed Concrete by the Rolling and Pressure Methods,”
Proceedings, ASTM International, Vol. 47, 1947, p. 833.
Tests run in accordance with these methods will yield accurate
[8] Burg, G. R. U., “Slump Loss, Air Loss, and Field Performance of
and useful results providing the operator follows the methods
Concrete,” Journal, American Concrete Institute, Vol. 80, No. 4,
carefully and care is taken to avoid procedural pitfalls or use July-Aug. 1983, pp. 332–339.
of methods inappropriate in certain situations, as described [9] Gay, F. T., “A Factor Which May Affect Differences in the
earlier. The specification writer should take cognizance of Determined Air Content of Plastic and Hardened Air-Entrained
these limitations and not ask more of the methods or the op- Concrete,” Proceedings, Fourth International Conference on
erators than they can provide. Cement Microscopy, Las Vegas, 28 March–1 April 1982, Interna-
The growing emphasis on high quality, durable concrete, tional Cement Microscopy Association, Duncanville, TX, pp.
coupled with advances in digital electronics, are changing this 296–292.
situation. Numerous in situ temperature measurement systems [10] Gay, F. T., “The Effect of Mix Temperature on Air Content
are available, and are employed for maturity methods under and Spacing Factors of Hardened Concrete Mixes with Stan-
ASTM C 1074. A complex, but potentially valuable test known as dardized Additions of Air-Entraining Agent,” Proceedings,
the Automatic Void Analyzer, or AVA, has been developed [22], Seventh International Conference on Cement Microscopy,
Fort Worth, 25–28 March 1985, International Cement
based on buoyancy of released air voids. This shows promise in
Microscopy Association, Duncanville, TX, pp. 305–315.
allowing determination not only of air content but also of air-
[11] Hover, K. C., “Some Recent Problems with Air-Entrained
void size in fresh concrete. There are, however, significant issues Concrete,” Cement, Concrete, and Aggregates, Vol. 11, No. 1,
of sampling reproducibility, time delay if used as an acceptance Summer 1989, pp. 67–72.
method, and experimental issues such as sensitivity to vibration, [12] Khayat, K. H. and Nasser, K. W., “Comparison of Air Contents in
which requires setting up the equipment remote from job sites Fresh and Hardened Concretes Using Different Airmeters,”
or using special vibration isolation techniques. Current use is Cement, Concrete, and Aggregates, Vol. 13, No. 1, Summer
generally for mix qualification, with job site control still by total 1991, pp. 16–17.
air content methods [23]. Evaluation of its reproducibility is cur- [13] Meilenz, R. C., Wolkodorff, V. E., Backstrom, J. E., and Flack,
rently underway, and possible biases between the results of the H. L., “Origin, Evolution, and Effects of the Air Void System in
method and actual hardened air parameters per ASTM C 457 Concrete, Part 1—Entrained Air in Unhardened Concrete,”
are not well understood. Some specifying agencies are currently Proceedings, American Concrete Institute, Vol. 30, No. 1, July
using the method. Good practice would indicate that different 1955, pp. 95–122.
mixes should have the AVA results calibrated against actual [14] Hover, K. C., “Analytical Investigation of the Influence of
Bubble Size on the Determination of Air Content in Fresh
ASTM C 457 prior to setting ranges in specifications, but this has
Concrete,” Cement, Concrete, and Aggregates, Vol. 10, No. 1,
not always been the case. Focus on this method is likely to bring Summer 1988, pp. 29–34.
significant advances in the understanding of its use and appro- [15] Roberts, L. R. and Scheiner, P., “Microprocessor-Based Linear
priate application in the near future. Traverse Apparatus for Air-Void Distribution Analysis,”
It is expected that the effort to improve the convenience Proceedings, Third International Conference on Cement
and rapidity of test methods will yield significant advances in the Microscopy, Houston, 16–19 March 1981, International Cement
next decade. For now, the methods discussed here provide ex- Microscopy Association, Duncanville, TX, pp. 211–227.
cellent quality control of fresh concrete when properly applied. [16] Ozyldirim, C., “Comparison of the Air Contents of Freshly Mixed
and Hardened Concretes,” Cement, Concrete, and Aggregates,
Vol. 13, No. 1, Summer 1991, pp. 11–17.
References [17] Roberts, L. R. and Gaynor, R. D., discussion of paper by
[1] Greening, N. A., “Some Causes for Variation in Required Ozyldirim, C., “Comparison of the Air Contents of Freshly Mixed
Amount of Air-Entraining Agent in Portland Cement Mortars,” and Hardened Concretes,” Cement, Concrete, and Aggregates,
Journal, PCA Research and Development Laboratories, Vol. 9, Vol. 13, No. 1, Summer 1991, pp. 16–17.
No. 2, May 1967, pp. 22–36. [18] Proceedings, Symposium on Measurement of Entrained Air in
[2] Whiting, D. A. and Nagi, M. A., Manual on Control of Air in Concrete, ASTM International, Vol. 47, 1947, p. 832.
Concrete, Portland Cement Association, Skokie, IL, 1998. [19] Britton, R. E., “Report of Investigation of Different Methods for
[3] Hover, K. C., “Air Content and Unit Weight,” in this publication. Determining the Amount of Air Entrained in Fresh Concrete,”
[4] Jeknavorian, A. A., “Air Entraining Admixtures,” in this Pennsylvania Slag Association, 11 April 1949.
publication. [20] Gaynor, R. D., personal communication, 1992.
[5] Kosmatka, S. H., Kerkhoff, B., and Panarese, W. C., Design and [21] Klieger, P., “Effect of Entrained Air on Concretes Made with So-
Control of Concrete Mixtures, Fourteenth Edition, Portland Called ‘Sand Gravel’ Aggregates,” Journal, American Concrete
Cement Association, Skokie, IL, 2002. Institute, Oct. 1948; Proceedings, Vol. 42, p. 629.
[6] Klein, W. H. and Walker, S., “A Method for Direct Measurement [22] Jensen, B. J., U.S. Patent No. 4,967,588, 6 Nov. 1990.
of Entrained Air in Concrete,” Journal, American Concrete [23] FHWA, Priority Market Ready Technologies and Innovations: Air
Institute, Vol. 42, p. 657. Void Analyzer, www.fhwa.dot.gov/rnt4u/ti/pdfs/airvoid.pdf.
10
Making and Curing Concrete Specimens

Joseph F. Lamond1

Preface The strength of concrete, in compression, tension, and

shear, or a combination of these, has in most cases a direct in-
THIS CHAPTER COVERS THE IMPORTANCE OF PRO- fluence on the load-carrying capacity of both plain and rein-
perty making and curing concrete test specimens in both the forced structures. In most structural applications, concrete is
field and the laboratory. This subject was covered in ASTM STP designed primarily to resist compressive stresses. In those cases
169 [1] in the chapter Static and Fatigue Strength authored by where strength in tension or in shear is of primary importance,
C. E. Kelser and C. P. Siess. In ASTM STP 169A [2], this became empirical relationships to the compressive strength are often
a separate chapter authored by T. B. Kennedy. In ASTM STP used. However, strength may not necessarily be the most criti-
169B [3], this chapter was authored by R. F. Adams. In ASTM cal factor in the overall performance of the concrete. For
STP 169C [4], this chapter was under the present authorship. example, the need for acceptable durability may impose lower
water-cement ratios than required to meet the strength re-
Introduction quirements. In such cases, the actual compressive strength of
the designed mixture may be in excess of structural require-
The 1914 Committee Report [5] is the basis for the presently ments. Specimens cast in the field are most often used to
accepted procedures for testing concrete cylinders and beams determine the compressive strength in accordance with ASTM
to determine the compressive or flexural strength. Two ASTM Test Method for Compressive Strength of Cylindrical Concrete
standards have been developed for making and curing test Specimens (C 39). The test results may be used as an indication
specimens. One is ASTM Practice for Making and Curing Con- of the verification of other required properties or characteris-
crete Test Specimens in the Field (C 31) and was originally tics of the concrete as delivered or designed. The most common
published in 1920 and updated periodically. The other is other use is to determine the splitting tensile strength in accor-
ASTM Practice for Making and Curing Concrete Test Speci- dance with ASTM Test Method for Splitting Tensile Strength of
mens in the Laboratory (C 192) and was originally published Cylindrical Concrete Specimens (C 496). Pavements are often
in 1944 and updated periodically. Since publication of ASTM designed for tensile stresses in flexure in the concrete slab. Traf-
STP 169C [4], C 31 has been revised six times and C 192 has fic loads have been found to induce critical tensile stresses in a
been revised four times. These two standards have had con- longitudinal direction at the top of the slab near the transverse
siderable use in concrete research and concrete construction. joint, and in the transverse direction near the longitudinal
Specimens have to be made and cured properly since depar- edges. Accordingly, a flexural strength test, either ASTM Test
tures from the standard procedures to make and cure speci- Method for Flexural Strength of Concrete (Using Simple Beam
mens will affect the test results. with Third-Point Loading) (C 78) or ASTM Test Method for Flex-
ural Strength of Concrete (Using Simple Beam with Center-
Applications Point Loading) (C 293), is used as an acceptance test for most
pavement concrete. However, due to inherent variability of the
The strength of concrete is one of its most important and useful flexural strength tests many agencies rely on compressive
properties and one of the most easily measured. The strength in- strength tests using an established correlation or empirical
dicated by the specimens is affected by many variables encoun- relationship to estimate flexural strength.
tered in making and curing test specimens. These include size of It is important to keep in mind that the test specimens in-
the aggregate, size and shape of the test specimen, consolidation dicate the potential rather than the actual strength of the con-
of the concrete, type of mold, capping procedure, curing, and crete in the structure. To be meaningful, conclusions on strength
temperature. The effect that any of these variables has on the ap- must be derived from a pattern of tests from which the charac-
parent strength of the specimen will often vary depending on teristics of the concrete can be estimated with reasonable accu-
the particular circumstances. Among the many who have writ- racy. An insufficient number of tests will result in unreliable con-
ten about factors that influence the strength of concrete are clusions. Statistical procedures provide tools of considerable
Price [6], Sparkes [7], ACI Manual of Concrete Inspection [8], value in evaluating results of strength tests. ACI Recommended
Walker and Bloem [9], and Richardson [10]. Practice 214 [13] discusses variations that occur in the strength

1
Consulting Engineer, Jeffersonton, VA 22724.

80
 LAMOND ON MAKING AND CURING CONCRETE SPECIMENS 81

of concrete, and presents statistical procedures and control piece reusable molds are made most commonly of plastic while
charts that are useful in the interpretation of these variations. assembled molds are made of steel, iron, or brass. The assem-
bled mold must be watertight; therefore, suitable sealants must
Testing Personnel be used where necessary to prevent leakage through the joints.
There is increasing emphasis and a requirement in many build- The single-use mold may be made of any material that passes
ing codes, political jurisdictions, project specifications, and the test requirements of ASTM C 470. The most prominent sin-
ASTM Specification for Ready-Mixed Concrete (C 94) that per- gle-use mold materials are plastic with treated paper products,
sonnel who make concrete test specimens be certified. Over metal molds, and other suitable material being used less fre-
59 000 technicians have been certified under the American quently. At least three single-use molds and reusable plastic
Concrete Institute’s certification program for “Concrete Field molds shall be selected at random from each shipment by the
Testing Technicians, Grade I.” Other equivalent certification purchaser to ensure that molds are in compliance with C 470.
programs that test technicians’ knowledge of the standard and Richardson [10] reported on a number of studies and ob-
ability to perform the test also exist and are used. servations on the rigidity, water absorption, and expansion of
Adams [3] discussed the importance that personnel are molds made with various kinds of materials. Reusable steel
properly trained in making specimens and the details of the molds not properly sealed had a tendency to leak at the joints.
various procedures are precisely followed. Only in this way can For high-strength concrete, compressive strength test results
meaningful and reproducible test results, which are not open may be lower with plastic molds than reusable steel molds [11].
to question, be obtained. The significant differences between When not properly waterproofed, treated paper products
trained and untrained personnel were reported by Wagner (cardboard molds) may suffer from absorption and elongation
[14]. Wagner’s study showed that strength tests made from the problems. All molds can suffer from being out of round. This
same concrete by trained personnel were higher and more uni- was noted even in some molds made with light-gage metal side-
form than those made by untrained personnel. walls. The shape tends to be oval at the unsupported tops for
In addition to proper training, supervising engineers and plastic and cardboard molds. The sidewalls of molds should be
technicians responsible for technicians making and curing test of a sufficient stiffness to prevent the mold from becoming out
specimens must be thoroughly familiar with the test procedures. of round. This can be attained with sufficient wall thickness
Personnel should have the required tools and equipment for alone or in combination with a stiffened top. The molds that re-
making and curing specimens. The standards must be reviewed sult in out-of-shape cylinders have an effect on the proper cap-
annually for any changes. Also, supervisors should review peri- ping and the specimens not meeting dimensional tolerances.
odically the making and curing of test specimens and the pro-
cedure to verify that it is being done correctly. When strength Making and Curing Test Specimens
test results fail to meet a specification requirement and rejection in the Field (ASTM C 31)
of the concrete is considered, the tests are almost always ques-
tioned, particularly if it can be shown that the person making the This practice is a definitive procedure for performing specific
test had not complied with all the details of the test procedure. operations to produce a test specimen. These specimens are
used in various test procedures to determine a property of con-
Molds crete. Specimens are fabricated using concrete delivered to a
project site sampled from the transportation unit prior to plac-
Specifications ing. Specimens are fabricated in cylindrical or beam molds and
The molds for casting concrete specimens must have the fol- consolidated by either rodding or vibration. The concrete slump
lowing properties: (1) made of a nonabsorbent material, (2) measured by ASTM Test Method for Slump of Hydraulic Ce-
nonreactive with the concrete (aluminum and magnesium are ment Concrete (C 143) is used to select the consolidation proce-
examples of reactive materials), (3) hold their dimensions and dure. If the slump is greater than 1 in. (25 mm), the concrete is
shape, and (4) be watertight. Compressive strength and split- either consolidated by rodding or vibration. If the slump is less
ting tensile test specimens are cast and hardened in a vertical than 1 in. (25 mm), specimens must be vibrated. An example of
position, with the length equal to twice the diameter. Flexural a concrete with a slump less than 1 in. (25 mm) would be roller-
strength specimens are rectangular beams cast and hardened compacted concrete. Roller-compacted concrete would be con-
with the long axis horizontal. Molds for casting cylindrical solidated using vibration procedures in ASTM Practice Making
specimens are covered in ASTM Specification for Molds for Roller-Compacted Concrete in Cylinder Molds Using a Vibrating
Forming Concrete Test Cylinders Vertically (C 470). Details for Table (C 1176) or ASTM Practice Molding Roller-Compacted
field-cast beam molds are covered in ASTM C 31. Details of Concrete in Cylinder Molds Using a Vibrating Hammer (C 1435).
molds for laboratory-cast flexural strength and freeze-thaw Other concretes that cannot be consolidated by rodding or vi-
beams are covered in ASTM C 192. Creep specimens are cylin- bration, or requiring other sizes and shapes of specimens to rep-
drical specimens cast horizontally. They are covered in ASTM resent the product or structure are not covered by this practice.
C 192 and ASTM Test Method for Creep of Concrete in Com-
pression (C 512). The requirements for molds may be covered Uses
in more than one standard; therefore, all the requirements ASTM C 31 states that specimens may be used to develop infor-
must be considered for the type of specimen being molded. mation for a number of purposes. There are two separate condi-
tions for using cylinders or beams: (1) they must be standard
Types cured and (2) they must be field cured. There have been instances
There are two types of molds, the reusable type and the single- where standard curing and field curing have been combined.
use type. The reusable molds are designed to be used more than This usually occurs when initial standard curing is misinter-
a single time. Reusable molds are made of nonabsorptive ma- preted as field curing. The two most frequent uses are standard
terials and constructed in one piece or several pieces. Single curing for acceptance testing and field curing for removal of
82 TESTS AND PROPERTIES OF CONCRETE

forms or shoring. When strength test results are used for accept- required. Chapter 8 in this volume contains workability infor-
ance testing, the cylinders or beams must be standard cured. Also, mation on the properties of fresh concrete. Chapter 9 in this vol-
they are standard cured when used to check the adequacy of mix- ume contains information on the effects of various air contents
ture proportions for strength or quality control. When strength on the properties of fresh concrete. And Chapter 13 in this vol-
test results are used for form or shore removal requirements, ume contains information on properties of hardened concrete.
cylinders or beams must be field cured. Also, they are field cured If concrete is air entrained, then the air must be controlled
for determination of whether a structure is capable of being put within tolerances. If air-entrained concrete is accidentally air-
in service, for comparison with test results of standard cured entrained, it has to be known and corrected because each
specimens or test results from various in-place test methods, or percent increase in air content reduces the compressive
to determine the adequacy of curing or protection procedures. strength. The concrete temperature test results may help
Various in-place test methods are discussed in ACI 228 [15]. Cur- explain some unusual strength results. The slump, air content,
ing procedures are discussed in the section on Curing Specimens. concrete temperature, and density test results should be used
to assure the concrete is controlled within required tolerances.
Samples
For evaluation of the test results by statistical procedures to be Specimen Sizes
valid, the data must be derived from samples obtained by means It is generally accepted that the diameter of the cylinder should
of a random sampling plan. Chapter 3 of this volume on sam- be at least three times the nominal size of the coarse aggregate.
pling contains information on a random sampling plan. For compressive strength specimens, a 6 by 12 in. (150 by 300
The sampling must be made after all on-site adjustments mm) cylinder is used when the aggregate is smaller than 2 in.
have been made to the mixture proportions including addition of (50 mm) or, when specified, 4 by 8 in. (100 by 200 mm) cylin-
total mixing water and admixtures. The procedures for obtaining ders can be used when the aggregate is smaller than 1 in. (25
representative samples of concrete at a construction project mm). For flexural strength specimens, the standard beam is 6
from different types of delivery equipment are covered in ASTM by 6 in. (150 by 150 mm) in cross section with a length of at
Practice for Sampling Freshly Mixed Concrete (C 172), unless an least 20 in. (500 mm). Other test methods may require test
alternative procedure has been approved. The size of the sample specimens when cast in the field to be a specific size.
for fabricating field test specimens is a minimum of 1 ft3 (28 L). If the aggregate is too large for the size of mold available,
The sample must be representative of the nature and condition the oversize aggregate is usually removed by wet screening as
of the concrete being sampled. The sample is collected by taking described in ASTM C 172. A larger specimen mold is used
two or more portions at regularly spaced intervals during dis- when specified and wet sieving is not permitted. In these cases,
charge of a stationary mixer, truck mixer, or agitator. The elapsed the diameter of the cylinder or the smaller cross-sectional
time between first and final portions should not exceed 15 min. dimension of the beam should be at least three times the nom-
The portions are combined and remixed to ensure uniformity inal size of the coarse aggregate in the concrete. Attention must
and transported to where the test specimens are to be made. be called to the fact that the size of the cylinder itself affects
Molding of the specimens must begin within 15 min after the observed compressive strength; for example, the strength
fabricating the composite sample. The time should be kept as of a cylinder 36 by 72 in. (920 by 1830 mm) may be only 82 %
short as possible and the sample should be protected from con- of the 6 by 12 in. (150 by 300 mm) cylinder [12].
tamination, wind, sun, and other sources of evaporation. The advantages of 4 by 8 in. (100 by 300 mm) cylinders are
Concrete is a hardened mass of heterogeneous materials as follows: 1) they are easier to fabricate, handle, and transport;
and its properties are influenced by a large number of vari- 2) smaller storage space is required; 3) less capping compound
ables related to differences in types and amounts of ingredi- is needed; and 4) higher strength concrete uses a smaller
ents, differences in mixing, transporting, placing, and curing. capacity compression machine. A cylinder smaller in size than
Because of these many variables, methods of checking the 6 by 12 in. (150 by 300 mm) will yield a somewhat greater com-
quality of the concrete must be employed. Strength test speci- pressive strength [6,16]. The difference in strength between 4
mens can only measure the potential strength of concrete in by 8 in. (100 by 200 mm) and 6 by 12 in. (150 by 300 mm)
the structure because of different size and curing conditions cylinders increases with an increase in the strength level of the
between the specimen and the structure. Therefore, multiple concrete [17]. Additional information evaluating cylinder sizes
test results based on a random pattern should be used as a has been published [18–20]. The present C 31 permits the use
basis for judging quality rather than placing reliance on only a of 4 by 8 in. (100 by 300 mm) cylinders and the specifier
few tests to check uniformity and other characteristics of determines it is appropriate, recognizing mixture proportion-
concrete. For this purpose, statistical methods given in ACI ing adjustments and any testing variability.
Recommended Practice 214 [13] should be used.
Richardson [10] pointed out that various specifications re- Making Specimens
quire different numbers of replicate cylinders to be tested at dif- Where the specimens are to be molded is important. It should
fering time intervals, usually 7, 14, and 28 days. It is important to be as near as practicable to where the specimens are to be
make sure that the concrete for a set of cylinders comes from a stored. The molds should be placed on a firm and level surface
single truck. A set of cylinders that does not come from the same that is free of vibrations and other disturbances. Select a small
truck will cause a considerable amount of consternation should tool to fill the molds with concrete that is representative of the
the 14-day measured strength be lower than the 7-day strength. sample and the batch. Select the appropriate rod for the size of
the specimen using a smaller rod for specimens less than 6 in.
Test Data on Specimens (150 mm) in diameter or width. Concrete should be placed in
After all adjustments have been made to the mix including total the mold to the required depth and with the required number
mixing water and admixtures, measure and record the slump, air of layers keeping the coarse aggregate from segregating as it
content, and concrete temperature. Also measure the density, if slides from the scoop, trowel, or shovel. If the specimen is
 LAMOND ON MAKING AND CURING CONCRETE SPECIMENS 83

rodded, carefully count the strokes for each layer. The rod day, 11 % at three days, and 18 % at seven days. Early-age re-
should not strike the bottom of the mold when rodding the first sults may be lower when stored near 60°F (16°C) and higher
layer. If the specimen is to be vibrated, determine the best vi- when stored near 80°F (27°C). For high strength concrete, heat
brator and the best uniform time of vibration for the particular generated during early ages may raise the temperature above
concrete. Vibration may reduce the air content of air-entrained the required storage temperature. Immersion in water satu-
concrete. If air content reduction is suspected, verification can rated with calcium hydroxide may be the easiest method to
be performed by determining the density of the specimen and maintain the required storage temperature.
comparing it to the calculated air-entrained concrete density at During the setting and initial hardening, the concrete can
the measured air content. Carefully follow the procedure in the be damaged by harsh treatment. For traffic-induced vibrations,
practice for rodding or vibrating specimens. The procedure is Harsh and Darwin [22] reported that wet mixtures exhibited as
not expected to produce optimum consolidation but is used in much as a 5 % loss in strength through segregation as opposed
order to permit reproducibility of results with different techni- to a 4 % gain in strength in dry mixtures due to improved
cians. Close any holes left by rodding or vibration after each consolidation.
layer is consolidated by tapping the outside of the mold.
The exposed concrete surface should be finished to pro- Field Curing
duce a flat even surface that is level with the rim or edge of the Field curing procedures are unique to each situation. These
mold and that has no depressions or projections larger than specimens will reflect the influence of ambient conditions on
1/8 in. (3 mm). It is the author’s experience that finishing the the properties of the concrete. Since each condition is differ-
exposed surface of a cylinder is one of the most violated ent, the method selected for moisture and temperature condi-
requirements. This causes poor capping and dimensional tion will be different. Even then, the effect of the moisture and
tolerances in cylinders. temperature would not be the same as on the actual concrete
Mark the specimens to positively identify them and the structure. This would be the case in a relatively massive struc-
concrete they represent. The specimens should be moved, if ture. They give little indication of whether a deficiency is due
necessary, to curing storage with a minimum amount of to the quality of the concrete as delivered or improper han-
handling and immediately after finishing. Cover the top of the dling and curing of the specimens. ACI Standard Practice 308
specimens with a sheet of plastic, seal them in a plastic bag, or [23] gives procedures for checking the adequacy of curing.
seal the top of the specimens with a plastic cap. Caps may leave
depressions in the concrete surface greater than 1/8 in. (3 mm) Transporting
in depth making capping for testing difficult. Wet fabric may Specimens transported to the laboratory for standard final cur-
be used to cover the specimens to help retard evaporation, but ing before 48 h should remain in the molds, then be demolded
the fabric must not be in contact with the surface of the and placed in laboratory standard final curing. Specimens
concrete or cardboard molds. shall not be transported until at least 8 h after final set. If the
specimens are not transported within 48 h the molds should be
Curing Specimens removed within 16 to 32 h and standard final curing used un-
This practice requires either standard curing or field curing til the specimens are transported. Transportation time should
methods. They are not interchangeable. If the specimens are not exceed 4 h. Specimens should be transported in such a
for checking adequacy of mixture proportions, for strength, manner that prevents moisture loss and exposure to freezing
acceptance testing, or quality control, standard initial and final temperatures, and protects them from jarring. Richardson [10]
curing method must be used. If specimens are for determining indicated that rolling and bumping around in the back of a
removal time of forms or shoring, when a structure may be put pickup truck could result in a 7 % loss of strength and drop-
into service, comparison with standard curing, or curing or ping cylinders from waist level can lower strength at least 5 %.
protection requirements, field curing method must be used. Cylinders and beams should be cushioned during transport
Standard curing is exposure of the specimens to standard and handled gently at all times. Care should be taken to ensure
conditions of moisture and temperature from the time of that the moisture condition of field-cured specimens is main-
fabrication to the time of testing. tained throughout the period of transportation.
Field curing is maintaining the specimens as nearly as
possible in the same moisture and temperature conditions as Standard Final Laboratory Curing
the concrete they represent. Standard final laboratory curing is at a temperature of 73.0

3.5°F (23.0
2.0°C) and a moist condition with free water main-
Standard Initial Curing tained on the surface of the specimens. Moist rooms and water
Control of standard curing conditions is very important since tanks are usually used for creating the moist environment. A
variations can dramatically affect the concrete properties and moist room is a “walk-in” storage facility with controlled tem-
test results. perature and relative humidity, commonly called a fog room
The standard initial curing period takes place in a moist when the prescribed relative humidity is achieved by atomiza-
environment with the temperature between 60 to 80°F (16 to tion of water. All fog rooms should be equipped with a record-
27°C) for up to 48 h. When the specified compressive strength ing thermometer. Water storage tanks constructed of noncor-
is 6000 psi (40 MPa) or greater, the standard initial curing tem- roding materials are also permitted. Automatic control of water
perature shall be between 68 and 78°F (20 to 26°C). It may be temperature and recording thermometer with its sensing ele-
necessary to create an environment during the initial curing ment is required in the storage water. The water should be clean
period to provide satisfactory moisture and temperature. In- and saturated with calcium hydroxide. Continuous running wa-
sufficient moisture during the initial curing can lower meas- ter or demineralized water may affect results due to excessive
ured strength. One study [21] showed that even at proper leaching of calcium hydroxide from the concrete specimens and
temperatures, air curing could lower the strength by 8 % at one should not be used in storage tanks. ASTM Specification for
84 TESTS AND PROPERTIES OF CONCRETE

Moist Cabinets, Moist Rooms, and Water Storage Tanks Used in required for molding the specimens. Machine-mixed concrete
the Testing of Hydraulic Cements and Concretes (C 511) covers should be mixed for 3 min after all the ingredients are in the
the requirements for fog rooms and water storage tanks used mixer followed by a 3-min rest and 2 min of final mixing. To
for standard curing. Price [6] stated that water-cured specimens eliminate segregation, deposit concrete onto a clean damp mix-
with a water-cement ratio of 0.55 were about 10 % stronger at 28 ing pan and remix by shovel or trowel. Perform the slump and
days than those cured in a fog room at 100 % relative humidity. temperature tests in accordance with ASTM C 143 and C 1064,
respectively. When required, perform the air-content test in ac-
Making and Curing Test Specimens cordance with ASTM C 231 or C 173 and the yield test in ac-
in the Laboratory (ASTM C 192) cordance with ASTM Test Method for Density (Unit Weight),
Yield, and Air Content (Gravimetric) of Concrete (C 138).
This practice, like ASTM C 31, is also a definitive procedure for Discard concrete used for determination of the air-content test.
performing specific operations that does not produce a test Make the specimens following the procedures in ASTM C 192,
result. The procedures are for a wide variety of purposes, such taking the precautions as previously discussed in the section on
as: (1) mixture proportioning for project concrete, (2) evaluation ASTM C 31. Care should be taken that specimens are cast and
of different mixtures and materials, (3) correlation with nonde- stored in accordance with the applicable test methods.
structive tests, or (4) providing specimens for research purposes.
Committee C9 has over 150 standards and C 192 is referenced Consolidation
in over 40 of these standards. Therefore, when using C 192 it The specimens are consolidated by rodding or internal and
necessary to coordinate with any other applicable standard. external vibration. The selection of the method is similar to
ASTM C 31 using the slump of the concrete as guidance for the
Equipment method to be used. However, a particular method of consoli-
The equipment needed in the laboratory includes the follow- dation may be required by the test method or specification for
ing: molds; rods or vibrators; mallet; small tools; sampling pan, which the specimens are being made. When vibration is per-
mixing pan, or concrete mixer; air-content apparatus; slump mitted or required, either internal or external vibration may be
cone; thermometer; and scales. Hand mixing is permitted but used. When using external vibration, care has to be taken that
not for air-entrained or no-slump concrete. Machine mixing is the mold is rigidly attached to the vibrating unit. Concrete with
preferred, especially when a laboratory regularly mixes low water contents such as roller-compacted concrete are
concrete. Scales for weighing batches of materials and con- covered in C 1176 and C 1435. Specimens consolidated with
crete should be checked for accuracy prior to use and be low water contents may require a surcharge weight on the
within acceptable tolerances. A sampling pan is required to specimen as they are consolidated with external vibration.
receive the entire batch discharged from the concrete mixer. Additional information on consolidating low-water-content
Conformance of the molds to the applicable requirements specimens is in Chapter 51 of this volume.
should be verified prior to mixing the concrete. Different tests
may require different molds. The dimensions of the molds also Finishing and Curing
vary for different tests and usually according to aggregate size. Finish the specimens as required. Cover and cure the specimens
Some test methods require specimens that are other than cylin- for 16 to 32 h prior to the removal of the molds. Concrete with
drical or prismatic in shape. They should be molded following prolonged setting time may require the molds not be removed
the general procedures in this practice. A program to deter- until at least 16 to 24 h after final setting time. Immediately sub-
mine the number of batches of concrete, number of specimens ject the specimens to standard curing conditions after removal
for all the various tests, and various test ages needs to be of the molds in moist room or water tanks to meet the require-
performed prior to laboratory mixing of the concrete. ments of ASTM C 511. Flexural test specimens must be stored
in water saturated with calcium hydroxide at standard curing
Materials Conditioning and Testing temperature for a minimum period of 20 h prior to testing.
Before mixing the concrete, all the materials must be at room
temperature in the range of 68 to 86°F (20 to 30°C) unless Evaluation
the design is being performed at other than room tempera- A precision statement of all the test methods included in the
ture. Store the cement as required and check it for fineness. standard is based on data from the concrete proficiency
If required, determine the relative density and absorption of sample program of the Cement and Concrete Reference
the coarse aggregate using ASTM Test Method for Density, Laboratory. Each laboratory should, as part of its quality
Relative Density (Specific Gravity), and Absorption of Coarse system, analyze their data against the values in this practice.
Aggregate (C 127) and fine aggregate using ASTM Test
Method for Density, Relative Density (Specific Gravity), and Conclusions
Absorption of Fine Aggregate (C 128). The moisture content
of the aggregates must be known before batching the con- The making and curing of concrete test specimens are cov-
crete. The weights of the cement, aggregates, admixtures, and ered by two ASTM standards: ASTM C 31 for field use, and
water must be known accurately prior to batching. Propor- ASTM C 192 for laboratory use. The field standard is used to
tioning concrete mixtures are covered in ACI Standard make specimens to comply with specification requirements
Practices 211.1 [24] and 211.2 [25]. for concrete used on construction projects. The failure to
meet specification requirements has resulted in many investi-
Mixing and Testing gations to determine the adequacy of in-place concrete,
Mix the concrete in a mixer that will provide a uniform, as-delivered concrete, and the making, curing, and testing of
homogeneous mixture in the mixing times required. The size concrete specimens. The laboratory standard is used to
of the batch has to be about 10 % in excess of the quantity develop mixture proportions for field concrete and research
 LAMOND ON MAKING AND CURING CONCRETE SPECIMENS 85

studies. The failure of the laboratory-developed mixtures to [11] “State-of-the-Art Report on High Strength Concrete” ACI
perform in the field has caused considerable problems. Committee 363, American Concrete Institute, Farmington Hills,
Research studies on test specimens made in the laboratory MI, 1997.
have been questioned on many occasions when one [12] “Concrete Manual,” U.S. Department of Interior, Bureau of
researcher’s results are different from those of another Reclamation, U.S. Government Printing Office, 1975.
researcher on the same type of study. [13] “Recommended Practice for Evaluation of Strength Test Results
of Concrete,” ACI Committee Report 214, American Concrete
Concrete test specimens should be made, cured, and
Institute, Farmington Hills, MI, 2002.
tested to obtain accurate and representative results. ASTM C 31
[14] Wagner, W. K., “Effect of Sampling and Job Curing Procedures
or C 192 and the standards they reference should be followed. on Compressive Strength of Concrete,” Materials Research and
Standards, Vol. 3, No. 8, Aug. 1963, p. 629.
[15] “In-Place Methods to Estimate Strength of Concrete,” ACI
References Committee 228, American Concrete Institute, Farmington Hills,
[1] Kesler, C. E. and Siess, C. P., “Static and Fatigue Strength,” MI, 1995.
Significance of Tests and Properties of Concrete and Concrete [16] Gonnerman, H. F., “Effect of Size and Shape of Test Specimens
Aggregates, ASTM STP 169, ASTM International, West Con- on the Compressive Strength of Concrete,” Proceedings,
shohocken, PA, 1956. American Society for Testing and Materials, Vol. 25, Part 2,
[2] Kennedy, T. B., “Making and Curing Test Specimens,” Signifi- 1925, pp. 237–250.
cance of Tests and Properties of Concrete and Concrete-Making [17] Malhotra, V. M., “Are 4  8 Inch Concrete Cylinders as Good as
Materials, ASTM STP 169A, ASTM International, West Con- 6  12 Cylinders for Quality Control of Concrete,” Journal,
shohocken, PA, 1966. American Concrete Institute, Jan. 1976.
[3] Adams, R. F., “Making and Curing Test Specimens,” Significance [18] Forstie, D. A. and Schnormeier, R., “Development and Use of 4
of Tests and Properties of Concrete and Concrete-Making by 8 Inch Concrete Cylinders in Arizona,” Concrete Interna-
Materials, ASTM STP 169B, ASTM International, West Con- tional, Farmington Hills, MI, July 1981, pp. 42–45.
shohocken, PA, 1978. [19] Pistilli, M. F. and Willens, T., “Evaluation of Cylinder Size
[4] Lamond, J. F., “Making and Curing Test Specimens,” Signifi- and Capping Methods in Compressive Strength Testing of
cance of Test and Properties of Concrete and Concrete-Making Concrete,” ASTM Journal Cement, Concrete & Concrete Aggre-
Materials, ASTM STP 169C, ASTM International, West Con- gates, Vol.15, No. 1, ASTM International, West Conshohocken,
shohocken, PA, 1994. PA, 1993, pp. 58–69.
[5] “Report of Committee on Specifications and Methods of Test [20] Burg, R. G., Caldarone, M. A., Detwiler, G., Jansen, D. C., and
for Concrete Materials,” Proceedings, National Association of Wilems, T. J., “Compressive Testing of HSC: Latest Technology,”
Cement Users (now American Concrete Institute), Detroit, MI, Concrete International, Farmington Hills, MI, August 1999, pp.
Vol. 19, 1914. 67–76.
[6] Price, W. H., “Factors Influencing Concrete Strength,” Proceed- [21] Bloem, D. L., “Effect of Curing Conditions on the Compressive
ings, American Concrete Institute, Detroit, MI, Vol. 47, 1951, pp. Strength of Concrete Test Specimens,” National Ready Mixed
417–432. Concrete Association, Publication No. 53, Silver Spring, MD,
[7] Sparkes, F. N., “The Control of Concrete Quality: A Review of 1969.
the Present Position,” Proceedings, Cement and Concrete [22] Harsh, S. and Darwin, D., “Traffic-Induced Vibrations and
Association, London, 1954. Bridge Deck Repairs,” Concrete International, Vol. 8, No. 5, May
[8] ACI Manual of Concrete Inspection, SP-2, 9th ed., American 1986, pp. 36–42.
Concrete Institute, Farmington Hills, MI, 1999. [23] “Standard Practice for Curing Concrete,” ACI Committee 308,
[9] Walker, S. and Bloem, D. L., “Studies of Flexural Strength of American Concrete Institute, Farmington Hills, MI, 2001.
Concrete, Part 2, Effects of Curing and Moisture Distribution,” [24] “Standard Practice for Selecting Proportions for Normal,
Proceedings, Highway Research Board, Washington, DC, Vol. Heavyweight, and Mass Concrete,” ACI Committee 211.1,
36, 1957. American Concrete Institute, Farmington Hills, MI, 1991.
[10] Richardson, D. N., “Review of Variables that Influence Mea- [25] “Standard Practice for Selecting Proportions for Structural
sured Concrete Compressive Strength,” Journal of Materials in Lightweight Concrete,” ACI Committee 211.2, American
Civil Engineering, May 1991. Concrete Institute, Farmington Hills, MI, 1998.
11
Time of Setting

Bruce J. Christensen1

Preface History
SCRIPTURE PREPARED THE ORIGINAL VERSION OF A number of methods for monitoring the time of setting of con-
this document, ASTM STP 169, in 1956 [1]. Kelly extended this crete have been investigated over the years. Some of the earli-
work with ASTM STP 169A in 1966 [2], which was subsequently est methods were developed in the late 50s and are discussed
updated by Sprouse and Peppler in 1978 with ASTM 169B [3]. In in detail in the previous edition of this publication [4]. The
1994, Dodson prepared the most recent update, that being ASTM intent of this document is to review the current ASTM method,
STP 169C [4]. In the preparation of this chapter, the contents of noting the advantages and disadvantages of the method, as well
the previous editions were drawn upon, but every attempt was as to highlight some of the other methods that have emerged
made to minimize duplication of information from the previous or are in the process of being developed.
edition. The author acknowledges the contributors of the previ-
ous editions and their summaries of the literature on this topic. Current ASTM Method
The current edition will review and update the topics as addressed
by the previous authors, provide up-to-date references and focus Basics of the Method
specifically on new technologies that have been developed. The The most widely used method for determining the time of setting
review period has been limited primarily to contributions made of concrete is ASTM C 403/C 403M—Standard Test Method for
during the last decade. Attempts have been made to uncover all Time of Setting of Concrete by Penetration Resistance [5]. As
relevant research, but undoubtedly important advances were not clearly stated in the title of the method, the time of setting is de-
covered or were unknowingly omitted in this review. termined from changes in the penetration resistance of a speci-
men as a function of time. In particular, a sample of mortar is ob-
Introduction tained by wet sieving a portion of the concrete on a 4.75 mm sieve
to remove the coarse aggregate fraction or it can be a prepared
Hydraulic cementitious compositions are unique in that in their mortar. The mortar fraction is placed into a rigid, non-absorptive
early stages they are moldable and castable materials that sub- container and stored at a specified temperature. Periodically, the
sequently are transformed into rigid structures with useful en- resistance to penetration of the mortar fraction is obtained by us-
gineering properties. This structural transformation occurs as ing a series of pins of successively decreasing diameter injected
the result of chemical reactions between the cementitious to a depth of 25 mm (1 in.). Bleed water is removed from the sur-
material and water, a process referred to as hydration. The hy- face of the mortar fraction on a regular basis. Pin penetrations
dration process results in the consumption of free water and are spaced in such a way as not to be influenced by previous pene-
formation of an interlocking network of hydrating particles, giv- tration locations and at a minimum distance from the perimeter
ing rise to stiffening, and subsequently a hardening behavior, of the container. Resistance to penetration is determined by di-
which is defined as “setting.” With portland cement-based com- viding the applied load by the surface area of the respective pin.
positions, this process normally occurs in a matter of hours These resistance values are then plotted as a function of hydra-
from the time of contact of cement and water. Because the hy- tion time and fitted by regression analysis to determine the times
dration process is complex and consists of a series of chemical of initial and final setting.
reactions, the process is affected by changes in the content, the The resistances to penetration corresponding to the times
chemistry and/or the surface area of the cementitious compo- of initial and final setting are arbitrarily defined as 3.5 MPa
nent, temperature, water content, and the presence of chemical (500 psi) and 27.6 MPa (4000 psi), respectively. The time of ini-
and mineral admixtures in concrete mixtures. These changes ul- tial setting is defined to be the time at which the concrete is no
timately affect the time at which finishing and curing opera- longer mobile, even by the application of vibration. Initial set
tions can be initiated, steam-curing can be applied, cold joints is often the time at which steam-curing is applied in precast op-
can be avoided, and load can be applied to the structure. There- erations, as well as generally regarded as the time at which fin-
fore, it is important to have methods by which the rate of setting ishing operations can begin. There is not complete agreement
of the cementitious composition can be quantified. on these definitions, though. The British Standards Institute BS

1
Vice President, R&D, Degussa Construction Chemicals.
86
 CHRISTENSEN ON TIME OF SETTING 87

5075 defines the limit for placing and compaction at 0.5 MPa
(72 psi) [4]. Abel and Hover observed that the time to begin fin-
ishing operations, which is observed in practice as the time at
which the boot of an adult male leaves an imprint approxi-
mately 6 mm deep in a fresh concrete surface, occurs at a pen-
etration resistance of approximately 0.1 MPa (15 psi) [6]. Ad-
ditionally, finishing operations on concrete slabs performed
using a finishing machine are understood to begin as soon as
measurable values of penetration resistance are obtained on
companion mortar specimens [7]. Therefore, evidence cur-
rently exists to suggest that finishing operations generally be-
gin much sooner than the time corresponding to initial setting
as determined by ASTM C 403/C 403M.
The time of final setting is generally regarded as the time
at which the concrete is no longer deformable. This signals the
end of the finishing window, at which time the surface of the
setting concrete can no longer be manipulated and also corre-
sponds to the time at which curing operations can begin, such
as the application of a curing compounds, water, wet burlap,
or the like. Measurements of the compressive strength of the
concrete specimens can be made at this time and have been
reported to be on the order of 0.7 MPa (100 psi) [4]. Fig. 2—Pocket penetrometer used to perform the ASTM C
The equipment typically used in the lab to determine the 403/C 403M method when a bench-top style is not available.
times of setting is a bench-top penetrometer like that shown in
Fig. 1. It consists of a base platen with a vertical post, to which
is attached a spring mechanism with a meter to display the ap-
plied force. The meter contains a floating needle, which acts as Markings corresponding to different levels of penetration re-
a marker of the maximum force delivered during the penetra- sistance (not force, since the pin size is fixed) are inscribed on
tion step. A lever attached to the side of the spring mechanism the side of the device. While holding the device by the spring-
is used to slowly inject the appropriate-sized pin into the plas- loaded head and manually inserting into the plastic mortar
tic mortar sample until the maximum penetration distance is specimen, the head travels downward. When the maximum
achieved. These instruments are fairly rugged, easy to use, and, penetration distance is achieved, the device is extracted from
if regularly calibrated, produce very reproducible results. the mortar specimen and penetration resistance determined
A smaller version that is often used in the field is com- from the location of the sliding collar. As one might expect, the
monly referred to as a pocket penetrometer. This penetrome- device is much more operator dependent than the bench-top
ter, shown in Fig. 2, is a small device that consists of a single unit, particularly due to the difficulty in maintaining the pen-
pin with a spring-loaded head assembly and a sliding collar. etrometer in an orientation perpendicular to the surface of the
specimen during penetration, and as such produces results
with a wider variability. This variability is further expanded be-
cause only one diameter pin is typically used, thereby limiting
the number of useful penetration events that can be executed
to get a good statistical sampling. Finally, since the application
of sufficient force to achieve penetration at the later stages of
the hardening process is very difficult, the device only provides
information on time of initial setting and not final setting. Nev-
ertheless, this device does have the advantages of being easily
transportable and can provide a quantifiable estimate of the
rate of hardening under many conditions.

Manipulation of the Data

An example set of data that was obtained from a bench-top pen-
etrometer is shown in Table 1. At the time of the development
of the method, these data were plotted manually on semi-loga-
rithmic graph paper and the times of setting determined by
hand-fitting. With the availability of personal computers and
spreadsheet software, it is now a relatively easy task to plot and
analytically determine the best function to fit the data. Figure
3 shows the penetration resistance versus elapsed time for the
specimen in Table 1 (control), as well as two additional sam-
ples. Equations are shown for fits made using either an expo-
nential or a power law function. Fitting a set of at least six data
Fig. 1—One style of bench-top penetrometer used to with one of these two functions and removing any outliers gen-
perform the ASTM C 403/C 403M method. erally results in a correlation coefficient that is greater than
88 TESTS AND PROPERTIES OF CONCRETE

Fig. 3—Penetration resistance versus elapsed time for mortar fractions from three
different concrete mixtures. Dotted lines and boxes indicate exponential function fits
to the data; solid lines and boxes represent power law fits to the data.

0.98. Once an adequate equation is determined, the times of ASTM C 191 or ASTM C 266 methods used to quantify the time
initial and final setting can be determined by inputting either of setting of cement pastes. For ASTM C 191, a Vicat needle of
3.5 MPa (initial) or 27.6 MPa (final) for the penetration resist- one dimension is used throughout the test, while for ASTM C
ance and calculating the corresponding elapsed time. Table 2 266, one pin size is used for determination of initial set and an-
shows the results of the calculation on this dataset. other size for final set [8,9]. Subsequently, ASTM C 403/C 403M
The precision and bias statistics of this method are shown is used in such specifications as ASTM C 494/C 494M to evalu-
in Table 3. These values were obtained using the inch-pound ate the effect of chemical admixtures on the time of setting [10].
system and hand-fitting procedures in 1973 and have remained Another advantage is the fundamentally similar mechanism of
unchanged since that time. The data were collected during a penetration to that used in the field by finishers to determine
round-robin test program in which five different laboratories when to begin finishing operations. The method is very straight-
participated. Evaluations were performed on prepared mor- forward to perform, utilizes relatively inexpensive equipment
tars for ease of preparation, in place of mortar wet-sieved from ($1–2 K), and is quickly understood by a technician.
concrete as specified in the method.
Disadvantages of the Method
Advantages of the Method The first disadvantage of the method is the requirement of wet-
One of the main advantages of this method for determining the sieving the plastic concrete to obtain the mortar fraction. This
time of setting is that it is based on multiple quantitative meas- process can be laborious, especially when the slump of the
urements on the same specimen. The use of multiple data concrete is low and/or the water-to-cement ratio (w/c) of the
points improves the precision of the time of setting, as deter- mixture is low. In the laboratory, where dozens of concrete
mined by this method. In addition, because the penetrations are mixtures may be evaluated in a day, this can represent a sig-
of various diameters, which are significantly larger than the nificant investment in time and effort to perform the method.
largest particle in the mortar specimen, the influence of edge In the field, wet sieving requires transporting a large sieve and
effects is minimized. These characteristics are in contrast to capturing vessel to be able to perform the method.

TABLE 1—Penetration Resistance Data for the Control Mixture

Elapsed Time (min) Penetration Force (N) Pin diameter (mm) Surface Area of Pin (mm2) Penetration Resistance (MPa)

251 176 20.3 323 0.54

281 176 14.3 161 1.1
311 370 14.3 161 2.3
341 299 9.1 65 4.6
386 396 6.4 32 12.4
416 343 4.5 16 21.4
446 546 4.5 16 34.1
 CHRISTENSEN ON TIME OF SETTING 89

TABLE 2—Calculation of Time of Setting for the Three Specimens shown in Fig. 3
Fitting Function Time of Setting (Initial or Final) Accelerated (minutes) Control (minutes) Retarded (minutes)

Exponential Initial 251 333 478

Final 336 430 583
Power Law Initial 248 337 489
Final 340 446 602

A second disadvantage is the requirement to continuously began to stiffen, penetration measurements were made at ten
monitor the specimen. Under normal conditions, the speci- different elapsed times between the time of mixing and the
mens can remain unattended for the first few hours after con- time of final setting. At each time interval, punches were made
tact of cement and water. Thereafter, though, the specimens at five equally-spaced locations along the diameter of the same
must be continuously monitored and tested at successively de- container, and the sample was then discarded (all punches
creasing time intervals until the time of final setting has been were made with the same operator on the same penetrometer
exceeded. Automated systems, such as the robotic system using the same fitting equation). As a result, the time of initial
shown in Fig. 4, exist and can perform the method on multiple and final setting could be determined as a function of position
specimens without human intervention. Unfortunately, this in the specimen. Using a fit to all ten data points for each po-
equipment requires a significant capital investment (
$80 K as sition, a plot was prepared and is shown in Fig. 5. The trend is
of 1999) and is impractical for many users. towards a shorter time of setting at the center of the container
Another disadvantage that has been already mentioned as compared to the perimeter, as expected, but the magnitude
briefly is the lack of correlation of the results from the method of the difference depends upon the stage of hydration. The dif-
to observations with field concrete. Finishing operations are ferences in the time of initial setting are small with position,
generally initiated earlier than the time of initial setting as in- while the maximum difference in the time of final setting is on
dicated by the method. This issue, though, appears to be a mat- the order of 40 min. This is significant for specifications such
ter of definition, and not a fundamental flaw in the method. as that of an ASTM C 494/C 494M Type F water reducer, where
Hence, useful information can still be obtained by comparing the maximum allowable delay in time of setting of the admix-
the relative setting characteristics of different concrete mix-
tures by penetration resistance.

Modifications to the Method

The basic aspects of the current method appear to have re-
mained essentially unchanged for more than three decades,
which is indicative of the usefulness of the method. Neverthe-
less, there is always room for improvement. One observation
that many practitioners have commented upon in recent years
is a belief that the operator has the ability to skew the values of
initial and final time of setting by varying the location at which
the pin is penetrated into the specimen. While there is general
agreement among the C09.23.1 committee members that this
is indeed true, no data could be found to substantiate the hy-
pothesis. Therefore, the following test was recently performed
in the laboratory to investigate this phenomenon.
One large batch of mortar was prepared and transferred
into ten identical cylindrical containers (approximately 18 cm
in diameter). As the hydration proceeded and the specimens

TABLE 3—Precision of ASTM C 403/C 403M

Coefficient of Variation, %

Single Operator, Multilaboratory

Multiday, Three Average of
Time of Setting Individual Results Three Tests

Initial 7.1 5.2

Final 4.7 4.5
Fig. 4—Automatic penetrometer with ability to monitor
eight specimens simultaneously.
90 TESTS AND PROPERTIES OF CONCRETE

Fig. 5—Time of setting of mortar specimens as a function of the punch location

on cylindrical specimens.

ture-treated concrete mixture is 90 min from the control. The Additional Methods for Evaluating
thermal loss from the perimeter of the container is likely re- Time of Setting
sulting in less activation of the hydration process than in the
center of the specimen. This gradient from inside to the out- Other Penetration Methods
side of the container is more significant as the hydration As mentioned previously, one of the limitations of the ASTM C
process proceeds, further substantiating the observation of 403/C 403M method is that the mortar must be wet sieved from
wider variation at the time nearer to final setting. the concrete. To eliminate this step, Abel and Hover proposed the
As a result of these observations, additional tests are use of a modified penetrometer with a much larger surface area
planned to investigate the extent of this variation in time of set- than the pins typically used in the current method [6]. The sur-
ting as a function of different cements and chemical admixture face area is similar to that of the base of an adult work boot and
compositions. In parallel, efforts are underway to modify the the penetrometer foot can be used directly on the surface of a
method. The current proposal is to limit the locations for pene- concrete slab. This apparatus is referred to as the “Finisher’s
tration measurements to a ring of a fixed distance from the Foot” and is shown in Fig. 6. Two different feet are used, the
outer perimeter of the container. smallest of which is still several times larger than the diameter of

Fig. 6—Finisher’s foot apparatus being using on a concrete slab to determine

penetration resistance (after Abel and Hover [6]).
 CHRISTENSEN ON TIME OF SETTING 91

from this plot. First is the observation that the penetration meas-
urements can be obtained much sooner on the concrete slab
than on the mortar. Second is the smooth transition during the
change from the large foot to the small foot, as well as during
the transition from measurements on the slab to the mortar
specimen. Probably of most significance, which is not apparent
from the plot alone, is the authors’ observation that the time to
begin finishing operations occurred at penetration resistances
on the order of 0.1 to 0.2 MPa, which is prior to the first pene-
trations on the mortar specimens. While some of these obser-
vations are specific to the reported projects and some of the ef-
fects are likely due to the differences in sizes of the specimens,
the time of “initial” setting by this method is much earlier than
predicted by ASTM C 403/C 403M. Interest in standardizing this
procedure is significant and is currently going through the bal-
loting process as an ASTM Standard Test Practice.
Another limitation of ASTM C 403/C 403M that has been
mentioned is the need to continuously monitor the specimen
during the hardening process. While this can be addressed
Fig. 7—Penetration resistance versus time after batching with a sophisticated and expensive robotic system, this re-
for field concrete. Open symbols are generated using the quirement can also potentially be overcome by using a continu-
“finisher’s foot”; closed symbols are from C 403 measurements ous penetration measurement. Sohn and Johnson discussed an
on mortar wet sieved from the same concrete (after Abel and apparatus that could be used to continuously monitor the hard-
Hover [6]). ening process of cement-based materials during microwave
curing [11]. Figure 8 shows the details of the arrangement.
They investigated multiple penetration rates and interpreted
a typical coarse aggregate. The penetration depth in this method the data in terms of hardening rate at various temperatures
is modified and limited to 6 mm. This depth is chosen to simu- and hardness values. Results were limited to investigations on
late the depth at which the work boot of a finisher will sink into cement pastes and mortars.
a bull-floated slab when beginning subsequent finishing opera- Further evaluations of this method were carried out in the
tions. The abbreviated depths may possibly result in less edge ef- laboratory on prepared mortars by modifying the orientation
fect from interaction with the coarse aggregate. of the apparatus and comparing the results to those obtained
In essentially the same manner as the ASTM C 403/C 403M by ASTM C 403/C 403M [12]. Different penetration rates, as
method, the penetrometer foot is pressed into the plastic con- well as specimens with a range of setting times, were investi-
crete to the prescribed depth, and the force necessary to achieve gated. An example comparison of the results of the two meth-
the penetration is recorded and converted to penetration re- ods is shown in Fig. 9. The correlation for the mixtures inves-
sistance. An example plot of penetration resistance versus batch tigated at this penetration rate was quite good; however,
time is shown in Fig. 7. The open symbols on the lower left por- further work was considered necessary to fully investigate the
tion of the plot are obtained on the concrete slab, while the method. Among the limitations of the method, the mortar frac-
closed symbols in the upper right portion of the plot are ob- tion still needs to be sieved from the concrete mixture. While
tained on companion mortar specimens sieved from the same in principle a similar system can be designed with a large pene-
concrete mixture. A couple of significant features are apparent trometer for use on concrete, economically impractical loads

Fig. 8—Continuous penetration apparatus discussed by Sohn and Johnson

[11]. Reprinted with permission.
92 TESTS AND PROPERTIES OF CONCRETE

Fig. 9—Comparison of initial and final times of setting determined by the

continuous penetration versus that obtained by C 403 [12]. A penetration
rate of 6.4 mm/hr was used on prepared mortar specimens.

would be required. Another limitation of the method is that the alter that correlation. Equipment is now available commer-
apparatus must be dedicated to one specimen throughout the cially that allows testing of multiple specimens at the same
hardening process, so multiple pieces of equipment are neces- time, but the cost is significant (
$20 K). This equipment is an
sary for use in a development laboratory. excellent tool for laboratory investigations and for pre-qualifi-
cation of field mixtures, but is not transportable to the field.
Thermal Methods Another limitation of this method is that the cell size is small
Because the hydration reaction of portland cement is exother- so that only pastes and mortars can be tested.
mic, researchers and practitioners have used this attribute as a One of the simplest approaches is merely to embed a ther-
means of assessing the state of the hardening process [13–17]. mocouple wire into the center of the concrete specimen and
Isothermal calorimetry is one technique that is very useful for monitor the temperature as a function of time. This condition
quantifying the energy liberated in this process, as well as a is often semi-adiabatic and the results are heavily dependent
very good tool for elucidating the reactions of the calcium upon the size of the specimen, environmental conditions, and
sulfate and other mineral phases. An example plot of energy mixture design. Nearly adiabatic conditions can be achieved by
liberated versus hydration time is shown in Fig. 10. In general, placing the specimen in an insulating chamber, which is useful
the time corresponding to the onset of the rise in the curve af- in some applications, but may not translate well to the setting
ter the induction period is considered to correspond closely characteristics of a slab or other structure that is not fully
with the time of initial setting, but many factors in the field can insulated. Nevertheless, portable equipment with the ability to

Fig. 10—Typical isothermal calorimetry curve for hydrating cement pastes (20 °C).
 CHRISTENSEN ON TIME OF SETTING 93

Fig. 11—Insulating chamber with cavities for concrete specimens used in pro-
posed thermal method. Thermocouple tips are imbedded in the base of the cavity.

read multiple thermocouples and that are relatively inexpen- sulating cavities, but no insulation is placed on the top surface
sive and convenient to use are now available. As such, investi- of the specimens. Thermocouples are attached permanently to
gations have been performed recently to further develop the the interior base of the cavities, thereby eliminating the need of
appropriate sample storage conditions, with the intent of im- embedding and removing the thermocouple wires every time a
proving the correlation to ASTM C 403/C 403M [14]. specimen is tested. An example plot of temperature versus time
This author has proposed a sample configuration and data for a specimen tested in this configuration is shown in Fig. 12.
manipulation technique for temperature versus time measure- Determination of the first and second derivatives of tempera-
ments that provides good correlation with ASTM C 403/C 403M ture versus time, and plotting the data versus time, results in the
[14]. The method is intended for use in the laboratory and al- curves shown in Figs. 13 and 14. The time corresponding to the
lows the use of a concrete sample, instead of mortar sieved from maximum in these plots is then compared to the times of initial
plastic concrete. An example of the sample configuration is and final setting determined on companion mortar specimens
shown in Fig. 11. The concrete mixture is placed into 15 cm
(wet sieved from the same concrete) using ASTM C 403/C 403M.
15 cm plastic cylinders, which are subsequently placed into in- These comparisons for three different concrete mixtures are

Fig. 12—Example of the temperature versus time plot obtained for a specimen
placed in one of the cavities shown in Fig. 11.
94 TESTS AND PROPERTIES OF CONCRETE

Fig. 13—Plot of the first derivative of temperature with time (from Fig. 11) as
a function of time. Time of the maximum appears to correlate with time of final
setting determined from C 403.

shown in Fig. 15. Interestingly, the correlation between the time Ultrasonic Methods
of final setting and the time of the maximum in the first deriv- A number of investigations have been published on the use of
ative versus time is reasonable. The time of the maximum in the ultrasonic waves to follow the setting of concrete [17–27].
second derivative versus time occurs earlier than the time of ini- Some techniques rely upon the propagation of the waves
tial setting by ASTM C 403/C 403M and seems to correlate bet- through the material, hence the specimen must be thin, due to
ter with the time at which the penetration resistance on the mor- its lossy nature. Others utilize reflection of the wave off of the
tar specimen is approximately 1.25 MPa (200 psi). Further work near surface, where the response is independent of the thick-
on a range of mixture designs is necessary to further validate ness of the specimen. Typically, either shear waves or com-
the method, but initial observations are very promising. Never- pression waves are used, though shear waves are only sup-
theless, because this method addresses many of the operational ported in solids, so assessing the extent of shear wave
deficiencies of ASTM C 403/C 403M (wet sieving and constant absorption or reflection can be used as a basis for investigating
human monitoring), the consensus of the C09.23.1 committee the transition from a fluid to a solid. Subramaniam et al. have
has been to move forward with standardization of the method. characterized this relationship as a wave reflection factor

Fig. 14—Plot of the second derivative of temperature with time (from Fig. 11)
as a function of time. Time of the maximum appears to occur slightly before the
time of initial setting determined from C 403.
 CHRISTENSEN ON TIME OF SETTING 95

Fig. 15—Comparison of the time at the maximum in the 1st or 2nd derivative of
time plot versus time of setting from ASTM C 403/C 403M. Data from three dif-
ferent mixtures with a range of setting times are presented, one of which is based
on information in Fig. 12.

(WRF), which changes with hydration time as shown in Fig. 16 Electrical Methods
[20]. They have observed that the time corresponding to the de- Investigations concerning changes of the electrical character-
crease in the WRF exhibits a linear relationship with the time istics of hydrating systems have continued since the last edition
of initial setting determined by ASTM C 403/C 403M [19]. of this chapter [28–32]. Many of the recent investigations uti-
One of the obvious advantages is that this technique is non- lize impedance spectroscopy (IS) as the technique of choice. IS
destructive, thus is suitable for real concrete structures. WRF allows easy separation of electrode effects, sweeps a range of
can also be used at times beyond the setting regime to charac- frequencies to find the true bulk resistance (or inversely the
terize compressive strength development, and hence can be conductance), and provides a means to determine the dielec-
used to characterize the concrete mixture throughout the entire tric properties of the material with time. An example plot of
hydration process. Continuous data acquisition also allows for bulk resistance versus hydration time is shown in Fig. 17. A
automation, so a large portion of the manual effort can be re- common feature is the inflection point, where the resistance
moved. A disadvantage of the reflection method is that a trans- begins to increase significantly with time, concurrent with the
ducer is required for each sample tested and the setup for mul- end of the dormant period and renewed hydration reactions.
tiple specimens is somewhat costly (
$10 K for an 8 channel Attempts by this author to correlate the time of the inflection
system). Nevertheless, the method appears to be quite promis- point from mortars with the time of initial or final setting from
ing and further development in this area could be quite fruitful. ASTM C 403/C 403M have not shown a clear relationship. An

Fig. 16—Wave reflection factor (WRF) versus hydration time for a concrete spec-
imen. Point “A” is where the WRF begins decreasing and corresponds closely with
time of initial setting (after Subramaniam et al. [20]). Reprinted with permission.
96 TESTS AND PROPERTIES OF CONCRETE

Fig. 17—Example of changes in bulk resistance of a mortar prism as a function

of hydration time. Vertical line indicates time corresponding to the inflection point
in the curve.

example of the trends is shown in Fig. 18. The specimens, tion, as well as the connectivity of the pore network, both of
which were retarded, exhibited significant bleed water, which which can change independently [28,29]. Therefore, this
may have adversely affected the test results. Torrents et al. in- method may not be sufficiently sensitive for some applications.
vestigated cement pastes with electrical means and the Vicat
needle [31]. Reasonable correlation was found between the re- Rheological Methods
gion defined as “beginning,” but poor correlation for the re- Struble and coworkers appear to be the primary group in-
gion referred to as the “end.” vestigating the relationships between rheological changes
A disadvantage of the equipment for making IS measure- and the setting characteristics of portland cement systems
ments is equipment cost (
$15 K), which could limit its wide- during hydration [33–35]. With the use of a constant stress
spread use. Furthermore, the resistance of the specimen has rheometer in a regime below the yield stress, they observed
shown to be dependent upon the resistance of the pore solu- two regions of yield stress in cement pastes. The first corre-

Fig. 18—Comparison of time of setting determined by ASTM C 403/C 403M

versus the time of the inflection point in the electrical resistance versus time plot.
 CHRISTENSEN ON TIME OF SETTING 97

lated with the induction period and the other with the ac- References
celeratory period, with the time of the transition correspon-
ding to the time of initial setting by the Vicat needle [33]. In [1] Scripture, E. W., “Setting Time,” Significance of Tests and Prop-
another study, they developed a dynamic rheology test and erties of Concrete and Concrete Aggregates, ASTM STP 169,
ASTM International, West Conshohocken, PA, 1956, pp. 53–60.
compared the oscillatory shear behavior to the features in
[2] Kelly, T. M., “Setting Time,” Significance of Tests and Properties
ASTM C 403/C 403M and C 191 [34]. Good agreement was
of Concrete and Concrete Aggregates, ASTM STP 169A, ASTM
observed up to the torque limit of the rheometer, but they
International, West Conshohocken, PA, 1966, pp. 102–115.
were unable to follow the rheological behavior to the time
[3] Sprouse, J. H. and Peppler, R. B., “Setting Time,” Significance of
of initial setting as defined by either of the ASTM methods. Tests and Properties of Concrete and Concrete Aggregates,
These results suggest that a higher torque apparatus would ASTM STP 169B, ASTM International, West Conshohocken, PA,
be necessary to make comparisons to ASTM C 403/C 403M. 1978, pp. 105–121.
In addition, measurements must still be performed on ce- [4] Dodson, V. H., “Time of Setting,” Significance of Tests and
ment paste instead of concrete and delicate equipment is Properties of Concrete and Concrete-Making Materials, ASTM
required. Bunt discussed methods for characterizing the STP 169C, Klieger and Lamond, Eds., ASTM International, West
rheology of calcium aluminate pastes, in addition to thermal Conshohocken, PA, 1994, pp. 77–87.
and penetration methods, but comparisons between the [5] ASTM Test Method for Time of Setting of Concrete Mixtures by
methods were not discussed [36]. Penetration Resistance (C 403/C 403M), Annual Book of ASTM
Standards, Vol. 04.02, ASTM International, West Con-
shohocken, PA, 2003, pp. 228–233.
Other Methods [6] Abel, J. D. and Hover, K. C., “Field Study of the Setting Behav-
Methods such as X-ray diffraction, NMR, and computer model- ior of Fresh Concrete,” Cement, Concrete and Aggregates, Vol.
ing are also discussed in the literature as a means of assessing 22, No. 2, 2000, pp. 95–102.
the setting characteristics of hydrating systems [37–40]. X-ray [7] Bury, M. A., Bury, J. R., and Martin, D., “Testing Effects of New
diffraction allows characterization of the gypsum/hemihydrate Admixtures on Concrete Finishing,” Concrete International,
reactions and the formation of portlandite. NMR can be used to Vol. 16, No. 1, 1994, pp. 26–31.
determine the physical state of the water, thereby providing [8] ASTM Test Method for Time of Setting of Hydraulic Cements by
some indication of the extent of hydration. Computer modeling Vicat Needle (C 191), Annual Book of ASTM Standards, Vol.
allows the possibility to simulate the extent of hydration at 04.01, ASTM International, West Conshohocken, PA, 2002, pp.
which the solid particles are percolated and form a three-di- 179–184.
mensional matrix, thereby resulting in initial setting. These ap- [9] ASTM Test Method for Time of Setting of Hydraulic Cements by
Gillmore Needles (C 266), Annual Book of ASTM Standards, Vol.
proaches are tools of most interest to the researcher, and at
04.01, ASTM International, West Conshohocken, PA, 2002, pp.
present are focused on mechanistic studies rather than routine
215–217.
measurements of time of setting.
[10] ASTM Standard Specification for Chemical Admixtures for
Concrete (C 494/C 494M), Annual Book of ASTM Standards, Vol.
Summary 04.02, ASTM International, West Conshohocken, PA, 2003, pp.
269–277.
Variables such as temperature, w/c, type of cement, fineness, [11] Sohn, D. and Johnson, D. L., “Hardening Process of Cement-
presence of chemical or mineral admixtures, and cement con- Based Materials Monitored by the Instrumented Penetration
tent all affect the rate of setting of concrete and have been re- Test: Part I: Neat Cement Paste and Mortar,” Cement and
ported on previously [4]. Quantifying the setting characteris- Concrete Research, Vol. 32, 2002, pp. 557–563.
tics of hydraulic mixtures is therefore an important part of [12] Daczko, J. A. and Cordova, R. “Development and Evaluation of
Automated Rate of Hardening Equipment,” Master Builders
determining when finishing operations can begin, steam can
Internal Report, TS-CM-001, 1999.
be applied to precast elements, and/or load can be applied to
[13] Carino, S. J., “The Use of Temperature as an Indicator of Setting
structures. The current ASTM C 403/C 403M method is Time of Mortar Specimens,” Undergraduate Research Project,
widely used as a means of assessing the setting characteristics Cornell University, Ithaca, NY, January 1995, 38 pp.
of concrete mixtures, both in the laboratory and in the field. [14] Christensen, B. J., “Determination of Time of Setting of
This method has proven to be robust, having remained rela- Concrete Using Temperature Measurements of Semi-Insulated
tively unchanged for more than three decades. Ease of use, Concrete Specimens,” to be submitted to the Journal of ASTM
relatively low cost, and suitability for lab and field have been International.
favorable attributes. [15] Pinto, R. C. A. and Hover, K. C., “Application of Maturity
Like any method, though, C 403 has some deficiencies. Approach to Setting Times,” ACI Materials Journal, Vol. 96, No.
New methods have been investigated to address some of 6, 1999, pp. 686–691.
these limitations, as well as to provide additional information [16] Carino, N. J., “Nondestructive Testing of Concrete: History and
to the researcher or the practitioner. Some of these include Challenges,” Concrete Technology: Past, Present and Future,
the “finisher’s foot” method for measurements of penetration ACI SP-144, 1994, pp. 623–678.
resistance directly on concrete, as well as continuous pene- [17] Valic, M. I., “Hydration of Cementitious Materials by Pulse
Echo USWR: Method, Apparatus and Application Examples,”
tration measurements on mortars. Others include the use of
Cement and Concrete Research, Vol. 30, No. 10, 2000, pp.
changes in the thermal wave propagation or rheological char- 1633–1640.
acteristics with time. Expectations are that some of these [18] Chotard, T., Fargeot, D., Gault, C., Gimet-Breart, N., Bonnet, J.
methods will be standardized in the next decade, and be in- P., and Smith, A., “Application of Ultrasonic Testing to Describe
cluded in ASTM manuals by the time of writing the next edi- the Hydration of Calcium Aluminate Cement at the Early Age,”
tion of this chapter. Other methods, either newly discovered Cement and Concrete Research, Vol. 30, 2001, pp. 405–412.
or overlooked in this writing, will undoubtedly become [19] Rapoport, J., Popovics, J. S., Subramaniam, K. V., and Shah, S. P.,
known as well. “Using Ultrasound to Monitor Stiffening Process of Concrete
98 TESTS AND PROPERTIES OF CONCRETE

with Admixtures,” ACI Materials Journal, Vol. 97, No. 6, 2000, Using Impedance Spectroscopy,” Journal of the American
pp. 675–683. Ceramic Society, Vol. 75, No. 4, 1992, pp. 939–945.
[20] Subramaniam, K. V., Mohsen, J. P., Shaw, C. K., and Shah, S. P., [30] Torrents, J. M. and Pallas-Areny, R., “Measurement of Cement
“Ultrasonic Technique for Monitoring Concrete Strength Gain Setting by Impedance Monitoring,” Proceedings of the IEEE In-
at Early Age,” ACI Materials Journal, Vol. 99, No. 5, 2002, pp. strumentation & Measurement Technology Conference, Vol. 2,
458–462. Piscataway, NJ, 1997, pp. 1089–1093.
[21] Subramaniam, K. V., Lee, J., and Christensen, B. J., “Monitoring [31] Torrents, J. M., Roncero, J., and Gettu, R., “Utilization of
the Settling Behavior of Cementitious Materials Using One- Impedance Spectroscopy for Studying the Retarding Effect of a
Sided Ultrasonic Measurements,” Cement and Concrete Superplasticizer on the Setting of Cement,” Cement and
Research, Vol. 35, No. 5, pp. 850–857. Concrete Research, Vol. 28, No. 9, 1998, pp. 1325–1333.
[22] Tavossi, H. M., Tittmann, B. R., and Cohen-Tenoudji, F., “Ultra- [32] Ng, S. T. C., Shapiro, G., and Mak, S. L., “Monitoring Setting and
sonic Characterization of Cement and Concrete,” Review of Early Hydration of Cement Using Dielectric Spectroscopy,” Fifth
Progress in Quantitative Nondestructive Evaluation, Vol. 18, CANMET/ACI International Conference on Recent Advances in
Plenum Publishing, New York, 1999, pp. 1943–1948. Concrete Technology, ACI SP 200, American Concrete Institute,
[23] Gimet, N., Fargeot, D., Gaillard, J. M., Smith, A., Gault, C., and Farmington Hills, MI, 2001, pp. 167–182.
Bonnet, J. P., “Ultrasonic Assessment of Portland Cement at [33] Struble, L. J. and Lei, W. G., “Rheological Changes Associated
Early Stages of Hydration,” Journal of Materials Science Letters, with Setting of Cement Paste,” Advanced Cement Based
Vol. 18, 1999, pp. 1335–1337. Materials Journal, Vol. 2, No. 6, 1995, pp. 224–230.
[24] Reinhardt, H. W., Grosse, C. U., and Herb, A. T., “Ultrasonic [34] Struble, L., Kim, T. Y., and Zhang, H., “Setting of Cement and
Monitoring of Setting and Hardening of Cement Mortar—A Concrete,” Cement, Concrete, and Aggregates, Vol. 23, No. 2,
New Device,” Materials and Structures, Vol. 33, 2000, pp. 2001, pp. 88–93.
580–583. [35] Struble, L. J., Zhang, H., and Lei, W. G., “Oscillatory Shear
[25] D’Angelo, R., Plona, T. J., Schwartz, L. M., and Coveney, P., “Ul- Behavior of Portland Cement Paste During Early Hydration,”
trasonic Measurements on Hydrating Cement Slurries: Onset of Concrete Science and Engineering, Vol. 2, 2000, pp. 141–149.
Shear Wave Propogation,” Advanced Cement Based Materials, [36] Bunt, N. E., “Advanced Techniques for Measuring Rheology of
Vol. 2, No. 1, 1995, pp. 8–14. Cement-Based Refractories,” American Ceramic Society Bul-
[26] Mueller, D. T., Lacy, L. L., and Boncan, V. G., “Characterization letin, Vol. 73, No. 3, 1994, pp. 81–84.
of the Initial, Transitional, and Set Properties of Oilwell [37] Bentz, D. P., Garboczi, E. J., Haecker, C. J., and Jensen, O. M.,
Cement,” Proceedings of the Society of Petroleum Engineers “Effects of Cement Particle Size Distribution on Performance
Annual Technical Conference and Exhibition, SPE 36475, Properties of Portland Cement-Based Materials,” Cement and
Denver, CO, October 1996, pp. 613–623. Concrete Research, Vol. 29, No. 10, 1999, pp. 1663–1671.
[27] Boumiz, A., Vernet, C., and Cohen-Tenoudji, “Mechanical Prop- [38] Garcia-Guinea, J., Martin-Ramos, D., Palomo, A., and Sanchez-
erties of Cement Pastes and Mortars at Early Ages: Evolution Moral, S., “Simultaneous Optical Stimulation and X-Ray Dif-
with Time and Degree of Hydration,” Advanced Cement Based fraction of Crystalline Phases During the Setting and Hardening
Materials, Vol. 3, No. 3/4, 1996, pp. 94–106. of Gypsum and Cement,” ZKG International, Vol. 54, No. 7,
[28] Christensen, B. J., Coverdale, R. T., Olson, R. A., Ford, S. J., 2001, pp. 404–411.
Garboczi, E. J., Jennings, H. M., and Mason, T. O., “Impedance [39] Apih, T., Lebar, A., Pawlig, O., and Trettin, R., “Continuous
Spectroscopy of Hydrating Cement-Based Materials: Measure- Monitoring of the Zinc-Phosphate Acid-Base Cement Setting
ment, Interpretation and Application,” Journal of the Ameri- Reaction by Proton NMR,” Journal of Applied Physics, Vol. 89,
can Ceramic Society, Vol. 77, No. 11, 1994, pp. 2789–2804. No. 12, 2001, pp. 7784–7790.
[29] Christensen, B. J., Mason, T. O., and Jennings, H. M., “Influence [40] Brooks, J. J., “Prediction of Setting Time of Fly Ash Concrete,”
of Silica Fume on the Early Hydration of Portland Cements ACI Materials Journal, Vol. 99, No. 6, 2002, pp. 591–597.
12
Bleed Water

Steven H. Kosmatka1

Preface and durability of the concrete near the surface. Excessive bleed-
ing also delays finishing as finishing should not proceed with ob-
THE SUBJECT OF BLEEDING WAS BRIEFLY AD- servable bleed water present. On the other hand, lack of bleed
dressed in the first edition of ASTM STP 169, published in water on concrete flat work can sometimes lead to plastic
1956. Ivan L. Tyler, manager of the Field Research Section of shrinkage, cracking, or a dry surface that is difficult to finish.
the Portland Cement Association, concisely described the gen- The first reported case of bleeding in North America was
eral significance of tests for bleeding in his article on Unifor- in 1902 during the construction of the stadium at Harvard
mity, Segregation, and Bleeding in the Freshly Mixed Concrete University [1,2]. During placement, up to 2/3 m of bleed water
section of ASTM STP 169. ASTM STP 169A and ASTM STP developed. Up to 150 mm of concrete was removed from the
169B did not address bleeding. ASTM 169C had a chapter sim- top of each lift prior to the sequential placements in order to
ilar to the one presented here. The effects of concrete ingredi- remove the less durable and weaker concrete. Even with the
ents on bleeding as well as the significance of bleeding with high degree of bleeding, this structure survived the elements
modern concretes are presented in this chapter. This chapter for over 100 years and will be serviceable for many years to
also reviews the standard ASTM test methods on bleeding and come (Fig. 2). Structures in which severe exposures exist and
provides data on the bleeding characteristics of a variety of ce- in which porous concrete was not adequately removed have
ment pastes, mortars, and concretes. not performed as well as structures from which the bleeding-
damaged concrete was properly removed.
Introduction During the construction of massive structures, such as
deep foundations, tall walls, or dams, bleeding became of con-
Bleed water is the clear water that can gradually accumulate at cern in early concrete projects. To study and help control
the surface of freshly placed concrete, mortar, grout, or paste bleeding, a variety of bleeding tests were developed. These tests
(Fig. 1). Bleed water is caused by sedimentation or settlement of will be discussed later under the section on Test Methods. By
solid particles (cement and any aggregate) and the simultaneous understanding the process of bleeding, Powers [3] and others
upward migration of water. This upward migration of water and provided means to control bleeding and today bleeding is
its accumulation at the surface is called bleeding, also referred rarely a problem.
to as water gain, weeping, and sweating in some countries. A Bleeding can occur at any time during the transportation,
small amount of bleeding is normal and expected on freshly handling, and placing of concrete, as well as shortly after place-
placed concrete. It does not necessarily have an adverse effect ment. Most of the discussion in this chapter will focus on con-
on the quality of the plastic or hardened concrete. However, ex- crete after it is placed in a form and will no longer be agitated.
cessive bleeding can lead to some performance problems with This chapter will address the effects of concrete ingredients
plastic or hardened concrete. With proper mix proportioning, and placing practices on bleeding and the effect of bleeding on
mixture ingredients, placing equipment, and proper construc- various concrete properties. Much of the discussion pertains to
tion practices, bleeding can be controlled to a desirable level. cement paste or mortar as bleeding in concrete is a direct func-
tion of the paste or mortar bleeding properties.
Significance
Fundamentals of Bleeding
Bleeding is not necessarily a harmful property nor is excessive
bleeding desirable. Because most concrete ingredients today A fresh concrete mixture is merely a mass of concrete ingredi-
provide concrete with a normal and acceptable level of bleeding, ents that are temporarily suspended due to the agitation and
bleeding is usually not a concern and bleeding tests are rarely mixing of the material. Once the agitation stops, the excess
performed. However, there are situations in which bleeding water rises through the plastic mass to the surface or, more
properties of concrete should be reviewed prior to construction. appropriately, the solid ingredients settle. Although the actual
In some instances lean concretes placed in very deep forms have volume of the total ingredients does not change, the height of
accumulated large amounts of bleed water at the surface. This the hardened concrete is less than the original plastic height as
not only creates a placing problem but also reduces the strength the bleed water will come to the surface and evaporate away.

1
Staff Vice President, Research and Technical Services, Portland Cement Association, Skokie, IL 60077-1083.

99
100 TESTS AND PROPERTIES OF CONCRETE

Fig. 1—Bleed water on the surface of a freshly placed concrete slab.

The accumulation of water at the surface of a concrete As bleeding proceeds, the water layer at the surface main-
mixture can occur slowly by uniform seepage over the entire tains the original height of the concrete sample in a vessel, as-
surface or at localized channels carrying water to the surface. suming that there is no pronounced temperature change or
Uniform seepage is referred to as normal bleeding. Localized evaporation. The surface subsides as the solids settle through
channels of water coming up through the concrete, sometimes the liquid (Fig. 3). Fig. 4 illustrates a typical bleeding curve re-
carrying fine particles, is termed channel bleeding and usually oc- lating subsidence of the surface to time. The initial subsidence
curs only in concrete mixes with very low cement contents, high occurs at a constant rate, followed by a decreasing bleeding
water contents, or concretes with very high bleeding properties. rate (Fig. 4).

Fig. 2—Harvard University stadium after 90 years of service (photo courtesy of Tim Morse).
 KOSMATKA ON BLEED WATER 101

Fig. 3—Demonstration of bleeding or settlement of ce-

ment particles in cement paste with water-cement ratios by
weight of 0.3, 0.7, and 2.2 (left to right). All cylinders contain Fig. 4—Typical bleeding curve for concrete illustrating
250 mL of paste and were photographed 1 h after the paste surface subsidence with respect to time [3].
was mixed and placed in the cylinders. Observe the accumu-
lation of bleed water for pastes with the higher water-ce-
ment ratios. paste within the concrete is greater than it is for paste alone. This
is due to the greater unit weight of the concrete. Consequently,
In the interval between the beginning of subsidence the hydraulic force induced by the aggregate in the concrete dis-
and setting, there are three primary zones describing the rupts the paste structure more so than in paste alone. Because
nature of the bleeding process (Fig. 5). These are the zones of of this, channeled bleeding develops in concrete sooner than it
(1) clear water at the sample surface, (2) constant water content does in paste alone and it develops at lower water-cement ratios
or density, and (3) compression. Figure 5 is a simplified version for concrete than for paste. The rate of bleeding in a concrete
of a five-zone analysis of the bleeding process of paste presented mixture is controlled by many variables that will be discussed
by Powers [3]. In the zone of constant water content, the water- later under the section on Effects of Ingredients on Bleeding.
to-cement ratio and density are essentially constant, even Bleeding capacity is the quantity of bleed water that a par-
though some water is moving through the zone. The compres- ticular concrete, mortar, grout, or paste mixture can release to
sion zone is a transition zone where the paste is being densified. the surface with respect to a certain depth. It is usually expressed
The water-to-cement ratio is also being reduced and the solid in terms of the settlement or change in height of the paste- or
particles represent a lesser volume than when originally placed. mortar-solid’s surface per unit of original sample height (or, in
The paste densifies until it stabilizes and stops settling or bleed- other words, the ratio or percentage of the total decrease in sam-
ing. At equilibrium, the paste achieves a stable volume and the ple height to the initial sample height). Bleeding capacity can
degree of consolidation in this fresh state dictates the hardened also be expressed as a percent of the mix water. Figure 7 illus-
properties of the paste, such as strength and durability. trates the bleeding capacity for pastes with a range of cements
Bleeding is often analyzed in terms of bleeding rate and and water-cement ratios. Table 1 shows bleeding capacities for
bleeding capacity. Bleeding rate is the rate at which the bleed concrete and mortar. Figure 8 demonstrates these data in com-
water moves through the plastic concrete, mortar, or paste. parison to paste. Bleeding capacity is directly related to the wa-
Bleeding rate can be expressed in terms of cubic centimetres ter and paste content of concrete mixtures. Higher water con-
of bleed water per second per square centimetre of sample sur- tents especially increase bleeding capacity. As expected, bleeding
face, centimetres per second, millimetres per second, or other capacity is closely related to bleeding rates (Fig. 9).
applicable units. The bleeding rate of pastes with different ce-
ments and water-to-cement ratios is shown in Fig. 6. The rate Duration of Bleeding
of bleeding is controlled by the permeability of the plastic
paste. As the solid particles settle, the flow of water is con- The length of time that the concrete bleeds depends upon the
trolled by the permeable space or capillaries between particles. depth of the concrete section as well as the setting properties
The bleeding rate of concrete is less than that of paste alone of the cementitious materials. A thin slab of concrete will set-
with the same water-cement ratio. However, this would be ex- tle or bleed for a shorter period of time than a deep section of
pected as the aggregate in the concrete replaces some of the vol- concrete. Likewise, a concrete that sets up quickly will bleed
ume of the paste. The velocity of the water moving through the much less than a concrete that takes many hours to set up.
102 TESTS AND PROPERTIES OF CONCRETE

Fig. 5—Illustration of the process of bleeding in cement paste.

Most bleeding occurs during the dormant period, when ce- Effects of Bleeding on Plastic Concrete
menting materials have little to no reaction. The dormant pe-
riod is commonly around an hour. However, chemical and Volume Change
mineral admixtures as well as different compositions and fine- Combining cement, water, and aggregates in a mixer creates a
nesses of cements can greatly affect the dormant period. Fig- disbursed and suspended state of particles in plastic concrete.
ure 10 illustrates the increase in bleeding with increased paste This suspended state is not stable because the heavier particles
height and dormant period. of cement and aggregate are forced downward through the

Fig. 6—Range in relationship between bleeding rate and Fig. 7—Range in relationship between bleeding capacity
water-cement ratio of pastes made with normal portland ce- (total settlement per unit of original paste height) and water-
ment and water. The range is attributed to different cements cement ratio of pastes made with normal portland cement
having different chemical compositions and finenesses [4]. and water. The range is attributed to different cements hav-
ing different chemical compositions and finenesses [4].
 KOSMATKA ON BLEED WATER 103

TABLE 1—Bleeding Capacities of Concretes and Mortars [3]

Mix by Mass, Water-Cement
Cement: Sand: Ratio by Mass, Approximate Bleeding Paste per Unit
Gravel w/c Slump, mm Capacity,
H Volume, p (
H)/ p

CONCRETE (AGGREGATE: 75 m TO 19 mm)

1:0.8:1.2 0.31 102 0.009 0.446 0.020
1:1.2:1.8 0.38 213 0.011 0.374 0.028
1:1.6:2.4 0.43 203 0.013 0.323 0.042
1:1.6:2.4 0.40 119 0.009 0.314 0.028
1:1.9:2.85 0.49 203 0.012 0.304 0.041
1:2.4:3.60 0.53 229 0.013 0.266 0.041

MORTAR (AGGREGATE: 75 m TO 4.75 mm)

1:0.8 0.34 ... 0.018 0.688 0.028
1:1.2 0.38 ... 0.019 0.614 0.034
1:1.6 0.41 ... 0.019 0.570 0.034
1:2.0 0.45 ... 0.019 0.506 0.037

lighter water by gravity. The downward movement of the solid The small amount of settlement or volume reduction is not
particles continues until settlement ceases when the particles of concern for most general construction practices or applica-
are in contact with one another and densify. As the surface of tions. However, applications in which concrete is being placed
the disbursed solids is replaced by water, the volume of solid under an item that it must support, such as a machine base,
matter decreases. Although the total volume of materials is should have little to no bleeding to prevent the formation of a
relatively constant, the volume after bleeding will be less than void between the concrete surface and the object the concrete
that of the original plastic mixture. is to support. Bleeding also increases the risk of plastic settle-
The total amount of settlement is proportional to the depth ment cracks over embedded items such as reinforcing steel.
of the freshly placed concrete. Settlement can occur even
though bleed water is not observed at the surface. This is be- Postbleeding Expansion
cause on many occasions, such as warm windy days, the rate of Following the bleeding period, expansion occurs within the
evaporation is sufficient to remove the bleed water as it comes paste. This postbleeding expansion is caused by a combination
to the surface. The amount of volume reduction is clearly of physical and chemical reactions occurring during the first
demonstrated in Fig. 4 in terms of settlement of the surface. stages of setting. In effect, the gel coating on cement grains, dis-
rupted during hydration, may exert enough pressure to cause
the sample to increase in volume. Most of this expansion oc-
curs within the first day. Typical one-day expansions range
from 0.05 % to 1 % for portland cement pastes at a water-ce-
ment ratio of 0.38 by weight [5,6]. Expansion beyond the first

Fig. 8—Bleeding capacities of mortars and concretes per Fig. 9—Relationship between bleeding rate and bleeding
unit of paste in the mix compared with the bleeding capacities capacity for cement paste using a variety of cements. Approxi-
of neat pastes of the same water-cement ratio. Concrete mix- mately 100 data points were used to develop the range [4].
ture proportions are by mass [3].
104 TESTS AND PROPERTIES OF CONCRETE

TABLE 2—Post-Bleeding Expansion of

Cement Pastea [5]
Observed Rate of
Age Interval, h Expansion, %/h Total Expansion, %

0 to 1 1/2 0 0
1 1/2 to 2 1/2 0.10 0.10
2 1/2 to 3 1/2 0.04 0.14
3 1/2 to 4 1/2 0.04 0.18
4 1/2 to 5 1/2 0.02 0.20
5 1/2 to 6 1/2 0.02 0.22
6 1/2 to 23 1/2 0.01 0.38

a
Type I cement was used with a water-cement ratio of 0.38 by mass. The age of
the paste at the end of the bleeding period was 1 h 12 min. The age at the
beginning of the expansion was 1 h 30 min. Expansion is expressed as the percent
of the depth of the sample.

Fig. 10—Bleeding time versus height of paste sample for

cements with different dormant periods [4].
Plastic Shrinkage
Plastic shrinkage, sometimes called setting shrinkage in older
literature, is shrinkage that occurs before the concrete has hard-
day can be expected to be less than half of that which occurs ened. This shrinkage results from a loss of free water in the
within the first day and is not likely to add more than 0.05 % mixture. The water loss and resulting shrinkage is caused by a
expansion. Table 2 illustrates the postbleeding expansion of a combination of loss of free water from the concrete due to
cement paste. The bleeding period ended at 1 h and 12 min bleeding and surface evaporation as well as consumption of the
and the expansion began at 1 h and 30 min. It must be realized water by the cement during hydration (autogenous and chemi-
that the amount of expected expansion for concrete would be cal shrinkage). The amount of bleeding and surface evaporation
much less than that for paste alone or even nondetectable. It predominantly control the amount of plastic shrinkage. If the
must also be realized that the samples in Table 2 were sub- rate of evaporation at the sample surface exceeds the bleeding
merged continuously during the test. Otherwise, autogenous rate, plastic shrinkage occurs. The three factors most signifi-
shrinkage opposes the expansion as the cement hydrates. cantly contributing to autogenous shrinkage are a concrete’s

Fig. 11—Plastic shrinkage cracks in concrete (photo courtesy of the Portland Cement Association).
 KOSMATKA ON BLEED WATER 105

chemical shrinkage, amount of bleeding, and time of harden- velocity, concrete temperature, and bleeding characteristics of
ing. Autogenous shrinkage is very small and usually does not the concrete (Fig. 12). At this stage, the concrete has obtained a
contribute significantly to total plastic shrinkage for concretes small amount of rigidity yet is unable to accommodate the rapid
with water-cement ratios greater than 0.42 [28]. The evapora- volume change induced by plastic shrinkage. Consequently,
tion and removal of the bleeding water from the concrete cre- tensile stresses develop and plastic shrinkage cracks form.
ates tensile stresses near the surface. These tensile stresses can The best way to prevent plastic shrinkage cracking is to
pull the concrete away from the form as well as form plastic prevent surface evaporation. Plastic shrinkage cracks can pene-
shrinkage cracks in the concrete that resemble parallel tears trate from one fourth to full depth of a concrete slab. Although
(Fig. 11). plastic shrinkage cracks may be unsightly, they often do not
The disappearance of the water from the surface of the con- hinder performance of nonreinforced concrete. Plastic shrink-
crete indicates when the rate of evaporation has exceeded the age cracks can also be reduced by the use of concrete with
bleeding rate. Shortly after this time, plastic shrinkage and higher bleeding characteristics as well as the use of fibers,
cracking occur. The time required to obtain this condition is evaporation retarders, shades, windbreaks, and plastic sheets
controlled by the air temperature, relative humidity, wind or wet burlap covering the slab.

Fig. 12—Nomograph demonstrating the effect of concrete and air temperatures,

relative humidity, and wind velocity on the rate of evaporation of surface moisture
from concrete. Evaporation rates exceeding 1 kg/m2/h are prone to induce plastic
shrinkage cracks in concrete (adapted from Ref [7]).
106 TESTS AND PROPERTIES OF CONCRETE

Senbetta and Bury [33] demonstrated that plastic shrink- Effect of Bleeding on Hardened Concrete
age cracking can be controlled in cold weather by lowering the
temperature of fresh concrete, thus minimizing the differential Strength and Density
between the temperature of the concrete and the ambient tem- The strength of hardened concrete is directly related to the wa-
perature. This results in a lower rate of evaporation of bleed ter-cement ratio. As the solid particles in the paste or concrete
water from the concrete surface. The use of a freezing-weather settle, they squeeze some of the water out of the paste, espe-
admixture can be used to offset slower setting time associated cially in the lower part of the placement. This lowers the water-
with a cooler concrete temperature. cement ratio and increases the strength. Because the degree of
Concretes with high cementing materials contents consolidation or settlement is not uniform throughout the
(around 500 kg/m3) or that use silica fume or low water-to- height of a sample (more consolidation at the bottom than the
cementitious material ratios (less than 0.42) aggravate plastic top), the strength can be expected to be slightly higher at the
shrinkage crack development. bottom than at the top.
The differential consolidation effect is usually identified by
Water-Cement Ratio an increase in concrete density. For example, a caisson place-
The water-cement ratio of a concrete mixture before bleeding is ment in the Chicago area illustrates the consolidation/density
higher than after bleeding. As illustrated in Figs. 3 and 5, a den- effect. The top, middle, and bottom of a 12-m-deep caisson had
sification of particles occurs. As the solid material compresses, densities of 2400, 2435, and 2441 kg/m3, respectively [29].
some of the water leaves the concrete and rises to the surface at Hoshino [10] demonstrated the effect of bleeding on strength,
which time it evaporates away. This applies to most of the illustrating strength increase with depth for normal-strength
concrete depth. However, where excessive bleeding occurs, the concretes with water-to-cement ratios of 0.6 and 0.7 by weight.
water-cement ratio of the surface concrete may actually be in- Matsushita and Sue [30] demonstrated that in the occurrence
creased and that particular concrete should be removed in of a cold joint, the strength of an element can be increased by
extreme situations. In addition, if the concrete is sealed or trow- removing bleed water from the lower layer of concrete.
eled before the bleed water comes to the surface, the water may With regular concrete, a differential settlement between
be trapped under the surface creating a weakened zone with a the paste and aggregates occurs. The aggregate can settle only
higher water-cement ratio than the rest of the concrete. In most until point-to-point contact (bridging) between aggregates
applications, normal bleeding is beneficial in reducing the ac- occurs. The paste continues to settle in between the stabilized
tual water-cement ratio of most of the concrete mixture in place. aggregate particles. This phenomenon also contributes to a
weakening of the paste due to bleeding, reduction of cement
Thixotropic Mixtures particles, and increase in water content in the upper portion of
Thixotropic mixtures of concrete, mortar, or grout have low the concrete. Separation between paste and aggregate due to
bleeding properties. They exhibit a cohesive nature and, be- the accumulation of bleed water around aggregate particles
cause of the ingredients in the mixture, they have little to no also reduces strength.
bleeding. This is illustrated in the left cylinder in Fig. 3, which
contains a paste with thixotropic properties. Thixotropic mix- Paste-Aggregate Bond
tures are important in applications where the volume of the Bleed water can accumulate under and alongside coarse
fresh concrete or mortar must equal the volume of the hard- aggregate particles (Fig. 13). This is especially prone to hap-
ened concrete or mortar. An example would be in the use of
supporting grout under machine base plates [8].

Placing and Finishing

Normal bleeding usually does not interfere with the placing and
finishing of concrete mixtures. However, excessive bleeding of
low-cement-content mixtures may cause undesirable early
segregation during transportation and placement. Some minor
bleed water is desirable to help keep the surface paste moist and
help provide lubrication during the finishing of concrete; how-
ever, bleeding should not be so excessive as to interfere with the
finishing operation. Concrete must never be finished with visible
bleed water on the surface as such practices promote dusting,
scaling, blisters, and other surface defects [9]. Usually bleed
water is allowed to evaporate away before finishing commences.
Concrete placed on a sub-base of low permeability, such as
clay, plastic sheeting, metal-deck forms, bituminous concrete,
or vapor barriers, bleeds noticeably more than concrete placed
on a granular base. Therefore, special care and planning
should be used when differences in bleeding rates are caused
by isolated vapor barrier locations.
If excessive bleeding occurs between lifts in a wall place-
ment, the water and about 100 mm of surface concrete should
be removed prior to placing the next lift. This practice, although
not usually needed for concrete with normal bleeding, removes
Fig. 13—Cross section view of concrete illustrating bleed
bleeding-induced low-strength concrete of poor durability (see
water accumulation along a coarse aggregate particle.
the sections on Durability and Removing Bleed Water).
 KOSMATKA ON BLEED WATER 107

pen when differential settlement occurs between the aggregate However, excessive water collection and void development can
and paste. Once the aggregate can no longer settle, the paste reduce bar embedment strength, paste-steel bond, and possibly
continues to settle allowing bleed water to rise and collect under promote corrosion of the steel at void locations—especially in the
the aggregate. Bleed-water channels also tend to migrate along presence of moisture, carbonation, or chlorides. Corrosion can
the sides of coarse aggregate. This reduction of paste-aggregate occur because the steel is not in contact with the corrosion
bond reduces concrete strength. This condition can partially be resistive paste. Welch and Patten [11] demonstrated the effect of
reduced by revibrating the concrete after some bleeding has bleeding or settlement on the bond stress of reinforcing steel.
occurred. A broken fracture face of a horizontally oriented For plain and deformed bars, bond stress was reduced with
crack from a vertical placement demonstrates the reduced settlement. As expected, top bars developed less bond than
paste-aggregate bond caused by bleeding. The upper portion bottom bars because more bleed water accumulated under top
will have half-embedded aggregate particles, whereas the lower bars than bottom bars (Fig. 15). Some differential bond stress
face (opposing face) will expose the socket or imprint of the can also be attributed to differential strength development
aggregate in the paste. This reveals the poor bond at the between the top and bottom of the specimen.
underside of the aggregate particle.
Sometimes, if the concrete’s surface mortar sets faster Durability
than the rest of the concrete due to hot weather conditions, Concrete mixtures are designed to be durable in the environ-
some of the coarse aggregate particles might settle, leaving a ment to which they are exposed. A normal, small amount of
small air void above the aggregate particle. Another phenome- bleeding does not reduce durability; however, excessive bleed-
non is the presence of flat particles near the surface that inhibit ing can have a serious effect if special precautions are not
bleed water from entering mortar that is above the aggregate. observed (Fig. 16). Durability and concrete’s resistance to
Upon rapid evaporation, this mortar dries out quickly and does aggressive chemicals, chlorides, acids, and sulfates are directly
not have the strength or durability of the surrounding mortar. related to the permeability and water-cement ratio of the
This can result in a condition called “mortar flaking” over the concrete. An increase in water-cement ratio and permeability
coarse aggregate particles. caused by excessive bleeding would reduce freeze-thaw, deicer-
scaling, and sulfate resistance, and allow aggressive materials
Paste-Steel Bond to enter the concrete.
Bleed water is prone to collect under reinforcing steel and The relationship between bleeding and durability is
other embedded items (Fig. 14). This is because not only does demonstrated in a study of plasticized, flowing concrete expe-
bleed water have a tendency to collect under large objects in riencing excessive bleeding. Bleed channels extending 12 mm
concrete, but as the concrete settles, the concrete pulls away below the surface and areas of high water-cement ratio were
from the steel leaving an air void and an easily accessible lo- observed by petrographic analysis. The weakened surface layer
cation for water to collect. and high water-cement ratio, induced by bleeding, contributed
A minor collection of bleed water or settlement under rein- to poor deicer-scaling resistance [12]. Weakened surface layers
forcement is not detrimental to strength development in the bar can sometimes delaminate upon exposure to impact or
because most of the stress is applied to the bar deformations. abrasion. Dusting of a concrete surface can also develop.

Fig. 15—The effect of concrete settlement on the bond

strength of horizontally embedded deformed bars using ASTM
Test Method for Comparing Concretes on the Basis of the Bond
Fig. 14—Imprint in concrete illustrating the collection of Developed with Reinforcing Steels (C 234). Bar diameter was 19
bleed water voids under a smooth steel bar held firmly in a mm. Concrete slump was 50 to 125 mm. The top and bottom
horizontal position during and after placement [2]. bars were 75 and 230 mm above the sample base [11].
108 TESTS AND PROPERTIES OF CONCRETE

Fig. 16—Deterioration at top of lifts in a dam from which the porous-nondurable

concrete was not adequately removed prior to sequential lift placement. The poor qual-
ity concrete at the lift surface was caused by excessive bleeding and deteriorated due to
frost action.

Corrosion of reinforcement is more likely when bleed wa- particular example, increased degrees of subsidence were
ter collects under reinforcement steel. Also, if the upper few formed by increasing the rate of evaporation over the concrete.
centimetres of a deep concrete section become significantly Contrary to common belief, this figure illustrates that in-
porous due to excessive bleeding, carbonation, chlorides, air, creased bleeding or subsidence can actually improve scale re-
and moisture would more readily reach the steel, thereby sistance for this nonair-entrained concrete. It must be realized,
inducing corrosion. however, that this concrete was finished after the bleed water
Grouted prestressing strands can experience corrosive at- had evaporated away from the surface and after maximum
tack, particularly corrosion pitting, in the presence of grout bleeding had occurred [14].
bleed water and air. The corrosion occurs in a sealed environ- The scale resistance of a concrete surface can be jeopard-
ment despite the high pH of the bleed water and low levels of ized whenever the plastic material near the surface has its wa-
chlorides. The corrosion mechanism is most likely a carbona- ter content changed or its ability to transmit bleed water is
tion reaction involving carbon dioxide in the air and calcium hy-
droxide in the grout bleed water. This reaction creates regions
of low pH at the surface of the bleed water where the passivity
of the steel is lost and corrosion subsequently occurs [31].
Bleeding channels can carry lightweight materials to the
surface that can reduce abrasion resistance as well as allow
popouts to form. A chemical analysis of bleed water indicates
that channeled or uniform bleeding can leach alkalies up from
the concrete to near the surface [13]. Upon evaporation of the
water, an accumulation of alkalies can develop in the upper
few millimetres of the surface. If alkali-aggregate-reactive par-
ticles are in this zone, the increased concentration of alkalies
can aggravate alkali-aggregate reactivity (AAR), possibly result-
ing in popouts or surface cracking. A similar accumulation of
alkalies can develop if the surface is sealed too early during
finishing. Landgren and Hadley [27] found that AAR-induced
popouts can be minimized by wet curing as early as possible,
protecting concrete from drying prior to finishing, and avoid-
ing hard troweling.

Scaling Fig. 17—The relationship between scaling and subsidence.

The relationship between bleeding and scaling depends upon The rate of scaling  numerical scale rating divided by the num-
the placing, finishing, and curing practices. Figure 17 illus- ber of cycles. Higher scaling values indicate more scaling. Con-
trates the relationship between scaling and subsidence. In this crete was wood floated after maximum subsidence [14].
 KOSMATKA ON BLEED WATER 109

reduced. Usually, this is achieved by reducing the water-cement

ratio or increasing the cement content of the surface material.
This can be done by finishing the slab prior to the accumula-
tion of bleed water, or by sprinkling dry cement onto the slab
to take up excess water to facilitate finishing.
Steel troweling early seals the surface much more than
wood floating. The water rises to the surface and hits the sur-
face stratum of more impermeable cement paste. It cannot
penetrate through and merely accumulates under the surface.
This accumulation of water either creates a weakened zone of
paste or, in some cases, actually creates a water void beneath
the surface. Upon freezing, this weakened zone or water-filled
void can scale off the surface. Figure 18 illustrates that final
finishing prior to the completion of bleeding significantly
increases scaling.

Mortar Flaking
Mortar flaking resembles scaling or a flat popout. It is identi-
fied by a loss of mortar over flat coarse aggregate particles at
the surface. Large flat coarse aggregate particles block the mi-
gration of bleed water to the mortar over the aggregate. If Fig. 18—Effect of time of final finish on scale resistance of
rapid evaporation occurs, as with an unprotected surface on a concrete after 100 cycles of test. Higher scaling values indicate
hot windy day, the mortar drys out, shrinks slightly more than more scaling. All samples received a wood float strikeoff. Con-
the surrounding mortar, and, due to a lack of water for hydra- crete with a final wood finish at 50 % subsidence scaled much
tion, does not develop adequate strength for frost resistance. more than concrete finished after bleeding stopped or con-
Upon repeated freezing in a wet condition, the surface mortar crete not receiving a final finish [14].
deteriorates and exposes the underlying coarse aggregate, usu-
ally with a flat surface parallel to the concrete surface (Fig. 19).
Further deterioration to the surrounding mortar usually does Surface Delamination
not occur, as it received the necessary water for proper Surface delamination here refers to the separation of a large
strength gain. This condition should not be confused with area of surface mortar from the base concrete (Fig. 20). The 1/4-
popouts caused by aggregate that swells excessively upon to 1-cm-thick surface delamination can occur in sizes ranging
water saturation or freezing. from 10 to 100 cm in diameter. The remaining exposed surface

Fig. 19—Mortar flaking over coarse aggregate (photo courtesy of the Portland Cement Association).
110 TESTS AND PROPERTIES OF CONCRETE

Fig. 20—Delaminated surface caused by early finishing that trapped

bleed water under the surface (photo courtesy of The Aberdeen Group).

resembles a scaled surface with coarse aggregate exposed. The reducing the water-cement ratio. The addition of cement to fa-
cause of surface delamination is the accumulation of bleed wa- cilitate finishing also reduces the water-cement ratio. These
ter under the surface creating a void or weakened zone (Fig. 21). two practices both reduce the settling rate of the surface and
Upon the freezing of water in the void or weakened zone, the make it more impermeable. This allows planes of weakness to
surface delaminates in sheet-like form. In some instances, de- develop and bleed water to accumulate under the surface,
lamination can occur with interior slabs not exposed to freez- forming a void. Finishing operations should be delayed as long
ing simply because of the large void under the finished surface. as the setting time will permit, and the sprinkling of cement on
The consolidation of a surface by floating and troweling to the surface should be avoided to minimize the risk of devel-
too early squeezes the water out of the top surface layer oping a plane of weakness or void beneath the surface. Early

Fig. 21—Illustration of weakened zone or bleed water void under a finished surface.
 KOSMATKA ON BLEED WATER 111

Fig. 22—Blisters (photo courtesy of the National Ready Mixed Concrete Association).

steel troweling is especially prone to trapping bleed water be- because of a potential change in water-cement ratio. A concrete
neath the surface. Jana and Erlin [34] show project examples that bleeds enough to increase the water-cement ratio at the
of delamination caused by trapped bleed water and they pres- surface will create a lighter-colored surface. Consequently, a
ent circumstances leading to the entrapment. concrete placement with nonuniform bleeding can possibly re-
Delaminated areas, still intact, can be located by chaining, sult in blotchy-colored areas of light and gray. In wall place-
hammering, or by electro-mechanical sounding procedures all ments, sand streaks can form as the bleed water collects and
outlined in ASTM Practice for Measuring Delaminations in rises along the form face. As the bleed water moves upward
Concrete Bridge Decks by Sounding (D 4580). Delaminated ar- along the form in long bleed-water channels, it washes away
eas produce a hollow sound upon impact and can often be some of the paste leaving behind a somewhat sandy appear-
lifted off the base concrete with a knife or screwdriver. ance (Fig. 24).

Blisters
Blisters are small bubbles of water that form under the surface
during finishing (Fig. 22). They usually occur during or shortly
after steel troweling, but before bleeding has stopped. If punc-
tured while the concrete is plastic, water will usually squirt out.
Spaced a few centimetres or decimetres apart, blisters are usu-
ally 1 to 10 cm in diameter, 1/4 to 1 cm thick, and visibly rise
above the surface. They can form by the accumulation of wa-
ter under the surface at particular locations—often at the top
end of a bleed-water channel (Fig. 23). Formation of blisters is
usually an indication that the surface was finished or closed up
too early. They are more apt to occur on interior steel-troweled
floors. Blisters can also form due to an excess of air in the con-
crete. An excess of fines or a lack of adequate vibration can
also trap air under the finished surface [9].

Surface Appearance
Uniform bleeding on flat work should not affect the color of
the surface; however, concretes placed adjacent to one another Fig. 23—Cross section of a blister, illustrating the bleed
that have different bleeding rates or different bleeding prop- water void under the surface (photo courtesy of the Portland
erties can induce a color change in the surface primarily Cement Association).
112 TESTS AND PROPERTIES OF CONCRETE

Fig. 24—Sand streaks along a wall caused by excessive bleed water rising along the form (photo
courtesy of The Aberdeen Group).

Effects of Ingredients on Bleeding The amount of SO3 that can be leached from cement in a
short time has been found to correlate rather well to bleeding.
The individual ingredients and the amount of each ingredient An increase in SO3 reduces bleeding [4]. The correlation be-
in concrete have a major effect on the bleeding characteristics tween water-soluble alkalies in the cement and bleeding is not
of concrete. This section will briefly review the effects of some good. The trend is that an increase in alkalies reduces bleeding;
of these common ingredients and their proportions. however, other factors such as the precipitation of gypsum and
SO3 content probably have an overshadowing effect.
Water Content and Water-Cement Ratio
The water content and water-cement ratio predominantly con-
trol the bleeding of concrete (Fig. 25). Any increase in the
amount of water or the water-to-cementitious material ratio re-
sults in more available water for bleeding. Both bleeding ca-
pacity and bleeding rate are increased with increased water
content. A one-fifth increase in water content of a normal con-
crete mixture can increase bleeding rate more than two and a
half times [3].

Cement
The type, content, chemistry, and fineness of cement can all in-
fluence bleeding properties. As the fineness of the cement in-
creases, the amount of bleeding decreases (Fig. 26). Increases
in cement content, as it relates to the reduction of water-
cement ratio, also reduces bleeding (Fig. 27).
Because the chemical and physical properties of cement
are interdependent on one another in how they affect bleeding,
it is difficult to isolate the effect of a particular property or
chemical compound. Only reactions that occur during the mix- Fig. 25—Relationship between water-cement ratio and
ing period or bleeding period will affect the bleeding rate. Fig- bleeding of concrete. Bleeding is expressed as a percent of mix
ure 10 illustrates the effect of cements with different dormant water [15].
periods on bleeding.
 KOSMATKA ON BLEED WATER 113

Fig. 26—Effect of cement fineness by Wagner turbidim- Fig. 27—Relationship between cement content and
eter on bleeding capacity of paste. Note that Wagner values bleeding of concrete. The increased cement content reflects a
are a little more than half of Blaine values [4]. decreased water-cement ratio. Bleeding is expressed here as a
percent of mix water [15].

The correlation between bleeding and heat liberation However, each material and different source all have varying
demonstrates that bleeding can be reduced by a higher degree effects.
of initial chemical reactivity occurring with the cement shortly Fly ash usually reduces bleeding. Table 3 compares the
after mixing. Tricalcium aluminate has a high degree of reac- bleeding of concretes containing ten different fly ashes with re-
tivity and heat-producing capacity and is considered an influ- spect to their water reduction and performance in comparison
ence on bleeding properties. to two controls. Most of the fly ashes reduce bleeding com-
pared to the concretes with cement only. Class C fly ashes re-
Supplementary Cementing Materials duce bleeding much more than Class F ashes in this study. The
Fly ash, slag, silica fume, rice husk ash, and natural pozzolans ability of fly ashes to reduce bleeding appears to be in their
can reduce bleeding by their inherent properties and by in- ability to reduce the water demand in the concrete to achieve
creasing the amount of cementitious materials in a mixture. a particular slump. This is not always the case, as can be

TABLE 3—Bleeding of Concretes With and Without Fly Ash [16]a

Change in Mixing
Water-Cementing Water Requirement
Fly Ash Class of Fly Ash Bleeding, % Bleeding, mL/cm2 Material Ratio by Compared to 307
Identification (ASTM C 618) of Mix Water of Surface Mass, W/CM kg/m3 Control, %

A C 0.22 0.007 0.40 7.0

B F 1.11 0.036 0.42 2.3
C F 1.61 0.053 0.42 2.3
D F 1.88 0.067 0.45 4.6
E F 1.18 0.035 0.41 4.7
F C 0.13 0.004 0.40 7.0
G C 0.89 0.028 0.42 2.3
H F 0.58 0.022 0.44 2.3
I C 0.12 0.004 0.42 2.3
J F 1.48 0.051 0.43 0
Average of
Class C 0.34 0.011 0.41 ...
Class F 1.31 0.044 0.43 ...
Control Cement
Mixtures Content, kg/m3
1 307 1.75 0.059 0.43 ...
2 282 2.42 0.080 0.48 ...

a
Concretes had a slump of 75 to 100 mm and air content between 6 and 7 %. Test mixtures contained 75 % cement and 25 % fly ash by mass of ce-
mentitious material, based on a 307 kg/m3 cementitious material content. Control mixtures contained no fly ash. Bleeding tested as per ASTM C 232.
114 TESTS AND PROPERTIES OF CONCRETE

In a study on calcined kaolin clay in mass gravel concrete,

pozzolan dosages at 30 and 50 % by volume of cementitious
material resulted in bleeding capacities of 9.4 and 6.9 %, re-
spectively. The control mix with 120 kg/m3 of portland cement
had 11.9 % bleeding [18].
Silica fume can greatly reduce bleeding, primarily due to
its extreme fineness. Compared to a control with 20 mL of
bleeding in samples 40 cm high, one study demonstrated that
concretes with silica fume at dosages of 3, 7, and 13 kg/m3 had
bleeding capacities of 12, 8, and 3 mL, respectively [19]. Silica-
fume concretes with very low water-to-cementing materials ra-
tios essentially have no bleed water available to rise to the
surface. Consequently, such concrete mixtures are prone to
plastic shrinkage cracking if proper precautions are not taken
to reduce or eliminate surface evaporation while the concrete
is in the plastic state.
Rice husk ash (also called rice hull ash) reduces bleeding
Fig. 28—Comparison of bleeding concrete versus fly ash proportionately with the amount of ash in the paste. The fine-
mortar water requirement of ASTM C 618 [16]. ness of the material is primarily responsible for the reduction
of bleeding [20].

Aggregate
observed with Fly Ash H, which increased water demand and Ordinary variations in aggregate grading have little effect on
yet still reduced bleeding. Fly Ash J, which has no effect on wa- the bleeding of concrete. This assumes that there is no appre-
ter, also demonstrated a reduction in bleeding. However, a di- ciable change in the minus 75-m material. Table 4 demon-
rect correlation between concrete bleeding and the water strates that the specific surface area of the sand, or the fine-
requirement of the ASTM Specification for Fly Ash and Raw or ness, has little effect on the bleeding rate of mortar at four
Calcined Natural Pozzolan for Use as a Mineral Admixture in different ranges of water-cement ratio. However, aggregates
Portland Cement Concrete (C 618) mortar test exists (Fig. 28). that contain a high amount of silt, clay, or other material pass-
The retardation effect of the fly ashes did not correlate with ing the 75-m sieve can have a significant effect in reducing
bleeding [16]. the bleeding of concrete. This is not surprising as the aggregate
Ground granulated blast furnace slags can increase bleed- represents only a small amount of the surface area within a
ing capacity and have little effect on bleeding rate compared to concrete mixture. For example, consider a concrete mixture
mixes with portland cement only. The effect of slag on bleed- with proportions of one part cement to six parts aggregate
ing is primarily due to the fineness of the slag [17]. (coarse plus fine) by mass. The total surface area of the aggre-
Natural pozzolans, such as calcined clay or ground di- gate is only 5 % of the total surface area of the concrete
atomite, usually reduce bleeding. The primary influence is re- mixture. The surface area of coarse aggregate is essentially
lated to the pozzolan’s fineness and its effect on water demand. negligible [3]. Similar results have been found by other

TABLE 4—Effect of the Specific Surface Area of Sand on the Bleeding Rate
of Mortars [3]
Water-Cement Ratio Specific Surface Area
Reference Number by Mass Bleeding Rate, 106 cm/s of Sand, cm2/cm3

47 0.384 30 86
51 0.393 36 99
55 0.402 37 113
59 0.393 32 126
48 0.431 43 86
52 0.443 50 99
56 0.452 47 113
60 0.439 48 126
49 0.477 52 86
53 0.490 63 99
57 0.508 60 113
61 0.490 60 126
50 0.646 114 86
54 0.668 121 99
58 0.693 113 113
62 0.668 112 126
 KOSMATKA ON BLEED WATER 115

Fig. 30—Effect of high range water reducers on concrete

bleeding when used to reduce water content. Mix C is the
Fig. 29—The effect of entrained air on bleeding rate of paste control and Mixes 1 to 4 contain HRWR admixtures inducing
[3]. water reductions of 13, 10, 10, and 17 %, respectively. ASTM C
232 bleeding expressed in percent of mix water. Nominal
cement content is 223 kg/m3 and control water content is 157
kg/m3 [15].
researchers, including a negligible reduction in bleeding with
reduced particle size [10]. Aggregates that increase water de-
mand, such as crushed rock, tend to increase bleeding due to
the higher water content in the mix [32]. ing. The effect of high-range water reducers or plasticizers is
Aggregates that significantly increase paste content, such similar to that of normal-range water reducers. A reduction in
as gap-graded aggregates, can also increase bleeding, especially water content through use of a water reducer results in re-
if the combined aggregate grading is missing the 2.36-mm to duced bleeding. When high-range water reducers are used to
9.5-mm sizes. reduce the free water content in concrete significantly, they
Järvenpää [32] determined the relationship between likewise dramatically decrease the bleeding of concretes as il-
aggregate fineness and pore area and bleeding in conjunction lustrated in Fig. 30. These data compare concretes of equal
with absorption tendencies and water-retaining capacity of slump.
aggregate. As the threshold for the ability of an aggregate to Table 5 illustrates the effect of plasticizers on increasing
hold mixing water is passed, the bleeding tendency quickly slump without influencing the water-cement ratio and also
increases. compares a high-slump versus low-slump control. At a constant
water-cement ratio, the concretes with high-range water reduc-
Chemical Admixtures ers have slightly more bleeding than the low-slump control, but
The most predominantly used admixtures are air-entraining significantly less bleeding than the high-slump control. At
admixtures and water reducers. In laboratory and field appli- equivalent water-cement ratios, the flowing concretes with the
cations, it has been observed that entrained air reduces bleed- plasticizers bleed more than the control, but significantly less
ing. The relationship between air content and bleeding rate for than the control of the higher water-cement ratio and high
paste is illustrated in Fig. 29. Bruere found that not only do the slump [22].
air-entraining agents reduce bleeding by inducing an air-void Figure 31 illustrates the effect of plasticizers on concrete
system in the concrete, but also the air-entraining admixture it- bleeding with two different cements used to make flowing con-
self can slightly reduce bleeding rates [21]. crete. All the concretes have the same nominal water and ce-
Because water reducers reduce the water content of a con- ment contents. The admixture was added to increase slump to
crete mixture, it is expected that they likewise reduce the bleed- between 175 and 225 mm. These mixes had more bleeding

TABLE 5—Bleeding of Concretes With and Without Plasticizers (ASTM C 232

Bleeding Test) [22]
Water-Cement Water Content, Bleeding, % by Bleeding, mL/cm2
Mix Identification Ratio by Mass kg/m3 Slump, mm Mass of Mix Water of Surface

Control 1 0.47 143 75 1.09 0.031

Control 2 0.58 171 215 3.27 0.143
Melamine 0.47 143 215 1.59 0.060
sulfonate
Naphthalene 0.47 144 225 1.50 0.059
sulfonate
116 TESTS AND PROPERTIES OF CONCRETE

Fig. 31—Effect of plasticizers on bleeding of flowing concrete. The

nominal cement, water, and air contents for all mixes were 323 kg/m3, 161
kg/m3, and 6  1 %, respectively. The initial slump was 75 to 125 mm and,
after the admixture addition, the slump was 175 to 25 mm. The increased
fluidity increased bleeding. Mix C is the control and Mixes N, M, B, and X
contain different plasticizers [12].

than the flowing concrete in Table 5 because of a higher initial Special bleed-reducing admixtures based on cellulose de-
water content. Figure 31 demonstrates that increased fluidity rivatives, water absorbing resins, or various other chemical for-
induced by the plasticizer increased bleeding and it demon- mulations can significantly reduce or eliminate bleeding.
strates the effect of different cement and admixture combina- These admixtures make concrete mixtures very cohesive and
tions on bleeding. Excessive bleeding in plasticized, flowing thixotropic and therefore are very effective at preventing seg-
concrete can reduce surface durability and deicer-scaling re- regation. These admixtures are rarely used for normal con-
sistance [12]. Excessive bleeding can be avoided by optimal ma- struction but are very effective for special applications.
terial selection and mixture proportioning.
Calcium chloride is commonly used as an accelerator in Effects of Placement Conditions on Bleeding
nonreinforced concrete. Sodium chloride and calcium chlo-
ride both can reduce bleeding significantly. The influence of Placement Size and Height
accelerating salt such as calcium chloride may be largely due The depth of a concrete placement directly affects the amount
to the effect of its acceleration of setting. The effect of calcium of bleeding. As shown in Fig. 32, the amount of bleeding
chloride on bleeding time and bleeding capacity is illustrated increases with sample depth. This phenomenon has been ob-
in Table 6. Calcium chloride tends to make cement paste more served especially in placements of low-cement-content con-
susceptible to channeling. Little is known about how the dif- crete mixtures in deep dam sections, where many centimetres
ferent chemical compounds and chemical admixtures affect of water can accumulate over a lift just a few metres deep.
bleeding, as bleeding is not a required test in ASTM admixture The shape of vertical forms can produce sufficient strain
specifications. in the plastic concrete to induce cracks or zones of weakness

TABLE 6—Effect of Calcium Chloride on Bleeding Rate and Bleeding Capacity [4]a
ASTM C 150 Admixture Bleeding Time, Bleeding Rate, Bleeding
Cement Type Addition min 106 cm/s Capacity,
H

I none 71 163 0.085

CaCl2 38 146 0.040
II none 71 139 0.072
CaCl2 46 103 0.039
III none 46 129 0.049
CaCl2 26 89 0.020

a
Amount of CaCl2 was 1 % of the cement mass and about 2.1 % of water. The water-cement ratio was 0.466 by mass.
 KOSMATKA ON BLEED WATER 117

Weather Conditions
Weather conditions can have a significant effect on bleeding.
Figure 33 illustrates how the increase in wind velocity and, con-
sequently, rate of evaporation significantly increase the
amount of surface subsidence. The wind velocity and increased
evaporation greatly increase the capillary force at the surface,
pulling the bleed water out of the concrete.
The bleeding rate increases with an increase in tempera-
ture; however, the bleeding capacity tends to be nearly con-
stant. In a test on 25 different cements, going from 23.5 to 32°C
resulted in a 20 % increase in the rate of bleeding for cement
paste. This is primarily due to the decrease in water viscosity
with an increase in temperature. In general, there was little
change in bleeding capacity with the change in temperature
for most of the cements studied [4].
Klieger illustrated that changes in temperature for con-
crete did not affect the subsidence of the surface [14]. At a
wind velocity of zero for nonair-entrained concrete, the subsi-
Fig. 32—Relationship between the amount of bleeding
dence at temperatures of 10, 23, and 32°C was 0.502, 0.413,
and sample height for concrete [10].
and 0.454 mm, respectively. The average subsidence was 0.456
mm for this concrete. Whiting found bleeding of concrete at 23
and 32°C to be 2 and 3.2 % by mass of mix water for normal
due to bleeding and settlement. Deep and narrow placements concrete. The addition of plasticizers in some cases reduced
in which the walls are not parallel are more prone to settle- bleeding at higher temperatures [12].
ment strain. These situations can create stresses within the
plastic concrete upon settlement and bleeding. For example, Consolidation and Revibration
assume that the settlement along a vertical face is constant. The Surface or internal vibration should not significantly affect
settlement along an inward incline would be less than that the amount of bleeding. However, some studies indicate
along the vertical face. This would create some possible move- that a 2- to 3-s vibration period can slightly increase the
ment from one side of the form to the other, inducing shear bleeding capacity; however, the use of an internal vibrator in
strain and possibly faults to form depending on the angle of concretes from 20 s to 10 min tends to reduce bleeding
the slope, distance between the wall faces, the depth of the capacity. This is partly due to the reduced volume of the
placement, rate of filling, and bleeding capacity of the con- matrix as well as the decreasing degree of dispersion of the
crete. The opposite situation occurs when the concrete parti- aggregate and the expulsion of some of the entrapped air.
cles settle away from an outward incline where a layer of water Normal field vibration would not be expected to greatly affect
can develop and collect along the sloped surface. Such condi- bleeding [23].
tions can create localized bleed channels along the form re-
sulting in sand streaks.
Settlement can also form zones of weakness in forms that
have areas of significantly reduced cross section. For example,
if a T section is filled in one placement, faulting can develop up
from the corners of the narrow form. Obviously, the way to
avoid this problem is to fill the narrow portion first and then
place the rest of the concrete after the concrete in the narrow
section has settled. The greater the bleeding capacity of the
concrete the greater the tendency of such faults and arches to
form. These faults can be eliminated by revibration.

Impermeable Sub-bases
Concrete placed on sub-bases of low permeability such as plas-
tic sheet vapor barriers, bituminous, or clay appear to bleed no-
ticeably more than concretes placed on granular sub-bases. All
bleed water must rise to the surface when concrete is placed
on an impermeable sub-base. When concrete is placed on a
granular material, some of the bleed water can flow out the
bottom of the slab into the granular material. Bleeding can be
reduced by placing 50 to 75 mm of compacted granular fill on
the impermeable surface. This practice also helps control slab
curling. Construction in which only parts of a placement are
on a vapor barrier will need special care in finishing practices
due to nonuniform bleeding in order to provide a uniformly
durable surface. Fig. 33—Effect of wind velocity on subsidence [14].
118 TESTS AND PROPERTIES OF CONCRETE

the bleeding to solve particular problems. Limits on the amount

TABLE 7—Effect of Revibration on Bleeding of bleeding are not available; however, acceptable bleeding rates
and Strength of Concrete [24]a can be determined by test and correlated to field experience.
Interval Before Bleeding, % of Compressive
Revibration Mix Water Strength, MPa Increasing Bleeding
The easiest way to increase the bleeding is to increase the
0 2.9 28 amount of water in the concrete mixture as well as reduce the
1h 3.5 32 amount of the fines in the sand and amount of cementing ma-
2h 3.4 39 terials. Bleeding may need to be increased to help prevent plas-
3h 3.2 31 tic shrinkage cracking or to improve a concrete’s ability to be
4h 2.8 29 finished in dry weather.

Reducing Bleeding
a
The cement was an ASTM C 150 Type I. The aggregate was natural sand and Bleeding may need to be reduced for a variety of reasons, in-
crushed trap rock with a maximum size of 25 mm. The cement content was 307 cluding to facilitate finishing operations, minimize the forma-
kg/m3. The slump of the concrete was 75 mm. Compressive strength was estimated tion of weak concrete at the top of lifts, reduce sand streaking
by impact hammer.
in wall forms, minimize bleed-related corrosion in grouted ten-
dons, or to stabilize the hardened volume with respect to the
After concrete has completed its settlement, it can bleed plastic volume of the concrete. The most important ways of re-
and settle again upon internal revibration of the concrete. This ducing bleeding in concrete are as follows:
is illustrated in Table 7 where consistent degrees of additional 1. Reduction of the water content, water-cementitious mate-
bleeding were obtained for revibration intervals up to 4 h. Also rial ratio, and slump.
note that the compressive strength was also increased. This 2. Increase the amount of cement resulting in a reduced wa-
would be expected as removal of some of the bleed water ter-cement ratio.
would have resulted in a lower water-cement ratio [23]. 3. Increase the fineness of the cementitious materials.
4. Increase the amount of fines in the sand.
Remixing 5. Use supplementary cementing materials such as fly ash,
Just as revibration allows bleeding to occur after concrete is slag, or silica fume.
revibrated, remixing of the cementitious system after it has set- 6. Use blended hydraulic cements.
tled has a similar effect with little change in bleeding charac- 7. Use chemical admixtures that reduce water-cement ratio
teristics. Table 8 illustrates that particular cement pastes or by other means are capable of reducing the bleeding of
remixed within the first hour bleed about the same amount as concrete.
before remixing with only slightly reduced bleeding rates and 8. Use air-entrained concrete.
capacities compared to the first settlement. This demonstrates
that concrete in a ready mixed truck for some period of time Removing Bleed Water
can, after remixing, resume its normal bleeding characteristics. Once the observation of bleed water has become a noticeable
problem, there is little time or opportunity to change the con-
Controlling Bleeding crete mixture, especially once the concrete is in the forms.
Therefore, certain techniques can be used to remove excess
Usually, little attention need be given to controlling bleeding of bleed water from a concrete surface if normal evaporation is
concretes made of normal ingredients at normal proportions. not satisfactory. If the concrete is placed within an enclosure,
However, this section will provide some guidance as to adjusting the temperature of the air can be increased and large fans can

TABLE 8—Bleeding Tests on Remixed Pastes [5]a

Time Between
Duration of Initial Mix and End
Rest Period,b min Bleeding Rate, 106 cm/s Bleeding Capacity,
H Bleeding, min of Bleeding, min

0 194 0.122 55 55
15 189 0.113 55 70
30 196 0.125 57 87
45 192 0.106 52 97
60 185 0.103 48 108
90 172 0.090 47 137
120 167 0.075 45 165

a
The water-cement ratio was 0.469 by mass. The schedule for the initial mixing was: 2 min mix, 2 min wait, followed by 2 min mix. The final remixing
was 30 s continuously. The depth of the paste was 36 mm. The temperature was 23.5°C. The results for remixed pastes are averages of two or three
tests.
b
This is the period after the initial mixing, at the end of which the final remixing was done.
 KOSMATKA ON BLEED WATER 119

be used to evaporate away some of the water. However, care ASTM C 940 is used to test the bleeding of grout, where
must be taken when using this technique to not remove so 800 mL of grout are placed into a 1000-mL graduated cylinder
much bleed water that the evaporation exceeds bleeding, re- and the cylinder is covered to prevent evaporation. The upper
sulting in potential plastic shrinkage crack development. surfaces of the grout and bleed water are recorded at 15-min
Vacuum dewatering can be accomplished by special intervals for the first hour, and hourly thereafter until bleeding
equipment that uses a filter mat placed on the surface of the stops. The final bleeding is recorded as the amount of bleed wa-
concrete. As the mat settles with the surface, bleed water is ter decanted as a percentage of the original sample volume.
forced up through the filter and the water is removed by a vac- ASTM C 941 is used to determine the water retentivity of
uum pump. Because this method consolidates a slab of con- grout under vacuum atmosphere. The test essentially deter-
crete more than would normally occur due to normal bleed- mines the relative ability of grout to bleed, or retain its water,
ing, the end product is a concrete with a higher strength, lower under pressure.
permeability, and a more abrasion-resistant surface. For verti- Several nonstandard test methods have also been used
cal applications such as wall forms, special liners can be placed for analyzing the bleeding properties of concrete. One of the
in the form prior to concrete placement that uniformly drain first was the Powers Float Method (Fig. 36). The float method
away the bleed water without the formation of sand streaking. was used to measure settlement of bleeding by measuring the
For normal concrete slabs experiencing occasional exces- subsidence of the small area at the top of the sample. The
sive bleeding that is not removed due to evaporation, a squeegee float consisted of a disk of lucite or bakelite to which a
or garden hose can be used to drag the water off of the surface straight glass fiber was mounted like the mast on a sailboat.
so that finishing can commence. Finishing should continue only The disk had a diameter of about 13 mm and a thickness of 3
after it has been ascertained that bleeding has nearly or com- mm. Movement of the float was measured with a micrometer
pletely stopped. If a finisher waits too long for the bleed water microscope. Because of the thixotropic structure of the ce-
to evaporate away, the concrete can harden and will not be able ment paste, the float remains in a relatively fixed position at
to be finished. For precast concrete elements experiencing set- the top of the paste during settlement. A layer of water is
tlement or bleeding problems, a centrifuge can be used to com- placed on the surface immediately after the float is installed
pact the concrete and remove the bleeding by centrifugal force. to prevent the development of capillary forces from evapora-
tion. The water depth is about 6 mm deep. The float is used
Test Methods both for paste, mortar, and concrete samples. However, for
concrete, the sample diameter was at least 500 mm. The float
Three standard ASTM test methods can be used to analyze the method was used in most of the early major findings on the
bleeding properties of concretes, mortars, pastes, and grouts. bleeding properties of paste and concrete. This test method
These are ASTM Test Methods for Bleeding of Concrete (C was used in most bleeding research at the Portland Cement
232), ASTM Test Method for Bleeding of Cement Pastes and Association in the 1930s to 1950s.
Mortars (C 243—discontinued in 2001), and ASTM Test Method
for Expansion and Bleeding of Freshly Mixed Grouts for Pre- Special Applications
placed-Aggregate Concrete in the Laboratory (C 940). Another
related ASTM standard is ASTM Test Method for Water Reten- Under some conditions, it is desirable to eliminate the bleeding
tivity of Grout Mixtures for Preplaced-Aggregate Concrete in capacity as much as possible. These are usually situations in
the Laboratory (C 941). which the surface level of the plastic volume of concrete, mor-
ASTM C 243 is based on a test method published in 1949 tar, or grout must be maintained. This would include, for ex-
[25]. The ASTM C 243 test apparatus is shown in Fig. 34. The ample, the construction of superflat floors where nonuniform
sample container is filled with mortar or paste. The collecting bleeding would interfere with surface tolerance.
ring is then placed to penetrate the sample. The burette and fun- In other applications, grout for preplaced-aggregate con-
nel are filled with carbon tetrachloride or 1,1,1-trichloroethane, crete must have a low level of bleeding in order to provide ade-
which are liquids that are denser than water. As the sample set- quate strength development within the entire mass. Grout used
tles and bleed water rises through the carbon tetrachloride or in post-tensioning ducts must minimize the formation of voids
1,1,1-trichloroethane to the top of the burette, bleeding meas- due to bleed water in order to prevent corrosion of tendons
urements are taken at regular intervals until settlement stops. within the duct. Grout placed under base plates for machinery
Bleeding rate is reported as cubic centimetres of bleed water or column applications must not allow bleed water to form
per square centimetre of sample surface per second. Bleeding voids. As the grout provides the support for these elements,
capacity is reported as cubic centimetres of water per cubic cen- bleeding must essentially be eliminated. Certain admixtures
timetres of paste or mortar. can be added to grout to prevent such bleeding and actually
ASTM C 232 has two procedures—one with vibration and induce a small amount of expansion to help eliminate bleed-
one without. Method A allows the concrete to be undisturbed water voids. Often, priority grouts are used for grouting base
in a container with a volume of about 0.014 m3 and the accu- plates [8].
mulation of bleed water is removed from the surface (Fig. 35).
Method B uses a similar sample and apparatus, along with a Mathematical Models
clamped-on cover, with vibration to help consolidate the con-
crete mixture. The sample is temporarily tilted when the water Ever since Powers’s work in the 1930s on bleeding, several re-
is drawn off to facilitate water collection. During the first 40 searchers, including Powers, developed several mathematical
min, water readings are taken every 10 min and every 30 min models for the bleeding rate and bleeding capacity for paste,
thereafter until bleeding stops. Bleeding is reported as millil- mortar, and concrete. Many of these relationships, pertaining
itres of bleed water per square centimetre of surface, or as a to the permeability of the paste, were based on Darcy’s law and
percent of the mix water. Poisseuille’s law. Stoke’s law was also used in empirical rela-
120 TESTS AND PROPERTIES OF CONCRETE

Fig. 34—ASTM C 243 test apparatus to determine the bleeding properties of

paste and mortar.
 KOSMATKA ON BLEED WATER 121

Conclusions
Bleeding is a fundamental property of concrete. By under-
standing its influences on plastic and hardened concrete prop-
erties and by understanding the effects of ingredients and
ingredient proportions on bleeding, bleeding can be economi-
cally controlled. With proper control, bleeding should not hin-
der concrete construction or adversely influence concrete
strength or durability.

References
[1] Johnson, L. J., Lecture Notes on Materials of Engineering,
Harvard Engineering School, Harvard University, Cambridge,
MA, 1926.
[2] Mardulier, F. J., The Bleeding of Cement: Its Significance in
Concrete, Paper No. 67, ASTM International, West
Conshohocken, PA, 1967.
[3] Powers, T. C., The Bleeding of Portland Cement Paste, Mortar
Fig. 35—ASTM C 232 test for bleeding of concrete; and Concrete, RX002, Portland Cement Association, Skokie, IL,
Method A without vibration. The container has an inside July 1939.
diameter of about 255 mm and a height of about 280 mm. The
[4] Steinour, H. H., Further Studies of the Bleeding of Portland
container is filled to a height of about 255 mm. The container
Cement Paste, RX004, Portland Cement Association, Skokie, IL,
is covered to prevent evaporation of the bleed water.
Dec. 1945.
[5] Steinour, H. H., Tests of Special Cements for Basic Research
Program, Major Series 290, Portland Cement Association,
tionships. Based on Poisseuille’s law of capillary flow, Powers Skokie, IL, April 1941.
and Steinour developed equations relating to the initial con- [6] Brownyard, T. L. and Dannis, M. L., Expansion of Pastes, Mortars
and Concretes During the First 24 Hours, Junior Series 354,
stant bleeding rate of paste. Unfortunately, the bleeding rate
Portland Cement Association, Skokie, IL, 1941.
diminishes with time. In addition, the early equations to deter-
[7] Menzel, C. A., “Causes and Prevention of Crack Development in
mine bleeding capacity assumed that the bleeding process was
Plastic Concrete,” Proceedings, Portland Cement Association,
not influenced by hydration of the cement. Skokie, IL, 1954, pp. 130–136.
As opposed to Powers’s consideration of bleeding as [8] Kosmatka, S. H., Cementitious Grouts and Grouting EB111,
merely sedimentation, Tan et al. [26] used a self-weight consoli- Portland Cement Association, Skokie, IL, 1990.
dation model to express the relationship between the different [9] Panarese, W. C., Cement Mason’s Guide, PA122, Portland
characteristics of the bleeding process. By using the consolida- Cement Association, Skokie, IL, 1990.
tion model, Tan et al. were able to obtain very good agreement [10] Hoshino, M., “Relationship Between Bleeding, Coarse Aggre-
with the bleeding process, even beyond the period of initial gate, and Specimen Height of Concrete,” ACI Materials Journal,
constant bleeding rate. March-April 1989, pp. 185–190.
Järvenpää [32] developed a bleeding model that considers [11] Welch, G. B. and Patten, B. J. F., “Bond Strength of Reinforce-
mix proportions, paste content, aggregate shape, aggregate ment Effected by Concrete Sedimentation,” Journal, American
porosity, aggregate surface area, fines surface area, and ad- Concrete Institute, Vol. 62, No. 2, Feb. 1965, pp. 251–263.
mixture effects. [12] Whiting, D. and Dziedzic, W., Effects of Conventional and High-
Range Water Reducers on Concrete Properties, RD107, Portland
Cement Association, Skokie, IL, 1992.
[13] Kobayashi, K. and Uno, Y., “Chemical Composition of Bleeding
Water on Fresh Concrete,” Seisan Kenkyu, Tokyo, Vol. 41, No. 2,
1989, pp. 128–130.
[14] Klieger, P., The Effect of Atmospheric Conditions During the
Bleeding Period and Time of Finishing on the Scale Resistance
of Concrete, RX072, Portland Cement Association, Skokie,
IL, 1956.
[15] Whiting, D., Effects of High-Range Water Reducers on Some
Properties of Fresh and Hardened Concretes, RD061, Portland
Cement Association, Skokie, IL, 1979.
[16] Gebler, S. H. and Klieger, P., Effect of Fly Ash on Some of the
Physical Properties of Concrete, RD089, Portland Cement
Association, Skokie, IL, 1986.
[17] Wainwright, P. J. and Rey, N., “Influence of Ground Granulated
Blast Furnace Slag (GGBS) Additions and Time Delay on the
Bleeding of Concrete,” 22(4) Cement and Concrete Composites,
2000, pp. 253–257.
[18] Saad, M. N. A., Pacelli de Andrade, W., and Paulon, V. A.,
Fig. 36—Powers float test for measuring bleeding. “Properties of Mass Concrete Containing an Active Pozzolan
Made from Clay,” Concrete International, July 1982, pp. 59–65.
122 TESTS AND PROPERTIES OF CONCRETE

[19] Scheissl, P. and Schmidt, R., “Bleeding of Concrete,” RILEM [27] Landgren, R. and Hadley, D. W., Surface Popouts Caused by Al-
Colloquium of Properties of Fresh Concrete, Paper No. 4, The kali-Aggregate Reaction, RD121, Portland Cement Association,
European Cement Association, Brussels, 1990. Skokie, Illinois, 2002, 20 pp.
[20] Hwang, C. L. and Wu, D. S., “Properties of Cement Paste [28] Holt, Erika E., Early Age Autogenous Shrinkage of Concrete, VTT
Containing Rice Husk Ash,” Fly Ash, Silica Fume, Slag, and 446, Technical Research Centre of Finland, Espoo, 2001, 196 pp.
Natural Pozzolans in Concrete, SP-114, American Concrete [29] Burg, R. G. and Fiorato, A. E., High-Strength Concrete in Massive
Institute, Detroit, MI, 1989, pp. 733–762. Foundation Elements, RD117, Portland Cement Association,
[21] Bruere, G. M., “Mechanism by which Air-entraining Agents Skokie, Illinois, 1999, 24 pp.
Affect Viscosities and Bleeding Properties of Cement Paste,” [30] Matsushita, H. and Sue, Y., “A Study on Relationship Between
Australian Journal of Applied Science, Vol. 9, Commonwealth Condition of Placing on Consolidated Fresh Concrete and Oc-
Scientific and Industrial Research Organization, Melburn, 1958, currence of Cold Joint,” Zairyo, Vol. 50, No. 8, 2001, pp.
pp. 349–359. 851–856.
[22] Gebler, S. H., The Effects of High-Range Water Reducers on the [31] Bricker, M. D. and Schokker, A. J., Corrosion from Bleed Water
Properties of Freshly Mixed and Hardened Flowing Concrete, in Grouted Post-Tensioned Tendons, RD137, Portland Cement
RD081, Portland Cement Association, Skokie, IL, 1982. Association, Skokie, Illinois, 2005, 77 pp.
[23] Powers, T. C., The Properties of Fresh Concrete, Wiley, New [32] Järvenpää, H., Quality Characteristics of Fine Aggregates and
York, 1968. Controlling their Effects on Concrete, Acta Polytechnica
[24] Vollick, C. A., “Effects of Revibrating Concrete,” Proceedings, Scandinavica, Civil Engineering and Building Construction
American Concrete Institute, Vol. 54, 1958, pp. 721–732. Series, Ci 122, Finnish Academy of Technologies, Espoo, 2001,
[25] Valore, R. C., Jr., Bowling, J. E., and Blaine, R. L., “The Direct and 243 pp.
Continuous Measurement of Bleeding in Portland Cement [33] Senbetta, E. and Bury, M. A., “Control of Plastic Shrinkage
Mixtures,” Proceedings, ASTM International, Vol. 49, 1949, Cracking During Cold Weather Concreting Using a Freezing
p. 891. Weather Admixture,” Concrete International, March 1991,
[26] Tan, T. S., Wee, T. H., Tan, S. A., Tam, C. T., and Lee, S. L., “A Con- pp. 48–53.
solidation Model for Bleeding of Cement Paste,” Advances in [34] Jana, D. and Erlin, B., “Delamination: The Sometime Curse of
Cement Research, Vol. 1, No. 1, Thomas Telford Services Ltd., Entrained Air,” Concrete Construction, Hanley Wood, Addison,
London, Oct. 1987, pp. 18–26. IL, January 2005, pp. 101–107.
 PART III
Hardened Concrete
13
Concrete Strength Testing

Celik Ozyildirim1 and Nicholas J. Carino2

Preface crete taken at job sites, either to determine the adequacy of the
mixture proportions developed for the particular job or to
THE TOPICS OF STRENGTH AND RELATED TESTING check for changes in strength that could indicate quality con-
were covered in the previous editions of the ASTM STP 169 se- trol problems for the concrete supplier or changes in ambient
ries by C. E. Kesler and C. P. Seiss, “Static and Fatigue Strength” conditions. When appropriately cured in place, these specimens
[1]; by C. E. Kesler, “Strength” [2]; by K. N. Derucher, “Strength” can also be used to determine when the concrete in the struc-
[3]; and by P. M. Carrasquillo, “Concrete Strength Testing” [4]. ture is sufficiently strong for application of construction loads,
The current chapter augments the topics addressed by these removal of formwork, or application of prestressing. In cases
previous authors by introducing new developments since STP where the strength of the in-place concrete is in doubt, speci-
169C was published. This chapter provides information on the mens may be cut from existing concrete placements and tested
nature of concrete strength, preparation of test specimens, test for strength, or nondestructive tests may be performed to esti-
procedures, significant factors affecting test results, strength re- mate the strength from previously established correlations.
lationships, nondestructive and in-place strength testing in- Unfortunately, concrete strength is not an absolute prop-
cluding maturity and temperature-matched curing, strength un- erty. Results obtained by testing a given concrete will depend
der combined stresses, and fatigue strength. on specimen shape and size, specimen preparation, and the
loading method. Therefore, existing standard test methods
Introduction must be followed for reliable results.

The most common concrete property measured by testing is Nature of Concrete Strength
strength. There are three main reasons for this. First, the
strength of concrete gives a direct indication of its capacity to This section discusses the nature of concrete strength and
resist loads in structural applications, whether they be tensile, those factors involved in the fabrication of the specimens that
compressive, shear, or combinations of these. Second, influence the measured strength. Factors involved in the test-
strength tests are relatively easy to conduct. Finally, correla- ing of concrete specimens are discussed in later sections.
tions can be developed relating concrete strength to other Reaction of cement with water forms hardened cement
concrete properties that are measured by more complicated paste, which binds together coarse and fine aggregate to form
tests. Caution should be exercised, however, when strength is a solid mass. Hardened paste consists of poorly crystallized
used to estimate other properties based on empirical correla- hydrates of various compounds, referred to collectively as gel,
tions. When a nonstrength property is of primary interest, that crystals of calcium hydroxide, unhydrated cement, and air
property should be measured directly. For example, when voids [5]. Concrete can be considered as a two-phase compos-
penetrability is of concern, a test related directly to penetra- ite, consisting of cement paste and aggregate, and its behavior
bility should be conducted, rather than relying on a strength under load is a result of the interaction of the two phases and
correlation that can be misleading. the interfacial regions between them. Failure of concrete
Strength tests of concrete specimens are used for three occurs as a result of the development of a network of micro-
main purposes: cracks that grow in length with increasing load to the point
• For research. where the concrete cannot support further load. The coarse
• For quality control and quality assurance. aggregate particles act as inclusions that can both initiate and
• For determining in-place concrete strength. arrest crack growth. The latter feature is beneficial in reduc-
In research, tests are used to determine the effects of vari- ing the brittleness of concrete.
ous materials or mixture proportions on the strength of con- Before external load is applied to concrete, fine cracks
crete. Strength tests are used to obtain reference values when exist in concrete at the interface of coarse aggregate and
other characteristics of concrete are being studied, such as abra- cement paste due to mechanical property differences and the
sion resistance or strength of structural members. In construc- occurrence of shrinkage or thermal strains. These pre-existing
tion, strength tests are conducted on specimens made from con- microcracks are responsible for the low tensile strength of

1
Principal Research Scientist, Virginia Transportation Research Council, Charlottesville, Virginia.
2
Consultant, Gaithersburg, Maryland.
125
126 TESTS AND PROPERTIES OF CONCRETE

Fig. 1—Stress-strain curve of concrete in compression and stages of microcracking.

concrete. As external load is applied, existing microcracks are strength of the cement paste. Various factors affect paste
stable up to about 30 % of the ultimate load, at which point strength. Probably the most important factor is paste density,
interfacial cracks begin to increase in length, width, and quan- which in turn depends highly on the water-cementitious mate-
tity. When 70–90 % of the ultimate strength is reached, cracks rials ratio (w/cm). As the w/cm is increased, the density of the
penetrate into the bulk paste leading to continuous larger paste decreases, and so does its strength. The nature of the ce-
cracks until the concrete cannot support additional load [5]. mentitious materials also affects paste strength. For example,
The shape of the compressive stress-strain curve of con- the cement type affects paste strength, as well as its strength-
crete is related to the formation and growth of microcracks. gain characteristics. The use of ground granulated blast-furnace
The process of microcracking and its relationship to the stress- slag and pozzolans, such as fly ash and silica fume, affect both
strain curve is illustrated in Fig. 1. Up to about 30 % of the the paste strength and strength gain. Chemical admixtures can
ultimate strength, the stress-strain curve is linear. When exist- be used to alter paste performance; for example, water-reduc-
ing interfacial microcracks begin to propagate, the curve starts ing admixtures reduce paste viscosity (make it flow more read-
to deviate from linear behavior; deviation from linearity ily), and other admixtures can be used to regulate time of set-
increases as more interfacial cracks are formed. When micro- ting and strength development. The entrainment of air without
cracks penetrate into bulk cement paste, deviation from lin- reducing the w/cm will decrease the strength due to the in-
earity increases at a faster rate. As ultimate strength is crease in voids in the paste. Finally, curing temperature and
approached, interfacial and bulk paste microcracks join to moisture conditions have a marked effect on the paste strength.
form continuous cracks parallel to the direction of loading. At High early curing temperatures, while producing higher early
some point, the extent of cracking is so great that the concrete strengths, will reduce strengths at later ages due to the forma-
cannot support additional load, and subsequently the stress tion of less dense and nonuniform hydration products [5]. Fur-
required for additional strain decreases. ther, hydration will continue only as long as free moisture is
Based on the above microcracking process, it is clear that available for reaction with the unhydrated cement. Thus, if con-
the ultimate strength of concrete is related strongly to the crete is allowed to dry prematurely, hydration will cease before
 OZYILDIRIM AND CARINO ON CONCRETE STRENGTH TESTING 127

desired properties are attained. In poorly cured members, con- ratory (C 192/C 192M). ASTM Test Method for Compressive
crete may never reach the strength obtained from standard- Strength of Concrete Cylinders Cast in Place in Cylindrical
cured cylinders at 28 days [6–8]. Molds (C 873) provides procedures for obtaining cylinders cast
In most concretes, aggregate strength is higher than the and cured in concrete slabs. The first two of these standards are
concrete strength and does not adversely affect the strength discussed in detail in the chapter by Lamond in this publication.
of concrete. In some lightweight and high-strength concrete, ASTM C 873 involves cylinders that are cast in special molds
however, aggregate strength may limit concrete strength. placed within a structural slab, so that the cylinders receive cur-
The strength of coarse aggregates is determined mainly by ing identical to that of the surrounding concrete in the structure
mineralogy and density. Beyond this, however, a smaller-sized until the time of test. Specimens made according to ASTM C 873
aggregate may have strength advantages in that internal will be referred to as push-out cylinders.
weak planes may be less likely to exist, and the smaller particles ASTM C 192/C 192M specifies procedures for preparing
reduce stress concentration effects at paste-aggregate interfaces. test specimens in the laboratory and is generally used in
Factors involving both the paste and aggregate will affect research and mixture proportioning studies. This procedure is
the strength of the paste-aggregate interface. The bond between considered to be the “ideal” condition for specimen prepara-
paste and aggregate particles will be stronger for smaller-sized tion, and will produce the most consistent results due to the
aggregates, which have a higher surface area per unit volume. high degree of control involved.
Also, a rough, angular surface texture such as exists in crushed Specimens prepared in the field according to ASTM C
aggregates increases interfacial bond strength [9]. The use of 31/C 31M have multiple uses. Those given standard moist cur-
pozzolans such as fly ash or silica fume results in denser hy- ing are used to check the adequacy of the mixture proportions
dration products and increases the strength of the paste-aggre- as supplied to the job, monitor the producer’s quality control,
gate interface [10,11]. Coatings, such as clay, on the aggregate and indicate changes in materials and other conditions
surface reduce interfacial strength. When concrete bleeds, the [15,16]. Specimens given curing similar to that of the struc-
rising bleed water is often trapped beneath coarse aggregate tural component are used to indicate the in-place strength of
particles, thus weakening the interfacial zone. concrete prior to form removal or application of construction
Other factors that affect the strength of concrete include loads. Even though efforts are made to provide equivalent cur-
degree of mixing of the constituent materials and consolidation ing to both the concrete in the structure and the molded speci-
of the concrete. For a given set of materials, there is an optimum mens, differences in strength are expected due to differences
proportion of ingredients that yields the most economical mix- in consolidation and early-age temperature histories. Push-out
ture with the desired properties. Once the proportions have cylinders (ASTM C 873) are also used to determine in-place
been selected, the materials must be mixed adequately to concrete strength, but the curing of push-out cylinders is more
achieve a homogeneous mixture. The concrete must then be like that of the structure than for molded specimens stored on
placed in the molds and thoroughly consolidated; this can be the structure according to ASTM C 31/C 31M. Push-out cylin-
achieved through rodding or vibration. Inadequate consolida- ders typically have length-diameter ratios (L/D) less than the
tion will reduce strength. Self-consolidating concrete, however, standard value of 2, and a strength correction factor has to be
does not require any consolidation effort during placement applied to the measured strength, as is done for cores.
[12]. Excessive bleeding and segregation of fresh concrete can Another technique to estimate the in-place strength is to
also lead to reduced strengths, especially in the top portions of use the temperature-matched curing (TMC) technique [17]. In
test specimens, which will have a higher w/cm due to the rise of a temperature-matched curing system, thermocouples inserted
bleed water. The use of cardboard or plastic cylinder molds in into the concrete mass monitor the temperature rise as it
preparing specimens has been reported to result in lower cures and control the temperature of specimens placed in a
measured strengths compared with the use of rigid steel molds water bath or in molds with heating elements. In effect, the
[13,14]. In practice, however, single-use plastic or cardboard ambient temperature surrounding the concrete specimens
molds are used widely for economic reasons. Also, concrete that matches the temperature history of the concrete mass as it
has been damaged due to mishandling of specimens or inad- cures. Such a curing system could also be used to follow a pre-
vertent application of loads may also show reduced strength. established temperature history of a structural element. It has
been found that temperature-matched curing allows for the
Preparation of Test Specimens best estimation of in-place strength [18].

Concrete strength tests are conducted on both molded speci- Specimens from Existing Structures
mens and specimens cut from existing structures. Although Procedures for obtaining strength test specimens from existing
testing procedures are similar, the significance of the infor- hardened concrete are specified in ASTM Test Method for Ob-
mation obtained can be quite different depending on speci- taining and Testing Drilled Cores and Sawed Beams of Concrete
men preparation and handling prior to testing. Following are (C 42/C 42M). Drilled cores can be tested either in compression
brief descriptions of current standard procedures for prepar- or splitting tension, and sawed beams are tested in flexure. In
ing test specimens up to the time of testing. addition to placing dimensional requirements on specimens,
ASTM C 42/C 42M requires that test specimens be comprised of
Molded Specimens intact, sound concrete, as free of flaws as the particular con-
Current ASTM test methods for measuring strength call for struction will allow. The standard also specifies moisture con-
specimens in the shape of cylinders or beams. Preparation of ditioning before testing (to be discussed subsequently).
these specimens in the field is governed by ASTM Practice for Generally, drilled cores or sawed beams are obtained when
Making and Curing Concrete Test Specimens in the Field (C doubt exists as to the strength of the concrete as placed. This
31/C 31M), and under laboratory conditions by ASTM Practice can be due to low strength test results during construction or
for Making and Curing Concrete Test Specimens in the Labo- signs of distress in the structure. Also, cut specimens are useful
128 TESTS AND PROPERTIES OF CONCRETE

if strength information is required for older structures or if The loading to which the concrete member has been sub-
service loads are to be increased above original design levels. All jected may also affect the measured core strength. Cores taken
other factors being equal, the strength of these specimens is from highly stressed regions, where microcracking is likely to
most likely to be representative of the strength of the concrete have occurred, have lower strength than those from unstressed
in the structure. Cutting of these specimens is, however, costly, regions [19,24].
and the drilling or sawing processes may introduce variables af- The inclusion of reinforcing steel in the specimen is not
fecting strength test results. permitted by ASTM C 42/C 42M since there are insufficient
Although the concrete in drilled or sawed specimens is data to demonstrate the effect of embedded steel on measured
more likely to be representative of that in the structure than strength. The effect of reinforcement on compressive strength
molded specimens, one must be aware of various factors that is variable. The presence of reinforcement in the tensile region
are likely to affect the strength of the concrete samples thus of flexural beams or in splitting tensile strength specimens has
obtained. Excess voids in a particular concrete sample, due to pronounced effects on measured strengths.
poor consolidation, will cause strength reductions. Further, Moisture conditioning of cores affects measured strengths.
the process of drilling or sawing of specimens may cause Early work showed that specimens tested dry had higher
some damage that may affect strength test results, and this fac- strengths than cores with two days of moist conditioning prior
tor may become more pronounced as the ratio of cut surface to testing [22]. Recent work on the effect of moisture condi-
to specimen volume increases [19]. The resultant strength tioning practices has shown similar results leading to changes
reductions have been reported to be greater in higher-strength in ASTM C 42/C 42M [25,26]. The moisture conditioning proce-
concretes [20]. dures adopted in 2003 are intended to preserve the moisture of
Besides the drilling or sawing processes, the location and the drilled core and to minimize the effects of moisture gradi-
orientation of the cut specimens in the structure will affect ents introduced by wetting during drilling and specimen prepa-
strength test results. Specimens cut from the top of a concrete ration. Cores are to be placed in sealed plastic bags or nonab-
placement have lower strengths than those cut from the bot- sorbent containers after the surface drill water evaporates or
tom of the placement due to bleeding, segregation, and curing within 1 h of drilling. If water is used for subsequent end prepa-
effects [5,14,19,21,22]. Bloem [22] reported that cores taken ration, the specimens are sealed again after surface drying.
from the top of an 8-in. (203-mm) slab tested approximately Cores are kept in the sealed condition for at least five days after
5 % weaker than those from the bottom when the slab was being last wetted and before testing. Unless specified otherwise,
subjected to good curing. The difference increased to 15 %, cores are to be tested within seven days of drilling.
however, when the slab was poorly cured. Studies on high-
strength concrete [23] indicated no effect of core elevation on Compressive Strength Test Procedures
strength test results, which can be attributed to the fact that
high-strength concretes generally exhibit little bleeding. Cores Compressive strength testing of molded concrete cylinders
drilled horizontally have been shown to yield lower strengths prepared according to ASTM C 31/C 31M or C 192/C 192M is
than those drilled vertically [5,14,19,21]. As shown in Fig. 2, specified in ASTM Test Method for Compressive Strength of
this is attributed to the alignment of weak interfacial regions Cylindrical Concrete Specimens (C 39/C 39M). Testing proce-
parallel to the loading direction in cores drilled horizontally, dures for cores (C 42/C 42M) and push-out cylinders (C 873)
due to accumulation of water under coarse aggregate particles also refer to ASTM C 39/C 39M for measuring compressive
due to bleeding. strength. All of these standards specify tolerances on specimen

Fig. 2—Planes of weakness due to bleeding: (a) axis of specimen vertical and
(b) axis of specimen horizontal.
 OZYILDIRIM AND CARINO ON CONCRETE STRENGTH TESTING 129

Effect of Specimen End Conditions

ASTM C 39/C 39 M states that the ends of cylindrical speci-
mens to be tested must not depart from perpendicularity with
the specimen axis by more than 0.5° (approximately 1/8 in. in
12 in. or 1 mm in 100 mm), and that the ends must be plane
to within 0.002 in. (0.050 mm). If the specimen does not meet
these tolerances, the ends shall be sawed or ground to meet
those tolerances, or capped in accordance with either ASTM C
617 or ASTM C 1231. ASTM C 31/C 31M and C 192/C 192M on
specimen preparation allow for depressions or projections
on finished surfaces of cylinders and beams of up to 1/8 in.
(3.2 mm). Formed surfaces of beams are to be smooth and
plane such that the maximum deviation from the nominal
cross section shall not exceed 1/8 in. (3.2 mm) for cross-
sectional dimensions of 6 in. (152 mm) or more, or 1/16 in.
(1.6 mm) for smaller dimensions. ASTM C 42/ 42M allows pro-
jections of up to 0.2 in. (5 mm) from the core ends and
requires that ends of cores do not depart from being perpen-
dicular to the specimen axis by a slope of more than 1:8D or
(1:0.3D), where D is the average core diameter in inches (or
millimetres). The purpose of the latter requirement is to avoid
exceeding maximum cap thickness specified in ASTM C 617.
The purpose of specifying end condition requirements of
planeness and perpendicularity is to achieve a uniform trans-
fer of load to the test specimen. Surface irregularities will lead
to local concentrations of stress even in specimens that are
Fig. 3—Frictional restraint at the ends of cylinders results
capped to meet the planeness requirements [15]. The effect of
in state of triaxial compression shown as shaded region.
cylinder end conditions prior to capping on strength test
results has been reported by several authors [27–30]. In gen-
eral, specimen ends that do not meet the specified require-
geometry, end conditions, and specimen moisture condition at ments prior to capping cause lower strength test results, and
time of testing. In addition, if the end condition tolerances are the degree of strength reduction increases for higher-strength
not met by the concrete specimens, the test methods require concretes.
grinding or sawing of the ends to meet the requirements, or ASTM C 617 covers procedures for capping with materials
capping with bonded caps according to ASTM Practice for that bond to the cylinder ends. Freshly molded cylinders may
Capping Cylindrical Concrete Specimens (C 617) or with be capped with a neat cement paste that is allowed to harden
unbonded elastomeric caps (to be described) according to with the concrete. It is important to keep the cement paste caps
ASTM Practice for Use of Unbonded Caps in Determination of moist, since they are susceptible to drying shrinkage and possi-
Compressive Strength of Hardened Concrete Cylinders (C ble cracking. Hardened cylinders and drilled cores may be
1231). All of these factors will influence measured strengths as capped with either high-strength gypsum cement or sulfur mor-
discussed in following sections. tar. The requirements on the strength of the capping material
Ideally, the result obtained from a compressive strength and the thickness of the caps for concrete strengths below and
test would be a direct indication of the concrete’s ability to above 7000 psi (50 MPa) are given in ASTM C 617. Sulfur mor-
withstand a uniaxial compressive force. The stress state in tars are permitted if allowed to harden at least 2 h before testing
standard compressive strength test specimens is, however, for concrete with strength less than 5000 psi (34.5 MPa). For
more complex than uniaxial compression. Friction between concrete strengths of 5000 psi or greater, sulfur mortar caps
the bearing faces of the testing machine and the test specimen must be allowed to harden at least 16 h before testing, unless a
restrains the specimen laterally, thereby inducing lateral com- shorter time has been shown to be suitable. The caps shall be
pression in the specimen ends as shown in Fig. 3 [5,14]. The plane to within 0.002 in. (0.05 mm), shall not depart from per-
portion of the cross section under triaxial compressions pendicularity with the specimen axis by more than 0.5°, and
decreases with distance from the specimen ends. At distances shall not be off-center with respect to the specimen axis by more
from the specimen ends of about 0.8 of the diameter, the tri- than 1/16 in. (2 mm). Generally, capping materials must be at
axial effects are negligible [5]. Thus, for cylindrical or pris- least as strong as the concrete, unless there are data showing
matic specimens having an aspect ratio of two, cross sections satisfactory performance even though the cube strength of the
at mid-height should be free of the effect of the end restraint. capping material is lower than the compressive strength of the
Details of this effect are covered in subsequent discussion of concrete. This is possible because bonded caps are under a state
the effect of length-diameter ratio. of triaxial compression and can withstand higher stresses than
the unconfined cube strength [31]. Caps on hardened concrete
Factors Affecting Compressive Strength specimens should be approximately 1/8 in. (3 mm) thick, but no
more than 5/16 in. (8 mm) thick, and well bonded to the speci-
In order to understand better the significance of the require- men end. Thick caps can reduce the measured strength [32].
ments of the test procedures, the effects of various factors on Unbonded caps have a thickness of 1/2
1/16 in. (13
2 mm]
compressive strength test results are discussed. and the diameter is not more than 1/16 in. (2 mm) smaller than
130 TESTS AND PROPERTIES OF CONCRETE

the inside diameter of the retaining ring. Required hardness of the capping material when bonded caps are used. Traditionally,
the pads depends on the strength level of the concrete, with it has been believed that the resulting cylinder strength is re-
higher-strength concrete requiring harder pads. The bearing lated to the strength of the capping material. Research has
surface of the retaining rings shall be free of gouges, grooves, shown, however, that the measured cylinder strength of a given
or indentations. concrete is related more closely to the elastic modulus of the
The use of sulfur mortar is the most common bonded cap- capping material rather than strength [31]. The capping mate-
ping technique. Sulfur capping compounds must be kept rial must have a high elastic modulus to distribute the applied
molten at a temperature of approximately 265°F (130°C). load uniformly to the ends of the specimen. Ideally, the elastic
These compounds have a range of melting temperatures, modulus of the capping material should be similar to that of the
above and below which they become viscous and difficult to concrete. In a study of different capping materials [32], it was
pour [15]. Further, volatilization of the sulfur occurs upon found that neat cement paste caps with cube strengths of 12 000
heating; thus, ASTM C 617 restricts the reuse of any material psi (85 MPa) and thicknesses up to 1⁄4 in. (6 mm) exhibited
that has been used five times. Retrieval of capping material strengths similar to specimens with ground ends for concrete
from specimen ends introduces oil and other contaminants strengths up to 17 000 psi (120 MPa). With sulfur mortar caps,
that can reduce the fluidity and strength of the material. however, reduction in strength was observed with 3/16-in. (5-
Although bonded capping satisfies the end condition mm) thick caps.
requirements for compressive testing, the capping operation is
tedious and can be hazardous in the case of molten sulfur Effect of Specimen Size
mortar. An alternative is the unbonded capping method cov- Historically, the standard field molded cylinder has been 6 in.
ered by ASTM C 1231, which can be used for testing concrete (152 mm) in diameter and 12 in. (305 mm) in length. The
with compressive strengths between 1500 and 12 000 psi. This smaller 4 by 8 in. (102 by 203 mm) cylinder size is permitted
system, which is illustrated in Fig. 4, consists of elastomeric when specified and provided that the nominal maximum size
pads, commonly made of neoprene, confined within metal of aggregate (NMSA) does not exceed 1/3 the cylinder diame-
retaining rings. The elastomeric pads conform under load to ter. The 4 by 8-in. cylinder is being used more commonly. The
the specimen end surfaces, thereby distributing the applied smaller specimens require less material to make and are much
load uniformly. The use of retaining rings is essential to easier to handle: a 6 by 12-in. (152 by 305-mm) concrete cylin-
restrict lateral flow of the pads that would otherwise induce der weighs about 30 lb (13.6 kg), compared with about 9 lb (4
lateral tension in the specimen ends, thereby reducing the kg) for a 4 by 8-in. cylinder. In addition, the use of concretes
apparent compressive strength. It has been found that the use having increasingly higher strengths requires that testing
of unbonded pads is a convenient and efficient capping machines have higher load capacities when 6 by 12-in. cylin-
method that produces test results comparable to those ders are used. Because 4-in. diameter cylinders have only 44 %
obtained with bonded caps [33,34]. The elastomeric pads can of the cross-sectional area of 6-in. cylinders, high-strength con-
be reused until physical damage is observed. ASTM C 1231 crete can be tested with existing testing machines. In cases of
includes information on the maximum uses of pads and the massive concrete placements, such as dams, the use of large
qualification of unbonded cap systems. Pairs of cylinders are aggregate sizes requires the use of larger-diameter specimens
made, and one is tested after grinding or capping while the in order to maintain a diameter-aggregate size ratio of at least
other is tested with unbonded caps. At least ten pairs of cylin- 3 to 1, or wet-sieving to remove larger aggregate sizes.
ders are made at both the highest and the lowest strength lev- It is commonly accepted that as specimen size increases,
els for which the unbonded caps are to be used. the measured concrete strength and the variation in test results
The measured cylinder strength of concrete is affected by decrease [5,7,14,35–39]. As shown in Fig. 5, the magnitude of
the size effect decreases with increasing specimen diameter [7].
The reasoning behind the size effect is that the strength of a
concrete specimen will be governed by the weakest part of that
specimen, and that the probability of the occurrence of large
flaws increases as specimen size increases. Further, it has been
reported that the strength difference due to specimen size in-
creases as concrete strength increases [36]. Drilled cores will
follow the same trend of increasing strength with decreasing
specimen size for larger-diameter cores. For small diameter
cores, however, the ratio of cut surface to specimen volume be-
comes significant, and it is possible that coring damage will
cause strength reduction for decreasing diameters below 4 in.
(102 mm) [19,40]. A review of available data indicated that 4-in.
cylinders, on average, result in about 4 % higher measured
strength compared with 6-in. cylinders [41,42]. It has been
shown, however, that this strength difference can be reduced if
4-in. cylinders are molded by using two layers instead of three
when consolidation is by rodding [43].
Although the testing of smaller specimens is more con-
venient, precision of strength determination should not be
sacrificed. The within-test variability of 4-in. cylinders is
Fig. 4—Unbonded cap system for testing concrete cylin- about 20 % greater than that of 6-in. cylinders [42]. Equal
ders in compression. precision of average strength can be obtained if the number
 OZYILDIRIM AND CARINO ON CONCRETE STRENGTH TESTING 131

Fig. 7—Overlapping regions of triaxial compressive

stresses in specimen with L/D  1.
Fig. 5—Effect of cylinder size on measured compres-
sive strength [7].
Effect of Length-Diameter Ratio
of 4-in. cylinders is increased to 1.5 times the number of 6-in. Standard test cylinders have a length-diameter (L/D) ratio of
cylinders. two, but L/D values of capped specimens as low as one are
permitted by ASTM C 39/C 39M and ASTM C 42/C 42M. The
Effect of Diameter-Aggregate Size Ratio actual length-diameter ratio will influence the apparent
Current specifications for molded specimens and push-out strength of the specimen being tested. As shown in Fig. 6,
cylinders require that the minimum specimen dimension be at measured strengths increase as the L/D value decreases
least three times the NMSA. Gonnerman [37] reported that [5,14,37,45–47]. This is attributed to the effect of the end
test results were satisfactory for specimens with diameter-ag- restraint due to the friction between the testing machine bear-
gregate size ratio of two. He reported difficulty, however, in con- ing blocks and the test specimen ends. As L/D decreases, a
solidating the specimens so that they were homogeneous, and greater proportion of the specimen is in a state of triaxial com-
thus recommended the minimum ratio of 3 to 1. For molded pression. For example, Fig. 7 illustrates the conditions in a
specimens, larger-sized aggregates may be removed by hand specimen with L/D equal to one. For L/D values between 1.5
picking or by wet sieving so that smaller specimen dimensions and 2.5, measured strengths are within
5 % of the strength
may be used. It has been reported, however, that the practice of for L/D  2. For L/D values below 1.5, measured strengths
removing larger aggregate sizes from concrete will result in increase markedly as shown in Fig. 6 [7].
higher compressive strengths [44]. For drilled cores, the prefer- ASTM C 39/C 39M and C 42/C 42M provide correction
able condition is that the core diameter is at least three times factors to be applied to strength test results obtained from
the maximum nominal aggregate size used in the concrete specimens (molded or cores) having an L/D value between
placement. This condition may be relaxed by the specifier of 1 and 1.75. These correction factors are shown in Fig. 8.
tests provided the core diameter is at least twice the maximum According to these test methods, the correction factors are
size of coarse aggregate. applicable to specimens of normal-density concrete, light-
weight concretes with densities between 100 and 120 lb/ft3
(1600 and 1920 kg/m3), concrete that is dry or soaked at the

Fig. 6—Relationship between measured compressive

strength and length-diameter ratio [7]. Fig. 8—L/D strength correction factors.
132 TESTS AND PROPERTIES OF CONCRETE

time of testing, and concrete strengths between 2000 and bearing block must not rotate when the specimen is being
6000 psi (14 and 42 MPa). For strengths higher than 6000 psi loaded, and ASTM C 39/C 39M prohibits the use of heavy
(42 MPa), correction factors may have larger values [48]. grease to lubricate the ball-seat assembly. Placing the speci-
men off-center with respect to the loading axis by only 1⁄2 in.
Effect of Specimen Moisture Condition (13 mm) can cause strength reductions of 10 %. A flexible
The moisture condition of the specimen at the time of testing testing machine results in sudden failures of test specimens
can have a significant influence on measured strengths. In especially for high-strength concrete [42]. The longitudinal
general, specimens have 5–20 % lower compressive strengths stiffness of the testing machine, however, does not appear to
when tested in a moist condition than they would if tested in have a significant effect on measured strength [42].
a dry condition [5,8,19,21,22,25,49]. The higher strength of dry
specimens is attributed to increased strength of secondary Effect of Loading Rate
bonds within the paste structure. It has been postulated that as The measured strength of concrete specimens increases as the
a specimen dries, the outer surface attempts to shrink, there- rate of loading increases [5,7,14,16,39,50,51]. Abrams reported
by inducing lateral compression on the specimen interior, that higher-strength concretes were more affected by loading
which increases its apparent compressive strength [49]. A rate, and that the ultimate strength was unaffected by rapid
specimen that is wetter in the outer region will have lower loading up to 88 % of the ultimate load followed by loading at
compressive strength. a standard rate [51]. The dependence of ultimate strength on
When choosing the specimen moisture condition for test- loading rate is thought to be related to mechanisms of creep
ing, the purpose of the test must be considered, as well as the and microcracking [5,14,39]. This would appear to be in agree-
effect that the moisture condition will have on the test results. ment with the observation that when subjected to a sustained
Standard-cured cylinders for acceptance testing are tested in a load of approximately 75 % of its ultimate capacity obtained
moist condition according to ASTM C 39/C 39M. On the other using ASTM C 39/C 39M, concrete will eventually fail with no
hand, field-cured cylinders and push-out cylinders are tested in further load application.
their “as received” condition. In core testing, if results are to ASTM C 39/C 39M requires that test specimens be loaded
be compared with standard-cured cylinder test results, then a at a constant rate of movement between the bearing blocks.
moist condition may be preferred; however, if determination Prior to 2002, ASTM C 39/C 39M specified the following load-
of the in-place strength of the concrete is desired, the speci- ing rates:
men should be tested as closely as possible in the moisture • For screw-type machines, approximately 0.05 in. (1.3
condition that exists in the structure. ASTM C 42/C 42M mm)/min running idle.
specifies moisture conditioning that preserves moisture of the • For hydraulically operated machines, a loading platen
drilled core and provides a reproducible moisture condition movement that produces a stress rate between 20 and 50
that minimizes the effects of moisture gradients [25,26] intro- psi/s (0.14 and 0.34 MPa/s).
duced by wetting during drilling and specimen preparation. Studies showed that at the extremes of the permissible
range of loading rate, non-negligible strength differences
Effect of Loading Direction versus could occur [42]. On average, the faster loading rate produced
Casting Direction about 2.2 % greater strength. These findings lead to narrowing
Molded concrete cylinders are tested parallel to their casting the loading rate range to 35
7 psi/s (0.25
0.05 MPa/s) in
direction. Beams and drilled cores, however, may be tested the 2004 version of ASTM C 39/C 39M. In addition, the 2004
either parallel or perpendicular to the casting direction, version revised the requirements for screw-driven or servo-
depending on the circumstances involved. In general, speci- controlled loading machines. The user is required to establish
mens tested in the same direction as cast will yield higher the rate of platen movement that will result in the prescribed
strengths than those tested perpendicular to it. As illustrated loading rate. This will require trial and error until sufficient
in Fig. 2, the difference in measured strength is attributed to experience is gained.
the occurrence of weak paste-aggregate interfaces aligned per- The designated rate of movement shall be maintained
pendicular to the casting direction due to water gain under from about 50 % of the expected ultimate load until the ulti-
coarse aggregate particles [5]. Cores tested parallel to the cast- mate load is attained. A higher rate of loading is not prohibi-
ing direction may have about 8 % higher strengths than those ted up to a load that is about 50 % of the expected ultimate
tested perpendicular to the casting direction [5]. load. The higher load rate shall be applied in a controlled man-
ner so that the specimen is not subjected to shock loading. The
Effect of Testing Machine Characteristics loading rate should not be adjusted as the load approaches the
ASTM C 39/C 39M gives required features for the loading ultimate strength.
apparatus to be used in compressive strength testing. Among
these are the capacity for smooth and continuous load appli- Significance of Compressive Strength
cation; accurate load measurement; and two bearing blocks, Test Results
one being spherically seated and one being solid, both of
which must satisfy further requirements of surface planeness, The compressive strength of concrete is of primary impor-
minimum and maximum diameters, and other features. Fail- tance in structural applications because design procedures
ure to meet these requirements has been shown to reduce require this property. As has been discussed, many factors
strength test results [16,30]. affect the measured compressive strength of concrete test
The spherically seated bearing block must be free to specimens. There are also differences between the conditions
rotate to accommodate any small deviation from parallelism that exist within a structure and those within a test specimen.
of the ends; strength reductions of up to 20 % have been These include restraint, loading conditions, and long-term
observed with no spherical seating [30]. The spherically seated effects such as creep and shrinkage. Thus results from
 OZYILDIRIM AND CARINO ON CONCRETE STRENGTH TESTING 133

standard compressive strength tests may not represent in- Flexural Strength Testing
place behavior. Tests of standard-cured specimens are used as ASTM C 78 and ASTM C 293 are similar in that they both
a basis for quality control of batching, mixing, and delivery involve testing a simply supported prismatic beam. For both,
operations; determination of compliance with specifications; failure initiates in the tensile faces of the beams. As shown in
and for evaluating effectiveness of admixtures and other con- Fig. 9, they differ, however, in the loading configuration.
stituent materials. ASTM C 78 specifies application of load to the specimen at the
Tests of field-cured specimens are intended to provide third points along the span, resulting in maximum bending
information on the in-place concrete. There are, however, limi- moment in the middle third of the specimen. ASTM C 293
tations on the ability of field-cured test specimens to be repre- specifies load application at midspan, which is the only cross
sentative of in-place concrete. Temperature-matched curing section subjected to maximum moment.
eliminates some of these limitations, but not all of them For both test methods, the beam is to be tested on a span
because molded specimens are not consolidated the same as that is within 2 % of being three times its depth. Sides shall be
in-place concrete. at right angles to the top and bottom faces, and surfaces in con-
Drilled cores provide representative samples of in-place tact with the loading or reaction blocks should be flat. The spec-
concrete. Several factors, however, contribute to the uncer- imens should be tested on their sides as molded. If 1-in. (25-mm)
tainty of measured core strength as being truly representative or longer gaps in excess of 0.004 in. (0.1 mm) exist between the
of the in-place strength. These factors include, among others, specimen surfaces and the loading or reaction blocks at no load,
presence of moisture gradients resulting from water-cooled the specimen surfaces should be ground or capped. If the spec-
drilling or saw cutting of ends, undefined damage introduced imen surfaces are within 0.015 in. (0.38 mm) of being plane,
by the core removal process, and differences in size and L/D leather shims may be used. The loading apparatus must apply
value compared with standard molded specimens. load perpendicular to the face of the beam without eccentricity,
and reactions must be parallel to the direction of load applica-
Tensile Strength Test Procedures tion. Further requirements are specified for the testing appara-
tus to ensure the supports and loading blocks are free to move
There are currently no standardized test procedures for deter- to maintain uniform distribution of load over the beam width
mining the direct tensile strength of concrete, that is, the as tested. The first half of the load may be applied rapidly; af-
strength under uniaxial tension. This is due to the difficulty terwards, the loading rate is required to produce a tensile stress
involved in inducing pure axial tension within a specimen between 125 and 175 psi/min (861 and 1207 kPa/min) in the ex-
without introducing localized stress concentrations. Knowl- treme fiber of the beam. The performance characteristics of
edge of the tensile strength of concrete is important because fiber-reinforced concrete are determined using the flexural test.
it determines resistance to cracking. Therefore, several test The load-deflection curve that is obtained can be used to calcu-
procedures have been developed to indicate indirectly the ten- late the residual strength or toughness at specified values of
sile strength of concrete. These include: midspan deflection. Flexural testing of FRC is, however, dis-
• ASTM Test Method for Flexural Strength of Concrete cussed in the chapter by Tatnall in this publication.
(Using Simple Beam with Third-Point Loading) (C 78)
• ASTM Test Method for Flexural Strength of Concrete Splitting Tensile Strength
(Using Simple Beam with Center-Point Loading) (C 293) ASTM C 496 gives requirements for the testing apparatus, test
• ASTM Test Method for Splitting Tensile Strength of Cylin- specimen geometry, and load application for determining
drical Concrete Specimens (C 496) splitting tensile strength of cylindrical specimens. The testing
Some agencies have adapted ASTM Test Method for machine should meet the requirements of ASTM C 39/C 39M.
Direct Tensile Strength of Intact Rock Core Specimens (D The cylinder is placed on its side and subjected to a diametric
2936) to measure direct tensile strength of concrete. compressive force along its length. If either the upper or lower

Fig. 9—Bending moment diagrams for center-point loading and third-point

loading.
134 TESTS AND PROPERTIES OF CONCRETE

test method for rock cores, ASTM D 2936. In it, cylindrical test
specimens are loaded in axial tension through the use of metal
caps bonded to the specimen ends. The direct tensile strength
is determined by dividing the axial load at failure by the speci-
men cross-sectional area. Care is required during specimen
preparation to ensure application of a tensile load with mini-
mum eccentricity. This test procedure is not widely used for
concrete specimens and ASTM Subcommittee C09.61 is not
pursuing standardization at this time (2004).

Factors Affecting Flexural Strength

The formulas in ASTM C 78 and ASTM C 293 for computing flex-
ural strength (modulus of rupture) are based on several as-
sumptions that are approximations when testing concrete
beams to failure [53]. One assumption is that the concrete be-
haves as a linear-elastic material throughout the test, which is
not true at stresses approaching failure. The flexural stress
equations apply to long, shallow beams, whereas the actual test
specimens are short and deep. The failure stress calculated
using the two test methods is higher than the actual extreme
Fig. 10—Stress distribution on diametrical plane of
fiber stress due to the simplifying assumption that the stress dis-
cylinder for load distributed over a width of 1/12 cylinder
diameter [52].
tribution over the depth of the beam is linear. It is likely, how-
ever, that compared with the variability inherent to concrete
strength, this approximation is not significant. Various other
bearing block of the testing machine is shorter than the cylin- factors have been found to affect flexural strength test results
der, a bearing bar or plate shall be used that is at least as thick obtained using either third-point or center-point loading. These
as the distance from the edge of the machine bearing block to will be discussed.
the end of the cylinder, at least 2 in. (51 mm) wide, and plane
to within 0.001 in. (0.025 mm). Plywood bearing strips at least Effect of Specimen Dimensions
as long as the cylinder and 1/8 in. (3.2 mm) thick and 1 in. (25 ASTM C 78 and ASTM C 293 require the test specimen to
mm) wide are placed between the specimen and the loading have a span of three times its as-tested depth. While the stan-
faces to accommodate minor surface irregularities. The load is dard beam dimensions are 6 in. by 6 in. by 20 in. (152 mm
applied so as to produce a spitting tensile stress rate between by 152 mm by 507 mm), tested on an 18-in. (456-mm) span,
100 and 200 psi/min (690 and 1380 kPa/min). other beam dimensions are permitted depending on maxi-
Failure of the cylinders occurs along a vertical plane con- mum coarse aggregate size. The minimum specimen dimen-
taining the specimen axis and the applied load. As shown in sion must be at least three times the nominal maximum size
Fig. 10, the loading configuration of this test method induces of coarse aggregate, and the ratio of width to depth as-
a nearly uniform tensile stress perpendicular to the vertical molded must not exceed 1.5. For specimens prepared in the
plane over approximately three-fourths of the specimen diam- laboratory, larger coarse aggregate particles may be removed
eter. The actual stress distribution on the vertical plane is by hand-picking or wet-sieving.
affected by the width of the uniformly applied load [52]. The For a constant beam width and test span, the apparent
value of the uniform tensile stress at failure is the splitting ten- flexural strength of specimens tested in third-point loading
sile strength and is approximated by the following equation: decreases as the depth of the beam increases; however, flex-
2P ural strengths seem to be independent of beam width for a
fsp   (1) given depth and span [54]. For a constant beam cross section,
LD
the effect of span length is unclear. Kellerman [55] reported
where that strength decreased for both center-point and third-point
loading as span length increased. Reagel and Willis [54], on
fsp  splitting tensile strength the other hand, reported no effect of span length on strength
P  maximum load during test for third-point loading.
D  cylinder diameter
L  cylinder length Effect of Specimen Size
It is commonly agreed that as the size of the test specimen
The regions of the specimen in the vicinity of the loading increases, flexural strengths decrease [54–57]. This has been
strips are subjected to large compressive stresses. Tensile found to be true for both center-point and third-point loading,
rather than compressive failure occurs because the stress state and is due to the size effect as was discussed for compressive
at the loading strips is triaxial compression, allowing the con- strength test specimens. Variability of test results decreases
crete to resist the high compressive stresses [52]. with increasing specimen size [57].

Direct Tension Test Effect of Coarse Aggregate Size

A proposed test method for direct tensile strength of concrete For the same w/cm, the flexural strength of concrete beams
or mortar specimens is an adaptation of an existing standard is higher for smaller coarse aggregate sizes. This was found
 OZYILDIRIM AND CARINO ON CONCRETE STRENGTH TESTING 135

to be true even when the use of the smaller coarse aggregate at a location other than midspan corresponds to a lower ex-
resulted in a mixture with a higher w/cm to maintain worka- treme fiber stress than exists at midspan. Thus, when failure
bility [55]. This effect should be considered when removing stress is calculated using the bending moment at the fracture
larger aggregate particles from concrete by hand-picking or plane, the flexural strength of the specimen is lower than when
wet-sieving. Variability of test results is lower with smaller calculated using the bending moment at midspan [55,57].
coarse aggregate size. ASTM C 293, however, does not require making note of the lo-
cation of fracture. ASTM C 78, on the other hand, requires that
Effect of Loading Rate if fracture occurs outside of the maximum moment region but
As with compressive strength, the apparent flexural strength within a distance of 5 % of the span length, the flexural strength
of specimens increases with loading rate [56]. A linear rela- is based on the bending moment at the location of the fracture
tionship between flexural strength and the logarithm of plane. If fracture occurs outside of the maximum moment re-
applied stress rate has been observed. gion by a distance of more than 5 % of the span length, the test
results are to be discarded.
Effect of Moisture Condition
Flexural strength test results are sensitive to the specimen mois- Factors Affecting Splitting Tensile Strength
ture condition at time of testing. When specimens are tested in
a drying condition, the apparent flexural strength is lower than In theory, the application of a line load perpendicular to the
when tested in a saturated condition [5,14,58,59]. Strength re- axis of a cylinder and in a diametrical plane produces a uni-
ductions of up to 33 % have been reported [58]. When the sur- form tensile stress perpendicular to that plane. For actual test-
face of the specimen is allowed to dry rapidly, it attempts to ing of a concrete cylinder there are several departures from
shrink, but this shrinkage is restrained by the specimen core. the theoretical case [52]. First, the theory applies to a homo-
This induces tensile stresses in the surface and, if the induced geneous material, which concrete is not. Secondly, concrete is
tensile stresses approach the tensile strength of the concrete, not linear elastic as is assumed in the analysis. Third, the load
cracks will develop in the outer surface of the specimen. When is not applied along a line, but rather is distributed within a
cracks exist due to drying, they act as stress concentrators and strip. The strip loading results in large compressive stresses
also reduce the effective cross section of the test specimen. If perpendicular to the diametrical plane, near the surface of the
cracks do not develop, the tensile stresses in the near-surface specimen under the loading strips. Nevertheless, the splitting
concrete due to drying act as a preload condition; that is, the tensile strength test is reasonably easy to conduct, and results
tensile stresses due to drying and those due to applied test load provide comparative values, even if direct tensile strength val-
are cumulative, thereby resulting in a lower applied load to ues are not obtained. Several factors that affect splitting ten-
cause failure. The drying condition discussed occurs rapidly un- sile strength test results are discussed.
der normal ambient conditions if care is not taken to keep the
specimen surfaces moist. Due to the significant effect of surface Effect of Specimen Length and Diameter
moisture condition on test results, flexural strength specimens For a given diameter, cylinder length does not seem to affect
that are used to indicate the strength of the concrete in-place test results, other than possibly reducing variability for longer
should be cured under conditions similar to the concrete struc- specimens [52]. Cylinders having a diameter of 4 in. (102 mm)
ture, but tested in a saturated condition. Note that the effect of were observed to have splitting tensile strengths that were
drying on flexural strength test results is opposite to that for roughly 10 % higher than those obtained from cylinders hav-
compressive strength test results. ing 6-in. (152-mm) diameters [52,60]. Variability of test results
decreases with increasing specimen diameter.
Effect of Center-Point Versus
Third-Point Loading Effect of Bearing Strips
As was discussed previously, the main difference between ASTM C 496 requires the use of bearing strips made of 1/8-
ASTM C 78 and ASTM C 293 is the location of load applica- in. (3-mm) thick plywood that are 1-in. (25-mm) wide and at
tion (refer to Fig. 9). In center-point loading (ASTM C 293), least as long as the cylinder. The purpose of these strips is to
the load is applied at the specimen midspan. In third-point conform to the specimen surface and distribute the load
loading (ASTM C 78), the load is applied at the third points from the loading block. Increasing the thickness of the bear-
along the test span. For the latter case, the middle third of the ing strips may cause strength reductions [52]. Steel bearing
beam span is subjected to maximum bending moment, and strips cause significant strength reductions, probably due to
thus maximum extreme fiber stress. In center-point loading, their inability to conform to the specimen surface. Subcom-
however, only the cross section at midspan is subjected to mittee C09.61 has rejected the use of alternative materials to
maximum moment and maximum extreme fiber stress. The 1/8-in. (3-mm) thick plywood unless comparative test data are
probability of having weak concrete in a region of highest provided.
stress is lower with the center-point loading compared with
the third-point loading. As a result, for a given beam size, Effect of Specimen Moisture Condition
flexural strengths obtained from third-point loading are lower It is not expected that drying of the specimen surface will
than those obtained from center-point loading [55–57]; dif- affect the measured splitting tensile strength as significantly as
ferences of 15 % are not unusual. Variability is also lower for it does flexural strength, since the specimen surface contained
third-point loading. within the failure plane is subjected to high triaxial compres-
Sometimes fracture of test specimens tested in center-point sive stresses. Indeed, it is possible that the effect of drying is
loading occurs at a location other than midspan. Since the bend- more similar to that in compressive strength cylinders, where
ing moment distribution along the beam span varies linearly the restrained shrinkage of the outer surface induces com-
from zero at the support to its maximum at midspan, fracture pression in the specimen interior.
136 TESTS AND PROPERTIES OF CONCRETE

Effect of Loading Rate the type of tension test [5]. General relationships between the
As with testing of specimens for compressive strength and compressive strength of concrete cylinders and the strengths
beams for flexural strength, higher splitting tensile strengths attained by various tension testing methods are discussed in
are obtained when the specimens are loaded at a more rapid the literature [5,7,8,14,39,62–64].
rate. The splitting tensile strength follows a power function
relationship of the compressive strength. As a result, the ratio
Relationship of Flexural Strength to Splitting of splitting tensile strength to compressive strength is not con-
Tensile Strength stant but decreases with increasing strength [65–68].

In the flexural strength tests, failure is controlled by the Nondestructive In-Place Strength Testing
strength of the concrete at the tension surface of the beam,
but in the splitting tension test failure can be initiated any- The tests described thus far involve separately made speci-
where in the portion of the diametrical plane that is in tension. mens that do not necessarily give direct information about the
Thus, based on the size effect principle, it is expected that split- actual concrete in the structure. Field-cured specimens (ASTM
ting tensile strengths would be lower than flexural strength. C 31/C 31M or ASTM C 873), cores, and temperature-matched
This has been shown to be the case, with an average ratio of cured specimens [69] are of some help in accounting for in-
splitting tensile strength to flexural strength for center-point place conditions; however, such methods either require pre-
loading of 0.65 [60]. planning, use of specialized equipment, or cause localized
damage to the structure. To avoid such limitations, a variety of
Significance of Tensile Strength Test Results in-place tests have been developed. Typically called “nonde-
structive,” these tests either do not damage the concrete or
There is currently no standard test method for determining result in superficial localized damage. Tests that cause no
the direct tensile strength of concrete. Indications of tensile damage allow for retesting at the same, or nearly the same,
strength are obtained through the use of flexural and splitting location. Thus, changes in properties can be monitored over
tension test methods. Strengths obtained from each test time. It is important to note that in-place strength tests do not
method are particular to that method, and cannot be used provide direct measures of compressive strength, which
interchangeably. The flexural strength test is more similar to means that relationships to other standard test results must be
the loading encountered in pavements than is the splitting ten- established [70]. Therefore, in-place tests should be accompa-
sion test. The former, however, is more sensitive to moisture nied by laboratory testing to develop the strength relationship
conditions, and thus more care must be taken to prevent beam before use in the field. These in-place tests are discussed in
surfaces from drying. The splitting tension test subjects the greater detail in the chapter by Malhotra in this publication.
major portion of a diametrical plane to tensile stresses, and A useful and relatively simple technique to estimate in-place
thus more closely resembles a direct tension test. When selec- strength is the maturity method, as described in ASTM Practice
tion of a concrete tension test is being made, the loading of the for Estimating Concrete Strength by the Maturity Method (C
concrete in the structure should be considered, as well as the 1074). The maturity method provides an approach for making
expected use of the test results. If the strength is to be used in reliable estimates of in-place strength by establishing a relation-
design calculations, the proper test must be performed to ship between concrete strength, age, and temperature. First, a
obtain the values required by the design formulas. For exam- strength-maturity relationship is developed by laboratory tests on
ple, splitting tensile strength is used in the design of light- the concrete mixture to be used. Then the temperature history of
weight concrete beams to resist shear [61]. Even so, the results the field concrete is recorded from the time of concrete place-
obtained are not direct tensile strengths, and may be more or ment to the time when the strength estimation is desired. This
less sensitive to factors that affect tensile strength. In addition, temperature history is then used to calculate the maturity index
similar to compressive strength test limitations, conditions in of the field concrete. Finally, using the calculated maturity index
the structure are much more complex than can be taken into and the strength-maturity relationship, the strength of the field
account using strength tests on small specimens. concrete is estimated. The maturity concept can also be used to
Also the variability of flexural strength test results is high estimate later-age potential strength based upon the early-age
due to sensitivity to details of fabrication, handling, curing, strength of standard-cured specimens. This application is cov-
and testing. Thus flexural strength tests may be problematic ered by ASTM Test Method for Developing Early-Age Compres-
for acceptance testing of concrete. As an alternative, a corre- sion Test Values and Projecting Later-Age Strengths (C 918) and
lation between compressive and flexural strength can be devel- is discussed in the chapter by Carino.
oped for a given concrete mixture, and compressive strength Another commonly used in-place test for strength is the
testing can be used for control and acceptance of concrete, pullout test described in ASTM Test Method for Pullout
with flexural strength data used for reference only. Strength of Hardened Concrete (C 900). In this method, a
metal insert is either placed into fresh concrete or installed
Strength Relationships into hardened concrete. When the in-place strength is desired,
the insert is pulled by means of a jack reacting against a bear-
Due to the convenience of performing compressive strength ing ring. The pullout strength is determined by measuring the
tests, empirical strength relationships have been developed so amount of force necessary to pull the insert from the concrete
that other strength properties may be estimated from results mass. These pullout strengths can be related to compressive
of compressive strength tests. The relationship between com- strength test results. Other in-place tests include the penetra-
pressive and tensile strength has been found to be influenced tion resistance test described in ASTM Test Method for Pene-
by many factors, including concrete strength level, coarse tration Resistance of Hardened Concrete (C 803) and the
aggregate properties, testing age, curing, air entrainment, and rebound hammer test described in ASTM Test Method for
 OZYILDIRIM AND CARINO ON CONCRETE STRENGTH TESTING 137

Rebound Number of Hardened Concrete (C 805). The report

of ACI Committee 228 provides guidance on the proper use of
these types of in-place tests [70].

Strength Under Combined States of Stress

In structures, concrete members are often subjected to com-
binations of compressive, tensile, and shear stresses. Thus, it
may be important to understand the behavior of concrete
under different stress states. ASTM Test Method for Deter-
mining the Mechanical Properties of Hardened Concrete
Under Triaxial Loads (C 801) provided a standardized method
for determining such behavior. According to the test method,
cylindrical test specimens are subjected to a lateral confining
stress in combination with an axial load. Thus specimens are
subjected to triaxial stress states such that two of the three
principal stresses are equal. The test method is analogous to
the triaxial test used for soil specimens. ASTM C 801 was with-
drawn in 2004 because the method was not being used.
A common use of data obtained from triaxial compressive
tests is to construct a Mohr failure envelope for determining
shear strength as a function of normal compressive stress. The
shear strength of the concrete is estimated by drawing a tan-
gent to the Mohr’s circles at failure obtained for various com-
binations of confining pressure and axial stress (see Fig. 11). Fig. 12—Biaxial stress failure envelope for normal-
strength concrete [71].
Triaxial states of stress are typically of significance in mas-
sive concrete structures. Biaxial states of stress are more com-
mon. The compressive strength of concrete subjected to bi- a biaxial stress state depended on the type of coarse aggregate
axial compression is higher than its uniaxial compressive [73]. As with normal-strength concrete, average strength in-
strength, and the tensile strength is not affected significantly creases of roughly 30 % above the uniaxial compressive
by the application of biaxial tensile stresses [5,14,71]. Figure strength were observed, ranging from a minimum of 26 % for
12 is an example of a biaxial stress failure envelope in which trap rock coarse aggregate to a maximum of 34 % for granite
stress combinations that lie within the envelope correspond to coarse aggregate. In addition, it was noted that the proportional
no failure. Research results have shown that the degree of con- limit for high-strength concrete as well as the discontinuity
finement at specimen loading surfaces had a significant effect point were at higher stress-strength ratios than for normal-
on biaxial stress test results, resulting in biaxial to uniaxial strength concrete, and increased with increasing biaxial stress
compressive strength ratios in the range of 1.1–3.5 [72]. Bi- ratio. The discontinuity point is the stress at which unstable pro-
axial tests conducted using brush platens seem to have gressive microcrack growth begins. The failure modes of the
removed the effect of end confinement to a great degree. As specimens indicated that fracture was controlled by a limiting
determined with these end conditions, an average increase in tensile strain perpendicular to the plane of loading.
compressive strength with biaxial compressive loading is Calixto [74] performed biaxial tension-compression tests
about 30 %, and occurs at a principal stress ratio of approxi- on high-strength concrete specimens. In general, the applica-
mately 0.5 [72]. tion of lateral tension caused a significant decrease in the axial
Higher-strength concretes are becoming more common in compression capacity of the concrete, whereas the application
construction. In biaxial compression tests performed on high- of lateral compression had a much less pronounced effect on
strength concretes, the increase in compressive strength under the axial tensile capacity. The increased capacity of concrete un-
der a biaxial or triaxial compressive stress state is significant in
that the lateral restraint provided in reinforced concrete struc-
tures through such means as steel spirals will enhance the
capacity of the concrete. Lateral tension, however, causes a
significant decrease in concrete’s compressive strength. The
higher stress-strength ratios required to initiate progressive, un-
stable crack growth in higher-strength concretes would seem to
provide an additional safety margin in these concretes when
subjected to the same sustained stress-strength ratios as for nor-
mal-strength concrete.

Fatigue Strength of Concrete

Fatigue strength is the greatest stress that can be sustained for
a given number of stress cycles without failure. In numerous
Fig. 11—Shear strength as a function of normal stress structural applications, such as bridge decks and pavements,
obtained from triaxial compression tests. concrete members are subjected to repeated applications of
138 TESTS AND PROPERTIES OF CONCRETE

Fig. 13—Schematic of cyclic load history and fatigue strength curve of concrete.

load at a level below the ultimate strength of the concrete. of the ultimate strength; the fatigue limit decreased with in-
Like most materials, concrete exhibits fatigue behavior; that is, creasing biaxial stress ratio. Similar to normal strength con-
when subjected to cyclic loading of a given level but below its crete, the fatigue life of high-strength concrete increases with
short-term static strength, it will eventually fail. Thus, for struc- decreasing maximum stress-strength ratio and stress range, and
tural design purposes, it is desirable to establish the response wet specimens fail at less cycles compared with dry specimens.
of concrete to fatigue loading. However, a standardized test Cyclic loading of the specimens below the fatigue limit in-
method is not available. creased the uniaxial static strength on the order of 40 %.
Concrete does not have a fatigue limit, that is, a ratio of ap- Whereas normal-strength concrete is reported to have a fatigue
plied stress to static ultimate strength (stress-strength ratio) be- limit of approximately 55 % of its static ultimate strength, the
low which the concrete can withstand an infinite number of corresponding value for high-strength concrete was found to be
loading cycles [5,14]. Therefore, the fatigue limit of concrete is slightly lower, in the range of 47–52 %.
taken as the maximum stress-strength ratio at which failure oc- Structures such as bridge decks and pavements undergo nu-
curs only after a large number of cycles, usually 107. In normal- merous load cycles under normal service conditions. Thus, it is
strength concrete, the fatigue limit thus defined is approxi- important to understand the response of concrete to cyclic load-
mately 55 % of its static ultimate strength. ing. Over 100 references are given in the ACI report on consid-
The fatigue strength of concrete is sensitive to several fac- erations for design of concrete structures subjected to fatigue
tors [75], most of which are associated with the details of the loading, including references on biaxial stress states and high-
applied loading. The number of cycles of loading to failure strength concrete [77]. Furthermore, since factors such as the
depends on the level of cyclic stress applied to the specimen. magnitude of applied load, the stress state, loading frequency, du-
As shown in Fig. 13, as the stress-strength ratio is increased, ration of rest periods, and concrete moisture condition are
the number of cycles to failure decreases [2]. Also, the range largely unknown under actual service conditions of the structure,
of maximum to minimum cyclic stress has an effect. As the information obtained from laboratory studies can only give gen-
difference between the two stresses increases, the number of eral information about a very complex phenomenon.
cycles to failure decreases.
The sequence of loading cycles is also significant. Miner’s
rule, which states that the effects of cyclic loads are cumula- Summary
tive, does not strictly apply to concrete. Rather, the fatigue life,
the number of cycles to failure, of a specimen is different if it This chapter has reviewed the standard practices and test
is subjected first to high stress-strength ratios followed by low methods used to measure concrete strength. These standards
stress-strength ratios, than if subjected to the reverse are under the jurisdiction of Subcommittee C09.61. The dis-
sequence. Cycling below the fatigue limit increases both the cussion has stressed the limitations of the results obtained
fatigue strength and the static strength by 5–15 % [5,14,76]. from tests on small specimens in estimating properties of the
When the frequency of load application is low, the fatigue life concrete in the structure. Nevertheless, with a thorough
is shortened compared with a higher frequency of load appli- knowledge of their limitations and the factors that affect meas-
cation. This is probably due to mechanisms of creep and crack ured strengths, results obtained from correctly performed
propagation. Rest periods of up to 5 min during fatigue test- standard test procedures will provide useful information.
ing also increase the fatigue life; increasing the length of the In concrete construction, strength tests have three main
rest period beyond 5 min has no additional beneficial effect. purposes: (1) research, (2) for acceptance of concrete and
Specimens subjected to a stress gradient also exhibit en- quality control by testing specimens subjected to standard-
hanced fatigue life over those subjected to a uniform stress curing conditions, and (3) for evaluating in-place properties of
distribution, possibly because the lower stressed regions concrete. While the tests discussed in this chapter appear to be
inhibit crack growth. Finally, specimens tested in a moist con- simple, strict adherence to specified procedures and toler-
dition exhibit lower fatigue life compared with those tested in ances is necessary for obtaining meaningful results and ensur-
a dry condition. ing reproducibility between laboratories. It is essential that the
In biaxial compression fatigue tests on high-strength con- purpose of the test be defined, and that those factors that will
crete, Nelson et al. [75] reported that biaxial compression de- influence test results be understood and controlled in accor-
creased the fatigue life for maximum stress levels below 76 % dance with the applicable standards.
 OZYILDIRIM AND CARINO ON CONCRETE STRENGTH TESTING 139

[22] Bloem, D. L., “Concrete Strength Measurement—Cores Versus

References Cylinders,” Proceedings, Vol. 65, ASTM International, West
[1] Kesler, C. E. and Seiss, C. P., “Static and Fatigue Strength,” Conshohocken, PA, 1965, pp. 668–696.
Significance of Tests and Properties of Concrete and Concrete [23] Yuan, R. L., Ragab, M., Hill, R. E., and Cook, J. E., “Evaluation
Aggregates, ASTM STP 169, ASTM International, West Con- of Core Strength in High Strength Concrete,” Concrete Inter-
shohocken, PA, 1956, pp. 81–93. national, Vol. 13, No. 5, May 1991, pp. 30–34.
[2] Kesler, C. E., “Strength,” Significance of Tests and Properties [24] Szypula, A. and Grossman, J. S., “Cylinder vs. Core Strength,”
of Concrete and Concrete-Making Materials, ASTM STP Concrete International, Vol. 12, No. 2, 1990, pp. 55–61.
169A, ASTM International, West Conshohocken, PA, 1966, [25] Fiorato, A. E., Burg R. G., and Gaynor, R. D., “Effects of Condi-
pp. 144–159. tioning on Measured Compressive Strength of Concrete
[3] Derucher, K. N., “Strength,” Significance of Tests and Proper- Cores,” Concrete Technology Today, Vol. 21, No. 3, December
ties of Concrete and Concrete-Making Materials, ASTM STP 2000, p. 3
169B, ASTM International, West Conshohocken, PA, 1978, [26] Bartlett, F. M. and MacGregor, J. G., “Effect of Moisture Con-
pp. 146–161. dition on Concrete Core Strengths,” ACI Materials Journal, Vol.
[4] Carrasquillo, P. M., “Concrete Strength Testing,” Significance 91 No.3, May-June 1994, pp. 227–236.
of Tests and Properties of Concrete and Concrete-Making [27] Richardson, D. N., “Effects of Testing Variables on the Compar-
Material, ASTM STP 169C, ASTM International, West Con- ison of Neoprene Pad and Sulfur Mortar-Capped Concrete
shohocken, PA, 1994, pp. 183–206. Test Cylinders,” ACI Materials Journal, Vol. 87, No. 5, 1990,
[5] Neville, A. M., Properties of Concrete, 4th ed., John Wiley & pp. 489–495.
Sons, Inc., New York, 1996. [28] Werner, G., “The Effect of Type of Capping Material on the
[6] Bloem, D. L., “Concrete Strength in Structures,” Proceedings, Compressive Strength of Concrete Cylinders,” Proceedings,
Vol. 65, American Concrete Institute, 1968, pp. 176–187. Vol. 58, ASTM International, West Conshohocken, PA, 1958,
[7] Price, W. H., “Factors Influencing Concrete Strength,” Pro- pp. 1166–1186.
ceedings, Vol. 47, American Concrete Institute, 1951, pp. [29] Troxell, G. E., “The Effect of Capping Methods and End Condi-
417–432. tions Before Capping Upon the Compressive Strength of
[8] Gonnerman, H. F. and Shuman, E. C., “Compression, Flexure Concrete Test Cylinders,” Proceedings, Vol. 41, ASTM Interna-
and Tension Tests of Plain Concrete,” Proceedings, Vol. 28, Part tional, West Conshohocken, PA, 1942, pp. 1038–1044.
II, ASTM International, West Conshohocken, PA, 1928, pp. [30] Gonnerman, H. F., “Effect of End Condition of Cylinder in Com-
527–564. pression Tests of Concrete,” Proceedings, Vol. 24, Part II, ASTM
[9] Blick, R. L., “Some Factors Influencing High-Strength Con- International, West Conshohocken, PA, 1924, pp. 1036–1065.
crete,” Modern Concrete, Vol. 36, No. 12, April 1973, pp. [31] Vichit-Vadakan, W., Carino, N. J., and Mullings, G. M., “Effect
38–41. of Elastic Modulus of Capping Material on Measured Strength
[10] Mehta, P. K., “Pozzolanic and Cementitious By-Products as of High-Strength Concrete Cylinders,” Cement, Concrete, and
Mineral Admixtures for Concrete—A Critical Review,” SP-79, Aggregates, CCAGDP, Vol. 20, No. 2, Dec. 1998, pp. 227–234.
Vol. I, American Concrete Institute, Detroit, MI, 1983, pp. [32] Lobo, C. L., Mullins, G. M., and Gaynor, R. D., “Effect of Cap-
1–46. ping Materials and Procedures on the Measured Compressive
[11] Neville, A., “There is More to Concrete than Cement,” Concrete Strength of High-Strength Concrete,” Cement, Concrete, and
International, Vol. 22, No. 1, Jan. 2000, pp. 73–74. Aggregates, Vol. 16, No. 2, Dec. 1994, pp. 173–180.
[12] Vachon, M., “ASTM Puts Self-Consolidating Concrete to the [33] Ozyildirim, C., “Neoprene Pads for Capping Concrete Cylin-
Test,” ASTM Standardization News, July 2002, pp. 34–37. ders,” Cement, Concrete, and Aggregates, Vol. 7, No. 1, 1985,
[13] Carrasquillo, P. M. and Carrasquillo, R. L., “Evaluation of the pp. 25–28.
Use of Current Concrete Practice in the Production of High- [34] Crouch, L. K. and Pearson, J. B., “Neoprene Capping for Static
Strength Concrete,” Materials Journal, Vol. 85, No. 1, 1988, Modulus of Elasticity Testing,” ACI Materials Journal, Vol. 92,
pp. 49–54. No. 6, Nov. 1995, pp. 643–648.
[14] Mindess, S. and Young, J. F., Concrete, Prentice-Hall, Inc., [35] Nasser, K. W. and Al-Manaseer, A. A., “It’s Time for a Change
Englewood Cliffs, NJ, 1981. from 6  12- to 3  6-in. Cylinders,” Materials Journal, Vol. 84,
[15] Concrete Testing-Alternative Methods Available, Concrete No. 3, 1987, pp. 213–216.
Construction, Addison, IL, April 1973, pp. 14–23. [36] Malhotra, V. M., “Are 4  8 Inch Concrete Cylinders as Good as
6  12 Inch Cylinders for Quality Control of Concrete?”
[16] Jones, R. and Wright, P. J. F., “Some Problems Involved in
Proceedings, Vol. 73, American Concrete Institute, 1976,
Destructive and Non-destructive Testing of Concrete,” Mix
pp. 33–36.
Design and Quality Control of Concrete, Proceedings, Cement
[37] Gonnerman, H. F., “Effect of Size and Shape of Test Specimen
and Concrete Association, London, May 1954, pp. 441–464.
on Compressive Strength of Concrete,” Proceedings, Vol. 25,
[17] Price, W. F. and Hynes, J. P., “In-Situ Strength Testing of High
Part II, ASTM International, West Conshohocken, PA, 1925,
Strength Concrete,” Magazine of Concrete Research, Thomas
pp. 237–255.
Telford Services Ltd., London, Sept. 1996, pp. 189–197.
[38] Peterman, M. B. and Carrasquillo, R. L., “Production of High
[18] Kehl, R. J., Constantino, C. A., and Carrasquillo, R., “Match- Strength Concrete,” Research Report 315–1F, Center for Trans-
Cure and Maturity: Taking Concrete Strength Testing to a portation Research, University of Texas at Austin, Oct. 1983.
Higher Level,” Report No. 1714-S, University of Texas, Center
[39] Mirza, S. A., Hatzinikolas, M., and MacGregor, J. G., “Statistical
for Transportation Research, 1998. Descriptions of Strength of Concrete,” Journal, Structural
[19] Bungey, J. H., The Testing of Concrete in Structures, 2nd ed., Division, American Society of Civil Engineers, Vol. 105, No. ST6,
Chapman and Hall, New York, 1989, pp. 94–110. June 1979.
[20] Malhotra, V. M., “Concrete Strength Requirements—Cores [40] Bartlett, F. M. and MacGregor, J. G., “Effect of Core Diameter
Versus In-Situ Evaluation,” Proceedings, American Concrete on Concrete Core Strengths,” ACI Materials Journal, Vol. 91,
Institute, Vol. 74, No. 4, 1977, pp. 163–172. No. 6, Sept. 1994, pp. 460–470.
[21] Suprenant, B. A., “An Introduction to Concrete Core Testing,” [41] Day, R. L., “Strength Measurement of Concrete Using Different
Civil Engineering for Practicing and Design Engineers, Vol. 4, Cylinder Sizes: A Statistical Analysis,” Cement, Concrete, and
No. 8, 1985, pp. 607–615. Aggregates, Vol. 16, No. 1, 1994, pp. 21–30.
140 TESTS AND PROPERTIES OF CONCRETE

[42] Carino, N. J., Guthrie, W., and Lagergren, E., “Effects of Testing [60] Melis, L. M., Meyer, A. H., and Fowler, D. W., “An Evaluation of
Variables on the Measured Compressive Strength of High- Tensile Strength Testing,” Research Report 432-1F, Center for
Strength (90 MPa) Concrete,” U.S. Department of Commerce, Transportation Research, University of Texas at Austin, Nov.
Technology Administration, NISTIR 5405, 1994. 1985.
[43] Carino N. J., Mullings, G. M., and Guthrie, W. F., “Evaluation of [61] “Building Code Requirements for Structural Concrete,” ACI
the ASTM Standard Consolidation Requirements for Preparing 318-05, American Concrete Institute, Farmington Hills, MI.
High-Strength Concrete Cylinders,” Special Publication SP-172, [62] Kesler, C. E., “Statistical Relation Between Cylinder, Modified
American Concrete Institute, 1999, pp. 733–768. Cube, and Beam Strength of Plain Concrete,” Proceedings,
[44] McMillan, F. R., “Suggested Procedure for Testing Concrete in Vol. 54, ASTM International, West Conshohocken, PA, 1954,
Which the Aggregate is More Than One-Fourth the Diameter pp. 1178–1187.
of the Cylinders,” Proceedings, Vol. 30, Part I, ASTM Interna- [63] Johnson, A. N., “Concrete in Tension,” Proceedings, Vol. 26,
tional, West Conshohocken, PA, 1930, pp. 521–535. Part II, ASTM International, West Conshohocken, PA, 1926,
[45] Chung, H. W., “On Testing of Very Short Concrete Specimens,” pp. 441–450.
Cement, Concrete, and Aggregates, Vol. 11, No. 1, 1989, [64] Popovics, S., Strength and Related Properties of Concrete: A
pp. 40–44. Quantitative Approach, John Wiley & Sons, Inc., New York,
[46] Kesler, C. E., “Effect of Length to Diameter Ratio on Compres- 1998.
sive Strength—An ASTM Cooperative Investigation,” Proceed- [65] Carino, N. J., and Lew, H. S., “Re-Examination of the Relation
ings, Vol. 59, ASTM International, West Conshohocken, PA, Between Splitting Tensile nd Compressive Strength of Normal
1959, pp. 1216–1229. Weight Concrete,” ACI Journal, Vol. 79, No. 3, May-June 1982,
[47] Tucker, J., Jr., “Effect of Length on the Strength of Compres- pp. 214–219.
sion Test Specimens,” Proceedings, Vol. 45, ASTM Internation- [66] Oluokun, F. A., Burdette, E. G., and Deatherage, J. H., “Split-
al, West Conshohocken, PA, 1945, pp. 976–984. ting Tensile Strength and Compressive Strength Relationships
[48] Bartlett, F. M. and MacGregor J. G., “Effect of Core Length-to- at Early Ages,” ACI Materials Journal, Vol. 88, No. 2, March-
Diameter Ratio on Concrete Core Strength,” ACI Materials April 1991, pp. 115–121.
Journal, Vol. 91, No. 4, July-August 1994, pp. 339–348. [67] Oluokun, F. A., “Prediction of Concrete Tensile Strength from
[49] Popovics, S., “Effect of Curing Method and Final Moisture its Compressive Strength Evaluation Relations for Normal
Condition on Compressive Strength of Concrete,” Proceedings, Weight Concrete,” ACI Materials Journal, Vol. 88, No. 3, May-
Vol. 83, American Concrete Institute, 1986, pp. 650–657. June 1991, pp. 302–309.
[50] Jones, P. G. and Richart, F. E., “The Effect of Testing Speed on [68] Gardner, N. J., “Effect of Temperature on the Early-Age
Strength and Elastic Properties of Concrete,” Proceedings, Vol. Properties of Type I, Type III, and Type I/Fly Ash Concretes,”
36, Part II, ASTM International, West Conshohocken, PA, 1937, ACI Materials Journal, Vol. 87, No. 1, Jan.-Feb. 1990,
pp. 380–392. pp. 68–78.
[51] Abrams, D. A., “Effect of Rate of Application of Load on the [69] Ozyildirim, C. and Gomez, J., “High-Performance Concrete in
Compressive Strength of Concrete,” Proceedings, Vol. 17, Part the Richlands Bridge in Virginia,” TRR 1698, Concrete 2000,
II, ASTM International, West Conshohocken, PA, 1917, pp. Transportation Research Board, 2000, pp. 17–23.
364–377. [70] “In-Place Methods to Estimate Concrete Strength,” ACI
[52] Wright, P. J. F., “Comments on an Indirect Tensile Test on Con- 228.1R-03, American Concrete Institute, Farmington Hills, MI,
crete Cylinders,” Magazine of Concrete Research, Vol. 7, No. 2005.
20, July 1955, pp. 87–96. [71] Carino, N. J., “The Behavior of a Model of Plain Concrete Sub-
[53] Goldbeck, A. T., “Apparatus for Flexural Tests of Concrete jected to Compression-Tension and Tension-Tension Biaxial
Beams,” Proceedings, Vol. 30, Part I, ASTM International, West Stresses,” Ph.D. thesis, Cornell University, Ithaca, NY, Aug.
Conshohocken, PA, 1930, pp. 591–597. 1974.
[54] Reagel, F. V. and Willis, T. F., “The Effect of the Dimensions of [72] Herrin, J. C., “Behavior of a High Strength Concrete Model Sub-
Test Specimens on the Flexural Strength of Concrete,” Public jected to Biaxial Compression,” Masters thesis, The University
Roads, Vol. 12, No. 2, April 1931, pp. 37–46. of Texas at Austin, Dec. 1982.
[55] Kellermann, W. F., “Effect of Size of Specimen, Size of Aggre- [73] Chen, R. L., Fowler, D. W., and Carrasquillo, R. L., “Behavior of
gate and Method of Loading Upon the Uniformity of Flexural High Strength Concrete in Biaxial Compression,” Research
Strength Results,” Public Roads, Vol. 13, No. 11, Jan. 1933, Report AF-3, Defense Technical Information Service, U.S. Air
pp. 177–184. Force, Dec. 1984.
[56] Wright, P. J. F. and Garwood, F., “The Effect of the Method of [74] Calixto, J. M. F., “Microcracking of High Strength Concrete Sub-
Test on the Flexural Strength of Concrete,” Magazine of jected to Biaxial Tension-Compression Stresses,” Masters thesis,
Concrete Research, No. 11, Oct. 1952, pp. 67–76. The University of Texas at Austin, May 1987.
[57] Carrasquillo, P. M. and Carrasquillo, R. L., “Improved Concrete [75] Nelson, E. L., Fowler, D. W., and Carrasquillo, R. L., “Fatigue of
Quality Control Procedures Using Third Point Loading,” High Strength Concrete Subjected to Biaxial-Cycle Compres-
Research Report 1119-1F, Center for Transportation Research, sion,” Research Report AF-4, Defense Technical Information
University of Texas at Austin, Nov. 1987. Service, U.S. Air Force, May 1986.
[58] Nielsen, K. E. C., “Effect of Various Factors on the Flexural [76] Nordby, G. M., “Fatigue of Concrete—A Review of Research,”
Strength of Concrete Test Beams,” Magazine of Concrete Proceedings, Vol. 55, American Concrete Institute, 1959, pp.
Research, No. 15, March 1954, pp. 105–114. 191–219.
[59] Johnston, C. D. and Sidwell, E. H., “Influence of Drying on [77] “Considerations for Design of Concrete Structures Subjected to
Strength of Concrete Specimens,” Proceedings, Vol. 66, Ameri- Fatigue Loading,” ACI 21R-74 (1996), American Concrete Insti-
can Concrete Institute, 1969, pp. 748–755. tute, Farmington Hills, MI, 2005.
14
Prediction of Potential Concrete Strength
at Later Ages

Nicholas J. Carino1

Preface 1996, the subcommittee was discharged and jurisdiction of its

standards was assigned to Subcommittee C09.61 on Strength
THIS CHAPTER DEALS WITH METHODS FOR ESTI- Testing.) The subcommittee developed two methods for esti-
mating the potential later-age strength of a concrete mixture mating the later-age strength of concrete specimens based
based upon the compressive strength measured on cylindrical upon early-age strength tests.
specimens at early ages. The results of the cooperative research One of these methods, ASTM C 684, involves testing
program leading to the development of ASTM Test Method for cylinders whose strength development has been accelerated by
Making, Accelerated Curing, and Testing Concrete Compression elevated curing temperatures. Depending on the specific
Test Specimens (C 684) are reviewed. This review is based on procedure that is used, the accelerated strength is measured at
Chapter 13 of ASTM STP 169B, which was written by M. H. Wills, ages ranging from 5 h to 48 h. A previously established rela-
Jr., and was titled “Accelerated Strength Tests.” The earlier text tionship between accelerated strength and standard-cured
was augmented by the current author in ASTM STP 169C by the strength is used to estimate the later-age strength under stan-
addition of information on the high-temperature and pressure- dard curing conditions based on the measured strength of ac-
accelerated test method that was added to ASTM C 684 in 1989. celerated test specimens.
In addition, the author expanded the chapter in ASTM STP 169C The other method, ASTM C 918, uses the maturity method
by including the basis of ASTM Test Method for Developing [1] and a previously established strength-maturity relationship
Early-Age Compression Test Values and Projecting Later-Age to estimate later-age strength based on the early-age strength of
Strengths (C 918). This current version updates changes to specimens whose temperature history has been measured.
ASTM C 684 and ASTM C 918 since ASTM STP 169C was pub- This chapter reviews the background of these two tech-
lished. niques and provides supporting information to the standard
test methods to assist persons who are contemplating specify-
Introduction ing these procedures in contract documents.

Rapid construction practices throughout the concrete industry Accelerated Curing Methods
have caused specifying agencies to require assessment of con-
crete strength at an age earlier than the traditional 28 days The first assignment for ASTM Subcommittee C09.02.09 was to
after placement. Currently, the later age is still specified for investigate the needs of the concrete industry and to study the
compression tests for acceptance of concrete as delivered to suitability of standardizing several accelerated testing proce-
the job site. During the 28-day period between the preparation dures being developed by King [2,3], Akroyd [4], Smith and
and testing of the specimens, a multistory building could go up Chojnacki [5], Malhotra and Zoldners [6], and Smith and Tiede
four or more floors. Many consider it precarious for construc- [7]. A positive response to this canvass led to a cooperative test
tion to proceed in this manner. Furthermore, expensive and program among nine laboratories to evaluate various proce-
costly delays are encountered when 28-day test results are low, dures (Table 1) involving immersion of specimens in hot or
because a field investigation may be necessary to verify the boiling water or autogenous curing in an insulated container
load-carrying capacity of the structure. Further delay is certain to accelerate strength development of concrete [8].
if the structure must be strengthened or replaced. Early as-
sessment of concrete quality is, therefore, essential for overall Experimental Program
construction economy and safety. Each of the accelerated testing procedures was conducted us-
Recognizing the need for alternative test methods for as- ing ASTM Type I (or Type II meeting Type I specifications) and
sessing the quality of concrete at early ages, ASTM Committee Type III portland cement. Cement contents were 265, 325, and
C9 on Concrete and Concrete Aggregates formed Subcommit- 385 kg/m3 (450, 550, and 650 lb/yd3). Sufficient air-entraining
tee C09.02.09 on Accelerated Strength Testing in 1964. (In admixture and mixing water were used to produce air contents

1
Consultant, Gaithersburg, MD.

141
142 TESTS AND PROPERTIES OF CONCRETE

the option of conducting one or more of Procedures D through

TABLE 1—Accelerated Testing Procedures G (Table 1). Despite the plans of the subcommittee, it was nec-
Procedure Description
essary to curtail some aspects of the experimental program.
Procedure C, involving the measurement of final setting time
A—Warm water Cylinders placed immediately in followed by boiling, had to be abandoned because it required
water at 35°C (95°F) too much overtime work to conduct. Consequently, Procedures
B—Modified boiling Standard curing for 23 h E, F, and G were also abandoned because they also required
followed by boiling for 3.5 h measurement of setting time and overtime work. Only four of
C—Final set and boiling At final setting (ASTM C 403/C the nine participating laboratories conducted Procedure D
403M),a cylinders boiled for 15 h (Autogenous Curing), which was optional. However, all labora-
D—Autogenous curing At 1-h age, place in insulated tories conducted Procedure A (warm water) and Procedure B
container for 46 h (modified boiling).
E—Initial set and hot At initial setting (ASTM C 403/C
water, 55°C (130°F) 403M), place in hot water for 15 h
F—Initial set and hot At initial setting, place in hot Accelerated Curing Apparatus
water, 75°C (175°F) water for 15 h The accelerated procedures involving the use of either hot or
G—Initial set and hot At initial setting, place in hot boiling water were conducted in thermostatically controlled
water, 90°C (195°F) water for 15 h tanks (Fig. 1). The tops of the cylinders were covered by about
100 mm (4 in.) of water. Water volume and heater capacity were
sized to prevent an appreciable reduction in the desired water
a
ASTM Test Method for Time of Setting of Concrete Mixtures by Penetration
temperature when the specimens were immersed. Specimens
Resistance (C 403/C 403M). were not placed in a tank that already contained specimens.
The autogenous curing container for Procedure D was
from 5.0 to 6.0 % and slumps from 50 to 75 mm (2 to 3 in.). Con- similar to that used by Smith and Tiede [7] to cure a single 152
cretes were made without a retarder or with a normal dosage by 305-mm (6 by 12-in.) cylinder. It was made by using a plas-
of a water-reducing retarder. Fine and coarse aggregates were tic container and polyurethane foam insulation (Fig. 2). The
graded to a No. 57 size, that is, 25.0 to 4.75 mm (1 in. to No. 4). polyurethane foam retained the heat of hydration of the ce-
Each laboratory used materials available in their locality with- ment, which accelerated the strength development of the con-
out interchange among laboratories. In all cases, compressive crete. Once curing was initiated, the container was not opened
strengths of the accelerated specimens were measured at ages until the end of the specified curing period.
of one day or two days and were compared with strengths of
standard-cured specimens at ages of 28 days and 364 days. Two Results
replicate specimens were tested at each age, and their averages The results of the cooperative testing program were summa-
were used in subsequent analyses. All specimens were prepared, rized by Wills [8], and only the main conclusions are
cured, and tested according to applicable ASTM standards. All highlighted here. The objective was to establish the nature of
materials conformed to appropriate ASTM specifications. the relationships between accelerated and standard-cured
The nine participating laboratories were asked to conduct strengths, and to examine whether these relationships were af-
Procedures A, B, and C listed in Table 1. Additionally, they had fected by factors such as the type of accelerated procedure,

Fig. 1—Hot or boiling-water accelerated curing tank.

 CARINO ON PREDICTION OF POTENTIAL STRENGTH AT LATER AGES 143

Fig. 3—Relationships between accelerated and standard-

cured strengths for Procedure A (Warm Water Method) using
Type I cement.

Procedure B (Modified Boiling)

After the cylinders were cast, the sealed molds were placed in
a standard moist room. At 23 h
15 min from the time of cast-
Fig. 2—Autogenous curing container. ing, the cylinders, including molds and covers, were immersed
in boiling water. Reduction in water temperature was limited
to 3°C (5°F), and the temperature was required to recover to
cement type, cement content, presence of set retarder, and lab- the boiling point in no more than 15 min. After boiling for 3.5
oratory. Only the results of Procedures A, B, and D are h
5 min, the cylinders were removed from the molds and al-
discussed here, because these procedures were eventually in- lowed to cool for about 45 min. They were capped with sulfur
corporated into the first version of ASTM C 684 in 1974. mortar, the sulfur was allowed to age for 1 h, and strengths
were measured at an age of 28.5 h
15 min. The modified boil-
Procedure A (Warm Water) ing procedure increased concrete strength between 1.1 and 2.1
Cylinders were cast in steel molds with tight-closing lids. Imme- times that measured after one day of standard moist curing.
diately after casting, the cylinder molds were immersed in the Figure 4 shows the results for concretes cured by Procedure
water bath at a temperature of 33 to 37°C (92 to 98°F), where B and made with Type III cement. The trends discussed for Pro-
they remained for a period of 24 h
15 min. Sulfur mortar caps cedure A are reemphasized. Most importantly, the laboratories
were applied to the cylinders and aged at least 1 h prior to meas- also obtained what seemed to be significantly different correla-
uring compressive strength at an age of 26 h
15 min. The tions. Again, this is attributed to each laboratory using locally
warm-water curing increased concrete strengths from 1.1 to 1.6 available materials. Within a given laboratory, however, good
times those achieved after one day of standard moist curing. correlations between accelerated strength and 28- or 364-day
Figure 3 shows the relationships between accelerated strengths were obtained [8]. Therefore, Procedure B was also
strength and 28-day strength for concretes made with Type I considered to have equal merit in assessing concrete quality.
cement. Within each laboratory, there were good correlations;
however, it appeared that each laboratory obtained signifi- Procedure D (Autogenous Curing)
cantly different correlations. Cylinders were molded in light-gage steel molds. One hour af-
Although not shown, the same general trends were ob- ter the start of mixing, the molded cylinders were sealed in
served with concretes made with Type III cement, except that plastic bags and placed inside the autogenous curing chambers
the relationships were at a higher strength level. Since the test (Fig. 2). They remained there for 46 h, then they were removed
procedures were carefully controlled, the results emphasized from the molds and allowed to cool for 45 min. Sulfur mortar
the effect of local materials, particularly the type of cement, on caps were applied and aged at least 1 h before strength was
the relationships between accelerated strength and standard- measured at an age of 49 h
15 min.
cured strength. Based upon the good correlations, however, it Only four laboratories performed this procedure, but their
was concluded that Procedure A can be used with a high de- data were sufficient to justify including autogenous curing in
gree of confidence to assess concrete quality when the tests are the subsequent ASTM standard. The two-day strengths after au-
made on mixtures produced from the same materials. togenous curing ranged from 1.4 to 2.5 times those obtained
144 TESTS AND PROPERTIES OF CONCRETE

tion but at a low rate of temperature increase, which is benefi-

cial to early hydration reactions. This previous fact, plus the ad-
ditional fact that higher curing temperatures are attained in this
procedure compared with Procedure A, may explain the high
values of accelerated strength for Procedure D.

Significance of Test Procedures

The subcommittee was convinced that laboratories, test proce-
dures, and cement types had significant effects on the correla-
tion between accelerated strength and standard-cured strength,
and the data were analyzed from that viewpoint [8]. It was as-
sumed that the correlation for each set of conditions could be
represented by a straight-line relationship as follows

Ŝ 28  B0  B1Sa (1)

where

S28  the standard-cured 28-day strength,

Sa  the accelerated strength for a particular procedure,
B0  the intercept, and
B1  the slope.

Fig. 4—Relationships between accelerated and standard-

For each laboratory, cement type, and procedure, the
cured strengths for Procedure B (Modified Boiling Method) best-fit values of B0 and B1 were obtained by least squares fit-
using Type III cement. ting. The resulting values of B0, B1, correlation coefficients
(r ), and residual standard deviations are summarized in Ta-
bles 2, 3, and 4 for Procedures A, B, and D, respectively. The
residual standard deviation (se) is a measure of the error be-
after an equal length of standard moist curing, which was the tween the data and the best-fit straight line and is computed
highest level of strength acceleration. as follows:
High concrete temperatures at early ages are detrimental


n
to the early hydration reactions of cement and to the develop-
ing paste microstructure. With the autogenous curing proce- se 
∑i (Sai  Ŝ 28)2 (2)

dure, the heat of hydration of the cement causes the accelera- n2

TABLE 2—Linear Regression Analysis Results for Procedure A

Laboratory Cement B0,a RSD,c
Number Type MPa B1a rb MPad

1 I 13.54 1.120 0.910 2.10

III 15.65 0.960 0.975 0.86
4 I 19.10 1.515 0.955 1.96
III 18.96 1.095 0.925 2.76
5 I 15.27 1.375 0.960 1.86
III 17.86 0.905 0.980 1.07
6 I 14.17 1.525 0.985 0.86
III 14.31 1.290 0.940 1.93
8 I 9.10 1.985 0.905 2.90
III 15.44 1.285 0.935 2.31
9 I 16.31 1.320 0.765 3.69
III 23.48 0.540 0.870 1.59
10 I 19.13 1.440 0.915 1.62
III 24.20 0.770 0.815 1.28
11 I 14.34 1.475 0.905 2.83
III 13.76 1.090 0.725 3.62
12 I 17.65 1.440 0.915 1.96
III 15.65 0.580 0.825 2.07

a
See Eq 1 for meaning of B0 and B1.
b
r  correlation coefficient.
c
RSD  residual standard deviation.
d
1 MPa  145 psi.
 CARINO ON PREDICTION OF POTENTIAL STRENGTH AT LATER AGES 145

TABLE 3—Linear Regression Analysis Results for Procedure B

Laboratory Cement B0,a RSD,c
Number Type MPa B1a r b
MPad

1 I 12.82 1.060 0.970 1.34

III 16.55 0.910 0.985 0.66
4 I 17.79 1.290 0.965 1.55
III 13.96 1.145 0.955 2.17
5 I 9.86 1.225 0.975 1.21
III 14.96 0.840 0.930 1.28
6 I 11.58 1.280 0.970 0.79
III 13.13 1.050 0.965 1.07
8 I 15.62 1.280 0.810 4.03
III 14.86 1.220 0.950 2.03
9 I 16.89 1.015 0.865 1.58
III 19.55 0.780 0.960 1.24
10 I 19.48 1.195 0.950 1.24
III 17.79 1.000 0.975 0.90
11 I 9.52 1.515 0.975 1.86
III 13.41 1.095 0.800 3.48
12 I 18.13 1.020 0.860 2.48
III 18.27 1.143 0.830 3.07

a
See Eq 1 for meaning of B0 and B1.
b
r  correlation coefficient.
c
RSD  residual standard deviation.
d
1 MPa  145 psi.

where had a much lower correlation coefficient and a higher residual

standard deviation. Therefore, the data for Laboratory 11 did
Sai  the average accelerated strength of the ith specimens,
not fit its straight-line relationship as well as the data for Lab-
Ŝ 28  the estimated 28-day strength corresponding to Sai, and
oratory 6.
n  number of pairs of strength values used in the regres-
After studying the residual standard deviations, it was con-
sion analysis.
cluded that all three procedures were equal in correlating ac-
For example, the regression equation for Laboratory 6 us- celerated and later-age strengths. Correlation coefficients were
ing Procedure A and Type I cement was quite high in most cases, as would be expected by the close fit
of the data to the linear relationships shown in Figs. 3 and 4 for
Ŝ 28  14.17  1.525 Sa (MPa) (3) each laboratory. Further, it was found that accelerated com-
pressive strength correlated with 364-day strength as well as
Note that the correlation coefficient had a high value of 0.985 with 28-day strength. Subsequently, the subcommittee started
and residual standard deviation had a correspondingly low work to develop a tentative test method that included Proce-
value of 0.86 MPa (125 psi). On the other hand, Laboratory 11 dures A, B, and D. The tentative test method was approved in
1971 as ASTM C 684-71T.

TABLE 4—Linear Regression Analysis Results Test Precision

for Procedure D While accelerated strengths measured by the nine laboratories
were significantly different for a given procedure, the variances
Laboratory Cement B0,a RSD,c (squares of standard deviations) of replicate strength measure-
Number Type MPa B1a rb MPad
ments seemed compatible and were therefore pooled across
laboratories and cement types for each procedure. The within-
4 I 11.45 1.000 0.990 0.96
III 3.79 1.230 1.000 0.41
batch and batch-to-batch coefficients of variation that were ob-
5 I 22.03 1.315 0.780 4.07 tained from the data analysis are shown in Table 5 [8]. These
III 22.65 0.875 0.960 1.48 values were used to prepare precision statements, according to
8 I 7.17 1.400 0.950 1.72 ASTM Practice for Preparing Precision and Bias Statements for
III 9.96 1.125 0.970 1.93 Test Methods for Construction Materials (C 670), for the three
12 I 16.86 0.930 0.900 2.10 procedures [8]. Because of the small differences between the re-
III 9.82 0.810 0.930 1.45 sulting statements, they were combined into a single precision
statement when ASTM C 684 was published as a tentative test
method. The autogenous curing procedure was called Proce-
a
dure C in ASTM C 684. Note that the precision statements deal
See Eq 1 for meaning of B0 and B1.
b
r  correlation coefficient.
only with the repeatability of the measured accelerated
c
RSD  residual standard deviation. strengths. A procedure for determining the precision of the es-
d
1 MPa  145 psi. timated standard-cured strengths is discussed subsequently.
146 TESTS AND PROPERTIES OF CONCRETE

schematic of the apparatus is shown in Fig. 6. Special molds

TABLE 5—Precision of Accelerated Strength with heating wires and insulation are used to prepare 76 by 152-
Tests [8] mm (3 by 6-in.) cylinders. After three molds are filled, they are
stacked in a compression testing frame, a compressive stress of
Within-Batch Batch-to-Batch 10.3
0.2 MPa (1500
25 psi) is applied, and the electrical
Coefficient of Coefficient of
heaters are turned on. The heaters raise and maintain the con-
Procedure Variation, % Variation, %
crete temperature at 149
3°C (300
5°F). After three hours,
Warm water 2.9 8.2
the heaters are turned off, the axial stress is maintained, and the
Modified boiling 3.0 8.5 specimens are allowed to cool for 2 h. The hardened cylinders
Autogenous curing 3.6 8.5 are extruded from the molds and tested for compressive
strength. Usually, capping materials are not needed because the
metal end caps result in sufficiently flat ends.
In 1980, the subcommittee was requested to modify the ex-
isting version of ASTM C 684 to permit the high temperature
Effect of Cement Chemistry and pressure (HTP) procedure as an alternative to the other
In order to explain the differences between laboratories, a pro- methods. Data were provided to the subcommittee to demon-
gram was conducted to determine the chemical or physical prop- strate that the method resulted in correlations similar to the
erties of cement, or both, that affect the one-day accelerated other methods [10]. The concrete mixtures were made with
strength of concrete. Similar concrete batches were made using Type I, III, and IV cements; water-cement ratios (w/c) between
eight Type I cements mixed with the same source of sand and 0.45 and 0.90; with and without fly ash, air-entraining agent, or
gravel. An attempt was made to hold the slump constant at 75 to water reducer; and with normal-density and low-density (light-
100 mm (3 to 4 in.). Four 152 by 305-mm (6 by 12-in.) cylinders weight) aggregates. The accelerated strengths were compared
were molded from each batch. Two cylinders were cured ac- with the strengths of standard-cured 152 by 305-mm (6 by 12-
cording to ASTM C 684, Procedure A, to obtain one-day acceler- in.) cylinders. The accelerated strengths were between 22 and
ated strengths and two cylinders were moist-cured at 23°C (73°F) 90 % of the 28-day strengths.
in 100 % relative humidity to obtain standard 28-day strengths. A task force of the subcommittee compared the residual
After all cylinders were tested, the physical and chemical standard deviations of correlations obtained with the HTP pro-
properties of each cement along with the corresponding con- cedure with those obtained by others using the standard pro-
crete strengths were analyzed to search for correlations. A cedures. The results of the comparison are shown in Table 6.
strong correlation was found between the sodium alkali (Na2O) The relationships for the HTP method were expressed as
content and the one-day accelerated strengths (Fig. 5). power functions as follows:
Several multiple correlations were also examined. For two
independent variables, only combinations involving Na2O pro- Ŝ 28  B0SaB1 (4)
duced useful correlations with one-day accelerated strength;
moreover, only loss on ignition (LOI) coupled with the Na2O
produced a better multiple correlation than Na2O alone.
Therefore, the variation between the accelerated strengths
for the different Type I cements was attributable mainly to
variations in alkali content. Undoubtedly, this caused the
principal variation between laboratories in the cooperative test
program since each used a different Type I cement.

High Temperature and Pressure Method

In 1978, a different accelerated strength procedure was pro-
posed that could produce test results within 5 h [9]. Accelera-
tion of strength development in the new procedure is achieved
by a combination of elevated temperature and pressure. A

Fig. 5—Relationship between alkali content and accel- Fig. 6—Schematic of apparatus for high temperature and
erated strength using Procedure A. pressure accelerated curing method.
 CARINO ON PREDICTION OF POTENTIAL STRENGTH AT LATER AGES 147

TABLE 6—Comparison of Accelerated Strength Correlations

Regression Number of RSD,a
Reference Procedure Cement Type Equation Points MPab

Bickley [11] Autogenous curing I Linear 43 2.35

IV Linear 147 2.46
I and IV Linear 68 2.08
Lapinas [12] Modified boiling I Linear 312 1.82
Roadway and Lenz [13] Modified boiling I Linear 219 3.66
I Linear 36 2.34
I Linear 76 2.51
V Linear 61 1.58
Bisaillon [14] Autogenous curing I Linear not given 1.48
Bisaillon et al. [15] Autogenous curing I Linear 213 2.30
Malhotra [16] Modified boiling I Linear 336 2.40
I Linear 40 2.64
I Linear 265 3.54
Nasser and Beaton [10] High temperature I Power function 171 3.59
and pressure III Power function 99 3.36
IV Power function 65 4.03

a
RSD  residual standard deviation.
b
1 MPa  145 psi.

Based on the comparisons, the task force recommended residual standard deviation. To account for the uncertainty in
that the HTP method should be added to ASTM C 684, be- the regression line, the confidence interval for the line is es-
cause the 5-h accelerated strength correlated reasonably tablished [17,18]. Then, for a new accelerated strength, the con-
well with standard-cured strength. Subsequently, a member fidence interval for the estimated average later-age strength
of the subcommittee reported on a comparative study of the can be detemined.
HTP method and the modified-boiling method (Procedure To illustrate the procedure, consider the 12 pairs of accel-
B).2 In this study, mixtures were made with fly ash from dif- erated and 28-day strengths given in the first two columns of
ferent sources, with w/c values of 0.4, 0.5, and 0.7, and with Table 7. Each number is the average strength of two cylinders.
an air-entraining agent. The accelerated strengths were cor-
related with standard-cured, 28-day strengths (Fig. 7). Power
functions were fitted to the data. The residual standard de-
viations were 1.21 MPa (175 psi) and 1.77 MPa (257 psi) for
Procedure B and the HTP method, respectively. Surpris-
ingly, the best-fit equations were very similar. This study pro-
vided further evidence of the suitability of the HTP method,
and steps were taken to incorporate this procedure into
ASTM C 684.
During the balloting process, concern was expressed that
the hydration reactions due to the high temperature and pres-
sure would not be representative of those due to normal cur-
ing. Thus it was felt that the accelerated strength from the HTP
method might not be indicative of the potential strength under
standard curing. Proponents of the method noted that the data
obtained by using a variety of materials showed good correla-
tions. In 1989, the HTP method was adopted as alternative Pro-
cedure D in ASTM C 684.

Estimation of Later-Age Strength

To estimate the potential later-age strength from a measured
early-age accelerated strength, the laboratory must first con- Fig. 7—Comparison of relationships between accelerated
duct enough tests to establish the regression equation and its and standard 28-day strengths.

2
V. M. Malhotra, private communication, Canada Centre for Mineral and Energy Technology, Ottawa, Ont., Canada, 11 Jan. 1985.
148 TESTS AND PROPERTIES OF CONCRETE

TABLE 7—Values Used in Sample Problem to Illustrate Estimated

95 % Confidence Interval for 28-day Strength
Accelerated 28-day Estimated Lower Upper
Strength, Strength, Strength, Confidence Confidence
Sa, MPaa S28, MPa Ŝ28, MPa Wi, MPa Limit, MPa Limit, MPa

12.06 33.71 33.86 1.78 32.07 35.64

12.15 34.33 33.96 1.76 32.20 35.72
12.96 35.23 34.92 1.54 33.38 36.46
13.85 35.05 35.99 1.33 34.66 37.31
15.19 37.74 37.58 1.09 36.49 38.67
16.09 37.21 38.65 1.03 37.62 39.67
17.08 40.71 39.82 1.06 38.76 40.88
18.15 40.97 41.10 1.21 39.89 42.30
18.24 41.96 41.20 1.22 39.98 42.42
18.42 41.60 41.42 1.26 40.16 42.67
20.12 45.73 43.44 1.67 41.77 45.11
21.28 42.50 44.82 2.00 42.82 46.82

CONFIDENCE INTERVAL FOR ESTIMATED STRENGTH AT ACCELERATED STRENGTH OF 17.00 MPa

17.00 39.73 1.05 38.68 40.78

16.29 38.89 1.02 37.87
17.71 40.57 1.13 41.70

a
1 MPa  145 psi.

Using ordinary least squares regression analysis, the best-fit re- where
gression equation for the data (Fig. 8) is
Wi  half-width of confidence interval at Sai,
Ŝ 28  19.51  1.19 Sa (MPa) (5) se  residual standard deviation for best-fit line (Eq 2),
F  value from F-distribution for 2 and n degrees of freedom
The residual standard deviation of the line, se, is 1.24 MPa and significance level 0.05,
(180 psi). n  number of data points used to establish regression line,
The 95 % confidence interval for the line [17,19] is Sai  selected value of accelerated strength,
constructed by calculating Ŝ 28 for selected values of Sa and Sa  grand average value of accelerated strength for all data
plotting Ŝ 28
Wi , points used to establish the regression line,
Saa  ∑(Sa)2(∑ Sa)2/n, and
Wi  se 2

F


1 (Sai  Sa)2
  
n Saa
(6)
Sa  values of average accelerated strength for each point
used to establish the regression line.

The third column in Table 7 lists the estimated average 28-

day strengths for the accelerated strengths in Column 1. The
value of Wi at each value Sa is listed in the fourth Column of
Table 7. Finally, columns five and six list the values of the lower
and upper 95 % confidence limits that are shown in Fig. 8. Note
that the width of the confidence limits is narrowest when Sai
equals Sa , because the second term under the square root sign
in Eq 6 equals zero.
Suppose that the average accelerated strength of two
cylinders made from similar concrete is 17.0 MPa (2460 psi).
From the regression equation, the estimated average 28-day
strength is 39.7 MPa (5760 psi). If the accelerated strength
were known without error, the 95 % confidence interval for
the average 28-day strength would be 38.7 to 40.8 MPa (5610
to 5920 psi) (see the bottom of Table 7). However, the accel-
erated strength has an uncertainty that is described by the
within-batch standard deviation. Assume that the compressive
Fig. 8—Confidence limits for the estimated 28-day strength strengths measured by the specific accelerated test method
based on a measured accelerated strength and the previously have a single-laboratory, within-batch coefficient of variation
established strength relationship. of 3.0 %. Therefore, the standard deviation, s, at an average
 CARINO ON PREDICTION OF POTENTIAL STRENGTH AT LATER AGES 149

strength of 17.0 MPa (2460 psi) is 0.51 MPa (74 psi). The 95 % an outgrowth of research performed in the late 1960s and
confidence interval for the average accelerated strength of early 1970s by Hudson and Steele [22,23] at the West Virginia
the two cylinders is Department of Highways, and it was adopted as ASTM C 918 in
1980. The basic principles of the maturity method are dis-
S
17.0
z0.025   17.0
1.96  0.51  0.707  17.0
0.7 MPa cussed by Malhotra in another chapter of this publication and
2 a more comprehensive review is also available [1].
where z0.025 is the value from the standard normal distribu- The motivation for the development of ASTM C 918 is
tion corresponding to 2.5 % of the area under the curve. Thus discussed first. Concrete mixtures cured at standard tempera-
the 95 % confidence interval3 for the average accelerated ture gain strength predictably, that is, there are fixed rela-
strength is 16.3 to 17.7 MPa (2360 to 2570 psi). Projecting the tionships between the strengths at early ages and at a later
limits of this interval to the lower and upper confidence lim- age. Attempts have been made, therefore, to estimate the
its of the regression line results in 37.9 to 41.7 MPa (5500 to later-age strength, such as at 28 days, by multiplying the early-
6050 psi) for the approximate 95 % confidence interval for age strength by an empirical factor. However, such estimates
the average 28-day strength. Each different measurement of have not been found to be reliable. As has been explained
accelerated strength produces a new confidence interval for [22,23], there are two pitfalls to practical application of this
the average 28-day strength. A personal computer is recom- simple approach: (1) on a construction project, it is not pos-
mended for implementing the preceding calculations for rou- sible to perform the early-age tests at precisely the specified
tine use. age, and (2) the early-age temperatures cannot be controlled
In 1999, ASTM C 684 was revised to provide explicit guid- as accurately as needed. The procedure in ASTM C 918 over-
ance on using statistical methods to estimate later-age strength. comes these limitations by requiring measurement of actual
A new paragraph was added under “Interpretation of Results,” temperature histories of the early-age specimens and by using
which stated that statistical methods were to be used to ac- a “maturity index,” rather than age, to relate to the level of
count for the various sources of uncertainty. It was also stated strength development. The meaning of “maturity index” is dis-
that all interested parties were to agree on the statistical pro- cussed in the next section.
cedures to be used and the criterion for acceptance testing. It To use ASTM C 918, the testing laboratory first establishes
was recommended that estimated strength corresponding to a relationship between the strength of concrete cylinders and
the lower 90 % limit could be used to compare with the speci- the maturity index. Subsequently, cylinders made from similar
fied standard-cured strength. A nonmandatory Appendix X2 concrete are tested at early ages. The maturity index from the
was added to provide a recommended statistical methodology time of molding each set of cylinders until the time of testing
based on the above statistical procedure, except that the 90 % is recorded. The early-age strengths and the corresponding
confidence interval (z0.05  1.645) was used in the illustrative values of maturity index are then used to estimate the later-age
example instead of the 95 % interval as presented above. strength based upon the previously established strength-
In summary, to estimate the average 28-day strength based maturity index relationship.
upon accelerated strength test results, a procedure is used that The following provides a brief review of the terminology
accounts for the uncertainty in the regression line and in the associated with the maturity method.
measured accelerated strength. It is insufficient to simply use
the regression equation to convert the accelerated strength to Concrete Maturity
an equivalent 28-day strength. Additional information on the In concrete technology, the term “maturity” refers to the extent
procedure presented in the example may be found in the ref- of the development of those properties that depend on cement
erences by Moore and Taylor [21] and in Miller [19]. Finally, it hydration and pozzolanic reactions. At any age, maturity de-
is emphasized that a particular regression equation is valid pends on the previous curing history, that is, the temperature
only for a specific test procedure and combination of materi- history and the availability of water to sustain the chemical
als. Each laboratory must conduct enough tests with a given set reactions. In the early 1950s, the idea was developed that the
of materials and a certain procedure to establish the regression combined effects of time and temperature could be accounted
line and its confidence limits before estimates of later-age for by using a “maturity function” to convert the temperature
strengths are possible. history to a “maturity index” that would be indicative
of strength development. In practice, the maturity index is ob-
Later-Age Strength Estimates Using The tained by using electronic instruments that monitor the con-
Maturity Method crete temperature and automatically compute the maturity in-
dex as a function of age. The maturity index can be displayed
Overview on a self-contained unit or downloaded to a computer. The
The other method developed by the original Subcommittee on reader is urged to read the appropriate section of Malhotra’s
Accelerated Strength Testing to estimate the later-age strength chapter in this publication and ASTM Practice for Estimating
from early-age tests is based upon the maturity method. Unlike Concrete Strength by the Maturity Method (C 1074) for back-
the accelerated strength methods, there are no special curing ground information on the maturity functions used to com-
requirements for this procedure, but it does require measuring pute the maturity index. In this chapter, the “equivalent age at
the temperature history of the test specimens. The method is 23°C” is used as the maturity index. While this index is not as

3
The 95 % confidence interval is often interpreted to mean that there is a 95 % probability that the true mean falls within the interval. However, the correct inter-
pretation is as follows: If 100 repeated samples are taken from the same population and the 95 % confidence intervals for the mean are computed in each case, 95
of the intervals would include the true mean. The 95 % confidence interval for the regression line has a similar interpretation: If 100 groups of data are taken from
the same population and the 95 % confidence intervals are computed for the regression equations, 95 of those intervals would include the true regression line for
the population. See Mendenhall and Sincich [20] for further explanations on the proper interpretation of confidence intervals.
150 TESTS AND PROPERTIES OF CONCRETE

well known in the United States as the “temperature-time fac- the average strengths. The values of a and b are obtained by
tor” (expressed in degree-hours or degree-days), equivalent age least squares fitting of Eq 7 to the data.
is the more meaningful quantity and is expected to be used To estimate the later-age strength of a similar concrete
more widely in the future. The equivalent age is the curing age mixture based on the results of subsequent early-age tests, the
at a specified temperature that would result in the same matu- maturity index of the early-age test specimens is monitored
rity as has occurred at the actual age under the actual temper- from the time of molding until the time of testing. This is con-
ature history. veniently done by casting a “dummy” cylinder into which the
The “strength-maturity relationship” for a specific con- sensor of the maturity meter is embedded. The instrumented
crete mixture is used to relate the strength development to the cylinder is exposed to the same environment as the cylinders
maturity index. The relationship can be used to estimate that will be tested for strength. The values of the early-age
strength development under different temperature histories, strength, Sm, and maturity index, m, are used along with the
as is covered by ASTM C 1074, or it can be used to estimate the previously established strength-maturity relationship to esti-
later-age strength under standard curing based upon a meas- mate the potential later-age strength.
ured early-age strength, as covered by ASTM C 918. During a project, the concrete-making materials will usu-
ally be the same from batch to batch, and the main purpose of
Strength-Maturity Relationship cylinder strength tests is to assure that the specified strength is
As mentioned earlier, knowledge of the relationship between achieved. The value of a in Eq 7 depends, however, on the
strength and the maturity index is a fundamental requirement water-cement ratio since it represents the strength at a partic-
to apply the maturity method. Over the years, a variety of em- ular maturity index. A change in the value of a shifts the
pirical equations have been proposed to represent such rela- straight line along the strength axis. It is not appropriate to
tionships. These have been reviewed elsewhere [1] and only the assume that the value of a obtained in the laboratory testing
function adopted in ASTM C 918 is discussed here. program is applicable to the cylinders being evaluated by early-
In 1956, Plowman [24] proposed that the strength of con- age testing. It is reasonable, however, to assume that the value
crete could be related to the maturity index by the following of b is applicable, because large changes in w/c are needed to
semi-logarithmic function change the value of b [23]. The purpose of the early-age results
is to establish the appropriate value of a as follows:
SM  a  b log M (7)
a  Sm  b log m (8)
where
where
SM  compressive strength at M,
M  maturity index, and m  maturity index at time of early-age test, and
a, b  regression constants. Sm  average early-age strength measured at maturity
index, m.
According to Eq 7, strength is a straight line function of the
logarithm of the maturity index. This is generally a reasonable By substituting Eq 8 into Eq 7, one obtains the following
approximation for strength development between 1 and 28 days “prediction equation” for estimating the strength at a later age:
under standard room temperature curing. The parameter a rep-
resents the strength at a maturity index of 1 (the logarithm of 1 SM  Sm  b (log M  log m) (9)
equals zero).4 The parameter b, which is the slope of the line,
represents the increase in strength for a tenfold increase in the where
maturity index. For example, if the equivalent age at 23°C in-
creases from one day to ten days, the strength increase equals M  maturity index at the later age when strength is to be es-
b. The values of a and b depend on the materials and mixture timated, and
proportions of the specific concrete [22–24]. SM  estimated strength at maturity index M.

Application Again, this derivation assumes that the value of b is not affected
To use ASTM C 918 to estimate the potential later-age strength by small variations in w/c.
at 28 days, for example, it is first necessary to establish the
strength-maturity relationship, that is, the “prediction equa- Example
tion,” for the concrete mixture. This is done by preparing stan- The following example illustrates the application of ASTM C
dard cylindrical specimens and subjecting them to standard 918. Assume that a set of cylinders are molded in the labora-
curing procedures. A maturity meter is used to monitor the ma- tory and that one of them is instrumented with a maturity me-
turity index of the specimens.5 Pairs of cylinders are tested at ter that computes the equivalent age at 23°C. Table 8 gives the
regular intervals (for example, at 1, 3, 7, 14, and 28 days),6 and average strengths (two cylinders) that are obtained at different
the corresponding maturity indexes are recorded along with values of the maturity index.

4
Note that the value of a depends on the units used for the maturity index. If the maturity index is expressed as the temperature-factor (degree-hours or degree-days),
the value of a will be a negative number that has no physical significance. If the maturity index is expressed in terms of equivalent age at 23°C in days, the value of
a is the strength at one day of curing at 23°C.
5
In the original version of ASTM C 918, the measurement of the ambient temperature history was considered adequate. However, revisions in 1993 called for moni-
toring the actual concrete temperature. The availability of relatively inexpensive maturity meters justifies this more precise approach.
6
Testing at ages that are related by a factor of two will result in approximately equal strength increments.
 CARINO ON PREDICTION OF POTENTIAL STRENGTH AT LATER AGES 151

such as at 28 days, will be eliminated in the near future. Thus

TABLE 8—Equivalent Age and Compressive the 28-day strengths of standard-cured specimens will proba-
Strength Values Used in Illustrative Example bly continue to be measured. By keeping records of estimated
and measured 28-day strengths of companion specimens
Equivalent Age Average Strength,
Age, days at 23°C, days MPa (psi)
from the same batch, one can continually correct and im-
prove the slope, b, of the prediction equation by using the fol-
1.0 0.91 9.44 (1370) lowing relationship [22]:

∑ (S28  Sm)
3.0 3.10 17.09 (2480)
7.0 7.25 21.77 (3160)
b   (11)
14.0
28.0
14.5
27.3
25.56 (3710)
29.28 (4250)
∑ (log M28  log m)
where

S28  measured standard-cured 28-day strength,

M28  maturity index corresponding to standard curing for 28-
These strengths have been plotted as a function of the log- days, and
arithm of the equivalent age (Fig. 9). The best-fit equation is Sm  measured early-age strength at maturity index m.

SM  10.25  13.29 log M, MPa (10) In addition, the differences between the estimated and
measured 28-day strengths can be used to calculate a confi-
Further assume that cylinders fabricated in the field were dence limit for future estimates, and thereby establish an ac-
subjected to “standard curing” as specified in ASTM Practice for ceptance criterion for future early-age results. First, the aver-
Making and Curing Concrete Test Specimens in the Field (C age difference, d between the measured and estimated 28-day
31/C 31M). After 24 h, two cylinders were removed from their strengths is computed
molds, capped, and tested for compressive strength. The aver- n
age strength of two cylinders was 10.0 MPa (1450 psi), and the ∑ (S28  SM )
i 1
corresponding equivalent age was 1.13 days based on a matu- d  
rity meter in the dummy cylinder. According to Eq 9, the esti- n
mated 28-day, standard-cured strength would be n
∑ di (12)
S28  10.0  13.29 (log 28  log 1.13) d  
i =1
n
 10.0  13.29 (1.4472  0.0531)
 10.0  18.5 where
 28.5 MPa (4130 psi)
SM  the estimated 28-day strength,
S28  measured standard-cured 28-day strength,
Interpretation of Results
di  the difference between the ith pair of strength values, and
The intent of early-age tests is to provide an early indication
n  number of pairs of strength values.
of the potential strength of the concrete sample. It is unreal-
istic to expect that the traditional standard tests at later ages,
The value of d is the “bias” of the prediction equation, and
it should be close to zero if the value of b is updated as new
data are accumulated. The standard deviation for the differ-
ence between the measured and estimated strengths is calcu-
lated as follows


n

sd  ∑
i=1
(di  d )2 (13)

(n  1)

The upper 95 % confidence limit for the average difference

between the measured and estimated 28-day strength is
sd
K  d  t0.95,n1 (14)
 n
where

t0.95,n1  value from the t-distribution at the 95 % point for

n  1 degrees of freedom

Fig. 9—Example of a strength versus maturity index From such past data, it is possible to establish an acceptance
relationship; the slope is used in the prediction equation to criterion for the concrete based on the estimated strength of a
estimate later-age strength based upon early-age strength. future early-age test. If the design strength is f28, the concrete
152 TESTS AND PROPERTIES OF CONCRETE

represented by the early-age results would be considered 6. For each cement type, accelerated strength correlated
acceptable if the following condition were satisfied: with 364-day strength as well as with 28-day strength.
Subsequent tests showed that significantly different
Sm  ƒ28  K (15) strength relationships were obtained by the laboratories be-
cause of variations in alkali content of the cements. The coop-
Before the 1993 version of ASTM C 918, information on in- erative study showed that accelerated strength testing by a sin-
terpretation of results was included in the non-mandatory ap- gle laboratory, using the same procedure and materials, results
pendix. In the 1993 revision, this information was moved into in an assessment of concrete quality at ages less than two days
the body of the standard. The definition of the factor K in the that is as reliable as that obtained after 28 days of standard
1993 version differed from Eq 14. In the 1997 revision of moist curing [8]. Appropriate statistical techniques, as dis-
ASTM C 918, Eq 14 was introduced, and ASTM C 1074 was ref- cussed in this chapter, should be used to establish a confidence
erenced for terminology related to the maturity method. Minor interval of the estimated potential, later-age strength.
changes were made in the 2002 revision of ASTM C 918. The other method (ASTM C 918) is based on the principle
of the maturity method, in which the early-age curing history
Precautions is converted to a maturity index indicative of the level of
The prediction equation given by Eq 9 assumes that concrete strength development. A previously established relationship is
strength is a linear function of the logarithm of the maturity in- also required to estimate the potential, later-age strength based
dex. Before implementing ASTM C 918, the user should verify on the measured early-age strength. In this case, the relation-
that this assumption is valid. As was stated, past experience ship is between strength and the maturity index. Unlike the ac-
shows that the linear equation is usually adequate for strengths celerated strength testing methods, ASTM C 918 does not re-
up to 28 days. There is no reason, however, why the procedure quire a precisely controlled curing procedure, provided
cannot be applied to estimate strength at ages later than 28 sufficient moisture is maintained during early ages to sustain
days. If the laboratory results reveal that the linear approxi- hydration. The specimens are cured according to the standard-
mation is not applicable, the basic principle of ASTM C 918 can curing procedures in ASTM C 31/C 31M. The only additional
still be applied provided an appropriate equation is used to rep- requirement is that the temperature history of the early-age
resent the actual strength-maturity relationship. Examples of specimens must be monitored to determine the maturity index
equations that may be applicable are given in Carino [1]. (refer to ASTM C 1074) at the time of testing. Early-age
strengths can be measured at any age 24 h after molding the
specimens.
Summary
The basis of ASTM C 918 is research performed at the
West Virginia Department of Highways [22,23]. In a statisti-
This chapter has discussed the basis of two test methods to es-
cally designed testing program, it was shown that the potential,
timate the potential later-age strength of concrete specimens
later-age strength could be estimated with sufficient precision
tested at early ages. The results of these early-age tests can pro-
and accuracy for quality control purposes. This method also re-
vide timely information on the concrete production process
quires use of statistical principles for a reliable lower-bound es-
for quality control. These procedures attempt to overcome the
timate of the potential strength.
inherent deficiencies of the traditional practice, which relies
In summary, test methods are available to overcome the
on strength tests at ages of 28 days or later, to judge the ade-
deficiencies of quality control programs based on the tradi-
quacy of concrete batches.
tional 28-day strengths. The potential benefits arising from
One of the test methods (ASTM C 684) involves subjecting
early-age testing should be considered carefully prior to setting
the early-age test specimens to specific curing conditions that ac-
up the quality control program to monitor the concrete pro-
celerate strength development. Four alternative accelerated cur-
duction process on major construction projects.
ing methods are specified that permit strength determinations
at ages ranging from about 5 h to 48 h after preparing the spec-
imens. Prior to using these measured accelerated strengths to es-
timate the later-age strength, laboratory testing is required to es- References
tablish the relationship between the accelerated strength and the [1] Carino, N. J., “The Maturity Method,” Chapter 5 in Handbook
standard-cured strength. Typically, the standard-cured strength on Nondestructive Testing of Concrete 2nd Edition, V. M.
is measured at 28 days, but other ages are possible, such as 56 Malhotra and N. J. Carino, Eds., CRC Press, Boca Raton, FL and
or 91 days, depending on the job specifications. ASTM International, West Coshohocken, PA., 2004.
The cooperative testing program leading to the initial de- [2] King, J. W. H., “Accelerated Testing of Concrete,” Proceedings,
velopment of ASTM C 684 was reviewed. Analysis of the data 50th Anniversary Conference of the Institution of Structural
from that program resulted in the following conclusions [8]: Engineers, London, Oct. 1958, pp. 376–381, 386–387.
1. Significantly different relationships were obtained by [3] King, J. W. H., “Accelerated Test for Strength of Concrete,”
Journal of Applied Chemistry (London), Vol. 10, June 1960,
the nine laboratories due to the use of different local
pp. 256–262.
materials.
[4] Akroyd, T. N. W., “The Accelerated Curing of Concrete Test
2. Different types of cements resulted in different relation-
Cubes,” Journal Institution of Civil Engineers (London), Vol. 19,
ships within the same laboratory. Paper No. 6441, May 1961, pp. 1–22.
3. The omission or addition of a normal amount of chemical [5] Smith, P. and Chojnacki, B., “Accelerated Strength Testing
retarder in the concrete did not affect the relationships. of Concrete Cylinders,” Proceedings ASTM, Vol. 63, 1963,
4. Batch-to-batch variations due to mixing on different days pp. 1079–1101.
did not have significant effects on the relationships. [6] Malhotra, V. M. and Zoldners, N. G., “Some Field Experience in
5. Correlation of accelerated strength with 28-day and 364- the Use of an Accelerated Method of Estimating 28-day
day standard strength ranged from good to excellent. Strength of Concrete,” Internal Report MP1 68-42, Minerals
 CARINO ON PREDICTION OF POTENTIAL STRENGTH AT LATER AGES 153

Processing Division Department of Energy, Mines and [22] Hudson, S. B. and Steele, G. W., “Prediction of Potential
Resources, Ottawa, Ont., Aug. 1968. Strength of Concrete from the Results of Early Tests,” Highway
[7] Smith, P. and Tiede, H., “Earlier Determination of Concrete Research Record, No. 370, Highway Research Board, 1971,
Strength Potential,” Highway Research Record, No. 210, 1967, pp. 25–36.
pp. 29–61. [23] Hudson, S. B. and Steele, G. W., “Developments in the Predic-
[8] Wills, M. H., Jr., “Early Assessment of Concrete Quality by tion of Potential Strength of Concrete from Results of Early
Accelerating Compressive Strength Development with Heat Tests,” Transportation Research Record, No. 558, 1975, pp.1–12.
(Results of ASTM’s Cooperative Test Program),” Journal of [24] Plowman, J. M., “Maturity and Strength of Concrete,”
Testing and Evaluation, Vol. 3, No. 4, July 1975, pp. 251–262. Magazine of Concrete Research, Vol. 8, No. 22, March 1956,
[9] Nasser, K. W., “A New Method and Apparatus for Accelerated pp. 13–22.
Strength Testing of Concrete,” in ACI SP-56, Accelerated
Strength Testing, V. M. Malhotra, Ed., American Concrete Bibliography
Institute, 1978, pp. 249–258.
American Concrete Institute, Farmington Hills, MI, ACI Committee
[10] Nasser, K. W. and Beaton, R. J., “The K-5 Accelerated Strength
214, “Use of Accelerated Strength Testing,” ACI Manual of
Tester,” Journal of the American Concrete Institute, Vol. 77, No.
Concrete Practice, ACI 214.1R-81(1986), 1986.
3, May-June 1980, pp. 179–188.
Brockenbrough, T. W. and Larason, R. R., “Early Strength Test for
[11] Bickley, J. A., “Accelerated Concrete Strength Testing at
Quality Control of Concrete,” Transportation Research Record,
the CN Tower,” in ACI SP-56, Accelerated Strength Testing,
No. 558, 1975, pp. 61–68.
V. M. Malhotra, Ed., American Concrete Institute, 1978,
pp. 29–38. Chin, F. K., “Strength Tests at Early Ages and at High Setting
Temperatures,” Transportation Research Record, No. 558, 1975,
[12] Lapinas, R. A., “Accelerated Concrete Strength Testing by
pp. 69–76.
Modified Boiling Method: Concrete Producer’s View,” in
ACI SP-56, Accelerated Strength Testing, V. M. Malhotra, Ed., Fãlcao Bauer, L. A. and Olivan, L. I., “Use of Accelerated Tests for
American Concrete Institute, 1978, pp. 75–93. Concrete Made with Slag Cement,” Accelerated Strength
Testing, ACI SP-56, V. M. Malhotra, Ed., American Concrete Insti-
[13] Roadway, L. E. and Lenz, K. A., “Use of Modified Boiling
tute, Farmington Hills, MI, 1978, pp. 117–128.
Method in Manitoba and Alberta, Canada,” in ACI SP-56,
Accelerated Strength Testing, V. M. Malhotra, Ed., American Ferrer, M. M., “Quality Control of Concrete by Means of
Concrete Institute, 1978, pp. 129–146. Short-Termed Tests at La Angostina Hydroelectric Project, State
of Chiapas, Mexico,” Accelerated Strength Testing, ACI SP-56, V.
[14] Bisaillon, A., “Accelerated Strength Test Results from
M. Malhotra, Ed., American Concrete Institute, Farmington Hills,
Expanded Polystyrene Molds with Emphasis on Initial Concrete
MI, 1978, pp. 51–73.
Temperature,” in ACI SP-56, Accelerated Strength Testing,
V. M. Malhotra, Ed., American Concrete Institute, 1978, Kalyanasundaram, P. and Kurien, V. J., “Accelerated Testing for
pp. 201–228. Prediction of 28-day Strength of Concrete,” Transportation
Research Record, No. 558, 1975, pp. 77–86.
[15] Bisaillon, A., Fréchette, G., and Keyser, J. H., “Field Evaluation
Malhotra, V. M., Zoldners, N. G., and Lapinas, R., “Accelerated 28-
of Expanded Polystyrene Molds for Self-Cured, Accelerated
Day Test for Concrete,” Canadian Pit and Quarry, March 1966,
Strength Testing of Concrete,” Transportation Research
pp. 51–54.
Record, No. 558, 1975, pp. 50–60.
Orchard, D. F., Jones, R., and Al-Rawi, R. K., “The Effect of Cement
[16] Malhotra, V. M., “Canadian Experience in the Use of the
Properties and the Thermal Compatibility of Aggregates on the
Modified Boiling Method,” Transportation Research Record, Strength of Accelerated Cured Concrete,” Journal of Testing
No. 558, 1975, pp. 13–18. and Evaluation, Vol. 2, No. 2, March 1974, pp. 95–101.
[17] Natrella, M. G., Experimental Statistics, NBS Handbook 91, Ramakrishnan, V. and Dietz, J., “Accelerated Methods of Estimating
National Bureau of Standards, Washington, D.C., Aug. 1963. the Strength of Concrete,” Transportation Research Record, No.
[18] Draper, N. R. and Smith, H., Applied Regression Analysis, 2nd 558, 1975, pp. 29–44.
ed., John Wiley & Sons, New York, 1981. Ramakrishnan, V., “Accelerated Strength Testing—Annotated
[19] Miller, R. G., Simultaneous Statistical Inference, 2nd ed., Bibliography,” Accelerated Strength Testing, ACI SP-56, V. M.
Springer-Verlag, New York, 1981. Malhotra, Ed., American Concrete Institute, Farmington Hills,
[20] Mendenhall, W. and Sincich, T., Statistics for Engineering and the MI, 1978, pp. 285–312.
Sciences, 3rd ed., Dellen Publishing Co., San Francisco, CA, 1992. Sanchez, R. and Flores-Castro, L., “Experience in the Use of the
[21] Moore, J. K. and Taylor, M. A., “Statistical Properties of Accelerated Testing Procedure for the Control of Concrete During
Techniques for Predicting Concrete Strength and Examples of Construction of the Tunnel ‘Emissor Central’ in Mexico City,” Ac-
Their Use,” in ACI SP-56, Accelerated Strength Testing, V. M. celerated Strength Testing, ACI SP-56, V. M. Malhotra, Ed., Amer-
Malhotra, Ed., American Concrete Institute, 1978, pp. 259–283. ican Concrete Institute, Farmington Hills, MI, 1978, pp. 15–28.
15
Freezing and Thawing

Charles K. Nmai 1

Preface weathering is determined by its ability to withstand the effects

of freezing and thawing in the presence of moisture.2
THE SUBJECT OF FREEZE-THAW DURABILITY The testing of concrete for resistance to freezing and thaw-
of concrete was covered in all four previous editions of the ing had its genesis in similar tests on building stones reported
ASTM Special Technical Publication on significance of tests as early as 1837 by Vicat [18]. It was only natural that the advent
and properties; the chapters on this subject in ASTM STP 169, of “artificial stone” in the form of portland cement concrete
ASTM STP 169A, ASTM STP 169B, and ASTM STP 169C were would result in corresponding evaluations. Few published tests
authored by C. H. Scholer and T. C. Powers, H. T. Arni, H. from the 19th century survive, but when freezing equipment be-
Newlon, Jr., and H. Newlon, Jr. and T. Mitchell, respectively. came commercially available for food processing in the 1930s,
Because little has changed in this subject area since the pub- the reporting of such testing increased. With the recognition
lication of ASTM STP 169C, this chapter includes only minor that tests conducted in a variety of freezers designed for other
changes, that is, an update of the reference list and a few re- purposes naturally gave variable results, several agencies such
visions in the text, to the version that appeared there. as the U.S. Bureau of Reclamation, the Corps of Engineers, the
Bureau of Public Roads (now the Federal Highway Administra-
Introduction tion), the Portland Cement Association, and the National Sand
and Gravel Association (now the National Aggregates Associa-
The purpose of this paper is to discuss the significance of cur- tion) constructed specialized equipment for testing concrete by
rently standardized testing procedures for evaluating the re- freezing and thawing. Descriptions of some of this equipment
sistance of concrete to weathering under service conditions. will be found in reports by the agencies [19–24]. These efforts
The published literature on the subject is extensive and only a marked the beginning of a systematic quest for standardization
somewhat superficial treatment can be given in a brief paper. and an understanding of factors influencing the resistance of
The reader should consult the similar papers in earlier editions concrete to freezing and thawing.
of this special technical publication by Scholer [1,2], Powers [3], There are currently two standard test methods under the ju-
Arni [4], and Newlon [5]. A monograph by Woods [6] presents risdiction of ASTM Committee C9 on Concrete and Concrete Ag-
a comprehensive picture of durability, and one by Cordon [7] gregates intended to aid in evaluating the resistance of concrete
discusses freezing and thawing in detail. Perhaps the most com- to freezing and thawing. In addition, several petrographic pro-
prehensive treatment of the influence of aggregates on resist- cedures are standardized that provide invaluable information
ance to freezing and thawing is that of Larson et al. [8]. An for predicting the resistance of concrete to freezing and thawing
extensive annotated bibliography on concrete durability pub- and for interpreting the results of exposure in either the labora-
lished by the Highway Research Board (HRB) in 1957 [9] with tory or under field conditions. Two older freezing and thawing
a supplement in 1966 [10] contains 534 citations. A more cur- methods have been discontinued. The presently standardized
rent annotated bibliography, with some 600 citations, was pub- methods as well as those discontinued have evolved over more
lished by the Strategic Highway Research Program in late 1992 than 60 years and reflect the inevitable compromise between the
[11]. The report of American Concrete Institute (ACI) Commit- need for rapid assessment of resistance to weathering and the
tee 201 on Durability of Concrete [12] contains valuable rec- difficulty of translating the results from accelerated laboratory
ommendations for the production of concrete that provides testing to the varied conditions encountered in field exposures.
resistance to the various destructive processes encountered in Of the currently standardized methods, ASTM Test
field exposures. ACI symposia on durability of concrete, in 1975, Method for Resistance of Concrete to Rapid Freezing and
1987, 1991, 1994, and 1997 resulted in 17, 111, 70, 66, and 81 Thawing (C 666) covers the exposure of specimens to cyclic
papers, respectively, on various theoretical and operational as- freezing and thawing. This test, the most widely used of those
pects of producing durable concrete [13–17]. available, is a consolidation of two earlier methods (ASTM C
In the absence of contact with aggressive fluids or incor- 290 and C 291) and provides for two procedures: Procedure A
poration of aggregates susceptible to detrimental expansion by in which both freezing and thawing occurs with the specimens
reaction with alkalies in cement, the resistance of concrete to surrounded by water, and Procedure B in which the specimens

1
Chief Engineer, Degussa Admixtures, Inc., Beachwood, OH 44122.
2
Attack by aggressive fluids and cement-aggregate reactions are treated elsewhere in this publication.

154
 NMAI ON FREEZING AND THAWING 155

freeze in air and thaw in water. ASTM Test Method for Scaling rently C09.67) with the major responsibility for proposing
Resistance of Concrete Surfaces Exposed to Deicing Chemicals methods for evaluating the resistance to freezing and thawing
(C 672) provides a procedure for evaluating the effect of deic- of aggregates in concrete and standardizing the methods of
ing chemicals on concrete and the effectiveness of modifica- tests. Drawing heavily on the results of the HRB cooperative
tions of the concrete or of surface coatings in mitigating the test series and the experience of several laboratories that had
detrimental influence of such chemicals. Two of the initially developed specialized equipment for conducting freezing and
approved standards were discontinued in 1971. These meth- thawing tests, ASTM approved four tentative test methods in
ods, ASTM Test for Resistance of Concrete Specimens to Slow 1952 and 1953 that later became standards. These methods
Freezing and Thawing in Water or Brine (C 292) and ASTM were largely representative of the methods and procedures
Test for Resistance of Concrete Specimens to Slow Freezing in then in use by the membership of Subcommittee III-0. The
Air and Thawing in Water (C 310), both of which provided for general characteristics of the methods are reflected in their ti-
a single cycle every 24 h, were dropped because of lack of use. tles: ASTM Test for Resistance of Concrete Specimens to Rapid
Another standard and a recommended practice were discon- Freezing and Thawing in Water (C 290), ASTM Test for Resis-
tinued in 2002, also because of lack of use. These are ASTM tance of Concrete Specimens to Rapid Freezing in Air and
Test for Critical Dilation of Concrete Specimens Subjected to Thawing in Water (C 291), ASTM Test for Resistance of Con-
Freezing (C 671), which employed a single cycle of cooling crete to Slow Freezing and Thawing in Water or Brine (C 292),
through the freezing point with specimens that are continu- and ASTM Test for Resistance of Concrete Specimens to Slow
ously wet, and the ASTM Practice for Evaluation of Frost Re- Freezing in Air and Thawing in Water (C 310). The two slow
sistance of Coarse Aggregates in Air-Entrained Concrete by tests were adopted to cover tests usually conducted in conven-
Critical Dilation Procedures (C 682). tional freezers with manual transfer of specimens between the
While much is still to be learned, an extensive body of freezing chamber and thawing tank. These methods were
knowledge has been developed that permits evaluations of dropped in 1971 because neither was in general use nor re-
concrete to aid in minimizing premature deterioration from quired by any other ASTM specification. In 1971, the two rapid
environmental factors. tests (ASTM C 290 and C 291) were combined as two proce-
While an extensive discussion of the theory and historical dures (A and B) in a single test (ASTM C 666).
development of freezing and thawing tests is beyond the scope When ASTM C 666 was originally adopted, deterioration
of this summary, a brief treatment is necessary. The reader of specimens was evaluated only by the resonant frequency
should consult the cited references for additional information method, that is, ASTM Test Method for Fundamental Trans-
on these subjects. verse, Longitudinal, and Torsional Frequencies of Concrete
Specimens (C 215). In 1984, length change was incorporated as
Historical Evolution an optional, additional means for that evaluation (length
As early as 1837, Vicat, in his famous Treatise on Calcareous change measurements are made with the apparatus described
Mortars and Cements, reported the results from experiments in ASTM Specification for Apparatus for Use in Measurement
by Brard [25] to distinguish the building stones that were in- of Length Change of Hardened Cement Paste (C 490).
jured by frost from those that were not [18]. These tests were In the third cooperative test series, initiated in 1954 and re-
conducted “by substituting for the expansive force of the con- ported in 1959 [31], soon after standardization of the four
gealing water, that of an easily crystallizable salt, the sulfate of methods, 13 laboratories participated using three concrete for-
soda.” In 1856, Joseph Henry reported testing, by 50 cycles of mulations: a concrete with good-quality coarse aggregate and
freezing and thawing, samples of marble used in the extension adequate air entrainment, one with good-quality coarse aggre-
of the U.S. Capitol [26]. gate and deficient air entrainment, and one with poor-quality
During the two decades before and those after the turn of coarse aggregate and adequate air entrainment.
the 20th century, various tests of stone, concrete, brick, and Arni [4] summarized the general conclusions from the
other porous materials were reported. In 1928, Grun [27] in 1944 and 1954 test series as follows:
Europe and Scholer [28] in the United States reported results 1. Methods involving freezing in water were more severe,
from accelerated freezing and thawing of concrete in the labo- that is, produced failure in fewer cycles, than were those
ratory. Following Scholer’s initial paper, accelerated labora- involving freezing in air.
tory testing greatly increased. In 1936 [29], 1944 [30], and 2. Rapid freezing was more severe than slow freezing when
again in 1959 [31], the Highway Research Board (HRB) Com- done in air but not when done in water.
mittee on Durability of Concrete reported results of coopera- 3. Rapid freezing and thawing in water (ASTM Test C 290)
tive freezing and thawing tests designed to identify factors that appeared to do the best job of detecting a difference be-
influence resistance of concrete to freezing and thawing. The tween concretes both of which had high durability.
first two cooperative test series were conducted before any 4. Only the slow freezing in air and thawing in water method
methods were standardized. The 1936 series concentrated on (ASTM Test C 310) was able to discriminate adequately be-
the influence of cement using mortar prisms incorporating ten tween concretes of low durability.
commercial cements. The 1944 series of tests used concrete 5. In general, the four methods tended to rate different con-
specimens. Both of these series emphasized the necessity for cretes in the same order of durability when there was a sig-
carefully regulating the methods of making and curing the nificant difference.”
specimens, the air content of the specimens, the degree of sat- One of the difficulties with the four methods that was em-
uration of the aggregates at the time of mixing the concrete, phasized especially by the HRB cooperative programs was poor
and the degree of saturation of the concrete at the time of repeatability and reproducibility of results within and between
freezing. laboratories. Good reproducibility generally was obtained only
In 1951, ASTM Committee C9 formed Subcommittee III-0 for concretes that were very high or very low in durability. For
on Resistance to Weathering (subsequently C09.03.15 and cur- concretes in the middle range, a wide spread in results was ob-
156 TESTS AND PROPERTIES OF CONCRETE

tained. While the variabilities were large in the middle range, was reproduced artificially in the laboratory environment.
they were amenable to establishing a precision statement that Powers and his co-workers, from their comprehensive study of
has been incorporated in ASTM C 666, as discussed later. the structure of cement paste in 1945 [38], advanced the hy-
As noted earlier, Subcommittee III-0 was formed initially pothesis that the destructive stress is produced by the flow of
in 1951 with the primary mission of developing freezing and displaced water away from the region of freezing, the pressure
thawing tests to be applied to the evaluation of coarse aggre- being due to the viscous resistance of such flow through the
gates. In 1961, the subcommittee reported that standardization permeable structure of the concrete. According to this theory,
of such tests was not warranted because of the high levels of when the flow path exceeds a critical length, the pressure ex-
variability associated with the existing methods. However, in ceeds the strength of the paste. Such flow would occur when
an unpublished report, the subcommittee did outline proce- the water content exceeds the critical saturation point. This
dures for using ASTM C 290 if evaluations of coarse aggregates concept is called the hydraulic pressure theory. The theory was
were required [32]. amplified in 1949 [39] to explain the beneficial influence of en-
Lack of a standardized test method for evaluating resist- trained air. Since the resistance to flow at a given rate is pro-
ance to freezing and thawing of coarse aggregates in concrete portional to the length of the flow path, the air bubbles were
coupled with theoretically based criticisms of the cyclic meth- conceived as spaces into which the excess water produced by
ods, particularly by Powers [33], as discussed later, led to the de- freezing could be expelled without generating destructive pres-
velopment of methods designed to determine the length of time sures. Powers calculated a critical dimension of the order of
required for an aggregate to become critically saturated in con- 0.25 mm (0.01 in.), a value that was approximately the same as
crete. Critical saturation was defined as when specimens ex- that suggested by Mielenz and his co-workers from experimen-
posed to continuous soaking and subjected to a cycle of cooling tal studies [40]. Powers’s initial studies suggested that the hy-
through the freezing point exhibited dilation greater than a pothesis advanced by Taber [39] to explain frost heaving of
specified value. This method was first used by the California soils did not apply to mature concrete. According to this hy-
Department of Highways [34] and later was refined and exten- pothesis, the stress is produced not by hydraulic pressure, but
sively evaluated by Larson and his co-workers [35]. The litera- by the segregation of ice into layers that enlarge as unfrozen
ture survey prepared by Larson et al., as part of the research, is water is drawn toward the region of freezing rather than
a particularly valuable reference on all aspects of freezing and forced away from it.
thawing studies related to aggregates [8]. As an outgrowth of Studies by Verbeck and Landgren [42], as well as those of
this work, ASTM C 671 was approved in 1971 along with ASTM Powers [43], make clear that the paste and aggregate should be
C 682, which provides guidelines for applying ASTM Test C 671 considered separately when explaining the resistance of con-
to the evaluation of coarse aggregates. ASTM C 671 and ASTM crete to freezing and thawing. This is because the paste not
C 682 were both dropped in 2002, again because neither was in only may become critically saturated by moisture from exter-
general use nor required by any other ASTM specification. nal sources but also must withstand pressure generated by wa-
More recently, agencies studying “D-cracking” of concrete ter expelled from the aggregate particles during freezing.
have generally found a modification of the older procedure The temperature at which water freezes in various pores
(ASTM C 666) to be useful in identifying coarse aggregates sus- within the paste decreases with the size of the pore so that,
ceptible to this type of deterioration [36]. According to Stark even if the concrete is at a uniform temperature throughout,
[37], for example, aggregates susceptible to causing distress in the water will be at various stages of conversion to ice. As the
concrete pavements can be identified by freezing and thawing water freezes, the solution in the pore becomes more concen-
concrete in water at a rate of two cycles per day. Equipment, trated. The existence of solutions of various concentrations in
specimen preparation, and procedures vary considerably, and the pores of the paste causes unfrozen water to move to the site
no standardized failure criterion is available. Agencies desiring of freezing in order to lower the concentration made higher at
to use these procedures for accepting or rejecting aggregates the freezing site than the more dilute solution of the unfrozen
develop failure criteria by relating the freeze-thaw test results water. This flow generates stress somewhat like osmotic pres-
to field performance. sures, hence, the designation “osmotic pressure hypothesis.”
Concern with surface mortar deterioration or deicer scal- Powers, in a subsequent summary of his and other re-
ing, particularly on highway and bridge deck pavements, led in search [44], concludes that all three of the theories, with some
1971 to standardization of ASTM C 672, which combined fea- modification, are required to account for the behavior of con-
tures of methods that had been developed and used by various crete subjected to freezing and thawing.
agencies for a number of years. The method uses blocks fabri- When freezing and thawing takes place in the presence of
cated to permit ponding of water on surfaces that are sub- deicing chemicals, localized failures of the exposed surfaces oc-
jected to freezing and thawing in the presence of various cur that is called “scaling,” or surface mortar deterioration. Ex-
deicing agents. cept where the concentration of deicing chemicals is high
Of the methods, ASTM C 666 continues to be the most enough to cause chemical attack, scaling results from freezing
widely used. Specialized equipment is commercially available and thawing. Various theories have been advanced to explain
for conducting the tests under controlled conditions, but the the increased severity of the damage as compared with freezing
essential elements of the method are those that have been used and thawing in water. Browne and Cady [45] have summarized
for more than a century. these theories and their own experiments. Although definitive
answers have not been obtained, the mechanism is probably
Theoretical Considerations most influenced by concentration gradients. In addition to the
Cyclic freezing and thawing tests were developed on a prag- lowering of the freezing temperature that accompanies in-
matic rather than a theoretical basis. It was assumed that the creased deicer concentration, flow of water from areas of lower
destruction resulted from the 9 % volume expansion accom- to those of higher concentrations generate stresses such as were
panying the conversion of water to ice, and that this process described earlier. Verbeck and Klieger [46], in work confirmed
 NMAI ON FREEZING AND THAWING 157

by others, found that deterioration was greater for intermediate The fundamental transverse frequency is determined with
concentrations of deicing chemical (3 to 4 %) than for lower or the specimens at a temperature of 5.0  1.6°C (41  3°F). Cal-
higher concentrations (to 16 %). Such findings can be explained culation of Pc assumes that the mass and dimensions of the
by generation of osmotic-like pressures accompanying the specimens remain constant throughout the test. While this as-
movement of water between areas of varying concentrations. sumption is not true in many cases because of disintegration, the
test is usually used to make comparisons between the relative dy-
Rapid Freezing and Thawing Tests namic moduli of specimens and Pc is adequate for the purpose.
The durability factor is calculated as
In 1971, ASTM C 290 and C 291 were combined as Procedures
PN
A and B in ASTM C 666. The method is designated as “rapid” DF  
M
because it permits alternately lowering the temperature of
specimens from 4 to
18°C (40 to 0°F) and raising it from
18 where
to 4°C (0 to 40°F) in not less than 2 h nor more than 5 h. Thus,
a minimum of four and a maximum of 12 complete cycles may DF  durability factor of the specimen;
be achieved during a 24 h period. The conventionally accepted P  relative dynamic modulus of elasticity at N cycles, %;
term of testing is 300 cycles, which can be obtained in 25 to 63 N  number of cycles at which P reaches the specified mini-
days. For Procedure A, both freezing and thawing occur with mum value for discontinuing the test or the specified
the specimens surrounded by water, while in Procedure B, the number of cycles at which the exposure is to be termi-
specimens freeze in air and thaw in water. For Procedure A, the nated, whichever is less; and
thawing portion is not less than 25 % of the total cycle time, M  specified number of cycles at which the exposure is to
while for Procedure B, not less than 20 % of the time is used be terminated.
for thawing. The time required for the temperature at the cen-
ter of any single specimen to be reduced from 3 to
16°C (37 Because of the danger of damage to specimen containers
to 3°F) shall be not less than one half of the length of the cool- and other parts of the equipment, testing usually is termi-
ing period, and the time required for the temperature at the nated when the relative dynamic modulus of elasticity falls
center of any single specimen to be raised from
16 to 3°C (3 below 50 %.
to 37°F) shall be not less than one half of the length of the heat- The scope of ASTM C 666 states that “both procedures are
ing period. For Procedure A, each specimen is surrounded by intended for use in determining the effects of variations in the
approximately 3 mm (18⁄ in.) of water during the freezing and properties of concrete on the resistance of the concrete to the
thawing cycles, while in Procedure B, the specimen is sur- freezing and thawing cycles specified in the particular proce-
rounded completely by air during the freezing phase of the cy- dure. Neither procedure is intended to provide a quantitative
cle and by water during the thawing phase. The requirements measure of the length of service that may be expected from a
for Procedure A are met by confining the specimen and sur- specific type of concrete.”
rounding water in a suitable container. The specimens are Procedure A is currently required in three ASTM specifi-
normally prisms not less than 75 mm (3 in.) nor more than 125 cations: namely, ASTM Specification for Air-Entraining Admix-
mm (5 in.) in width and depth, and between 275 and 405 mm tures for Concrete (C 260), ASTM Specification for Chemical
(11 and 16 in.) in length. Admixtures for Concrete (C 494), and ASTM Specification for
During the early years of testing by freezing and thawing, Chemical Admixtures for Use in Producing Flowing Concrete
laboratories constructed specialized equipment. Currently, sev- (C 1017). In these admixture specifications, the performance
eral manufacturers produce off-the-shelf or custom-built requirement is stated in terms of a “relative durability factor”
freeze-thaw equipment that meets the requirements of ASTM C calculated as
666. Capacities typically range from 18 to 80 specimens, al-
PN
though custom-built units have been designed for up to 120. DF or (DF1)  
300
Some of the equipment is designed for Procedure A only, while
the larger units typically can be used for both A and B. Some and


aspects of these machines have been subject to criticism, but
DF
they do meet the needs for rapid testing within practical limits. RDF    100
DF1
Deterioration of specimens is determined by the resonant
frequency method, ASTM C 215. The fundamental transverse where
frequencies are determined at intervals not exceeding 36 cy-
cles of exposure and are used to calculate the relative dynamic DF  durability factor of the concrete containing the admix-
modulus of elasticity ture under test;
DF1  durability factor of the concrete containing a reference


n 12
Pc  
n2
 100 admixture (or in the case of ASTM C 494, only an ap-
proved air-entraining admixture);
where P  relative dynamic modulus of elasticity in percent of the
dynamic modulus of elasticity at zero cycles (values of
Pc  relative dynamic modulus of elasticity, after c cycles of P will be 60 or greater since the test is to be terminated
freezing and thawing, %; when P falls below 60 %);
n  fundamental transverse frequency at 0 cycles of freezing N  number of cycles at which P reaches 60 %, or 300 if
and thawing; and P does not reach 60 % prior to the end of the test
n1  fundamental transverse frequency after c cycles of freez- (300 cycles); and
ing and thawing. RDF  relative durability factor.
158 TESTS AND PROPERTIES OF CONCRETE

TABLE 1—Within-Laboratory Precisions for Averages of Six Beams Tested in Accordance with
ASTM C 666
Procedure A Procedure B
Range of Average
Durability Factor Standard Deviation (1S) Acceptable Range (D2S) Standard Deviation (1S) Acceptable Range (D2S)

0 to 5 0.3 0.9 0.4 1.2

5 to 10 0.6 1.8 1.7 4.7
10 to 20 2.4 6.8 3.3 7.4
20 to 30 3.4 9.7 4.3 12.2
30 to 50 5.2 14.7 6.3 17.8
50 to 70 6.4 18.1 8.2 23.2
70 to 80 4.7 13.4 7.0 19.7
80 to 90 2.3 6.5 3.6 6.8
90 to 95 0.9 2.4 1.6 4.5
Above 95 0.4 1.3 0.8 2.3

ASTM C 260, C 494, and C 1017 require that the relative with such characteristics and, therefore, uniform precisions
durability factor of the concrete containing the admixture un- would not be expected for the entire range of durability factors.
der test be at least 80 when compared with the reference While ASTM C 666 is specified in the ASTM Specification
concrete. The value of 80 is not intended to permit poorer for Lightweight Aggregates for Structural Concrete (C 330), no
performance than the reference concrete, but rather to as- minimum durability factor values are given. The specification
sure the same level of performance, with appropriate recog- states that “in the absence of a proven freezing and thawing
nition of the variability of the test method. The value of 80 test satisfactory to the purchaser.” ASTM C 666 is similarly
was established before levels of precision were established for listed as a method of sampling and testing in ASTM Specifica-
ASTM C 666, but is consistent with the now-established pre- tion for Concrete Aggregates (C 33), but again no minimum
cision values. durability factors or other requirements are stated.
One criticism of rapid freezing and thawing tests has been As noted, the primary measure of deterioration is the rela-
variability of the results both within and between laboratories. tive dynamic modulus calculated from determinations of reso-
In response to the requirement by ASTM for precision state- nant transverse frequency. Length change is noted in ASTM C
ments in all test methods, ASTM Subcommittee C09.03.15 666 as an acceptable additional way of assessing deterioration.
(now C09.67) reviewed the data from major published studies. Mass loss is also sometimes used for such assessments, while de-
Statistical analyses of these data, as described by Arni [47], terminations of reductions in tensile or compressive strength
showed that the variability was primarily a function of the level are used only infrequently because such testing is destructive.
of durability factor of the concrete for the ranges of N and P The question of which measure of deterioration is best to
normally used. In 1976, a precision statement was added to use is complicated by the fact that the different available tests
ASTM C 666 for expected within-batch precision for both Pro- measure different things, and the manner and extent to which
cedures A and B. Values for standard deviation (1S) and the properties measured are related to freezing-and-thawing
acceptable range (D2S), as defined by ASTM Practice for damage, especially under natural conditions, are matters of dis-
Preparing Precision and Bias Statements for Test Methods for agreement. Thus, the particular measure used often depends on
Construction Materials (C 670), are given in Tables 2 and 3 of the philosophy of the laboratory using it and on the particular
ASTM C 666 for ranges of average durability factor in ten in- purpose for which the tests are being made. Mass loss measures
crements and numbers of specimens averaged from two loss of material or sloughing from the surface of the specimens.
through six. Values for six specimens (as required by ASTM C Length change is based on the fact that internal damage is ac-
260, C 494, and C 1017) are given in Table 1. companied by expansion rather than contraction during cool-
These values confirm the long-recognized facts that the vari- ing or by a permanent dilation after a freezing and thawing
ability is less for very good or very poor concrete than for con- cycle, as discussed by Powers [33]. Either mass loss due to
cretes of intermediate durability and that results from Proce- sloughing or reduction in resonant frequency may occur with-
dure A are somewhat less variable than those obtained from out accompanying length change, and, in Procedure A, mass
Procedure B. An example of earlier work showing the variability loss often occurs without a reduction in resonant frequency or
at different durability factors is given in Fig. 1. This figure, from expansion. Resonant frequency and expansion reflect internal
Cordon [7], shows durability factors of concretes containing a disruptions that are caused by unsound aggregates or deficient
variety of aggregates, and with varying cement contents, water- air-void characteristics, while mass loss reflects primarily sur-
cement ratios, and air contents. For very low and very high air face mortar deterioration.
contents, the measured durability factors showed relatively Users of ASTM C 666 have not reached a consensus on
small variations. At intermediate values of air content, a “transi- limiting values for the method’s measurement quantities. A
tion zone” reflected much greater variability of performance. 1987 survey of State and Canadian provincial highway agen-
One might conclude that, for intermediate levels of perform- cies indicated that seven used ASTM C 666 to qualify coarse
ance as indicated by durability factors, an undeterminable por- aggregates for construction on agency projects [48]. The
tion of the variability reflects the real variability of concretes seven had “seven different specification limiting values, based
 NMAI ON FREEZING AND THAWING 159

Fig. 1—Durability factors for concretes containing various aggregates, cement contents, and water-cement ratios,
as a function of air content (after Cordon [7]).

on the fact that each had a unique way of testing regarding reported to overcome the problem by some and not by others.
aggregate grading, moisture conditioning, cement contents, That the degree of restraint offered by the container influences
air contents, coarse aggregate contents, curing methods and the number of cycles required to reach a specified level of rela-
time, cycle length, number of cycles, method of measuring tive dynamic modulus is clear, as shown by Cook [49], who
deterioration, failure criteria, and number of specimens con- found that the number of cycles necessary to reach a given level
stituting a test.” Further, three agencies used durability factor, of durability factor was increased dramatically when the speci-
three used length change or another measure of expansion, mens were tested in rubber containers as compared with con-
and one used both for specification limits. A standard prac- tainers made of steel. One agency has minimized the container
tice, which would include research-supported limiting values, influence by using cylindrical specimens and surrounding them
is clearly desirable. with rubber boots [48].
One important potential influence on ASTM C 666 results Another important influence on the results is the degree of
is the effect of the container used to hold the specimen during saturation of the concrete and the aggregates both at the time
testing with Procedure A. Rigid containers have the potential to of mixing and throughout the course of the testing. The speci-
damage specimens and are not permitted. A note explains fur- fications citing the use of ASTM C 666 require that testing be-
ther. “Experience has indicated that ice or water pressure, dur- gin after 14 days of moist curing. (A note in ASTM C 666 itself
ing freezing tests, particularly in equipment that uses air rather says that the tests should be started when the specimens are 14
than a liquid as the heat transfer medium, can cause excessive days old “unless some other age is specified.”) The mixing pro-
damage to rigid metal containers, and possibly to the specimens cedures referenced in ASTM Test Method for Making and Cur-
therein. Results of tests during which bulging or other distor- ing Concrete Test Specimens in the Laboratory (C 192) require
tion of containers occurs should be interpreted with caution.” that coarse aggregates be immersed for 24 h prior to mixing.
This situation is particularly noticeable when the specimens are Thus, the comparatively high degree of internal saturation and
exposed with the long dimension vertical. Ice forms quickly on the early age at which testing begins result in a relatively severe
the open top rather than uniformly along the long dimension test when compared with field exposures in which a period of
of the container. Pressures from continued conversion of water drying normally occurs before exposure to freezing and thaw-
to ice between the specimen and container cannot be relieved, ing. Even a brief period of drying greatly improves resistance
with consequent bulging of the container. Various adjustments to freezing and thawing, since it is difficult to resaturate con-
have been made to mitigate this problem, including exposure of crete that has undergone some drying. Strategic Highway
the specimens in containers with the long dimensions horizon- Research Program researchers [11] confirmed the effect of
tal to provide a larger open area; inclusion of flexible windows, drying. They found that Procedure B results approach those
or corners, in containers; or inclusion of a rubber ball in the bot- from Procedure A if the specimens are wrapped in terrycloth
tom of containers to absorb the pressures. All of these have been during the test; this keeps the outer surface of the concrete
160 TESTS AND PROPERTIES OF CONCRETE

moist, that is, at or near saturation, rather than allowing it to ural cooling rates that seldom exceed 3°C/h (5°F/h). He also
dry out during the freezing-in-air portion of the test cycle. noted the significance of moisture conditioning of aggregates
Still another influence on freezing and thawing tests is the and concretes in the test methods that generally provided for
use of salt water (typically 2 % sodium chloride in water) as the only saturated concrete as compared with natural exposures
freezing-and-thawing medium. Some agencies have used salt where some seasonal drying is possible.
water since that is the medium encountered in many field situa- He stated that durability was not a measurable property
tions where water ponds on structures regularly treated with de- but that expansion that occurred during a slow cooling cycle
icers. Use of salt water in Procedure A increases the severity of when the concrete or its aggregates became critically saturated
the test beyond that attained with the use of fresh water. was measurable and would provide an indication of potential
Philleo has taken ASTM C 666 to task, recommending that resistance to damage by freezing and thawing. He proposed
it be “modified or replaced by a more realistic standard for that specimens be prepared and conditioned so as to simulate
judging the acceptability of concrete for field applications” field conditions and then be subjected to periodic slow-rate
[50]. He points out that most concrete undergoes both drying freezing and storage in water at low temperature between
and curing longer than 14 days before encountering its first freezing exposures. Concrete subject to frost damage should
freezing. Nevertheless, accommodating the many possible reach some critical saturation level, after which it would ex-
variations of saturation that might be encountered is impracti- pand on freezing. The length of time required to become criti-
cal so that the most consistently reproducible condition is that cally saturated would be compared with the field exposures to
of continued moist curing. be encountered. If the period during which freezing would be
Because the testing is initiated at a fixed age, considerable expected was less than the time to reach critical saturation,
variation of strength at the time of exposure to freezing and then no damage would be expected. As opposed to a single
thawing may be encountered with cements of different strength durability factor, the time to critical saturation could be more
gain characteristics for concrete with sound aggregates and sat- readily interpreted for various field exposures.
isfactory air-void characteristics. There is not a great body of The California Division of Highways was first to report, in
data on the influence on resistance to freezing and thawing of 1961, a practical application of Powers’s proposed method
strength at the time that exposure begins. Buck et al. [51] re- [34]. They developed specimen preparation and conditioning
ported tests indicating that, because of its relationship with the methods, testing and measuring techniques, and performance
amount of freezable water, a given level of maturity (strength) criteria. The method was used to evaluate several aggregates
was necessary to provide an acceptable degree of frost re- for a major highway construction project. The aggregates
sistance as indicated by a durability factor of 50 for concrete judged acceptable by the California procedure would have
containing satisfactory aggregates and entrained air. They cited been rejected by other conventionally accepted criteria. The
earlier work by Klieger [52], who reported similar findings in concrete is now more than 30 years old; sections have been
his studies of salt scaling. Consideration of strength at the time overlaid because they are structurally inadequate, but the
of initial freezing is particularly important in testing concrete freeze-thaw performance of all of the concrete has been re-
made with blended cements and pozzolanic supplementary ce- ported as continuing to be satisfactory.
mentitious materials that gain strength more slowly than con- The procedures proposed by Powers and the method de-
crete without such materials. This influence is minimized when veloped in California were extensively studied and refined by
the evaluations are made by comparing concretes made with Larson and others [35]. In 1971, ASTM C 671 was standardized.
similar materials as required in ASTM C 260 and C 494. This method provides for cylindrical specimens 75 mm (3 in.)
Visual examination of specimens during ASTM C 666 cy- in diameter and 150 mm (6 in.) long that are stored in water at
cling may also give warnings of the likelihood of popout prob- 1.7  0.9°C (35  2°F). At two-week intervals, the specimens
lems in a concrete. Popouts are shallow, usually conical spalls are cooled in water-saturated kerosene at a rate of 2.8 
of the concrete through aggregate particles due to internal 0.5°C/h (5  1°F/h). During the cooling cycle, the specimen is
pressure and can be attributed to defects in the aggregate. The placed in a strain frame to permit measurement of length
small size of the specimens in ASTM C 666 has been criticized change. A typical plot of length change versus temperature is
because a large piece of popout-producing aggregate in the shown in Fig. 2. Prior to critical saturation, the length change
center of a relatively small specimen could cause it to fail; in will proceed along the dashed curve without dilation. Critical
the field, the popout material would presumably only cause su- dilation is defined as a sharp increase (by a factor of 2 or more)
perficial surface defects [53]. between dilations on successive cycles. Highly frost-resistant
Cyclic freezing and thawing methods were developed and concrete may never exhibit critical dilation. ASTM C 682 was
applied only to laboratory-mixed concrete until 1975, when also standardized in 1971. This practice is based largely on the
they were extended to cores or prisms cut from hardened con- work of Larson and his coworkers. Procedures essentially in ac-
crete. Experience with testing of specimens from hardened cordance with ASTM C 671 have been used by Buck [54], and
concrete is limited; the results of a 1987 survey of State and he found that a specimen that is frost resistant will not show
Canadian provincial highway agencies gave no indication that increasing dilation with continuously decreasing temperature;
any were using anything other than specimens cast to size it will show some limited initial dilation as all moist specimens
specifically for freeze-thaw testing [48]. do, but dilation will not continue throughout the cooling. In ad-
dition to the ASTM C 671 criterion for critical dilation in terms
Dilation Methods of increase from one cycle to another, he suggested that a cri-
terion for critical dilation applicable to results of a single test
In 1955, Powers published a critical review of existing cyclic be as follows:
methods for freezing and thawing tests [33]. He was particu- (a) If the dilation is 0.005 % (  50 millionths) or less, the
larly critical of what he considered unrealistically high freezing specimen may be regarded as frost resistant, that is, the
rates of from 6 to 60°C/h (10 to 100°F/h) as compared with nat- dilation is not critical.
 NMAI ON FREEZING AND THAWING 161

Fig. 2—Typical length change and temperature charts.

(b) If the dilation is 0.020 % (  200 millionths) or more, the Scaling Resistance
specimen may be regarded as not frost resistant, that is,
critical dilation has been exceeded. In the early 1960s, it became apparent that the increasing use
(c) If the dilation is in the range between 0.005 and 0.020 %, of deicing chemicals as part of a “bare pavement” policy
an additional cycle or more should be run. adopted for the nation’s highways was being reflected in
ASTM C 671 and C 682 were not used extensively and, con- widespread surface scaling of pavements and bridge decks. It
sequently, were discontinued in 2002. The 1987 survey of State has long been known that dense, high quality concrete, with
and Canadian provincial highway agencies indicated that only adequate entrained air and with adequate curing and a pe-
one was using ASTM C 671 to any extent [48]. This agency has riod of drying before the first application of deicing agents,
subsequently stopped using the method. Although the appara- is essential in preventing damage [12]. Widespread scaling
tus is comparatively inexpensive and larger numbers of speci- demonstrated that all of these requirements were not being
mens can be processed than with ASTM C 666 equipment, sig- met consistently and brought forth a plethora of remedial or
nificant storage capacity is required. The procedures of ASTM preventive products including admixtures, surface treatments,
C 682 were extensive and complex, largely because of require- and curing agents. In 1971, ASTM C 672 was standardized.
ments designed to bracket a broad range of potential exposure The test is based on the experience of a number of agencies
conditions. It was suggested that aggregates and concrete be who used blocks that were fabricated to permit ponding of
“maintained or brought to the moisture condition representa- water on one surface and that could be exposed to freezing
tive of that which might be expected in the field.” However, it and thawing in the presence of deicing agents. The test “cov-
was noted that “aggregate moisture states other than dry or sat- ers determination of the resistance to scaling of a horizontal
urated are very difficult to maintain during preparation of concrete surface subject to freezing and thawing cycles in the
specimens. Reproducibility of overall test results is likely to be presence of deicing chemicals. It is intended for use in evalu-
affected adversely by variability in aggregate moisture.” While ating the surface resistance qualitatively by visual examina-
the complexity of the evaluation procedure limited its general tion. This test method can be used to evaluate the effect of
applicability, the procedures may have been justified where mixture proportioning, surface treatment, curing, or other
large projects or economic consequences of detailed aggregate variables on resistance to scaling. This test method is not in-
evaluation were warranted. ASTM C 671 included the state- tended to be used in determining the durability of aggregates
ment that it “is suitable for ranking concretes according to or other ingredients of concrete.” The report of ACI Commit-
their resistance to freezing and thawing for defined curing and tee 201 says, on the other hand, the “use of ASTM C 672 will
conditioning procedures.” However, both ASTM C 671 and C demonstrate the acceptability or failure of a given concrete
682 warned that the significance of the results in terms of po- mixture” [12].
tential field performance would depend upon the degree to The specimens must have a surface area of at least 0.045
which field conditions could be expected to correlate with m2 (72 in.2) and be at least 75 mm (3 in.) deep. The specimens
those employed in the laboratory. are placed in a freezing space after moist curing for 14 days
162 TESTS AND PROPERTIES OF CONCRETE

and air storage for 14 days. Provisions are included for appli- characteristics, and exposures that reduce the opportunity for
cations of protective coatings if desired at the age of 21 days. critical saturation. While adequate air entrainment with
The method calls for covering the surface with approximately proper air-void parameters protects the paste, it may not over-
6 mm (14⁄ in.) of a solution of calcium chloride and water hav- come the effect of aggregate that is susceptible to damage by
ing a concentration such that each 100 mL of solution contains freezing and thawing. Direct translation of results from labo-
4 g of anhydrous calcium chloride. Modifications of the deicer ratory freezing and thawing tests is difficult at best because
and application procedures, including freezing of water and of the variety of exposures encountered, but the currently ap-
addition of the solid deicer, are allowed where there is need to proved standards are very useful when properly conducted
evaluate the specific effect. and interpreted and have undoubtedly resulted in a signifi-
The specimens are cycled through a freezing environment cant improvement in the resistance of concrete to weathering
for 16 to 18 h, followed by laboratory air for 6 to 8 h. While the from natural forces.
method describes laboratory procedures, it has been used for
outdoor exposures as well.
The specimens are rated visually according to a scale of 0 References
(no scaling) to 5 (severe scaling) after 5, 10, 15, 25, and every [1] Scholer, C. H., “Durability of Concrete,” Report on Significance
25 cycles thereafter. Some investigators have measured the of Tests of Concrete and Concrete Aggregates, ASTM STP 22A,
mass of the detritus, but this is not required by the method. ASTM International, West Conshohocken, PA, 1943.
As noted in the method, the ratings are ranks, and as such [2] Scholer, C. H., “Hardened Concrete, Resistance to Weather-
may not be subjected appropriately to analyses based on the ing—General Aspects,” Significance of Tests and Properties of
calculation of averages and standard deviations or other tech- Concrete and Concrete Aggregates, ASTM STP 169, ASTM In-
niques that assume continuous distributions. If groups of simi- ternational, West Conshohocken, PA, 1955.
lar specimens are to be reported or compared with other [3] Powers, T. C., “Resistance to Weathering—Freezing and Thaw-
groups, such nonparametric quantities as the median and ing,” Significance of Tests and Properties of Concrete and Con-
range may be used. crete Aggregates, ASTM STP 169, ASTM International, West
Conshohocken, PA, 1955.
Experience with ASTM C 672 generally has been satisfac-
[4] Arni, H. T., “Resistance to Weathering,” Significance of Tests
tory for evaluating the variables for which it was developed.
and Properties of Concrete and Concrete-Making Materials,
ASTM STP 169A, ASTM International, West Conshohocken, PA,
Other Weathering Processes 1966.
[5] Newlon, H., Jr., “Resistance to Weathering,” Significance of
Other weathering processes that have been suspected of caus- Tests and Properties of Concrete and Concrete-Making Materi-
ing deterioration include heating and cooling and wetting and als,” ASTM STP 169B, ASTM International, West Conshohocken,
drying. As noted, chemical attack and alkali-aggregate reac- PA, 1978.
tions are treated elsewhere in this publication. Although cases [6] Woods, H., “Durability of Concrete Construction,” ACI Mono-
have been reported where deterioration was attributed to ag- graph No. 4, American Concrete Institute, Detroit, MI, 1968.
gregate with an abnormally low coefficient of thermal expan- [7] Cordon, W. A., “Freezing and Thawing of Concrete—
sion [55] or where the freezing and thawing resistance was Mechanisms and Control,” ACI Monograph No. 3, American
influenced by aggregates with different coefficients of thermal Concrete Institute, Detroit, MI, 1966.
expansion [56,57], “thermal incompatibility” is now generally [8] Larson, T., Cady, P., Franzen, M., and Reed, J., “A Critical Review
of Literature Treating Methods of Identifying Aggregates Sub-
believed to have at most a minor effect on concrete durability
ject to Destructive Volume Change When Frozen in Concrete
within the normal temperature range. Elevated temperatures
and a Proposed Program of Research,” Special Report 80, High-
such as are encountered in certain parts of nuclear construc- way Research Board, Washington, DC, 1964.
tion are beyond the scope of this paper. [9] “Durability of Concrete: Physical Aspects,” Bibliography 20,
While wetting and drying induce variations in moisture NAS-NRC Publication 493, Highway Research Board, Washing-
content that influence the resistance to freezing and thawing, ton, DC, 1957.
aggregate cracking from excessive drying shrinkage, and an in- [10] “Durability of Concrete: Physical Aspects: Supplement to Bibli-
creased concentration of dissolved salts, all of which reduce re- ography No. 20,” Bibliography 38, NAS-NRC Publication 1333,
sistance to weathering, the writers are not aware of cases where Highway Research Board, Washington, DC, 1966.
alternate wetting and drying per se have caused deterioration. [11] Snyder, M. B. and Janssen, D. J., “Freeze-Thaw Resistance of
Concrete—An Annotated Bibliography,” Report SHRP-C/UFR-
Summary 92-617, Strategic Highway Research Program, Washington, DC,
1992; available from Transportation Research Board, Washing-
ton, DC.
The resistance of concrete to weathering in the absence of
[12] “Guide to Durable Concrete,” Manual of Concrete Practice, ACI
chemical attack or detrimental cement-aggregate reactions de-
201.2R-92, Part 1, American Concrete Institute, Detroit, MI,
pends on its ability to resist freezing and thawing. Dry con- 1993.
crete will withstand freezing and thawing indefinitely, [13] “Durability of Concrete,” Publication SP-47, American Concrete
whereas highly saturated concrete exposed to particularly se- Institute, Detroit, MI, 1975.
vere conditions such as hydraulic head or very low tempera- [14] “Concrete Durability—Katherine and Bryant Mather Interna-
tures may be severely damaged in a few cycles. This damage tional Conference,” Publication SP-100, American Concrete In-
is more likely to occur when the air content of the concrete stitute, Detroit, MI, 1987.
is at the lower end of the recommended range. Research and [15] “Durability of Concrete,” Publication SP-126, American Con-
experience have shown that resistance to freezing and thaw- crete Institute, Detroit, MI, 1991.
ing requires a low water/cement (w/c) ratio, an adequate vol- [16] “Durability of Concrete—Third International Conference,” SP-
ume of entrained air with the proper void distribution and 145, American Concrete Institute, Farmington Hills, MI, 1994.
 NMAI ON FREEZING AND THAWING 163

[17] SP-170, American Concrete Institute, Farmington Hills, MI, [39] Powers, T. C., “The Air-Requirement of Frost Resistant Con-
1997. crete,” Proceedings, Highway Research Board, Vol. 29, 1949.
[18] Vicat, L. J., Treatise on Calcareous Mortars and Cements, (trans- [40] Mielenz, R. C., Wolkodoff, V. E., Burrows, R. W., Backstrom, J.
lated from the French by J. T. Smith), J. Weale, London, 1837. L., and Flack, H. E., “Origin, Evolution and Effects of the Air
[19] “Symposium on Freezing-and-Thawing Tests of Concrete,” Pro- Void System in Concrete,” Journal, American Concrete Insti-
ceedings, ASTM International, West Conshohocken, PA, Vol. 46, tute, Detroit, MI, July 1958, Aug. 1958, Sept. 1958, and Oct.
1946. 1958.
[20] Wuerpel, C. E. and Cook, H. K., “Automatic Accelerated Freez- [41] Taber, S., “The Mechanics of Frost Heaving,” The Journal of
ing-and-Thawing Apparatus for Concrete,” Proceedings, ASTM Geology, Vol. 38, No. 4, 1930.
International, West Conshohocken, PA, Vol. 45, 1945. [42] Verbeck, G. and Landgren, R., “Influence of Physical Character-
[21] Walker, S. and Bloem, D. L., “Performance of Automatic Freez- istics of Aggregate on the Frost Resistance of Concrete,” Pro-
ing-and-Thawing Apparatus for Testing Concrete,” Proceed- ceedings, ASTM International, West Conshohocken, PA, Vol. 60,
ings, ASTM International, West Conshohocken, PA, Vol. 51, 1960.
1951. [43] Powers, T. C., “The Mechanism of Frost Action in Concrete,”
[22] Arni, H. T., Foster, B. E., and Clevenger, R. A., “Automatic Stanton Walker Lecture No. 3, National Sand and Gravel Asso-
Equipment and Comparative Test Results for the Four ASTM ciation, National Ready-Mix Concrete Association, Silver Spring
Freezing-and-Thawing Methods for Concrete,” Proceedings, MD, 1965.
ASTM International, West Conshohocken, PA, Vol. 56, 1956. [44] Powers, T. C., “Freezing Effects in Concrete,” Durability of Con-
[23] Cordon, W. A., “Automatic Freezing-and-Thawing Equipment crete, Publication SP-47, American Concrete Institute, Detroit,
for a Small Laboratory,” Bulletin No. 259, NAS-NRC Publication MI, 1975.
768, Highway Research Board, Washington, DC, 1960. [45] Browne, F. P. and Cady, P. D., “Deicer Scaling Mechanisms in
[24] Cook, H. K., “Automatic Equipment for Rapid Freezing-and- Concrete,” Durability of Concrete, Publication SP-47, American
Thawing of Concrete in Water,” Bulletin No. 259, NAS-NRC Pub- Concrete Institute, Detroit, MI, 1975.
lication 768, Highway Research Board, Washington, DC, 1960. [46] Verbeck, G. J. and Klieger, P., “Studies of ‘Salt’ Scaling of Con-
[25] Hericart de Thury, “On the Method Proposed by Mr. Brard for crete,” Bulletin 150, Highway Research Board, Washington, DC,
the Immediate Detection of Stones Unable to Resist the Action 1957.
of Frost,” Annales de Chimie et de Physique, Vol. 38, 1828, pp. [47] Arni, H. T., “Precision Statements Without an Interlaboratory
160–192. Test Program,” Cement, Concrete, and Aggregates, Vol. 1, No.
[26] Henry, J., “On the Mode of Building Materials and an Account 2, 1979.
of the Marble Used in the Extension of the U.S. Capitol,” Amer- [48] Vogler, R. H. and Grove, G. H., “Freeze-Thaw Testing of Coarse
ican Journal of Science, Vol. 72, 1856. Aggregate in Concrete: Procedures Used by Michigan Depart-
[27] Grun, R., “Investigations of Concrete in Freezing Chambers,” ment of Transportation and Other Agencies,” Cement, Con-
Zement, Vol. 17, 1928. crete, and Aggregates, Vol. 11, No. 1, 1989.
[28] Scholer, C. H., “Some Accelerated Freezing and Thawing Tests,” [49] Cook, H. K., “Effects of Fluid Circulation and Specimen Con-
Proceedings, ASTM International, West Conshohocken, PA, tainers on the Severity of Freezing and Thawing Tests,” infor-
Vol. 28, 1928. mal presentation, Session 13, Annual Meeting of Highway Re-
search Board, Washington, DC, Jan. 1963.
[29] Mattimore, H. S., “Durability Tests of Certain Portland Ce-
ments,” Proceedings, Highway Research Board, Vol. 16, 1936. [50] Philleo, R. E., “Freezing and Thawing Resistance of High-
Strength Concrete,” NCHRP Synthesis No. 129, Transportation
[30] Withey, M. O., “Progress Report, Committee on Durability of
Research Board, Washington, DC, 1986.
Concrete,” Proceedings, Highway Research Board, Vol. 24, 1944.
[51] Buck, A. D., Mather, B., and Thornton, H. T., Jr., “Investigation
[31] Foster, B. E., “Report on Cooperative Freezing and Thawing
of Concrete in Eisenhower and Snell Locks St. Lawrence Sea-
Tests of Concrete,” Special Report No. 47, Highway Research
way,” Technical Report 6-784, Waterways Experiment Station,
Board, Washington, DC, 1959.
Vicksburg, MS, July 1967.
[32] ASTM Committee C9, minutes of December 1961 Meeting.
[52] Klieger, P., “Curing Requirements for Scale Resistance of Con-
[33] Powers, T. C., “Basic Considerations Pertaining to Freezing-and- crete,” Bulletin 150, Highway Research Board, Washington, DC,
Thawing Tests,” Proceedings, ASTM International, West 1957.
Conshohocken, PA, Vol. 55, 1955. [53] Sturrup, V., Hooton, R., Mukherjee, P., and Carmichael, T.,
[34] Tremper, B. and Spellman, D. L., “Tests for Freeze-Thaw Dura- “Evaluation and Prediction of Concrete Durability—Ontario
bility of Concrete Aggregates,” Bulletin 305, Highway Research Hydro’s Experience,” Concrete Durability—Katherine and
Board, Washington, DC, 1961. Bryant Mather International Conference, Publication SP-100,
[35] Larson, T. D. and Cady, P. D., “Identification of Frost-Suscepti- American Concrete Institute, Detroit, MI, 1987.
ble Particles in Concrete Aggregates,” NCHRP Report No. 66, [54] Buck, A. D., “Investigation of Frost Resistance of Mortar and
Highway Research Board, Washington, DC, 1969. Concrete,” Technical Report C-76-4, U.S. Army Engineer Water-
[36] Schwartz, D. R., “D-Cracking of Concrete Pavements,” NCHRP ways Experiment Station, Vicksburg, MS, 1976.
Synthesis No. 134, Transportation Research Board, Washington, [55] Pearson, J. C., “A Concrete Failure Attributed to Aggregate of
DC, 1987. Low Thermal Coefficient,” Proceedings, American Concrete In-
[37] Stark, D., “Characteristics and Utilization of Coarse Aggregates stitute, Vol. 48, 1952.
Associated with D-Cracking,” Living with Marginal Aggregates, [56] Callan, E. J., “Thermal Expansion of Aggregates and Concrete
ASTM STP 597, ASTM International, West Conshohocken, PA, Durability,” Proceedings, American Concrete Institute, Vol. 48,
1976, p. 45. 1952.
[38] Powers, T. C., “A Working Hypothesis for Further Studies of [57] Higginson, E. C. and Kretsinger, D. G., “Prediction of Concrete
Frost Resistance of Concrete,” Journal, American Concrete In- Durability from Thermal Tests of Aggregate,” Proceedings,
stitute, Detroit, MI, Feb. 1945. ASTM International, West Conshohocken, PA, Vol. 53, 1953.
16
Corrosion of Reinforcing Steel

Neal S. Berke1

Preface of corrosion products can cause spalling of the concrete cover,

and sufficient reduction in the cross section of the bar to
THIS VERSION OF CHAPTER 17 IN ASTM STP 169C IS render a structure incapable of safely supporting its design
an updated version of the chapter written by William F. Peren- loads.
chio. In ASTM STP 169B, this chapter was authored by P. Cady. Such chloride-induced corrosion is a serious problem.
In ASTM STP 169A, this chapter was authored by B. Temper. Durable repair procedures are available; however, all are
The first three sections remained as they were with a change in expensive. Significant improvements in methods of protecting
Ref 1 to provide a source of more information on the subject. embedded metals from corrosion at the design stage have been
Substantial changes are included in the following sections to re- made in recent years. This chapter will discuss the mechanisms
flect more knowledge that is now available on the use of corro- of corrosion, typical damage caused by it, investigative tech-
sion inhibitors, epoxy-coated steel, sealers, cathodic protection, niques for evaluating its extent, means of preventing it in new
and new steels. The main value of the chapter is to provide an construction, and remedial measures.
introduction to the field and numerous references for a more
in-depth understanding. The number of references has in- Mechanisms of Corrosion
creased from 28 to 53 reflecting the increase in information
available today. When a metal corrodes, it returns to its natural state, which
Note—as in the original versions, the effects of carbona- is usually the oxide or the hydroxide. Metals tend to do this
tion are not addressed as chloride is the major problem and because, in their metallic state, they are at a higher energy
design for chloride resistance will provide protection against level. Materials tend to seek lower energy levels; hence, the
carbonation [1]. Cracking can cause chlorides to reach rein- tendency to corrode. Few metals (except for the noblest
forcement earlier, resulting in corrosion starting at an earlier ones: platinum, gold, silver) are ever found in nature in the
time, so it should be minimized. metallic (elemental) state.
Because of this tendency to return to the state from which
Introduction they were refined, one might expect that metals would be un-
usable, quickly disappearing while in service. However, a pas-
Many metals, including reinforcing steel, are normally well sive film of tightly adhering corrosion products usually forms
protected against corrosion by embedment in high-quality on steel in alkaline environments, as in concrete, that protects
concrete. However, many of the bridges, parking garages, and the remaining steel from the corrosive conditions [1]. Chloride
marine structures in this country have been severely dam- ions can disrupt the passive film through a process known as
aged by corrosion of reinforcement and the resultant in- pitting and this is the initiation process of corrosion of steel in
creased volume. These and other structures have suffered concrete in the presence of chlorides [1]. A good overall treat-
damage due to the penetration of chloride in sufficient ment of the corrosion of steel in concrete is given in Ref 1.
amounts to the depth of the reinforcing steel. In the case of Any chemical action can be regarded as electrochemical,
bridges and parking structures, the source of the chloride is since it involves transfer or displacement of electrons. How-
usually deicing salts. In marine structures, it is seawater. In ever, the term is sometimes applied only to cases of corrosion
others, it may be chemical admixtures added to the concrete where anodes and cathodes are some finite distance apart,
that contain chlorides for set acceleration. making the flow of electrical current over measurable dis-
Cracks are not necessary for chloride penetration into the tances a part of the process [2]. Corrosion, of course, can take
concrete. Salts are capable of penetrating solid concrete along place on reinforcing steel that is exposed only to air and rain.
with the water in which they are dissolved. Variations in the This is, therefore, sometimes termed atmospheric corrosion
amount of chloride, or moisture, or oxygen at different points [3,4]. An electrolyte is not necessary; however, even a thin film
along a reinforcing bar, or between different bars, can cause of rainwater can act as one, carrying ions and completing the
voltage differentials to develop that greatly increase the rate of electrical circuit. In this case, the anodes and cathodes are
corrosion. The increased volume due to the development contained in the same bar, separated by only small distances.

1
Research and Development Fellow, Grace Construction Products, Cambridge, IL 02140.

164
 BERKE ON CORROSION OF REINFORCING STEEL 165

Concrete provides protection for embedded metals, if it is erates by the zinc going into solution and leaving electrons
high quality and if the depth of cover is about 2 in. (50 mm) or behind, which travel to the carbon cathode in the center of
more. However, if either the quality or the depth of cover, the cell. This flow of electrons excites the filament in the
or both, are low enough, corrosion of steel can occur when: flashlight bulb, producing light.
1. the concrete cover becomes carbonated, dropping the pH A more detailed and complete description of electrolytic
from about 13 to about 9; or cells is given by Uhlig [2].
2. chloride ions have penetrated into the concrete such that Extending this battery concept to a more practical prob-
the concrete adjacent to the steel contains at least 1.3 lb of lem, that of corrosion of reinforcing steel in concrete, consider
chloride ions/yd3 (0.77 kg of chloride ions/m3). the schematic diagram in Fig. 2. This figure depicts the typical
In the case of reinforcing steel, if the corrosion develops conditions found in a bridge deck or a floor in a parking struc-
because the cover concrete has become carbonated, the rate ture, in areas where deicing salts are used. The salts typically
of corrosion is relatively slow. If, however, the corrosion is in- consist essentially of sodium chloride.
stigated by the presence of chloride ions, the rate can be The top layer of steel is termed the anodic steel because
many times as fast as atmospheric corrosion [2] and it causes this is where most of the corrosion occurs. It becomes anodic
much greater damage. This type of corrosion involves voltage because it is much closer to the source of water and chloride
differences and transfer of electrons and electrical current, in ions; therefore, the surrounding concrete becomes inundated
addition to various chemical reactions. with more of each than does the concrete near the bottom
Chloride-induced corrosion is a very common cause of steel, which becomes cathodic. Therefore, the iron that makes
concrete deterioration in all types of structures along the sea up the majority of the steel composition goes into solution,
coast and in northern states where salt is used to remove ice
from roads, streets, parking structures, balconies on high-rise
buildings, etc. The damage occurs at a more rapid rate in
warmer climates because it is due primarily to chemical reac-
tions. Most chemical reactions double in rate for every 18°F
(10°C) increase in temperature [3].
A simplified electrolytic cell is shown in Fig. 1. In the
figure, a voltage difference has been set up by a low salt con-
tent surrounding the iron electrode on the left side and a high
salt content on the right. The difference in voltage could be in-
creased by bubbling oxygen (or air) into the left side. It is this
difference in voltage that drives the cell and causes the corro-
sion to proceed rapidly.
The primary reactions that take place at each electrode are
shown below the cell. Corrosion, or the conversion of metallic
iron into ferrous ions in solution, occurs at the anode, leaving
two electrons behind for every iron ion released. These remain
within the electrode and, if no suitable cathode is available, will
eventually stop the corrosion reaction. This is called anodic
polarization. If a cathode is available and is connected to the
anode by a material, such as metal or carbon, that is capable of
transferring electrons (carrying electrical current),2 these extra
electrons can be shed by the reaction shown. Oxygen and wa-
ter combine with the electrons to produce hydroxyl ions. The
electrical circuit is completed through the solutions surround-
ing the electrodes and the semi-permeable membrane by the
movement of ions such as OH
, Cl
, Na, K, and Ca.
The common (or now, old-fashioned) flashlight battery is
an example of an electrolytic cell that produces a usable
product, light. (This is called the Leclanche’ cell in physical
chemistry textbooks.) The case of the battery is made of zinc
metal and the center pole is made of carbon. Between them
is a mixture of manganese oxide, pulverized carbon, ammo-
nium chloride, and water. It is of paste consistency rather
than a solution, but still acts as an electrolyte. The battery op- Fig. 1—Electolytic cell.

2
Although rubbing two dissimilar metals together was found to attract bits of paper or pith, and sparks had been made to jump gaps, for over a hundred years pre-
viously, in 1729, Stephan Grey found he could transfer this “force” along a damp thread several hundred feet long and attract a feather at the other end. Because of
this capacity for flow, electricity was thought to be some kind of fluid, or two different fluids. Shortly thereafter, Charles Dufay named these two “vitreous” and
“resinous” electricity for the opposite charges produced by rubbing glass or resin.
In 1747, Benjamin Franklin, well known for his electrical experiments, suggested a “one-fluid system,” calling it positive or negative depending on the direction of
flow. Because he had no way of knowing which direction the current was flowing, he arbitrarily decided it was from positive to negative. One hundred years later,
J. J. Thompson demonstrated that electrons flow from negative to positive, but Franklin’s convention was so well established by then that we still refer to “current”
as flowing in the direction opposite to electron flow. (Taken from Maxey Brooks, “Why is a Cathode Called a Cathode,” Materials Performance, National Association
of Corrosion Engineers, Vol. 30, No. 6, Houston, TX, June 1991.)
166 TESTS AND PROPERTIES OF CONCRETE

Fig. 2—Electrolytic corrosion inside concrete.

leaving electrons behind. The electrons find their way to the very little deterioration was observed. The oldest concrete was
bottom mat by metallic continuity brought about by bent bars, also of the poorest quality. Therefore, carbonation progressed
bar chairs, etc. The electrons are then shed into the solution in rapidly, allowing atmospheric corrosion to take place. But the
the pores of the portland cement paste by combining with wa- same high porosity that allowed the rapid carbonation also
ter and oxygen to produce hydroxyl ions, as indicated by the re- provided space for the corrosion products, and disruptive pres-
action shown, the same as that in Fig. 1. Ionic conductance sures did not soon develop. Before carbonation, the steel was
within the concrete completes the circuit. passivated by the high pH of the concrete. After carbonation
The voltage difference between the top and bottom mats of and the resultant drop in pH, the passivation was lost.
steel are due in part to differences in concentration of water, Chloride-induced corrosion, by comparison, is usually
chloride ion, and oxygen. As previously stated, the top steel has much more destructive. It does not depend on carbonation and
more water and chloride in the cement pores surrounding it, can develop even in good quality concrete. All that is necessary
while the bottom steel has more oxygen because there is less wa- is a reasonable supply of water (75 % humidity appears to be
ter in the concrete at that depth and below. Oxygen (and other sufficient), and at least the threshold amount of chloride ions
gases) can diffuse into dry concrete much more easily than wet at the anode. The cathode must have good access to oxygen.
concrete because, with wet concrete, it must diffuse through The voltages that develop cause corrosion to advance so
pore water rather than pore air, a much slower process. quickly that cracking, delamination, and spalling occur in the
The preceding discussion is presented as a common ex- cover concrete. The deterioration, of course, occurs because
ample of macrocell corrosion. However, it is not meant to im- the solid volume of the corrosion products is many times the
ply that separations of anode and cathode are always so great. solid volume of the original metal, in the case of iron. This in-
Microcells can be set up along a given bar, separated only by crease in volume is largely due to the production of solids from
tiny fractions of an inch. Bits of mill scale can act as cathodes reaction of the metals with gases or liquids or dissolved solids.
that drive an adjacent anode in the steel [2]. Local chemical or When steel corrodes, it produces many types of corrosion
physical differences in or on the metal can also cause small products (collectively called rust), depending primarily on
voltage differences. Such localized cells can result in serious oxygen, chloride, and water availability. Most of the products
pit corrosion. are amorphous (no crystalline structure determinable by X-ray
diffractometry) ferrous (Fe) and ferric (Fe) oxides, hy-
Damage Caused by Corrosion droxides, chlorides and hydrates, and complexes of these.
The only usual crystalline substance detected is magnetite
Damage to concrete, in the form of cracking, delamination, (Fe3O4), but exposure to air and slow drying produces more
and spalling, can occur due to atmospheric corrosion; that is, crystalline oxides. Occasionally, someone new to X-ray analysis
the type brought on by carbonation of the cover concrete. interprets the pattern as simply magnetite, ignoring the char-
However, this is slow to develop and, depending on the exter- acteristic “amorphous hump” in the pattern that indicates that
nal environment, may never cause serious deterioration. The the rust is almost entirely amorphous. Crystalline corrosion
previous author was asked to visit Tiger Stadium at Louisiana products such as magnetite occupy two to three times the
State University to inspect the football stadium, parts of which space that the original iron (or steel) did, while the amorphous
were 70, 50, 40, and as little as 10 years old. The oldest portions products are generally more voluminous and variable, de-
were badly stained by reinforcing steel corrosion products, but pending on specific conditions.3

3
William G. Hime, personal communication with William Perenchio.
 BERKE ON CORROSION OF REINFORCING STEEL 167

its way to the bottom through cracks. After the electrical con-
duit corroded away, local anodes and cathodes became estab-
lished within the bottom mat of steel. These microcells devel-
oped rapidly because the distance that ions must travel
through the concrete to complete the cell is small; therefore,
the total current resistance is small.

The Peculiar Effect of Chloride Ion

Thus far, we have discussed only one ion as an instigator of cor-
rosion, the chloride ion. Corrosion specialists in general agree
that any ion of the halogen family would also be destructive.
However, chloride ions are the only ones of this group that are
normally found in large quantities in concrete.
Fig. 3—Concrete spall over a reinforcing bar in a parking The peculiar action of the chloride ion is not entirely
garage floor. understood. Some believe that, when the chloride ion concen-
tration becomes large enough, ferrous chloride, or a ferrous
chloride complex, is formed on the steel surface, replacing
the ferric oxide4 film that was stabilized by the high pH of the
Some examples of damage due to chloride-induced corro-
cement paste. Being more mobile (soluble) than the oxide,
sion are shown in Figs. 3 and 4. Figure 3 shows a spall over the
the chloride salt or complex moves away from the steel, expos-
top reinforcing mat in a parking structure. Such spalls com-
ing fresh iron to the now-corrosive local environment, and
monly start to occur after 5–15 years of service, depending on
instigating the electrochemical cell. Certainly, the presence
concrete cover and quality. Figure 4 shows cracking, leaching,
of the chloride, and the water that carried it in, increases the
and spalling on the underside of a floor slab in a 27-year-old un-
current-carrying capacity of the concrete, normally a very
derground parking structure that had been exposed to deicing
poor conductor.
salt. When it was built, the technology of corrosion of metals
Until the late 1970s, the most commonly used procedure
embedded in concrete was in its infancy; therefore, no effective
for testing concrete for chloride ion permeability was de-
anti-corrosion measures had been taken.
scribed in AASHTO T259, “Resistance of Concrete to Chloride
The very large spall near the center of the photo was
Ion Penetration.” In this method, concrete prisms are continu-
caused by a 4-in. (100-mm) zinc alloy electrical conduit, which
ously ponded with 3 % sodium chloride solution for 90 days.
corroded galvanically due to its proximity to the steel. (Gal-
At the end of this time, the chloride content at several depths
vanic corrosion is discussed in the chapter “Embedded Metals
is determined. The obvious disadvantage to this test is the time
and Materials Other Than Reinforcing Steel” in this volume.)
required. ASTM Test Method for Determining the Penetration
This may not appear to follow the description that attended
of Chloride Ion into Concrete by Ponding (C 1543) was ap-
Fig. 2, and indeed it does not. However, the cross section of this
proved in 2003 and is an updated version of AAHTO T259.
slab did follow that description very well when the structure
In 1977, the Federal Highway Administration entered into
was much younger. The top surface spalls were “repaired” with
a contract with Construction Technology Laboratories to de-
open-graded asphalt, which allowed the entry of large quanti-
velop both a rapid field and laboratory test procedure for de-
ties of water and salt into the top of the slab, which then found
termining chloride permeability of concrete. The results were
to correlate well with the 90-day ponding test. The rapid test
method is now described by AASHTO as the rapid chloride per-
meability test [6]. A version of this standard was also published
in the 1992 Annual Book of ASTM Standards as ASTM Test
Method for Electrical Indication of Concrete’s Ability to Resist
Chloride Ion Penetration (C 1202). It involves forcing chloride
ions to move through a small sample of concrete by applying
direct current voltage.
Subsequent work has shown that the rapid method has an
excellent correlation with chloride penetration as tested by
AASHTO T259, for a wide range of concrete quality and types.
However, some concretes that have altered electrical properties
can give false readings. In the past version of this chapter it was
believed that concrete containing silica fume, for example, is
rated at lower permeability than it should be [7–9], presumably
because the silica fume is such an active pozzolan. This ties up
the normally free sodium, potassium, and calcium ions within
Fig. 4—Severe corrosion of reinforcing steel and an insoluble calcium silicate hydrate structures, reducing the ca-
embedded electrical conduit on the underside of a parking pacity of the concrete for conduction of electricity. However, a
garage floor. more detailed look at the test procedure shows that sample

4
This stabilized layer causes “passivation” of the steel, but only in an uncarbonated portland-cement paste. According to some investigators, this passivation is due to
the formation of a very thin gamma Fe2O3 film that prevents further corrosion.
168 TESTS AND PROPERTIES OF CONCRETE

heating occurs, increasing current passed, so that higher steel against corrosion. Work [11] has shown the extremely im-
permeability concretes look worse than they really are [10]. portant effects of cover and low water/cement ratios in resist-
Thus, the test tends to overestimate the permeability of ordinary ing the ingress of the chloride ion into concrete, as shown in
concretes due to the heating effect. Very low permeability Fig. 5. The reason that 1 in. (25 mm) of cover was relatively in-
concretes such as those containing silica fume or other poz- effective in resisting chloride intrusion is considered to be due
zolans at low w/cm, do not heat up over the testing period so to the fact that 3/4-in. (19-mm) maximum size aggregate was
their currents don’t increase over time. Work is currently under used. Because of its intrinsically higher permeability and the
way to develop a modified version of ASTM C 1202 to address high permeability of the paste adjacent to some aggregates,
the heating and several other issues. Because of such situations, one large aggregate particle can easily “short-circuit” the dis-
it is best to interpret the test results carefully. The test is an tance between the concrete surface and the steel. This explains
indirect method for determining chloride permeability, done why two inches (50 mm) of cover is so much more effective
under artificial conditions of high saturation and applied volt- than one, and three is, comparatively, little better than two.
age. Its speed, however, makes the method attractive to those Other methods of decreasing the permeability of the con-
who need such information. crete cover are available. A synthetic latex admixture was first
A new bulk diffusion test, ASTM C 1556, was issued in used in a concrete overlay on a bridge deck in Michigan in the
2003. This method allows for a more accurate determination early 1960s. Since that time, because the latex proved to be so
of the diffusion coefficient for chloride into concrete, but re- effective in resisting moisture and chloride penetration, a great
quires a minimum of three months to get results. Another new deal of work has been done in the laboratory [7] and field to
method on water absorption, ASTM Test Method for Measure- develop the technology of the physical properties of latex-mod-
ment of Rate of Absorption of Water by Hydraulic Cement ified concrete.
Concretes (C 1585), is now available to add additional ability to Latex reduces the permeability of concrete in two ways.
model the ingress of chloride. The physical presence of a lattice of latex (or rubber) within
the concrete presents a more torturous route to water at-
Precautionary Steps Against Corrosion tempting to penetrate the concrete. And the reduction in mix
water, made possible by the solids portion of the latex that im-
Concrete water-cement ratio is especially important in deter- parts workability of its own to the concrete, results in lower
mining the rate of ingress of water, and therefore dissolved permeability due to a decreased water/cement ratio paste.
salts, into concrete. This ratio above all else in plain concrete is Another method of greatly reducing the amount of water
the most important factor [11]. Some recently introduced ad- required to make concrete workable is the addition of a high-
mixtures and additives are very effective in reducing further range water reducer (HRWR) (ASTM Specification for Chemi-
the permeability of concrete. These will be discussed in more cal Admixtures for Concrete C 494, Types F and G). These
detail in a later section. materials, first identified in the 1930s by Tucker [13], became
The effect of aggregates on the permeability of concrete is available in the United States in the mid-1970s, having been
generally to increase it. Because the aggregates occupy ap- developed for use in concrete almost simultaneously in
proximately three-quarters of the volume of concrete, and their Germany and Japan. They are extremely effective in reducing
permeabilities can be as much as 1000 times greater than that water requirements. For the first time, commercially prepared
of a high quality portland cement paste [12], they have a pro- concrete could be supplied with water/cement ratios as low
found effect on the protection concrete offers to reinforcing as 0.30 by weight, with good workability. All of the physical

Fig. 5—Chloride content profiles at 44 weeks for different w/c ratio concretes.
 BERKE ON CORROSION OF REINFORCING STEEL 169

properties of concrete with equal proportions, with or without metal. However, if large amounts of chloride are present in con-
the presence of a HRWR, are essentially identical. Recent crete containing the zinc, zinc hydroxy chloride can also form.
developments in HRWRs include the introduction of polycar- This occupies about 3.5 times the volume of the metal and, al-
boxylates, which are more effective at lower dosages [14]. though this is less than the iron products, it is still capable of
Soon after the HRWRs were introduced, another type of producing large expansive forces.
material, solid rather than liquid, appeared. This was silica Field evaluation of bridge decks reinforced with galva-
fume, an extremely fine form of silica (“finer than tobacco nized reinforcing steel showed good performance. Evaluated
smoke”) and, therefore, extremely active as a pozzolan, a bridges were subjected regularly to deicer salts for more than
byproduct of the manufacture of silicon metal and ferro-silicon 25 years and chloride concentration measured at the level of
alloys. Because of its fineness, 50 to 100 times finer than port- steel were scientifically higher than the corrosion threshold for
land cement, used as an additive by itself it would cause such a black steel. Remaining coating thicknesses of steel samples
great increase in concrete water requirement that the material taken from the bridges were within the range of recommended
would be of very low quality. However, in combination with a coating thickness for new reinforcement [31].
HRWR, concrete can be produced with permeabilities as low A more widely used material to use for surface protection
[7,8] as that of latex-modified concrete, at a fraction of the cost. of the steel is fusion-bonded epoxy [11,17]. Early work showed
The use of other supplementary cementing materials that, although the epoxy typically has small holes or holidays
(SCM) such as fly ash and ground granulated blast-furnace slag in the finished coating, and the coating can be damaged by
(GGBFS) reduce the permeability of concrete [15,16]. Incor- rough handling or bending of the bars, it can do a very good
poration of such materials in cement paste reduces the pore job of protecting the steel, particularly if all the steel in the
sizes, resulting in decreased permeability. structure is coated and the number of holes, or holidays, are
Another additive available as an anti-corrosion agent is minimized. Although high corrosion densities may occur at
calcium nitrite. This material acts as an anodic or passivating breaks in the coating, the rate of metal loss is moderate. Based
inhibitor, to increase the chloride threshold level for corrosion upon field experience in Florida and elsewhere [32–34], it is
initiation [17–19]. Later work showed that it can reduce corro- now realized that defects should be minimized and that con-
sion rates after initiation [20,21]. stant wet conditions can be more severe than laboratory expo-
Several papers [22–24] suggest using a combination of this sures with drying cycles.
calcium nitrite and silica fume. The philosophy was that the sil- The protection systems mentioned above can be combined
ica fume would slow the entry of water and the chloride ion, to provide improved performance. They are especially more
and the calcium nitrite would slow the corrosion rate after the effective at lower w/cm values [1].
chloride concentration reached the threshold level at the steel.
Several organic based inhibitors are now commercialized Prestressed Concrete
[25,26]. Some of these have dampproofing capabilities that
help to reduce chloride ingress. These materials are predomi- Generally, the same factors are involved in the corrosion of
nantly amine and amine and fatty acid derivatives. prestressing steel as in mild reinforcing steel. However, the
A different approach to the prevention of penetration of same percentage loss of metal can cause catastrophic failures
water and dissolved salts into concrete is to apply a surface in prestressed structures, while causing nothing more than
sealer after the concrete has hardened. A 1981 study [27] spalls and delaminations in normally reinforced structures.
investigated 22 materials used for this purpose. A screening test, Also, prestressing steel is subject to stress corrosion cracking, a
amounting to nothing more than immersion of 4-in. (100-mm) phenomenon that does not occur in mild steel. This is some-
concrete cubes in salt water for three weeks, was used to deter- times brought about, or at least is associated with, the presence
mine which of these appeared to have merit. Five were of certain ions. Two of these that can be present in concrete are
outstanding. These were tested further by subjecting them to the nitrate and bisulfide ions (NO3
and HS
) [35]. Chloride
wetting and drying, and either freezing and thawing cycles ions, in this case, often appear to be innocuous [35,36], except
or extremely severe exposure to ultraviolet light. This testing at high temperatures.
eliminated two materials, leaving a silane, an epoxy, and a Cases have been reported [36] where chloride levels have
methyl methacrylate as the best performers. The silane had the been high in prestressed concrete members, but no corrosion
added advantages that it could not be seen, it penetrated about is evident. However, it is generally accepted that corrosion of
1/8 in. (3 mm) into the concrete, and it did not restrict the prestressing steel is instigated by chloride ions.
outward movement of water vapor. Silanes can be effective but Research [11,37] has shed some light on this subject. In one
need to be reapplied on a routine basis (every 3 to 5 years) as study [11], the corrosion threshold of chloride for prestressing
found in SHRP [28]. However, one study indicated longer steel was found to be approximately six times that for mild steel,
performance [29]. The use of such effective sealers as those or 1.2 % by weight of the cement. Part or all of the reason for
identified by this study is sometimes more effective against the greater resistance to corrosion may be that stearate com-
corrosion than producing a very impermeable concrete [11]. pounds are used during the manufacturing process. Note that
However, as indicated in the SHRP study reapplication is this was the finding of only one study and that chloride induced
necessary under field conditions. corrosion in prestressing strand can occur at low chloride lev-
Another approach to preventing corrosion is to coat the els in the bulk concrete due to cracking or at the anchorage
steel. Zinc has been used for this purpose, and is somewhat ef- zones for post-tensioned structures [38]. Note that prestressing
fective; however, it must be used on all the steel within the steel is more susceptible to embrittlement and failure if the
structure rather than only that which is expected to become concrete or mortar pH drops [1].
anodic [11,30]. A study [37] of pretensioned and post-tensioned prestress-
The usual product that results from corrosion of zinc is zinc ing systems evaluated grouts, ducts, and anchorage systems.
oxide, which occupies only 50 % more space than the original Highlights of the results were that polyethylene ducts were more
170 TESTS AND PROPERTIES OF CONCRETE

effective in shielding the strand from corrosion than steel ducts ter saturation, and junction potentials can cause corrosion
were, but the plastic tended to wear through at bends during potentials to vary significantly for essentially the same corro-
stressing of the steel, exposing the steel to the surrounding con- sion rate [41]. Indeed, no corrosion has been found at much
crete. Traditional bare and galvanized steel ducts deteriorated lower half-cells than
0.23 V versus CSE [18]. Thus, half-cell
badly under chloride exposure, allowing chloride to enter the potentials are best used in contour maps and with subsequent
grout. Joints in duct material were a serious problem, but it was autopsy or corrosion rate measurements.
overcome by the use of shrink-fit tubing. Providing an excellent A magnetic rebar locator may be used to assess the
grout was largely unsuccessful, because of the minimal grout amount of concrete cover over reinforcing steel. This instru-
cover possible within the duct at the inside surfaces of bends. ment is useful in determining the potential for future corro-
The anchorages proved to be vulnerable behind the tradi- sion in various parts of a structure. It can also be used to locate
tional dry-pack mortar. These should be coated or covered with a near-surface bar for grounding the half-cell voltmeter.
epoxy. The epoxy coating on the strand was never breached, Ground penetrating radar, which is based on electromag-
even within the wedge grips. netic wave principle can be used to measure concrete cover
more effectively. Cover profiles can be obtained by scanning the
Assessing the Severity of Corrosion in Existing radar antenna along the surface of reinforced structures [42].
Structures
Concrete Cores
The simplest technique for assessing the present condition of a
structure deteriorated by corrosion is a visual survey [39]. Visual For detailed petrographic examination, chloride ion de-
survey data can be used to produce maps that show locations of termination, or compressive strength tests, concrete cores
cracks, spalls, and other features of deterioration. This is usually removed from the structure can be useful. The petrographic, or
followed by a delamination survey, done by striking the concrete microscopical, examination, ASTM Practice for Petrographic
surface with a metallic object and listening for hollow sounds. Examination of Hardened Concrete (C 856), can yield infor-
The visual and delamination survey results can be used to select mation on the quality of the cement paste, degree of curing,
areas for the in-depth studies described later, usually those areas stability of the aggregates, air content, and damage caused by
that are typical of the worst, moderate, and best conditions. chemical attack, freezing and thawing, corrosion, etc.
Half-cell potential surveys are very useful in determining
which areas are actively corroding [40]. The action of an Chloride Samples
electrochemical cell produces differences in electrical poten-
tial of the steel. By measuring these potentials with a half-cell When concrete cores are not removed, powder samples can
and a voltmeter on a grid pattern, diagrams that resemble be taken for chloride analyses. This is usually done with a ro-
contour maps can be constructed, with lines connecting tary hammer. Samples are taken at various depths, down to
points of equal potential, or voltage. Closely spaced lines are and slightly beyond the depth of the steel. Analysis can be
typically observed near areas of high corrosion activity. One done on an acid-soluble or water-soluble basis. Historically,
study [11] has shown that active corrosion is indicated wher- the acid-soluble technique (AASHTO T260 or ASTM C 1152)
ever the half-cell potential is more negative than
0.23 V ver- has been used, with full awareness of the fact that all of this
sus CSE. Figure 6 shows the relationship of half-cell potential chloride is not available to support corrosion, due to chemi-
to actual corrosion current that was determined in this study. cal combination with the cement or because it is tightly held
The test method is described in ASTM Test Method for Half- within aggregate particles. The results of water-soluble tests
Cell Potentials of Uncoated Reinforcing Steel in Concrete (C are greatly affected by the degree of grinding of the sample
876). However, the oxygen content, usually related to the wa- and by the length and temperature of leaching [43]. The

Fig. 6—Relationship between corrosion current and half-cell potential.

 BERKE ON CORROSION OF REINFORCING STEEL 171

currently accepted test for water-soluble chloride is de- immersed in the sea. Corrosion of the anode supplies electrons
scribed in ASTM Water-Soluble Chloride in Mortar and Con- to the steel, preventing its corrosion. A similar system was used
crete (C 1218). in a research project a decade ago [47] to protect a deteriorat-
ing interstate highway bridge deck. It was considered a success
by the researchers, but the technique has not gained popular
Repairs to Deteriorated Structures acceptance among transportation agencies.
Systems that employ impressed current, while not ex-
Bridges and parking structures in areas of deicer use, along tremely popular, have met with somewhat better success. The
with marine structures of all sorts, are constantly exposed to first instance of the use of such a system was in 1974, on a
corrosive conditions and, therefore, are the usual structures bridge deck in California [48]. Since that time, hundreds of
needing repair and rehabilitation. Typically, these repairs con- bridge decks have been so treated, but serious difficulties
sist of removal of spalled and delaminated concrete (suffi- remain [49]. Controlling current densities and maintaining
ciently below the top reinforcing steel so that it will be encap- uniform potentials continue to be basic problems, although
sulated within the repair concrete), replacement of the several viable anode systems have been developed. Impressed
concrete, and some means for restricting the entry of water current systems have not been put to use commercially on
and salt solutions. In most instances, long-term durability of prestressed structures due to the possibility of hydrogen gen-
such repairs has not been evaluated. In extreme cases, the eration [50], which can give rise to hydrogen embrittlement. A
structure must be demolished. In others, cathodic protection general early review of cathodic protection was presented by
may be a viable alternate. two British authors [51]. Additional information can be found
After the deteriorated concrete has been removed and in a paper by Broomfield and a SHRP Review [52,53], as well as
replaced, some form of waterproofing must be applied. If it is the proceedings from several NACE annual meeting symposia
not, new delaminations will soon form in the original concrete on cathodic protection [54].
surrounding the repairs. This is called the “ring anode effect.”
Starving the concrete for water is the most effective way to New Steels
reduce the corrosion rate, outside of cathodic protection. Over-
lays of specialty concretes, such as latex-modified, low-water ce- Stainless steel bars and cladding are now available and ASTM
ment ratio, or silica fume-containing, can be used, but they are is working on a standard. Results are significantly improved
expensive. Difficulties in achieving bond with the substrate over standard reinforcing bars in testing, but dependent upon
are sometimes encountered. Thin proprietary membranes of which alloys are used [55,56]. Though corrosion is reduced, it
polyurethane alone, or combined with an epoxy-sand wearing is not necessarily prevented [57].
course, are also used with good success. However, they are Steel processed to have untransformed nano-sheets of
expensive and must be maintained. austenite sandwiching dislocated martensite laths is available.
A simpler and less expensive method is the application of There are no carbon containing phases present, so cathodic
a penetrating sealer such as a silane or a siloxane. Silanes reaction rates are claimed to be low on this steel. The version
penetrate the surface more readily [44] and, therefore, are with the most promise contains over 8 % chromium [57].
expected to last longer. These materials polymerize within the
concrete and also combine with the siliceous portions of the References
cement and aggregates, becoming chemically bound adjuncts.
[1] Bentur, A., Diamond, S., and Berke, N. S., Steel Corrosion in
Because they are inside the concrete, ultraviolet light, a major Concrete—Fundamental and Civil Engineering Practice, E & FN
disintegrator of organic molecules, cannot reach them. Silanes Spon, London, 1998.
are now available as water emulsions or as 100 % silane, rather [2] Uhlig, H. H., The Corrosion Handbook, Wiley, New York, 1948.
than the more traditional alcohol solutions, doing away with [3] Schweitzer, P. A., Corrosion and Corrosion Protection
volatile organic compound concerns. Handbook, Marcel Dekker, New York, 1983, pp. 31 and 502.
Patch areas often fail due to shrinkage of the patching ma- [4] Mansfield, F., Corrosion Mechanisms, Marcel Dekker, New
terials as they are restrained by concrete around the patch [45]. York, 1987, p. 211.
Reducing the strength of the concrete in the patch or reducing [5] Clear, K. C., “Evaluation of Portland Cement Concrete for
shrinkage can help to improve performance. Permanent Bridge Deck Repair,” Report No. FHWA-RD-74-5,
Federal Highway Administration, February 1974.
[6] “Rapid Determination of the Chloride Permeability of
Cathodic Protection Concrete,” AASHTO T277, American Association of State
Highway and Transportation Officials.
Cathodic protection is truly the only known method of com- [7] Whiting, D., “Rapid Determination of the Chloride Permeabil-
pletely preventing corrosion of reinforcing steel embedded ity of Concrete,” Final Report No. HWA/RD-81/119, Federal
Highway Administration, Washington, D.C., August 1981.
in chloride-contaminated concrete. A metal is cathodically
[8] Whiting, D., “Permeability of Selected Concretes,” Permeability
protected when sufficient current is applied to polarize the
of Concrete, SP-108, American Concrete Institute, Detroit, MI,
cathodes to the open circuit potential of the anodes [46]. 1988, pp. 195–222.
This can be accomplished easily and safely, in constantly [9] Pfeifer, D. W., “Corrosion Protection for Concrete Structures:
wet exposures, with a sacrificial anode system. The technique The Past and the Future,” Civil Engineering Practice, Vol. 6, No.
has been used for many years on concrete piles by the state of 1, Boston Society of Civil Engineers, 1991.
Florida, and as part of the overall corrosion protection systems [10] Scali, M. J., Chin, D., and Berke, N. S., “Effect of Microsilica and
for offshore drilling platforms. An anode of zinc or magnesium Fly Ash Upon the Microstructure and Permeability of
is connected to the reinforcing cage through a lead wire, then Concrete,” Proceedings of the 9th International Conference on
172 TESTS AND PROPERTIES OF CONCRETE

Cement Microscopy, International Cement Microscopy Associa- National Research Council, Final Report C-101, SHRP-S/FR-92-
tion, TX, 1987, pp. 375–397. 107, Washington, DC, 1992.
[11] Pfeifer, D. W., Landgren, J. R., and Zoob, A., “Protective Systems [29] Perenchio, W. F., “Durability of Concrete Treated with Silanes,”
for New Prestressed and Substructure Concrete,” Federal Concrete International, November 1988, pp. 34–40.
Highway Administration, Report No. FHWA/RD-86/193, [30] Stark, D. and Perenchio, W., “The Performance of Galvanized
Washington, DC, April 1987. Reinforcement in Concrete Bridge Decks,” Final Report, Project
[12] Neville, A. M., Properties of Concrete, Wiley, New York, 1963. No. 2E-206, Construction Technology Laboratories, Skokie, IL,
[13] Tucker, G. R., U.S. Patent No. 2052586, September 1, 1936. July 1974–October 1975.
[14] Jeknavorian, A. A., Roberts, L. R., Jardine, L., Koyata, H., and [31] Nagi, M., Corrosion 2005, Paper No. 05264, NACE International,
Darwin, D. C., “Condensed Polyacrylic Acid-Aminated Polyether Houston, TX, 2005.
Polymers as Superplasticizers for Concrete,” 5th CANMET/ACI [32] Transportation Research Circular, “Epoxy-Coated Reinforcement
International Conference on Superplasticizers and Other in Highway Structures,” No. 403, March 1993, Transportation Re-
Chemical Admixtures in Concrete, V. M. Malhotra, Ed., ACI search Board, National Research Council, Washington, DC.
Special Publication, SP-173-4, 1997. [33] Kessler, R. J. and Powers, R. G., “Corrosion of Epoxy-Coated
[15] Helmuth, R., Fly Ash in Cement and Concrete, Portland Cement Rebar Keys Segmental Bridges,” Corrosion Report 88-8A,
Association, Skokie, IL, 1987. Florida Department of Transportation, 1988.
[16] Malhotra, V. M., “Properties of Fresh and Hardened Concrete [34] Sagues, A. A., “Mechanism of Corrosion of Epoxy-Coated
Incorporating Ground Granulated Blast Furnace Slag,” Supple- Reinforcing Steel in Concrete,” Report No. FL/DOT/RMC/0543-
mentary Cementing Materials for Concrete, V. M. Malhotra, 3296, University of South Florida, 1991.
Ed., Ottawa, Canadian Government Publishing Centre, 1987. [35] Verbeck, G. J., “Mechanisms of Corrosion of Steel in Concrete,”
[17] Virmani, Y. P., Clear, K. C., and Pasko, T. J., “Time-to-corrosion Corrosion of Metals in Concrete, SP-49, American Concrete
of Reinforcing Steel in Concrete Slabs. Volume 5: Calcium Institute, Detroit, MI, 1975, pp. 21–38.
Nitrite Admixture or Epoxy-Coated Reinforcing Bars as [36] Moore, D. G., Klodt, D. T., and Hensen, R. J., “Protection of Steel
Corrosion Protection System,” Report No. FHWA/RD-83/012, in Prestressed Concrete Bridges,” Report 90, National Coopera-
Federal Highway Administration, Washington, DC, 1983. tive Highway Research Program, Washington, DC, 1970.
[18] Berke, N. S. and Rosenberg, A., “A Technical Review of Calcium [37] Perenchio, W. F., Fraczek, J., and Pfeifer, D. W., “Corrosion
Nitrite Corrosion Inhibitor in Concrete,” Transportation Protection of Prestressing Systems in Concrete Bridges,” NCHRP
Research Record 1211, Transportation Research Board, Report 313, Transportation Research Board, Washington, DC,
Washington, DC, 1989, pp. 18–27. February 1989.
[19] Berke, N. S. and El-Jazairi, B., “The Use of Calcium Nitrite as a [38] Novokshchenov, V., “Condition Survey of Prestressed Concrete
Corrosion Inhibiting Admixture to Steel Reinforcement in Bridges,” Concrete International, Vol. 11, No. 9, September
Concrete,” in Corrosion of Reinforcement in Concrete, C. L. 1989, pp. 60–68.
Page, K. W. J. Treadaway, and P. B. Bamforth, Eds., Elsevier [39] Perenchio, W. F., “The Condition Survey,” Concrete Interna-
Applied Science, New York, 1990, pp. 571–585. tional, Vol. 11, No. 1, January 1989, pp. 59–62.
[20] Berke, N. S. and Hicks, M. C., “Predicting Long-term Durability [40] Escalante, E., Whitenton, E., and Qiu, F., “Measuring the Rate
of Steel Reinforcement in Concrete with Calcium Nitrite of Corrosion of Reinforcing Steel in 6Concrete—Final Report,”
Corrosion Inhibitor,” Cement and Concrete Composites, NBSIR 86-3456, National Bureau of Standards, Washington, DC,
Vol. 26, 2004, pp. 191–198. October 1986.
[21] Issacs, H. S. and Ryan, M. P., “Current Distribution and [41] Bennett, J., et al., Corrosion 92, Paper 191, NACE International,
Dissolution Mechanisms During Localized Corrosion of Steel in Houston, TX, 1992.
Alkaline Environments,” presented at Conference on [42] Nagi, M., “Corrosion Evaluation of Reinforced Concrete
Understanding Corrosion Mechanisms in Concrete: A Key to Bridges,” Structures Congress, American Society of Civil
Improving Infrastructure Durability, at the Massachusetts Engineers, ASCE, New York, 2005.
Institute of Technology, 27–31 July, 1997, Cambridge, MA. [43] Hope, B. B., Page, J. A., and Poland, J. S., “Determination of the
[22] Berke, N. S., Pfeifer, D. W., and Weil, T. G., “Protection Against Chloride Content of Concrete,” Cement and Concrete Research,
Chloride-induced Corrosion,” Concrete International, Vol. 15, No. 5, September 1985, pp. 863–870.
December 1988, pp. 44–55. [44] Kottke, E., “Evaluation of Sealers for Concrete Bridge
[23] Berke, N. S., Hicks, M. C., and Folliard, K. J., “Systems Approach Elements,” BRI/Reports/287PD010, Alberta Transportation and
to Concrete Durability,” Duability of Concrete, SP-170, Utilities, Edmonton, Alberta, Canada, August 1987.
American Concrete Institute, 1997, pp. 1293–1316. [45] Poston, R. W., Kesner, K. E., Emmons, P., and Vaysburd, A. M.,
[24] Berke, N. S., Hicks, M. C., Malone, J., and Rieder, K. A., “Holistic “Performance Criteria for Concrete Repair Materials, Phase II
Approach to Corrosion of Steel in Concrete,” Concrete: Laboratory Results,” Technical Report REMR-CS-57, U.S. Army
Material Science to Application, SP-206, American Concrete Corps of Engineers, April 1998.
Institute, 2002, pp. 137–155. [46] Mears, R. B. and Brown, R. H., “A Theory of Cathodic
[25] Elsener, B., “Corrosion Inhibitors for Steel in Concrete,” Protection,” Transactions, Electrochemical Society, 1938.
European Federation of Corrosion Publications, No. 35, W. S. [47] Whiting, D. and Stark, D., “Galvanic Cathodic Protection for
Manney and Son, Ltd., 2001. Reinforced Concrete Bridge Decks—Field Evaluation,” NCHRP
[26] Berke, N. S., Aldykiewicz, A. J., Jr., and Li, L., “What’s New in Report No. 234, Transportation Research Board, Washington,
Corrosion Inhibitors, Structure,” July/August 2003, pp. 10–12. DC, June 1981.
[27] Pfeifer, D. W. and Scali, M. J., “Concrete Sealers for Protection [48] Stratfull, R. F., “Experimental Cathodic Protection of a Bridge
of Bridge Structures,” NCHRP Report No. 244, Transportation Deck,” Transportation Research Record No. 500, Transportation
Research Board, Washington, DC, 1981. Research Board, Washington, DC, 1974, pp. 36–44.
[28] Whiting, D., Borge, O., Nagi, M., and Cady, P., “Condition [49] Schell, H. G. and Manning, D. G., “Research Direction in
Evaluation of Concrete Bridges Relative to Reinforcement Cathodic Protection for Highway Bridges,” Material
Corrosion—Vol. 5: Methods for Evaluating the Effectiveness of Performance, National Association of Corrosion Engineers,
Penetrating Sealers,” Strategic Highway Research Program, Houston, TX, October 1989, pp. 11–15.
 BERKE ON CORROSION OF REINFORCING STEEL 173

[50] Hope, B. B. and Poland, J. S., “Cathodic Protection and [54] NACE International, Houston, TX.
Hydrogen Generation,” ACI Materials Journal, Vol. 87, No. 5, [55] McDonald, D. B., Sherman, M. R., Pfeifer, D. W., and Virmani, Y.
September–October 1990, pp. 469–472. P., “Stainless Steel Reinforcing as Corrosion Protection,”
[51] Wyatt, B. S. and Irvine, D. J., “A Review of Cathodic Protection Concrete International, May 1995, pp. 65–70.
of Reinforced Concrete,” Materials Performance, National [56] Flint, G. N. and Cox, R. N., “The Resistance of Stainless Steel
Association of Corrosion Engineers, Houston, TX, December Partly Embedded in Concrete to Corrosion by Seawater,”
1987, pp. 12–21. Magazine of Concrete Research, Vol. 40, No. 142, 1988,
[52] Broomfield, J., “Field Survey of Cathodic Protection on North pp. 13–27.
American Bridges,” Materials Performance, Vol. 31, No. 9, [57] Darwin, D., Browning, J., Nguyen, T. V., and Locke, C., Jr.,
September 1992, pp. 28–33. “Mechanical and Corrosion Properties of a High-Strength, High
[53] “Cathodic Protection of Reinforced Concrete Elements—State Chromium Reinforcing Steel for Concrete,” SM Report
of the Art Report,” SHRP S337, Strategic Highway Research Pro- No. 66, South Dakota DOT Office of Research, Pierre, SD, 31
gram, National Research Council, Washington, DC, 1993. March, 2002.
17
Embedded Metals and Materials Other Than
Reinforcing Steel

Bernard Erlin1

Preface General Condition

THE INITIAL WORK ON THIS CHAPTER WAS DONE Moisture is usually necessary for the chemical degradation of
by Hubert Woods, a concrete consultant who is now deceased. any material. It can be in the form of water vapor, liquid water,
The chapter was published 27 years ago in ASTM STP 169A and, or solutions where water is the solvent. Concrete is never com-
with some modifications, in the subsequent two volumes, 169B pletely dry because it contains air voids and capillary voids that
and 169C. A current search of the literature for the effects of hold moisture in vapor form, sufficient in many cases, to main-
concrete on embedded nonferrous metals reveals little new in- tain a relative humidity of about 80 %, which is the moisture
formation than reported previously in this chapter. What has equilibrium of the paste. Furthermore, although a concrete
been located is now included. may be relatively dry, carbonation of the portland-cement
The basic principals of chemistry and physics do not paste will release water when carbonation occurs, as shown by
change. The response of materials to their environment is the the following:
same now as it was when this chapter was originally written.
And, in spite of lessons learned or, because of a peculiar envi- CaSiO2
xH2O  CO3 → CaCO3  SiO2  xH2O
ronmental exposure, a material may perform in an unusual
way. An understanding and caution should always be exercised Concrete is wet when mixed and completely saturated with wa-
in material usages. Perhaps that understanding and caution ter for some time after hardening. There usually is more mois-
has guided the use of embedded materials other than rein- ture than needed to hydrate the cementitious materials so there
forcing steel so that there is little about their adverse perform- remains water to be lost to the environment. The distance
ance reported in the literature. If that is the case, then a value through which free water must move to an external surface
of the principals that extend the service life of those embedded where it can evaporate dictates the residual moisture content of
materials has become a code of use. the concrete at any given time. The rate of drying also depends
on ambient humidity of air in contact with the concrete, but
Introduction even at low ambient humidities, a long drying time may be
needed to lower the concrete humidity to a point where corro-
Metals other than conventional reinforcing steel are some- sion of embedded materials is slowed dramatically or stopped.
times used in conjunction with concrete. Emphasis will be Thus, concrete that will eventually be dry may be at an internal
given here to the possible degradable aspects relative to their relative humidity long enough to cause corrosion of susceptible
use and conditions that may render them unserviceable. materials, and, of course, concrete exposed continuously or fre-
The materials to be described include metals and other in- quently to a damp environment may remain moist enough to
organic and organic substances. Among the metals are alu- support corrosion. The free moisture in concrete can stay in va-
minum, lead, copper and copper alloys, zinc, special alloys of por form or as pore solutions that facilitate transport of soluble
iron, monel metal, stellite, silver, and tin. Among the inorganic chemical substances, such as oxygen, calcium hydroxide, alka-
materials are glass, asbestos, and recycled concrete; and among lies, sulfates, and chlorides, toward embedded materials and of
the organic materials are a variety of plastics and wood and simi- any soluble corrosion products away from the materials. It also
lar cellulose materials. Fiber-reinforced concrete is gaining in increases the electrical conductivity of concrete, thus aiding any
use, so information about some fibers has been included. tendency for electrochemical corrosion of metals.
A general condition necessary for the chemical degrada- The relatively long drying time of concrete is not usually
tion of any material in concrete is exposure to moisture, and a appreciated. The rates of drying of 15  90  90 cm (6  36
short discussion on that aspect precedes the discussions of ma-  36 in.) slabs of normal-weight concrete are shown in Table 1
terial behavior. [1]. The table shows that at an environmental relative humidity

1
Petrographer, The Erlin Company, Latrobe, PA 15650.

174
 ERLIN ON EMBEDDED METALS 175

TABLE 1—Rates of Drying 15  90  90 cm

(6  36  36-in.) Slabs of Normal-Weight
Concretea
Drying Time to Reach Various Relative
Humidities in the Concrete Slab
Environmental
at Middepth, days
Relative
Humidity (RH) 90 % RH 75 % RH 50 % RH

10 18 80 620
35 30 110 840
50 36 240 …
75 36 … …

a
From [1].

of 50 %, 36 days are needed to lower the mid-depth relative hu-

midity to 90 %, and 240 days are needed to reach 75 %. Addi-
tional tests show that when the mid-depth relative humidity
reaches 90 %, the relative humidity at a point only 1.9 cm
(3/4 in.) from an exposed surface is 87 %. The moisture condi-
tion of concrete required to support active electrochemical
corrosion of aluminum and other susceptible metals is not
known with any accuracy; but it seems almost certain that such
corrosion could proceed, other conditions being favorable, at
90 % relative humidity and probably also at 60 % relative hu-
midity, though at a much lower rate.
Concrete made with lightweight aggregates dries more
slowly than concrete made with normal-weight aggregates. Fig. 1—Concrete spalled by corroding aluminum conduit
(from Ref 5).
Thicker sections, such as beams and columns, will dry more
slowly than thin sections, such as walls or elevated floor slabs.
Indeed, today lightweight aggregate is used for internal curing.
dium [4] and corrosion of aluminum balusters embedded in
Aluminum concrete have been reported [5]. Several dozen additional
cases of concrete cracking due to corrosion of embedded alu-
Aluminum has been given widespread attention because it has minum conduit, posts, and window frames in contact with
been the cause of problems in the past. These prompted labo- concrete have occurred. In every case, chloride was present. An
ratory studies designed to describe conditions necessary for its example of concrete spalled over corroded aluminum conduit
corrosion and methods for circumventing its distress. is shown in Fig. 1.
In fresh concrete, aluminum reacts principally with the al- The laboratory investigations by Monfore and Ost [6] show
kali hydroxides derived from hydration of portland cement. One that calcium chloride concentration, alkali content of portland
reaction product is hydrogen gas, and for this reason aluminum cement, electrical coupling of steel and aluminum, and the ratio
powder sometimes is used to form lightweight cellular concrete of steel area to aluminum area are all interrelated in the corro-
(an invention usually attributed to Thomas A. Edison, who de- sion of the aluminum and the subsequent cracking of concrete.
veloped it for insulation purposes). In smaller amounts, the gas In their studies, 15-cm (6-in.) concrete cubes were prepared us-
provides slight expansion of grout and has been used, for exam- ing cements of high and low alkali contents in which pieces of
ple, for bedding machinery base plates. Aluminum in rod, sheet, nominal 1.25-cm (1⁄ 2-in.) aluminum conduit was embedded 1.25
or pipe form reacts much less vigorously than the powdered cm (1⁄ 2 in.) from one face. C-shaped sheets of mild steel were also
metal because of the lesser surface area exposed to the alkalies, embedded in the cubes. The steel and aluminum were externally
and the reactions will continue until the metal is totally reacted. connected in some tests and not connected in others. Various
Tests carried out by Jones and Tarleton [2] indicate that amounts of calcium chloride additions were used. The cubes
the corrosion of embedded aluminum can crack concrete un- were removed from their molds at 24 h, coated with a curing
der unfavorable circumstances. It has also been shown that the compound, and then stored at 23°C (73°F) and 50 % relative
situation can be worse if concrete contains purposefully intro- humidity for 28 days and observed regularly for cracks. After 28
duced calcium chloride (chloride ions are a strong electrolyte), days, the aluminum pieces were removed, cleaned, and weighed.
and much worse if it also contains reinforcing steel that is in The principal results of the studies are given in Table 2. Several
direct contact with the embedded aluminum. important findings were noted:
Wright [3] described a case of corrosion of sufficient • All cubes that cracked contained calcium chloride.
severity to cause collapse of aluminum conduit in reinforced • Cubes containing 1 % flake calcium chloride (CaCl2

concrete made using calcium chloride. Extensive concrete 2H2O) by weight of cement did not crack, but metal losses
spalling over aluminum conduit in the Washington D.C. Sta- in these cubes were as high as, or higher than, those in
176 TESTS AND PROPERTIES OF CONCRETE

TABLE 2—Corrosion of 6063 Aluminum Conduit Embedded in 15-cm (6-in.) Concrete Cubes Stored
at 50 % Relative Humidity for 28 Daysa
Ratio of Steel Loss in
Cement CaCl2
2H2O, Area to Surface
Alkalies as % by weight of Aluminum Days to Thickness,
Cement Na2O, % cement Area Electrodes Cracking milsb

C 0.24 0 28 uncoupled no crack 0.16

2 14 coupled 5 0.92
2 28 coupled 4 1.2
4 7 coupled 3 1.6
4 14 coupled 3 2.3
4 28 coupled 3 2.4
0 28 uncoupled no crack 0.09
2 14 uncoupled no crack 0.07
2 28 uncoupled no crack 0.10
4 7 uncoupled no crack 0.07
4 14 uncoupled no crack 0.07
4 28 uncoupled no crack 0.04
0 0 ... no crack 0.09
4 0 ... no crack 0.04
D 0.89 0 28 coupled no crack 0.12
1 3.5 coupled no crack 0.54
1 7 coupled no crack 0.77
1 14 coupled no crack 1.0
1 28 coupled no crack 0.85
2 3.5 coupled 3 1.0
2 7 coupled 3 1.4
2 14 coupled 4 1.4
2 28 coupled 4 1.6
4 3.5 coupled 2 1.5
4 7 coupled 2 1.7
4 14 coupled 2 2.2
4 28 coupled 7 3.3
0 28 uncoupled no crack 0.33
2 14 uncoupled no crack 0.06
2 28 uncoupled no crack 0.06
4 7 uncoupled no crack 0.07
4 14 uncoupled no crack 0.09
4 28 uncoupled no crack 0.08
0 0 ... no crack 0.17
4 0 ... no crack 0.05

NOTE—Conversion factor: 1 mil  2.54  105 m.

a
From [5].
b
Calculated from weight losses.

other cubes that cracked. (Exposure to a damper environ- contents up to the end of the 28-day tests. When the total
ment might have caused cracking to occur.) electrical flow during 28 days was plotted against the amount
• With no calcium chloride and no electrical coupling, the of corrosion measured as loss in thickness, straight curves of
corrosion was greatest with the high alkali cement. varying shapes were obtained with the slopes depending on the
• With a 2 or 4 % calcium chloride addition, and with met- amount of calcium chloride. The steeper slopes occurred for
als electrically coupled, corrosion was slightly greater the higher amounts of calcium chloride.
when the higher alkali cement was used. Tests somewhat similar to those reported by Monfore and
• With calcium chloride and metals electrically coupled, cor- Ost were done on 31-cm (12-in.) concrete cubes by Wright and
rosion generally increased with increasing ratio of steel Jenks [7]. With electrically coupled steel and aluminum (area
area to aluminum area and invariably increased as the ratio 10:1), cracks did not occur when calcium chloride addi-
amount of calcium chloride increased. tions were not used. However, cracks did occur at various ages;
• Cubes that cracked did so within seven days. after 61 days with 1.1 % flake calcium chloride by weight of
In the case of coupled metals, considerable galvanic cement to eight days with 5.7 % flake calcium chloride by
currents occurred in the circuit connecting the aluminum and weight of cement.
steel. Some measured currents are shown in Fig. 2 as a McGeary [8] also found that: (a) for aluminum conduit elec-
function of time and calcium chloride content. Current flow trically coupled to steel, cracks did not occur in the encasing
progressively increased with increasing calcium chloride concrete when chlorides were not present; (b) the cause of
 ERLIN ON EMBEDDED METALS 177

patching using epoxy or epoxy polysulfide systems, or both;

and (d) engagement of an experienced patching specialist to
complete the repairs.
The results of these various investigations and of field ex-
perience show that reinforced concrete is likely to crack and
spall from corrosion of embedded aluminum if calcium chlo-
ride was used, and coatings that insulate the aluminum from
the concrete are commercially available and practical for use.
In view of chemical similarities between calcium chloride
and sodium chloride, it seems evident that the latter would also
facilitate corrosion of aluminum. Sodium chloride is the prin-
cipal constituent of sea salt, and it therefore seems prudent not
to use aluminum in concrete exposed in, or near, seawater. The
role of chloride when galvanic cells (e.g., steel and aluminum
in electrical contact) are present is that of an electrolyte, a
necessary part of a galvanic cell. Other salts that hydrolyze and
provide strong electrolytes should also be damaging.
Fig. 2—Effect of calcium chloride on galvanic current
(from Ref 5). Based upon laboratory studies and field experiences, it is
prudent not to use calcium chloride in concrete in which alu-
minum will be used or where concrete will be exposed to
corrosion of the aluminum was due unquestionably to galvanic chlorides.
cells activated when chlorides were present in the concrete; (c)
corrosion of the aluminum was of the cubic type, which is also Lead
referred to as intergranular corrosion; and (d) there was an
enrichment of chloride at the aluminum conduit surface. Lead has a high resistance to certain chemical actions, but in
Tests [6] were done to determine the effectiveness of contact with damp concrete it is attacked by the calcium hy-
several different coatings on aluminum embedded in 1.25-cm droxide component of portland cement hydration and be-
(1⁄ 2-in.) concrete cubes made with cement having an alkali comes converted to lead oxide or to a mixture of lead oxides.
content of 0.89 % and containing 4 % calcium chloride by If the dampness persists, the attack will continue. A lead pipe,
weight of cement. The results are shown in Table 3. The data for example, may be destroyed in a few years. If the lead is elec-
demonstrate that a silicone coating was ineffective; that a trically coupled to reinforcing steel in concrete, galvanic action
lacquer (Lacquer B) prevented cracking within 28 days but may accelerate the attack [9a], in which case the rate of corro-
permitted some corrosion; and that Lacquer C and Bituminous sion may be several millimeters per year.
Coatings A and D were each effective in preventing both Lead partially embedded in concrete, and thus partially ex-
corrosion and cracking. posed to air, is susceptible to corrosion because of the differen-
Other tests [8] indicate that the following coatings provide tial electrical potential that results. In the presence of water, the
resistance to galvanic corrosion: certain bitumens, epoxies, flu- air-exposed portion becomes a cathode with respect to the em-
idized-bed plastics, certain metallic pigmented coatings, and bedded portion that becomes an anode. A galvanic cell forms,
alkyd and phenolic materials. and corrosion and deterioration of the embedded lead result.
McGeary [8] found the following to be effective repair pro- A protective coating or covering should be used when lead
cedures for concrete cracked because of corroded aluminum: pipe or cable sheaths are to be embedded in concrete. Bitumi-
(a) removal of all loose concrete; (b) removal of all corrosion nous coatings have been used successfully. Synthetic plastic
products adjacent to the embedded aluminum surface; (c) and other organic coatings or sleeves, which are themselves

TABLE 3—Effect of Protective Coatings on Corrosion of Aluminum

Conduit Embedded for 28 Days in 15-cm (6-in.) Concrete Cubes
Containing 4 % Calcium Chloride and Steel Electrically Coupled to
the Aluminuma
Protective Loss in Surface
Coating Material Thickness, mil Days to Cracking Thickness, milb

None ... 3 2.5

Silicone ... 2 2.1
Lacquer B 1 no crack 0.47
Lacquer C 2 no crack nil
Bitumen A 5 no crack nil
Bitumen D 15 no crack nil

NOTE—Conversion factors: 1 mil  2.54  105 m; 1 in.  2.5 cm.

a
From [5].
b
Calculated from weight losses.
178 TESTS AND PROPERTIES OF CONCRETE

unaffected by damp concrete, are also appropriate. Lead in

contact with concrete should be protected by suitable coatings
or otherwise isolated from contact with the concrete [9b]. Also,
lead and lead salts very strongly retard portland cement
hydration.
There appears little, if any, likelihood of concrete itself be-
ing damaged by corrosion of lead because of the softness of the
metal and its capability for absorbing stress.

Copper and Copper Alloys

Copper will not corrode in concrete unless soluble chlorides are
present. Copper pipes are used successfully in concrete except
under unusual circumstances where ammonia is present [10].
Very small amounts of ammonia, and possibly nitrates, can Fig. 3—Corrosion and perforation of galvanized steel
cause stress corrosion cracking. Brass wall ties have reportedly from under concrete slab (from Ref 11).
failed by stress corrosion, and manganese bronze bolts have
sheared below their ultimate strength. However, because such
phenomena can occur under circumstances unrelated to con- embedded galvanized steel reinforcement has been used in
crete, it is not clear what role, if any, the concrete plays. Very lit- concrete exposed in marine and other environments where chlo-
tle systematic work has been reported on the behavior of copper rides are present (e.g., bridge decks and columns, and docks).
and its alloys in contact with concrete, probably because these Galvanized reinforcing steel has not always been successful in
metals have given satisfactory service under such conditions. stopping corrosion of the steel, and thus there are conflicting
Copper, brass, red brass, bronzes, aluminum-bronze, and reports about the benefit, if any, of the coating [13–15]. Stark
copper-silicon alloys embedded in concrete have good resist- [16] investigated embedded galvanized reinforcement in seawa-
ance to corrosion [11,12]. ter-exposed concretes that were from 7–23 years old. The con-
When copper is connected or adjacent to steel reinforce- crete cores examined were from mean tide and above high tide
ment and an electrolyte such as chloride is present, corrosion zones. Chloride concentrations to depths of 10.1 cm (4 in.) were
of the steel due to galvanic action is likely to occur. Thus, un- always exceedingly high at about the level of the galvanized bars
der such circumstances, it is desirable to insulate the copper [from 1.9–6.4 kg/m3 (3.2–10.7 lb/yd3)]. Concrete cover over the
with suitable coatings. galvanized bars was from 5.7–13.3 cm (214⁄ –514⁄ in.). The amount
of galvanized coating remaining was 92–100 %. A summary of
Zinc the data is given in Table 4. Unpublished data by Hime and Erlin
for galvanized bars threaded through brick hollow cores packed
Zinc reacts chemically with alkaline materials, but normally in
with mortar of 12-year-old masonry facades containing exceed-
concrete the reaction is superficial and may be beneficial to
ingly high chloride contents revealed that the galvanize coating
bond zinc (galvanized) to steel. The primary chemical reaction
was completely corroded. Galvanized wall ties of brick masonry
with calcium hydroxide is
facade construction can undergo similar corrosion when chlo-
Zn  Ca(OH)2 → CaZnO2  H2↑ (1) rides are present.
Galvanized corrugated steel sheets often are used as per-
When zinc is used in concrete, it is generally as a galvanized manent bottom forms for concrete roof, bridge deck, or floor
coating for steel. The reaction is self-limiting, and products of construction. Both satisfactory and unsatisfactory perform-
reaction are not voluminous; consequently, damaging stresses ances have been reported. Figure 3 shows one instance of mul-
are not created. tiple perforation and corrosion of such sheets under a roof
Although good concrete normally provides a nearly ideal slab [10]. Most of the corroded spots were dry to the touch, but
environment for protecting embedded steel from corroding, some were moist and had acidic pH values of 2.7–4.8. Analysis
of the corrosion protuberances indicated the presence of iron,
zinc, and chloride. Further investigation of this and other simi-
lar cases showed that in each instance calcium chloride was
TABLE 4—Galvanize Coating Remaining on used in the concrete so that chlorides were in direct contact
Reinforcing Bars After Indicated Years of with the galvanized sheets. This abetted the corrosion by
Concrete Exposure to Seawater chemical action on the zinc and by increased electrical con-
ductivity of the concrete. The chemical reaction is probably
Years of Age Galvanize Coating Remaining, %

7 98 Zn  CaCl2  2H2O → CaZnO2  2HCl  H2↑ (2)

8 100
10 95 This chemical action produces hydrochloric acid, which explains
10 96 the observed acidity of the corrosion product. It is probable that
10 99 this reaction takes place only after local depletion of calcium hy-
12 92 droxide by the reaction expressed in Eq 1. Other reported zinc
23 98 corrosion products are zinc oxide and zinc hydroxy-chloride.
Admixtures containing major amounts of chloride should
not be used in concrete that will be exposed to moisture and
 ERLIN ON EMBEDDED METALS 179

contains, or is in contact with, galvanized steel. It is also advis- cut from the panels and tested using accelerated methods in
able to keep chloride-containing solutions from permeating the laboratory had expansions of 0.2 % after three months of
the concrete. testing. The glass aggregate removed from the concrete, tested
Zinc galvanize can corrode when in contact with relatively using the procedures in the ASTM Test Method for Potential
fresh concrete. The corrosion phenomenon results in pitting of Reactivity of Aggregates (Chemical Method) (C 289), was found
the zinc surface due to reactions of the zinc and alkalies in the to be deleteriously reactive. Other reports [21] have also indi-
cement paste. Thus, zinc-coated steel forms, for example, in cated the deleterious behavior of reactive glass.
precast or cast-in-place concrete, may cause concrete surface Before glass is used in portland-cement concrete, it
disfiguration due to contact of the zinc. Contact of galvanize to should be tested using the methods provided in the Appendix
steel in concrete in the presence of chloride and other elec- of ASTM Specification for Concrete Aggregates (C 33), and
trolytes can cause galvanic corrosion. specifically those given in Paragraph X1.1.2 of ASTM C 289
Passivation of zinc by use of chromate dips has been re- and Paragraph X1.1.3 of the ASTM Test Method for Potential
ported to be effective in protecting galvanized products. The Alkali Reactivity of Cement-Aggregate Combinations (Mortar-
dips are solutions of sodium or potassium dichromate acidi- Bar Method) (C 227).
fied with sulfuric acid. Chromate dips on galvanized bars or Nondeleterious reactive glass is manufactured and is avail-
400 ppm chromate in the mixing water also prevent hydrogen able for use in concrete.
evolution in fresh concrete.
Concrete in which galvanized reinforcing steel is located Wood
close to galvanized forms has a tendency to stick to the forms.
A chromate treatment, such as previously described, has been Current trends in the use of new or unusual materials in con-
used as a method for avoiding that problem. crete are due to the emphasis on conservation of energy and
Use of corrosion inhibitors, such as calcium nitrite, im- utilization of wastes and by-products. Among the materials
proves the corrosion resistance of zinc in concrete [10a]. proposed for use in concrete is wood (including bamboo,
fibers, bark, jute, cotton, and rice stalks and hulls).
Other Metals Problems incidental to the use of natural cellulosic materi-
als have included adverse effects of sugars on concrete setting
The following metals have been reported to have good resist- and degradation of fibers due to the high concrete alkalinity.
ance to corrosion in concrete: stainless steels, chrome-nickel Further, high differential thermal coefficients of expansion of
steels, chromium-aluminum-silicon steels, cast silicon-iron, some of these materials and unaccommodative volume changes
alloyed cast iron, nickel, chrome-nickel alloys, iron-chrome- can cause cracks to develop. These factors are perhaps foremost
nickel alloys, monel metal, stellite, silver, and tin [12]. The in precluding the use of many cellulosic materials.
resistance of some of these metals to corrosion may be affected Additional problems incidental to the use of natural-
by the presence of “corrosion promoters” such as soluble cellulosic materials include swelling upon moisture absorption
chlorides. Monel metal and Type 316 stainless steel are well and subsequent shrinkage after drying, and chemical degrada-
known for their resistance to the effects of sodium chloride tion due to contact with calcium hydroxide solutions. Prior
and other constituents of seawater and should work well in treatment of the cellulosic materials, such as impregnation or
concrete. The 300 series of stainless steel will corrode in the coating techniques, or carbonation of the portland-cement
presence of chloride and when temperatures are above 140°F paste [22], are possible ways for improving their utilization.
(60°C). Special circumstances might justify the use of these Sawdust, wood pulp, and wood fibers have been incorpo-
more costly metals. rated in mortars and concretes, and timbers have been embed-
Nickel and cadmium-coated steel will not corrode in chlo- ded in or placed in intimate contact with concrete in composite
ride-free concrete if the coatings are continuous [17]. However, constructions. The use of fresh untreated sawdust, wood chips,
the corrosion resistance of these materials becomes question- or fibers in concrete commonly results in very slow setting and
able if chlorides are present in the concrete or in solutions that abnormally low strength because of interference with normal
permeate the concrete [18]. setting and hardening processes by carbohydrates, tannins, and
possibly other substances in the wood. The amount of such
Glass substances differs with wood species and from time-to-time with
place-of-origin within a single species. Softwoods generally give
Glass sometimes is embedded in mortar or concrete as artifi- less trouble in this respect than hardwoods.
cial aggregate [19,20] used for decorative or aesthetic pur- Many admixtures and wood treatments have been proposed
poses, as reinforcing as a substitute for steel, as wall blocks or or used to circumvent the influence of wood constituents on set-
tiles, and as frameless windows or lights. Some glasses are re- ting and hardening. Addition of hydrated lime to the mixture, in
active with alkalies in portland cement paste and form alkali- an amount equivalent to one third to one half of the cement by
silica gel. The resulting expansion may cause severe damage to volume, has been found effective in overcoming this action [23].
the glass or the concrete or both [21]. The treatment is usually effective with mixed softwoods, except
Whenever glass is to be used in concrete, it should always when a high proportion of larch or Douglas fir is present.
be tested to ensure that it would be chemically stable in the al- Five percent calcium chloride by weight of the cement is
kaline concrete environment and not cause deterioration due sometimes added, as well as hydrated lime. With woods of high
to alkali-silica reaction. For example, waste bottle glass used as tannin or carbohydrate contents, the addition of lime with or
aggregate in a decorative concrete facing of outside-exposed without calcium chloride is not effective. Other treatments that
pre-cast concrete panels caused sufficient expansion to warp have been suggested include soaking in sodium silicate solu-
the panels. The warping was due to expansion of the facing, tions, moistening the wood with 1 % sulfuric acid for 4–14 h
which was bonded to a non-expansive concrete backing. Prisms then neutralizing with “milk of lime,” and treating it with 37 %
180 TESTS AND PROPERTIES OF CONCRETE

TABLE 5—Resistance of Plastics to Strong Alkalies [26]

Class of Material Resistance

Polyethylene excellent
Polymethyl methacrylate poor
Polypropylene excellent to good
Polystyrene excellent
Polystyrene acrylonitrile excellent to good
Polytetrafluoroethylene excellent
Polytrifluorochloroethylene excellent
Polyvinyl chloride and polyvinyl chloride vinyl acetate excellent
(rigid)
Polyvinyl chloride and polyvinyl chloride vinyl acetate fair to good
(plasticized)
Saran (monofilament grade) fair to good
Epoxy (unfilled) excellent
Melamine (formaldehyde) poor
Phenol (formaldehyde) poor
Polyester styrene-alkyd poor
Urea (formaldehyde) poor

aluminum chloride solution or 50 % zinc-chloride solution in a which causes dissolution of lipins, and decomposition, chiefly of
rotary barrel with beater. A treatment found by Parker [23] to pentosans, to a smaller extent of lignin, and least of all to
be effective with all sawdust woods evaluated consists of the cellulose. The most suitable wood for embedment is said to be
following consecutive steps: (a) boiling in water, (b) draining pine or fir, preferably of a type with high resin content [24].
and washing with water, (c) reboiling with a 2 % water solution
of ferrous sulfate, and (d) draining and rewashing. Plastics
Concrete made with wood aggregate has considerably
greater volume change on wetting and drying, or simply with The use of plastics in concrete and concrete construction has
changes in external humidity, than concrete made with min- increased significantly. Plastic products are now being used as
eral aggregates. If the element is restrained, drying may lead to pipes, conduit shields, sheaths, chairs, waterstops, and joint
cracking. If drying is not uniform, the element may warp. The fillers. Their compatibility with concrete is thus important.
pretreatments mentioned previously have only a small The principal chemicals in concrete that could conceiv-
influence on these volume changes. Various methods have ably attack plastics are calcium hydroxide, sodium hydroxide,
been used to reduce the changes in volume with changes in and potassium hydroxide, which create a minimum pH of 12.4.
moisture. Some of these methods involve encasement of the The following plastic groups have excellent resistance to these
wood particles or of the finished product in a material of low alkalies at 24°C (75°F) [25]: polyethylene, styrene copolymer
permeability to moisture; but the details of such treatments rubber-resin blends poly (vinyl chlorides), Types I and II, and
and the results achieved have not, in general, been revealed. polytetrafluoroethylene.
Timbers embedded in concrete sometimes have been Another source, Ref [26], provides information on the re-
observed to deteriorate. The harm is done by calcium hydroxide, sistance of plastics to strong alkalies (Table 5).

TABLE 6—Some Applications of Fibers in Concrete

Type of Use Type of Fiber

Cast-in-place and precast bridge deck units; overlays; steel and polypropylene
structural bridge deck elements; pavement; dolosse;
boats; poles; tunnel linings; rock slope stabilization;
highway, street, and airfield pavements; sluices;
industrial floors
Maintenance and repairs to dams, slabs, pavements, steel and polypropylene
bridges, culverts, etc.
Industrial floors, slab overlays, pile caps, pavements steel, glass, and polypropylene
Miscellaneous small precast items (burial vaults, steps, glass and polypropylene
garden units, etc.)
Pipes, sheets, boards, fence posts, panels, piles, asbestos, glass cellulose, and
building panels polypropylene
 ERLIN ON EMBEDDED METALS 181

Fibers Glass fibers formulated with zirconia were thought to be

resistant to alkali degradation [34,35]. However, the loss of
In a broad sense, fibers used for reinforcing concrete are small fracture resistance caused by chemical changes within glass-
versions of conventional steel reinforcement, and they provide fiber reinforced concrete (which occurs prior to five-year ex-
a similar service. For fibers to be useful and effective, they posures in wet environments) was as dramatic as previously
must enhance the physical attributes of concrete and be experienced. The embrittlement associated with that exposure
durable. Among the desirable characteristics that fibers can im- is thought to result from reformation of cement hydration
part to concrete are increased flexural strength, increased re- products in and tightly around the fibers so that fiber encase-
sistance to fatigue and impact, and increased fracture resist- ment is enhanced and failures are due to fracture rather than
ance. Applications of fiber-reinforced concrete are shown in pullout of fibers [36]. That explanation does not entirely satisfy
Table 6. A report on fiber-reinforced concrete prepared by the all aspects of the situation, however, and further research is
American Concrete Institute [27] provides little information re- needed [37,38]. New glass fibers coated with chemically inert
garding their corrosion characteristics. coating have been used in an attempt to overcome the strength-
Among the materials that have been used as fibers in con- loss phenomenon.
crete are steel, glass, polyethylene, polypropylene, nylon, as-
bestos, and carbon. Organic fibers are considered to rely upon Organic Materials
mechanical interlock, and steel and glass upon chemical inter-
actions, to develop adhesion to the cement matrix. Steel fibers Polymer fibers, such as nylon and polypropylene, improve im-
have also been produced as crimped or deformed fibers so that pact strength of concrete, but not tensile or flexural strengths,
they mechanically interlock in addition to chemically bond. because they have a low modulus of elasticity. The polypropy-
Some glass fibers can corrode and also cause embrittlement as lene fibers are in common use. They increase impact strength
discussed in the section on Glass Fibers. and minimize the concrete potential for early cracking, such as
that due to plastic shrinkage.
Steel
Good durability of steel-fiber reinforced concrete has been re- Asbestos
ported [28,29]. However, steel fibers are subject to the same
type of corrosion as reinforcing steel, and thus the durability The use of asbestos in portland cement-based products is a
of concrete made with steel fibers and exposed in environ- thing of the past because of its potential carcinogenic effect.
ments chemically aggressive to the steel may be poor. The de- However, asbestos has been used in conjunction with portland
terioration of steel is enhanced, particularly by chlorides. It has cement since about 1900. Asbestos minerals are naturally
been reported that steel-fiber reinforced concrete used for occurring and include a variety of different materials that fall
pavements, bridge decks, and other similar usages where chlo- into two mineral groups: amphibole and chrysotile. The
ride-deicing agents are used, can suffer reductions in strength asbestos minerals may differ in composition but have the
[30,31]. If cracks are present, rusting will be initiated at loca- common composition of magnesium silicate, and alkalies—
tions where the fibers are exposed in the cracks. Obviously, magnesium and iron may freely substitute for each other.
chlorides should not be a component of admixtures used in Although the durability of embedded asbestos fibers is not
steel fiber-reinforced concrete, and the concrete should have as considered a problem, some corrosion of the fibers may occur
low a permeability as workability and water-ratio will permit. after prolonged periods because of chemical reactions with
In studies of steel-fiber reinforced beams in a simulated calcium hydroxide in the portland-cement paste. The chemical
seawater environment for eight years, rusting of fibers to reactions have been found to be topo-chemical and occur on
depths of 0.34 cm (11⁄ 6 in.) occurred [32]. Because of the lack of fiber surfaces and on cleavage planes. Corrosion products are
electrical continuity between the steel fibers, electro-chemical probably magnesium hydroxide, magnesium carbonate, mag-
corrosion cannot develop on a large scale. nesium silicate hydrates, and low lime calcium silicate hydrates
Steel fibers and steel filings are used in proprietary floor [39]. The alteration of the asbestos fibers does not adversely
toppings to enhance wear characteristics desirable in ware- affect properties of the concrete because the alteration is
houses. The volume percentage of metal used is significantly restricted to fiber surfaces.
greater than for steel fiber-reinforced concrete. Deterioration
of surfaces has resulted because of rusting of the fibers. In one Concrete
such case, chloride equivalent to a purposeful calcium chloride
addition (1 % by mass of portland cement) was present in the The current emphasis on recycling of materials and the razing
concrete on which a topping shake containing steel fibers had of old concrete structures and pavements have resulted in the
been applied. This situation points out the desirability of not use of old concrete as aggregates for new concrete. For such
using calcium chloride, either as an admixture to concrete mix- use, the old concrete is crushed and graded and used either
tures or, along with sodium chloride (rock salt), as deicing alone or blended with other aggregates [40–42]. Crushed
chemicals. concrete has had principal use as a base for concrete made
with conventional aggregates.
Glass Fibers The workability of concrete made with recycled concrete
aggregates is about that of concrete made with conventional
Glass fibers can be sensitive to alkalis in portland cement paste. aggregates. However, its compressive strength is about 75 %,
Early work using glass fibers reflected the adverse effects of al- and its modulus of elasticity is about 60 % of that for conven-
kali-silica reactivity [33]. tional concrete [43].
182 TESTS AND PROPERTIES OF CONCRETE

Buck reported that concrete made with recycled aggre- [11] Halstead, P. E., “Corrosion of Metals in Buildings. The Corrosion
gate had resistance to cyclic freezing [44]. Aggregates for use of Metals in Contact with Concrete,” Reprint No. 38, Great
in concrete must possess certain necessary physical and Britain Cement and Concrete Association; reprinted from
chemical characteristics, such as those required in ASTM C Chemistry and Industry, 24 Aug. 1957, pp. 1132–1137.
33. Because the concrete has seen prior service in a given [12] Rabald, E., Corrosion Guide, Elsevier, New York, 1951.
exposure and environment for lengthy periods does not [13] Stark, D. and Perenchio, W., “The Performance of Galvanized
Reinforcement in Concrete Bridge Decks,” Portland Cement
necessarily mean that it will perform similarly as concrete ag-
Association, Skokie, IL, Oct. 1975.
gregate in other service and exposures. For example, concrete
[14] Griffin, D. F., “Effectiveness of Zinc Coatings on Reinforcing
made using chemically unstable aggregate and used in an Steel in Concrete Exposed to a Marine Environment,” Technical
elevated structure where the exposure has been dry will prob- Note N-1032, U.S. Naval Civil Engineering Laboratory, Port
ably respond differently when used in a moist environment. Hueneme, July 1969, June 1970, and June 1971.
Concrete made with a calcium chloride addition, if used in [15] “Use of Galvanized Rebars in Bridge Decks,” Notice, No. 5,
reinforced nonchloride-containing concrete, could promote 140.10, Federal Highway Administration, Washington, DC, 9
corrosion of reinforcing steel. These are two examples that July 1976.
demonstrate the need for establishing suitable specifications [16] Stark, D., “Galvanized Reinforcement in Concrete Containing
and requirements to ensure that concrete made with recycled Chlorides,” Construction Technology Laboratories, Project No.
concrete will provide adequate service. Concrete from ZE-247, Portland Cement Association, Skokie, IL, April 1978.
building construction has been used for fill material as well [17] Freedman, S., “Corrosion of Nonferrous Metals in Contact with
as for concrete aggregate. Contamination of the crushed con- Concrete,” Modern Concrete, Vol. 36, Feb. 1970.
crete aggregate by adhering gypsum plaster from plastered [18] “Guide to Durable Concrete,” Journal, ACI Committee 201,
walls can cause internal sulfate attack. American Concrete Institute, Dec. 1977, pp. 591–592.
[19] “Waste Materials in Concrete,” Concrete Construction, Vol. 16,
No. 9, Sept. 1971, pp. 372–376.
References [20] Johnson, C. D., “Waste Glass as Coarse Aggregate for Con-
crete,” Journal of Testing and Evaluation, Vol. 2, No. 5, 1974,
[1] Abrams, M. S. and Orals, D. L., “Concrete Drying Methods and
pp. 344–350.
Their Effect on Fire Resistance,” Research and Development
Bulletin 181, Portland Cement Association, Skokie, IL, 1965. [21] Waters, E. H., “Attack on Glass Wall Tiles by Portland Cement,”
Report S-41, Australia Commonwealth Scientific and Industrial
[2] Jones, F. E. and Tarleton, R. D., “Effect of Embedding Alu-
Research Organization, Division of Building Research, 1956.
minum and Aluminum Alloys in Building Materials, Research
Paper 36, Great Britain Department of Scientific and Industrial [22] “Fiber Reinforced Concrete,” Centre de Recherches de Pont-a-
Research, Building Research Station, London, 1963. Morrison, French Patent No. 1,369,415, 14 Aug. 1964.
[3] Wright, T. E., “An Unusual Case of Corrosion of Aluminum [23] Parker, T. W., “Sawdust-Cement and Other Sawdust Building
Conduit in Concrete,” Engineering Journal, Canada, Vol. 38, Products,” Chemistry and Industry, 1947, pp. 593–596.
No. 10, Oct. 1955, pp. 1357–1362. [24] Dominik, W. and Hans, M., “Action of Cement on Wood,”
[4] “Spalled Concrete Traced to Conduit,” Engineering News— Przemsyl Cheniczny, Vol. 22, 1938, pp. 74–82.
Record, Vol. 172, No. 11, March 1964, pp. 28–29. [25] Seymour, R. B. and Steiner, R. H., Plastics for Corrosion-Resistant
[5] Copenhagen, W. J. and Costello, J. A., “Corrosion of Alu- Applications, Reinhold, New York, 1955.
minum Alloy Balusters in a Reinforced Concrete Bridge,” Ma- [26] Handbook of Chemistry and Physics, 41st ed., Chemical Rubber
terials Protection and Performance, Vol. 9, No. 9, Sept. 1970, Company, Cleveland, 1959–1960.
pp. 31–34. [27] “State-of-the-Art Report on Fiber Reinforced Concrete,”
[6] Monfore, G. E. and Ost, B., “Corrosion of Aluminum Conduit Journal, Title No. 70-65, ACI Committee 544, American
in Concrete,” Journal, Portland Cement Association, Re- Concrete Institute, Detroit, MI, Nov. 1973, pp. 729–744.
search and Development Laboratories, Vol. 7, No. 1, Jan. [28] Shroff, J. K., “The Effect of a Corrosive Environment on the
1965, pp. 10–22. Properties of Steel Fiber Reinforced Portland Cement,” Masters
[7] Wright, T. E. and Jenks, I. H., “Galvanic Corrosion in Concrete thesis, Clarkson College of Technology, Potsdam, New York,
Containing Calcium Chloride,” Proceedings, American Society Sept. 1966.
of Corrosion Engineers, Journal Structural Division, Vol. 89, ST [29] “A Status Report on Fiber Reinforced Concretes,” Concrete
5, Oct. 1963, pp. 117–132. Construction, Jan. 1976, pp. 13–16.
[8] McGeary, F. L. “Performance of Aluminum in Concrete [30] Bateson, G. B. and Obszarski, J. M., “Strength of Steel Fiber
Containing Chlorides,” Journal, American Concrete Institute; Reinforced Concrete Exposed to a Salt Water Environment,”
Proceedings, Vol. 63, Feb. 1966, pp. 247–264. Materials Performance, July 1977, Technical Note, p. 48, Paper
[9a] “Corrosion of Lead by Cement,” Concrete Construction No. 27, NACE Annual Conference, Houston, March 1976.
Engineering, Vol. 44, No. 11, 1949, p. 348; Dodero “Accelerat- [31] “Corrosion Behavior of Cracked Fibrous Concrete,” Materials
ing Reactions of the Corrosion of Lead in Cement,” Metaux & Systems and Science Division of the Construction Engineering
Corrosion, Vol. 25, No. 282, 1949, pp. 50–56. Research Laboratory, Project OK1, Task 03, Work Unit 006, 1975
[9b] Biczok, I., “Concrete Corrosion and Concrete Protection,” (unpublished).
Akadimiai Kiado, Budapest, 1964. [32] Lankard, D., private communications, formerly Battelle
[10a] Mange, C. E., “Corrosion of Galvanized Steel in Contact with Memorial Institute, now with Materials Laboratory, Inc., 1977.
Concrete Containing Calcium Chloride,” Preprint, National As- [33] Marek, R. et al., “Promising Replacements for Conventional
sociation of Corrosion Engineers, 13th Annual Conference, St. Aggregates for Highway Use,” HRB NCHRP, Final Report,
Louis, MO, 1957. Project 4–10, University of Illinois, Urbana, IL, Jan. 1971.
[10b] Berke, N. S., and Rosenburg, A., “Technical Review of Calcium [34] Ironman, R., “Stronger Market Seen for Glass-Fiber Concrete,”
Nitrite Corrosion Inhibitor in Concrete,” Transportation Re- Concrete Products, Jan. 1976, pp. 42–44.
search Record 1211, Transportation Research Board, Washing- [35] Tallentire, A. G., “Glass Fibre Cement Applications,” Precast
ton, DC, 1989, 18 pp. Concrete, Feb. 1977, pp. 95–97.
 ERLIN ON EMBEDDED METALS 183

[36] Grimer, F. J. and Ali, M. A., “The Strengths of Cements Times,” Proceedings, Symposium on Fiber Reinforced Cement
Reinforced with Glass Fibers,” Magazine of Concrete Research, and Concrete, RILEM, Paris, 1975, pp. 269–276.
Vol. 21, No. 66, March 1969, pp. 23–30. [40] “Rubble Recycling Saves Time, Energy, and the Environment,”
[37] Majundar, A. J., “Properties of Fiber Cement Composites,” Rock Products, Vol. 80, No. 5, May 1977, pp. 107–108.
Proceedings, Symposium on Fiber Reinforced Cement and [41] “Reusing Concrete,” Engineering News-Record, Vol. 186, No.
Concrete, RILEM, Paris, 1975, pp. 279–313. 22, June 1971, p. 15.
[38] Cohen, E. B. and Diamond, S., “Validity of Flexural Strength [42] “Recycled Slab is New Runway Base,” Highway and Heavy
Reduction as an Indication of Alkali Attack on Glass in Fibre Construction, Vol. 120, No. 7, July 1977, pp. 30–33.
Reinforced Cement Composites,” Proceedings, Symposium on [43] Frondistou-Yamas, S., “Waste Concrete as Aggregate for New
Fiber Reinforced Cement and Concrete, RILEM, Paris, 1975, pp. Concrete,” Journal, American Concrete Institute, Vol. 74, No. 8,
315–325. Aug. 1977, pp. 373–376.
[39] Opoczky, L. and Pentek, L., “Investigation on the ‘Corrosion’ of [44] Buck, A. D., “Recycled Concrete,” Highway Research Record No.
Asbestos Fibers in Asbestos Cement Sheets Weathered for Long 430, 1973, p. 108.
18
Abrasion Resistance

Karl J. Bakke1

Preface third type of wear is primarily grinding and cutting actions.

The action of the abrasive particles carried by the flowing wa-
ter, of course, is controlled largely by the velocity of the water,
IN PREPARATION OF THIS CHAPTER, THE CONTENTS the angle of contact, the type of abrasive material, and the gen-
of the 4th edition were drawn up. The author acknowledges the eral surrounding conditions. The fourth type of wear, cavita-
authors of the first four editions, H. L. Kennedy and M. E. tion, is caused by the abrupt change in direction and velocity
Prior, M. E. Prior, R. O. Lane, and Tony C. Liu, respectively. The of a liquid to such a degree that the pressure at some point is
current edition will review and update the topics as addressed reduced to the vapor pressure of the liquid. The vapor pockets
by the previous authors, introduce new technology that has so created, upon entering areas of high pressure, collapse with
been developed, and include up-to-date references. a great impact, which eventually causes pits or holes in the con-
crete surface. Discussion of cavitation resistance of concrete is
Definitions and Types of Abrasion beyond the scope of this paper; therefore, it will not be cov-
ered. An excellent source of information on cavitation erosion
ASTM Terminology Relating to Erosion and Wear (G 40) defines can be found in Refs 4 and 5.
abrasion as “wear due to hard particles or hard protuberances
forced against and moving along a solid surface.” Abrasion re- Factors Affecting Abrasion Resistance
sistance, according to American Concrete Institute (ACI) Com-
mittee on Cement and Concrete Terminology (ACI 116), is the Factors that may affect the resistance of concrete to abrasive
“ability of a surface to resist being worn away by rubbing and action should be considered in the design and construction of
friction” [1]. concrete surfaces that are to withstand abrasion due to rub-
Wear of concrete surface can be classified as follows [2]: bing, scouring, sliding, impact, scraping, attrition, percussion,
• Wear on concrete floors. gouging, or cutting from mechanical or hydraulic forces.
• Wear on concrete road surfaces due to heavy trucking, and Frequently, the failure of concrete to resist abrasion can be
automobiles, with and without studded snow tires or traced to cumulative effects such as soft aggregate, inadequate
chains (attrition, plus scraping and percussion). compressive strength, improper curing or finishing, or over-
• Wear on hydraulic structures such as dams, spillways, manipulation during finishing of concrete surface. The follow-
bridge piers, and abutments due to the action of abrasive ing discussion of significant factors relative to concrete resist-
materials carried by flowing water (erosion). ance to abrasion illustrates the importance of the proper
• Wear on concrete dams, spillways, tunnels, and other wa- selection, composition, and application of concrete based on
ter-carrying systems where high velocities and negative the specific type of service condition.
pressure are present. This is generally known as cavitation
erosion (cavitation). Quality of Aggregates
The first type of wear to concrete floors can vary widely Studies by Liu [6] and Laplante et al. [7] indicated that the
from a rubbing action to high impact and hard-wheel traffic of abrasion resistance of concrete is strongly influenced by the
forklifts. Concrete floor wear is greatly increased by the hardness of its coarse aggregate. The abrasion resistance can
introduction of foreign particles, such as sand, metal scraps, or be increased appreciably by the use of maximum amount of
similar materials. ACI 302.1R-04, Table 2.1 classifies floors, dense, hard coarse aggregates such as traprock, chert, granite,
including design and construction considerations based on the or metallic aggregate. For example, the abrasion loss of con-
intended use [3]. The second type of wear is caused by a rub- crete containing limestone aggregate has approximately twice
bing action, plus an impact-cutting type of wear. This is brought as much as that of the concrete containing chert [6].
about by the use of chains on automobile and truck tires or Abrasion tests carried out by Liu [6] indicated that no cor-
metal vehicle wheels. As the wheel revolves, it brings the metal relation existed between abrasion resistance of concrete and
into contact with the concrete surface with considerable im- the quality of the coarse aggregate as determined by ASTM Test
pact, a process that tends to cut the surface of the concrete. The Method for Resistance to Degradation of Small-Size Coarse

1
Sales Engineer, Kalman Floor Company, Evergreen, CO 80439, Subcommittee Chairman ASTM C09.62. Abrasion Testing.

184
 BAKKE ON ABRASION RESISTANCE 185

Aggregate by Abrasion and Impact in the Los Angeles Machine tested to measure compressive strength is not a measure of the
(C 131). As can be seen from Fig. 1, the Los Angeles abrasion slab surface strength [12].
losses are approximately equal for soft aggregate, such as lime-
stone, and relatively hard aggregate, such as chert. However, Mixture Proportioning
the abrasion losses of the concrete containing these aggregates Abrasion test results by many researchers clearly indicated
vary widely; the abrasion loss of the limestone concrete is that, for a given aggregate, the abrasion resistance of concrete
much more than that containing chert. A similar finding was increases with a decrease in water-cement ratio [6]. A maxi-
reported by Smith [8]. mum water-cement ratio of 0.45 has been specified by the U.S.
Schuman and Tucker [9] pointed out that the shape of ag- Army Corps of Engineers for concrete subject to abrasion in
gregate particles regulates the water requirements for placing hydraulic structures.
and finishing and has a direct influence on the abrasion resist- The effects of various admixtures, including air-entraining
ance of concrete. Angular to subangular-shaped aggregate is and water-reducing admixtures and retarders, on the abrasion
known to improve bond, usually resulting in increased abra- resistance of concrete could not be established conclusively on
sion resistance. the basis of the literature review. The general trend in most
For concrete subject to light-to-medium abrasion, good studies reflects an increase in abrasion loss proportional to in-
quality aggregates meeting the requirements of ASTM Spe- crease in air content for any given water-cement ratio (w/c). As
cification for Concrete Aggregates (C 33) are generally the compressive strength is adversely affected by increased air
acceptable. Heavy-duty floors and slabs exposed to more contents, so will the resistance of concrete to abrasion. Water-
severe abrasive action demand well-graded hard mineral reducing admixtures, by virtue of their positive effect on the
aggregates [2]. w/c and consequently on increases in strength, tend to produce
a concrete of improved abrasion resistance.
Compressive Strength
Witte and Backstrom [10], as researchers before and after Concrete Types
them, considered the compressive strength as one of the most Studies by Holland et al. [13] and Laplante [7] indicated that
important factors responsible for the abrasion resistance of adding condensed silica fume and high-range water-reducing
concrete. For the same aggregate and finishing procedure, the admixture to a concrete mixture greatly increases compressive
abrasion resistance of concrete increases with an increase in strength, which, in turn, increases abrasion resistance. The
compressive strength. For example, as shown in Fig. 2, the av- abrasion resistance of silica-fume concrete containing
erage abrasion resistance (the reciprocal of abrasion-erosion relatively soft limestone is similar to that of a high-strength
loss) for concrete containing limestone increases approxi- conventional concrete mixture containing a very hard chert
mately 44 % as the compressive strength increases from 20.7 aggregate [13]. Apparently, for the high-strength silica-fume
MPa (3000 psi) to 62.1 MPa (9000 psi) [6]. concrete, the hardened cement paste assumes a greater role in
However, abrasion resistance is a function of the water-ratio resisting abrasion and as such the aggregate quality becomes
at the surface and of the quantity of quality aggregate, not of the correspondingly less important. These very high-strength
compressive strength of concrete [11]. The 6-in.
12-in. cylinder concretes appear to offer an economical solution to abrasion

Fig. 1—Relationship between resistance of aggregate to abrasion and concrete

abrasion loss.
186 TESTS AND PROPERTIES OF CONCRETE

Fig. 2—Relationship between abrasion resistance and compressive strength.

problems, particularly in those areas where locally available ag- polymer portland cement concrete, methyl methacrylate poly-
gregate otherwise might not be acceptable. mer concrete, and vinyl ester polymer concrete), which all con-
Tests by the U.S. Army Corps of Engineers indicated that tained limestone aggregates, was investigated by Liu [21].
the abrasion resistance of concrete containing fly ash is Among these polymer concretes tested, the vinyl ester polymer
comparable to that of concrete without fly ash, as long as the concrete ranked first in abrasion resistance, followed by
concretes contain the same type of aggregate and have methyl methacrylate polymer concrete, polymer-impregnated
comparable compressive strength at the time of tests. concrete, and polymer portland cement concrete.
Shrinkage compensating concrete, when properly propor-
tioned and cured, has an abrasion resistance from 30 to 40 % Finishing Procedures
higher than portland cement concrete of comparable mix pro- Abrasion resistance of concrete is affected by the finishing
portions [14]. Other studies with shrinkage compensating ce- procedures used. Tests by Kettle and Sadegzadeh [22] indi-
ment have indicated similar results: Nagataki and Yoneyama cated that the abrasion resistance of concrete subjected
[15] and Klieger and Greening [16]. to power finishing is significantly higher than that subjected to
While the addition of steel fibers in the concrete mixture hand finishing. This is attributed to surface compaction and
would be expected to increase the impact resistance of con- to reduction of the w/c of the surface matrix.
crete, fiber-reinforced concrete is less resistant to abrasion Finishing techniques, including wood float, magnesium
than conventional concrete of the same aggregate type and wa- float, steel trowel, and hard-steel trowel finishes, are also
ter-cement ratio [17]. The abrasion losses of fiber-reinforced compared with abrasion resistance of concrete in a study by
concrete as determined by ASTM C 1138 were consistently Fentress [23]. The wood float tends to tear the surface and to
higher than those of the conventional concrete over wide displace the aggregate. The magnesium float, despite ease of
ranges of water-cement ratio and compressive strength. This is finishing, causes a rough-textured surface and a lowering in
attributed primarily to the fact that fiber-reinforced concrete abrasion resistance. Both the steel trowel and hard-steel trowel
generally has less coarse aggregate per unit volume of concrete produce smooth surfaces, closing any existing imperfections
than comparable conventional concrete. and providing excellent resistance to abrasion.
Polymer-impregnated concrete for highway bridges was The key to durable, cleanable concrete slabs can be
investigated by Fowler of the University of Texas [18]. Concrete achieved by proper finishing procedures and in a reduction in
of 34.5- to 44.8-MPa (5000- to 6500-psi) compressive strength, the w/c [24]. Combining these demands in one application,
non-air-entrained, treated with polymer showed considerable Ytterberg demonstrated that a deferred topping finish from
increase in abrasion resistance. Experimental studies by Liu which surface water is removed by a vibratory absorption
[6], Krukar and Cook [19], and Dikeou [20] also demonstrated process yields highest abrasion resistance [24]. The technique
remarkable improvements in concrete properties, particularly requires that after the water of workability has been removed,
in strength and abrasion resistance due to impregnation and the surface is blade-floated, and upon stiffening, troweled.
in-place polymerization of monomers in overlays and in ma- Abrasion tests carried out by Kettle and Sadegzadeh [22], the
ture concrete. U.S. Bureau of Reclamation [25], and the U.S. Army Corps of
The relative abrasion resistance of four different types of Engineers [21] confirmed that vacuum-treated concrete surfaces
polymer concretes (that is, polymer-impregnated concrete, are considerably more resistant to abrasion than surfaces with
 BAKKE ON ABRASION RESISTANCE 187

either nontreated lean or richer concrete. The improvement in [21]. However, problems in field application of surface coat-
abrasion resistance of vacuum-treated concrete is due principally ings have been reported. These have been due primarily to im-
to the reduction of water content in the concrete mixture. This proper surface preparation or thermal incompatibility be-
treatment is most effective on concrete with a high w/c. tween coatings and concrete. More recently, formulations have
Concrete wearing surfaces can be improved by the use of been developed that have coefficients of thermal expansion
cast-on hardeners or “dry shakes” consisting of cement/hard-ag- more similar to that of the concrete substrate [4].
gregate mixtures. Pockets of bleed water should be removed
prior to the application of the dry shake. After the dry shake is Abrasion Test Methods
applied to the concrete, it should be given enough time to absorb
water prior to floating. Once the dry shake has absorbed water, The techniques and test methods that have been employed for
it is power-floated and then steel troweled into the surface with- over a century to evaluate the abrasion resistance of concrete
out the use of additional water to produce a hard, dense topping have attempted with varied success to reproduce the typical
layer. Testing by the Portland Cement Association in 1970 forces detrimental to concrete surfaces. Preceding any stan-
demonstrated that abrasion resistance is significantly improved dardization of laboratory methods or of testing machines, re-
using cast-on surface hardeners with hard, steel-trowel, finish as liance was placed initially on the use of rattlers that provide
compared to ordinary concrete without surface hardeners [26]. abrasive action based on the tumbling of steel balls impacting
a test surface [33]. Numerous modified versions of the rattler
Curing principle continued to be the dominant test procedure along
Efficient curing increases the abrasion resistance. A correla- with the revolving metal disk pressed against a small concrete
tion of curing time and abrasion resistance reported by Sawyer specimen [34]. The drill-press-type abrasion machine [35] in
[27] involved a series of tests comprising a wide range of modified form still is used by highway departments and enjoys
cement contents, w/c, and incremental curing. From this study, some popularity because of its simple design.
it is apparent that marked improvement in abrasion resistance Currently, there are four standard ASTM test methods for
can be expected with extended curing time, especially for evaluating the resistance of concrete subjected to various types
surfaces composed of leaner concrete. of abrasive actions. The background, significance, and applica-
A laboratory program conducted by the California Divi- bilities of these ASTM test methods are presented here.
sion of Highways [28] considered the effect on abrasion of
such variables as slump, finishing, curing, and surface ASTM C 418
treatments using linseed oil. Test data indicated, among these
variables, that the greatest abrasion losses encountered were ASTM Test Method for Abrasion Resistance of Concrete by
associated with less-than-adequate curing procedures. Sandblasting (C 418) dates back to 1958 and is based on the
For concrete floor and slab construction using Type I principle of producing abrasion by sandblasting. This proce-
portland cement, at least five days of curing at concrete dure simulates the action of waterborne abrasives and abra-
temperatures of 21°C (70°F) or higher and a minimum of sives under traffic on concrete surface. It performs a cutting
seven days at temperatures of 10–21°C (50–70°F) are required action that tends to abrade more severely the less resistant
to assure adequate abrasion resistance [3]. components of the concrete. Adjustments in the pressure
used and the type of abrasive permit a variation in the sever-
Surface Treatment ity of abrasion that may be used to simulate other types of
Certain chemicals (magnesium and zinc fluorosilicates, abrasion.
sodium silicate, gums, and waxes) serve to prolong the life of This test method was modified by ASTM Subcommittee
older floors and are considered an emergency measure for C09.03.13 on Methods of Testing Concrete for Resistance to
treatment of deficiencies in relatively pervious and soft Abrasion from the original Ruemelin blast cabinet equipped
surfaces that wear and dust rapidly [29]. Treatment with with an injector-type blast gun with high-velocity air jet (Fig. 3).
fluorosilicates densifies and hardens surfaces and improves The control over such variables as gradation of the silica sand,
abrasion resistance of old, worn floors, but the improvement air pressure, rate of feed of the abrasive, distance of the nozzle
by the two types of fluorosilicates is not of equal magnitude from the surface, and the area of the shielded surface is critical.
[30]. While the magnesium fluorosilicate is more resistant to The 1958 standard specified carborundum or silica sand. This
rubbing action simulated best by a revolving-disk-type abrasion procedure was revised in 1964 calling for the explicit use of sil-
machine, the zinc fluorosilicate shows greater resistance to im- ica sand that allows for closer control of the abrasive action. The
pact-type wear as imposed by the dressing-wheel-type machine. 1976 revision specifies the use of oil-base modeling clay of
For poorly cured or porous concrete surfaces, beneficial known specific gravity for the filling of voids in the abraded con-
effects from treatment with linseed oil were demonstrated by crete surface. This permits accurate measurement of abrasion
an increase in abrasion resistance of as much as 30 % [31]. by volume displacement and replaces the old weight-loss deter-
Paints and coatings used to seal concrete surfaces and to mination. Rushing [36] compared the 1958 and 1964 proce-
protect the concrete from the attack of the environment or dures of ASTM C 418 in a laboratory study and found the later
chemicals [32] possess only limited abrasion resistance and revision permits a more accurate determination of abrasion
any test simulation to evaluate their effectiveness is difficult. loss. The current revision now allows for a ceramic nozzle in
However, some coatings, among them vinyl and heavy rubber, addition to cold rolled steel and hardened tool steel.
if properly bonded remain fairly resilient and effective in
protecting concrete from abrasive actions. Several types of ASTM C 779
surface coatings including polyurethane, epoxy-resin mortar,
furan-resin mortar, acrylic mortar, and iron-aggregate toppings For almost two decades, ASTM C 418 remained the only speci-
have exhibited good abrasion resistance in laboratory tests fied standard for testing concrete resistance to abrasion. From
188 TESTS AND PROPERTIES OF CONCRETE

Fig. 3—Sand blast cabinet.

1964 through 1971, ASTM Subcommittee C09.03.13 initiated a dressing-wheel machine, and (c) the ball-bearing machine. All
number of major test programs conducted by independent lab- three machines are portable and adaptable for laboratory and
oratories. The results of these comparative programs estab- in-place field abrasion testing.
lished such parameters as severity and reproducibility, leading The revolving-disk machine (Fig. 4) introduces frictional
to the selection of three abrasion machines and their inclusion forces by rubbing and grinding. Sliding and scuffing is
in one ASTM standard. ASTM Test Method for Abrasion Resis- accomplished by rotating steel disks in conjunction with abra-
tance of Horizontal Concrete Surfaces (C 779) covers three sive grit. The revolving-disk machine was redesigned by Mas-
testing procedures: (a) the revolving-disk machine, (b) the ter Builders Company, Cleveland, Ohio, and was essentially

Fig. 4—Revolving-disk abrasion test machine.

 BAKKE ON ABRASION RESISTANCE 189

patterned after a machine developed by the National Institute

of Standards and Technology. Among the three standard
methods, the revolving-disk machine provides the most re-
producible results. A supply of No. 60 silicon carbide abrasive
is fed to the disks at a rate of 4–6 g/min. A test period of 30
min generally produces significant wear on most concrete
surfaces, but it is recommended to extend the test period to
60 min if information on the longtime abrasion resistance is
desired. Davis and Troxell [37] reported that the depth of
wear for periods of 30–60 min is about the same for ordinary
slabs as for heavy-duty concrete surfaces. A plane surface re-
sulted in either case regardless of the variation in the quality
of the hardened cement paste or the toughness of the aggre-
gate that became exposed as the depth of wear increased. The
abrasive mode of this procedure best simulates wear by light-
to-moderate foot traffic and light-to-medium tire-wheeled traf-
fic or moving of light steel racks, etc.
In contrast to the disks, the dressing-wheel machine (Fig.
5) imparts high concentrated compressive forces with high
impact stresses. The dressing-wheel machine is similar to the
revolving-disk machine in appearance, except for three sets of
seven dressing wheels mounted on horizontal shafts that take
the place of the three rotating steel disks. However, no abra-
sive material is employed with this machine. Initial and in-
termediate measurements of the test path are taken with a Fig. 6—Ball-bearing abrasion test machine.
depth micrometer. Tracking of the dressing wheel normally
leaves a grooved path with the test surface being irregular
and fairly rough as harder aggregate particles stand out from
the softer aggregate particles and mortar that are abraded

more quickly. On ordinary-finished concrete slabs, the dress-

ing wheels produce a depth of wear more than double that
obtained with the revolving-disk machine for the same test du-
ration. Yet, approximately equal depths of wear are obtained
from both machines for the same test period when hard trow-
eled finished floors are tested. The coefficient of variation
established by the Berkeley study [37] for slabs abraded with
the dressing-wheel machine is several times as great as that
for the revolving-disk machine. These results are essentially
concurrent with a parallel study conducted in the Portland
Cement Association laboratories [26]. Abrasion of concrete
induced by the dressing-wheel machine closely simulates the
rolling, pounding, and cutting action of steel wheels or the
effect of studded tires.
Repeated dynamic loading through strong impact, com-
pressive forces, and high-speed rolling constitutes the abrasive
action of the ball-bearing machine (Fig. 6). The ball-bearing ma-
chine operates on the principle of a series of eight ball bearings
rotating under load at a speed of 1000 rpm on a wet concrete
test surface. Water is used to flush out loose particles from the
test path, bringing the ball bearing in contact with sand and
stone particles still bonded to the concrete surface, thus pro-
viding impact as well as sliding friction. During the test, abra-
sion readings are taken every 50 s with a dial micrometer
mounted directly to the supporting shaft allowing readings “on
the fly.” Readings are continued for a total of 1200 s or until a
maximum depth of 3.0 mm (0.12 in.) is reached. In its abrasive
severity, the ball-bearing machine exceeds both the revolving
disk and the dressing wheel, also producing the highest coeffi-
cient of variation among the three procedures (Table 1). The ef-
fects of the ball-bearing machine on concrete surfaces indicate
abrasive action becomes progressively more severe as the test
Fig. 5—Dressing-wheel abrasion test machine. continues, due to soft and hard spots causing the core barrel to
190 TESTS AND PROPERTIES OF CONCRETE

bounce to high angular speed. This test procedure has merit in

that the ball bearings are similar to rolling wheels and are
more typical of actual loadings by steel-wheel traffic to which
a concrete floor may be exposed.
Experience with testing of abrasion by ASTM C 779
demonstrates relatively good reproducibility over a fairly wide
range of concrete surface types and conditions when the
abrasive action is moderate, for instance, through attrition.
More severe abrasion due to percussion, impact, cutting, or
rolling creates erratic action and is less conducive to repro-
ducibility. In general, loss in test accuracy may develop due to
the following factors:
• Rapid wear of dressing wheels, revolving disks, and steel
balls and failure to replace them regularly as specified;
• Failure to remove dust and loose-abraded material
regularly from the test surface;
• Improper selection of a representative test area or
specimen;
• Differences in age levels of concrete at which test results
are compared;
• Arbitrary increase in the length of test or flow of sand; and
• Insufficient number of wear readings.
Excluding these variables, the within-laboratory precision
shown in Table 1 indicates the lowest coefficient of variation is
obtained with the revolving-disk method, increasing in order
with dressing wheels and ball bearings.

ASTM C 944
ASTM Test Method for Abrasion Resistance of Concrete or Fig. 7—Rotating-cutter drill press.
Mortar Surfaces by the Rotating-Cutter Method (C 944) gives
an indication of the relative abrasion resistance of mortar
and concrete based on testing of cored or fabricated speci- ing a constant load on the abrading cutter when using the
mens. This test method has been successfully used in the lever, gear, and spring system of a drill press has been elimi-
quality control of highway and concrete bridges subject to nated by placing a constant load of 98 N (22 lbf) directly
traffic. This method is primarily intended for use on the top upon the spindle that turns the cutter. General practice is to
ends of 152-mm (6-in.) diameter concrete cores, mortar speci- clean the surface occasionally during the test by blowing the
mens, or other samples of concrete of insufficient test area to dust off the test specimen. The reproducibility of test results
permit the conduct of tests by ASTM C 418 and C 779. This has been poor, with the single-operator coefficient of varia-
test method produces a much more rapid abrasive effect than tion of more than 20 %.
the other test methods. The test apparatus is a fairly simple
piece of equipment consisting of a rotating cutter and a drill ASTM C 1138
press or similar device with a chuck capable of holding and
rotating the abrading cutter at a speed of 200 rpm. Figure 7 ASTM Test Method for Abrasion Resistance of Concrete (Un-
shows a rotating-cutter drill press. The difficulty in maintain- derwater Method) (C 1138) was originally developed by Liu
[21] in 1980 for evaluating the resistance of concrete surfaces
subjected to the abrasive action of waterborne particles on
hydraulic structures such as stilling basins and outlet works.
TABLE 1—Within-Laboratory Precision for ASTM adopted this test procedure with some minor modifi-
Single Operator, ASTM C 779 cations in 1989. The apparatus consists of essentially a drill
press, an agitation paddle, a cylindrical steel container that
Coefficient of Acceptable Range houses a disk-shaped concrete specimen, and 70 steel grind-
Variation, percent of Two Results, ing balls of various sizes. The detailed cross-sectional view of
Procedure of meana percent of meana the test apparatus is shown in Fig. 8. The water in the
container is circulated by the immersed agitation paddle that
A. Revolving disk 5.51 15.6
B. Dressing wheel 11.69 33.1
is powered by the drill press rotating at 1200 rpm. The
C. Ball bearing 17.74 50.2 circulating water, in turn, moves the abrasive charges (steel
grinding balls) on the surface on the concrete specimen, pro-
ducing the abrasion effects. Testing, totaling 24 h, generally
a
produces significant abrasion in most concrete surfaces. The
These numbers represent respectively the 1S % and D2S % limits as described in
ASTM Practice for Preparing Precision and Bias Statements for Test Methods for
standard test consists of six 12-h test periods for a total of 72
Construction Materials (C 670), in the section on Alternative Form of the Precision h. Additional testing time may be required for concrete that
Statement. is highly resistant to abrasion. This test method can duplicate
 BAKKE ON ABRASION RESISTANCE 191

Fig. 8—Test apparatus, ASTM C 1138.

well the abrasive action of waterborne particles in the stilling Application of Test Methods
basins. This method is not, however, intended to provide a
quantitative measurement of length of service that may be As discussed in the previous section, ASTM C 418, C 779, C 944,
expected from a specific concrete. and C 1138 offer six distinct procedures that simulate various
The abrasion loss, as determined by the original Corps of degrees of severity and types of abrasive or erosive forces.
Engineers’ test method [21], was expressed as a percentage of Table 2 serves as a general guide for possible applications of
the original mass of the specimen. ASTM Subcommittee the six standard procedures for various categories of abrasion.
C09.03.13 revised the method by calculating the volume loss, This tabulation attempts to correlate the severity of in-place
or the average depth of wear, at the end of the test time wear with the abrasive action particular to each of the test
increment to eliminate differences in results resulting from the methods. However, it is necessary to bear in mind that there is
allowable variations in specimen size. The single laboratory co- overlap in the type of abrasion imparted by these test applica-
efficient of variation made from a single batch of concrete is tions and that a specific service condition may be reproduced
14 %. by more than one procedure.

TABLE 2—Application of Test Procedures

ASTM C 779
Type of Abrasion ASTM C 418 ABC ASTM C 944 ASTM C 1138

Foot traffic or light-to-medium tire-wheeled traffic, etc. X X

Forklift, heavy tire-wheeled traffic, automobile with chains,
heavy steel-wheeled traffic, or studded tires, etc. XX X
Abrasive erosion of waterborne particles on hydraulic structures X X
192 TESTS AND PROPERTIES OF CONCRETE

Conclusion [13] Holland T. C., Krysa, A., Luther, M. D., and Liu, T. C., “Use of Sil-
ica-Fume Concrete to Repair Abrasion-Erosion Damage in the
Kinzua Dam Stilling Basin,” Proceedings, Second International
Concrete materials are susceptible to deterioration due to the
Conference on Fly Ash, Silica Fume, Slag, and Natural Pozzolans
abrasive action of environmental and man-made factors. In in Concrete, Madrid, Spain, 1986.
addition, intrinsic conditions related to properties of concrete [14] ACI 223-98, Standard Practice for the Use of Shrinkage
and construction techniques contribute to the reduction in Compensating Concrete, Manual of Concrete Practice, Part I,
abrasion resistance. Proper selection of concrete-making ma- American Concrete Institute, Farmington Hills, MI.
terials, mixture proportion, curing, and finishing procedures [15] Nagataki, S. and Yoneyama, K., “Studies on Continually
are the basic requirements applying to any concrete sub- Reinforced Concrete and Prestressed Concrete Pavements
jected to abrasion from vehicular traffic or water-borne par- Made with Expansive Cement Concretes,” Klein Symposium on
ticles. Effects of variables on reduced abrasion resistance of Expansive Cement Concretes, ACI SP-38, American Concrete
concrete have proved to be cumulative, which may explain Institute, Farmington Hills, MI, 1973, pp. 131–163.
why only certain portions of a concrete surface may fail in [16] Klieger, P. and Greening, N. R., “Properties of Expansive
resisting abrasion while the remainder gives satisfactory Cement Concretes,” Proceedings, 5th International Symposium
on Chemistry of Cement (Tokyo 1968), Cement Association of
performance.
Japan, Tokyo, 1969, Paper IV -132, pp. 439–379.
There are four ASTM test methods, that is, ASTM C 418, C
[17] Liu, T. C. and McDonald, J. E., “Abrasion-Erosion Resistance of
779, C 944, and C 1138, for evaluating the resistance of con-
Fiber-Reinforced Concrete,” ASTM Cement and Aggregate, Vol.
crete subjected to various types of abrasive actions. These test 3, No. 2, Winter 1981.
methods serve to (1) evaluate, predict, or accept the quality of [18] Fowler, D. W., Houston, J. T., and Paul, D. R., “Polymer Impreg-
concrete surfaces; (2) evaluate specific effects of variables such nated Concrete for Highway Application,” Research Report No.
as concrete-making materials, curing, finishing procedures, 1141, Center for Highway Research, University of Texas, Austin,
surface hardeners, or coating materials; (3) compare various TX, February 1973.
types of concrete surfaces under simulated abrasion condi- [19] Krukar, M. and Cook, J. C., “The Effect of Studded Tires on
tions; and (4) verify products or systems to meet specifications. Different Pavement and Surface Texture,” Washington State
University, Pullman, WA, February 1973.
[20] Dikeou, J. T., “Radiation Polymerization of Monomers,” U.S.
Bureau of Reclamation, Washington, DC, January 1970.
References
[21] Liu, T. C., “Abrasion-Erosion Resistance of Concrete,” Technical
[1] ACI 116R-00, Cement and Concrete Terminology, ACI Manual of Report C-78-4, Report 3, U.S. Army Corps of Engineers,
Concrete Practice, Part 1, ACI Committee 116, American Waterways Experiment Station, Vicksburg, MS, July 1980.
Concrete Institute, Farmington Hills, MI, 2001. [22] Kettle, R. and Sadegzadeh, M., “The Influence of Construction
[2] ACI 201.2R-01, Guide to Durable Concrete, ACI Manual of Procedures on Abrasion Resistance,” Proceedings, Katharine
Concrete Practice, Part 1, ACI Committee 201, American and Bryant Mather International Conference on Concrete
Concrete Institute, Farmington Hills, MI, 2001. Durability, ACI SP-100, Vol. 2, American Concrete Institute,
[3] ACI 302.1R-04, Guide to Concrete Floor and Slab Construction, Farmington Hills, MI, 1987.
ACI Manual of Concrete Practice, Part 2, American Concrete [23] Fentress, B., “Slab Construction Practices Compared by Wear
Institute, Farmington Hills, MI. Tests,” Journal, American Concrete Institute, Farmington Hills,
[4] ACI 210R-87, Erosion of Concrete in Hydraulic Structures, ACI MI, July 1973.
Manual of Concrete Practice, Part 1, ACI Committee 210, [24] Ytterberg, R. F., “Wear Resistance of Industrial Floors of
American Concrete Institute, Farmington Hills, MI, 1990. Portland Cement Concrete,” Journal, American Society of Civil
[5] “Erosion of Concrete by Cavitation and Solids in Flowing Wa- Engineers, Civil Engineering, January 1971.
ter,” Laboratory Report No. C-342, U.S. Bureau of Reclamation, [25] “Erosion Resistance Tests of Concrete and Protective Coatings,”
Washington, D.C., 1947. Concrete Laboratory Report No. C 445, U.S. Bureau of
[6] Liu, T. C., “Abrasion Resistance of Concrete,” ACI Journal, Reclamation, Washington, DC, February 1952.
Proceedings, American Concrete Institute, Farmington Hills, MI, [26] Klieger, P. and Brinkerhoff, C. H., “Cooperative Study on
September–October 1981, Vol. 78, No. 5, pp. 341–350. Methods of Testing for Abrasion of Concrete Floor Surfaces,”
[7] Laplante, P. C., Aitcin, P. C., and Vezina, D., “Abrasion Resis- Research and Development Laboratories, Portland Cement
tance of Concrete,” Journal of Materials in Civil Engineering, Association, Skokie, IL, March 1970.
Vol. 3, No. 1, February 1991, pp. 19–28. [27] Sawyer, J. L., “Wear Test on Concrete Using the German Stan-
[8] Smith, F. L., “The Effect of Aggregate Quality on Resistance of dard Method of Test and Machine,” Proceedings, ASTM Interna-
Concrete to Abrasion,” Cement and Concrete, ASTM STP 205, tional, West Conshohocken, PA, Vol. 57, 1957, pp. 1143–1153.
ASTM International, West Conshohocken, PA, 1956. [28] Spellman, D. L. and Ames, W. H., “Factors Affecting Durability
[9] Schuman, L. and Tucker, J., Jr., “A Portable Apparatus for of Concrete Surfaces,” Materials and Research Department,
Determining the Relative Wear Resistance of Concrete Floors,” California Division of Highways, Sacramento, CA, Report No.
Research Paper No. RP 1252, National Bureau of Standards, M&R 250908-1, BRPD-3-7, 1967.
Washington, DC, 1939. [29] “Surface Treatments for Concrete Floors,” Concrete Informa-
[10] Witte, L. P. and Backstrom, J. E., “Some Properties Affecting the tion ST-37, Portland Cement Association, p. 2.
Abrasion Resistance of Air-Entrained Concrete,” Proceedings, [30] “The Effect of Various Surface Treatments Using Magnesium
ASTM International, West Conshohocken, PA, Vol. 51, 1951, p. and Zinc Fluorosilicate Crystals on Abrasion Resistance of
1141. Concrete Surfaces,” Concrete Laboratory Report No. C 819, U.S.
[11] Ytterberg, R. F., “Shrinkage and Curling of Slabs on Grade,” Bureau of Reclamation, Washington, DC, February 1956.
Concrete International Design & Construction, 3-Part Series, [31] Mayberry, D., “Linseed Oil Emulsion for Bridge Decks,” U.S.
April, May, June 1987. Department of Agriculture, Agriculture Research Service,
[12] ACI 360R-92 (Reapproved 1997) Design of Slabs on Grade, ACI Peoria, IL, September 1971.
Manual of Concrete Practice, Part 5, American Concrete Institute, [32] ACI 515.1R-79, Revised 1985, A Guide to the Use of Water-
Farmington Hills, MI. proofing, Dampproofing, Protective, and Decorative Barrier
 BAKKE ON ABRASION RESISTANCE 193

Systems for Concrete, ACI Manual of Concrete Practice, Part 5, [35] Harris, D. H., “Apparatus for Testing the Hardness of
ACI Committee 515, American Concrete Institute, Farmington Materials,” Chemical Abstract, Vol. 38, 1944, p. 1338.
Hills, MI, 1990. [36] Rushing, H. B., “Concrete Wear Study,” Report PB 183410,
[33] Scofield, H. H., “Significance of Talbot Jones Rattler as Test for Louisiana Department of Highways, Baton Rouge, LA, June
Concrete in Road Slabs,” Proceedings, Highway Research 1968.
Board, Washington, DC, 1925, p. 127. [37] Davis, R. E. and Troxell, G. E., “Methods of Testing Concrete for
[34] Guttman, A., “Abrasion Tests on Concrete,” Chemical Abstract, Resistance to Abrasion—Cooperative Testing Program,”
Vol. 30, 1936, p. 5750. University of California, Berkeley, CA, October 1964.
19
Elastic Properties, Creep, and Relaxation

Jason Weiss1

Preface the load is removed. This, in its simplest form, describes a ma-
terial’s elasticity and the concept can be simply illustrated by
A CHAPTER ON ELASTIC PROPERTIES OF CONCRETE considering a spring. When the spring is pushed or pulled it
first appeared in ASTM STP 169 (1956) authored by L. W. Teller changes length and when sthe load is removed it returns to its
of the U.S. Bureau of Public Roads (now FHWA). Robert E. original position (Fig. 1a). Hooke proposed this concept in
Philleo authored the chapters appearing in ASTM STP 169A 1678 by stating that the deformation of a body () was linearly
(1966) and ASTM STP 169B (1978) under the title “Elastic Prop- proportional to the force that is applied (F ). Over time this pro-
erties and Creep.” The chapter on elastic properties was portionality constant (K ) was termed as a spring constant that
reprinted in ASTM STP 169C as it appeared in ASTM STP 169B. can be written mathematically as
This version is based largely on these earlier chapters; however,
this chapter has been modified to include information on non- F  K
destructive testing, elastic modulus inferred from other test
methods, and early age creep and relaxation measurements. where
Importance of Elastic Properties and Creep
F  the applied force or load
Engineers need to be able to compute deflections of structures, K  the spring constant
to compute stresses from observed strains, to proportion   the deformation
sections, and to determine the quantity of steel required in re-
inforced concrete members. In each of these calculations engi- In actuality, the spring coefficient can be thought of as a
neers need material properties that can relate stress and strain. “stiffness” of the material/structure that is dependent on the
These properties are commonly referred to as the elastic prop- given geometry of body that is loaded as well as the properties
erties. Strictly speaking, the stress-strain response of concrete is of the material. To overcome this limitation the force can be
F
nonlinear and inelastic; however, it is frequently assumed that written in terms of the stress (i.e.,   , where A is the cross-
for low load levels (stresses less than 50 % of the strength and A
sectional area) and the deformation can be written as a strain
strains less than 1000
 in compression and 100
 in tension)

the relationship between stress and strain can be described us- (i.e.,   , where L is the original specimen length) which
L
ing a linear relationship. In this linear relationship the elastic gives rise to the more familiar form of Hooke’s Law for uni-
modulus describes the ratio of the change in stress and change axial loading where stress is related to strain through a pro-
in strain. While the elastic properties can be used to describe portionality constant which is referred to as the elastic modu-
the initial deformation under loading, concrete can exhibit in- lus or Young’s Modulus (Fig. 1b). The elastic modulus is
creased deformations over time due to the presence of a sus- essentially only a function of the material and as such it is in-
tained load. Creep describes the slow, progressive deformation dependent of specimen size and geometry, thereby making it a
of a material under a sustained loading. Relaxation describes material property.
the slow reduction in stress over time due to a system displace-
ment. This chapter will review some of the basic elastic proper-
  
ties, compare these properties with other construction materi-
als, illustrate why concrete may or may not be elastic, discuss
common test methods for obtaining elastic properties and the where
stress strain response, discuss creep and relaxation, and discuss
potential applications and future needs.   the applied stress
E  the elastic modulus
Background Information About Elastic   the strain
Constants and Properties
When stress is applied in a given direction, there are changes
When a load is applied to a material body it deforms. For many in the dimension of the perpendicular directions. The magni-
materials the body will return to its original dimensions after tudes of the lateral strains are different for different materials.
1
Associate Professor, Purdue University, West Lafayette, Indiana.

194
 WEISS ON ELASTIC PROPERTIES, CREEP, AND RELAXATION 195

Fig. 1—a) Conceptual illustration of the relationship between force and

deformation using a stiffness constant (K), and b) the linear relationship between
stress and strain using the elastic modulus (E).

Thus, two parameters are required to describe the elastic 0.5 for nearly all materials. It can be seen that Poisson’s ratio
behavior of a material. The parameters may take many forms, for concrete is lower than that of most metals and it is consis-
but the two most commonly used are elastic modulus and Pois- tent with many ceramic materials.
son’s ratio. Poisson’s ratio is defined in ASTM Standard It is important to note at this time that these two elastic
Terminology Relating to Methods of Mechanical Testing (E 6) properties (i.e., E and
) are generally used to describe the re-
as “the absolute value of the ratio of transverse strain to the sponse of many materials since a more general relationship
corresponding axial strain resulting from uniformly distrib- can be written to account for cases of multi-axial loading.
uted axial stress below the proportional limit of the material.” When an element is subjected to simultaneous normal stresses
The transverse strains are opposite in direction to the axial in each of three axial directions (x, y, z ), the resulting strain
strains and can be described using the following relationship. components can be obtained from the following equations
A new material property (Poissons Ratio,
) is introduced in where the subscripts refer to actions in a specific direction.
the following equation to relate the axial and lateral strains:
1
  [ 
(y
z)]
Axial E
Lateral  
Axial  

E 1
y  [y 
(x
z)]
E
where
1
z  [ 
(x
y)]
E
Axial  the applied stress in the axial direction

 Poisson’s Ratio It is also important to note that once two elastic properties
Axial  the strain in the axial direction are known (i.e., in this case E and
) any other elastic property
E  the elastic modulus, and can be determined. For example, since the elastic modulus and
Lateral  the strain in the lateral direction. Poisson’s ratio are known they can be used to calculate the
shear modulus (G ). The shear modulus, also called the modu-
Table 1 shows typical values for the elastic modulus and lus of rigidity or torsional modulus, is the ratio of shear stress
Possion’s ratio for mature concrete and other commonly used to shear strain. Shear stress is defined in ASTM E 6 as “the stress
construction materials. It can be seen that the elastic modulus or component of stress acting tangential to a plane,” and shear
for concrete is lower than that of most metals while it is slightly strain is defined as “the tangent of the angular change between
higher than that of wood. Poisson’s ratio falls between 0 and two lines originally perpendicular to each other.” It can be

TABLE 1—Elastic Modulus and Poisson’s Ratio of Commonly Used

Construction Materials
Elastic Modulus Poisson’s Ratio

Material x106 psi GPa

Steel, Grade A36 30 207 0.3

Iron 9.6 to 25 66 to 169 0.26 to 0.31
Aluminum 10 to 10.5 69 to 72.5 0.33
Copper 14 to 22 97 to 150 0.30 to 0.35
Concrete 3 to 6 21 to 42 0.18
Wood, Parallel to Grain 1.6 to 2.0 11 to 13.8

Wood, Perpindicular to Grain 0.08 to 0.10 0.55 to 0.69

196 TESTS AND PROPERTIES OF CONCRETE

shown that for an elastic material the following relationship

exists between Young’s modulus of elasticity in shear, and Pois-
son’s ratio:
E

   1
2G
where

 Poisson’s ratio
E  Young’s modulus of elasticity
G  modulus of elasticity in shear

In addition to measuring elastic properties by applying a

mechanical load and measuring the deformation it is impor-
tant to note that the elastic properties can describe the natural
frequency of vibration. The natural frequency of vibration in
an elastic body is proportional to the square root of either the
elastic modulus or the shear modulus, depending on the mode
of vibration. In addition, the velocity with which a compression
wave travels through an elastic body is proportional to the Fig. 2—A stress-strain response of a normal strength
square root of the elastic modulus. concrete in compression.
Also it should be noted that the deformation exhibited by
many materials depends on numerous factors including the A typical compressive stress-strain response of concrete is
magnitude of the load, the rate at which it is applied, and the illustrated in Fig. 2 [1]. While a line has been superimposed on
elapsed time after the load application that the observation is the stress-strain diagram for low stress and strain values it
made. This response is generally known as rheological behav- quickly becomes apparent that the manner in which its modu-
ior. While instantaneous effects are referred to as the elastic re- lus of elasticity is defined is somewhat arbitrary. It can be seen
sponse, time-dependent deformations are commonly referred that in Region I of this diagram the relationship between the
to as creep or relaxation. stress and strain is relatively linear. This can be seen to occur
at stress levels below approximately 50 % of the peak strength
Elastic Properties of Concrete and at strains lower than approximately 1000
. It can be seen
that at higher stresses and strains (Region II) the response be-
Several standard tests exist to determine the mechanical re- comes nonlinear due to the development of cracks between the
sponse of concrete. Compressive testing is typically deter- aggregates and cracks in the paste. After the stress reaches a
mined in accordance with ASTM Standard Test Method for maximum value, the stress strain curve is observed to begin to
Compressive Strength of Cylindrical Concrete Specimens (C soften due to the opening of cracks in Region III [2].
39) to determine the peak strength and ASTM Standard Test It should be noted that concrete has neither a definite pro-
Method for Static Modulus of Elasticity and Poisson’s Ratio of portional limit nor an elastic limit. As a result, various forms of
Concrete in Compression (C 469) to determine the static elas- the modulus are illustrated on the stress-strain curve in Fig. 3.
tic modulus in compression. No standard test currently exists ASTM E 6 defines several different moduli as follows:
to assess direct tensile strength; however, the flexural strength 1. Initial Tangent Modulus—The slope of the stress-strain
can be determined using ASTM Standard Test Method for Flex- curve at the origin.
ural Strength of Concrete (Using Simple Beam with Third- 2. Tangent Modulus—The slope of the stress-strain curve at
Point Loading) (C 78) and ASTM Standard Test Method for any specified stress or strain.
Flexural Strength of Concrete (Using Simple Beam with 3. Secant Modulus—The slope of the secant drawn from the
Center-Point Loading) (C 293). Additionally, ASTM Standard origin to any specified point on the stress-strain curve.
Test Method for Flexural Toughness and First-Crack Strength 4. Chord Modulus—The slope of the chord drawn between
of Fiber-Reinforced Concrete (Using Beam with Third-Point any two specified points on the stress-strain curve.
Loading) (C 1018) is commonly used to test the flexural tough-
ness and first crack strength of fiber-reinforced concrete, Modulus of Elasticity in Compression
though it would also be possible to use this test to determine Since structural concrete is designed principally for compres-
the elastic modulus. sive stresses, by far the greatest amount of work on the elastic
The elastic modulus is one of the most commonly used pa- properties of concrete has been done on concrete in compres-
rameters to describe a material even though many materials do sion. The only ASTM test method for the static modulus of elas-
not exhibit a truly linear stress-strain relationship. Two addi- ticity of concrete, ASTM C 469, is a compressive test method. It
tional terms are generally used to describe the limits of elastic stipulates a chord modulus between two points on the stress
behavior: (1) proportional limit and (2) elastic limit. The pro- strain curve defined as follows: the lower point corresponds to
portional limit is defined in ASTM E 6 as “the greatest stress a strain of 50 millionths (i.e., 50
) and the upper point cor-
which a material is capable of sustaining without any deviation responds to a stress equal to 40 % of the strength of concrete
from proportionality of stress to strain (Hooke’s law).” The at the time of loading. The lower point is near the origin but
elastic limit is “the greatest stress which a material is capable far enough removed from the origin to be free of possible ir-
of sustaining without any permanent strain remaining upon regularities in strain readings caused by seating of the testing
complete release of the stress.” machine platens and strain measuring devices. The upper
 WEISS ON ELASTIC PROPERTIES, CREEP, AND RELAXATION 197

to span an adequate sample of the material. It must not, how-

ever, encroach on the ends of the specimen. This limitation is
established because restraint occurs where the specimen is in
contact with the steel platens of the testing machine. As a result
the strains near the ends of the specimen may differ somewhat
from strains elsewhere in the specimen. ASTM C 469 specifies
that the gage length shall be not less than three times the maxi-
mum size of aggregate nor more than two thirds the height of
the specimen. Half the specimen height is said to be the pre-
ferred gage length.
A convenient device for measuring the strains is a com-
pressometer, such as the one illustrated in Fig. 4. The upper
yoke is rigidly attached to the specimen, whereas the lower
yoke is free to rotate as the specimen shortens. The pivot rod
and dial gage are arranged so that twice the average shorten-
ing of the specimen is read on the dial. This type of device was
used in the first comprehensive investigation of modulus of
elasticity by Walker [4], and it is cited in ASTM C 469 as an ac-
ceptable device. It should be noted, however, that other proce-
dures may exist [5,6]. Because the test is intended to measure
Fig. 3—Various forms of static modulus of elasticity. only time-dependent strains, it is important that the specimen
be loaded expeditiously and without interruption. For this
purpose, an automatic stress-strain recorder is helpful but not
essential. Figure 4 illustrates the use of a linear variable differ-
point is taken near the upper end of the linear behavior and ential transformer (LVDT) displacement transducer in which
near the maximum working stress that is assumed in most de- the deformation, instead of being observed on a dial gage, is
signs. Thus, the determined modulus is approximately the av- indicated and recorded.
erage modulus of elasticity in compression throughout the Although the standard test method is not concerned with
working stress range. the behavior of concrete at stresses above 40 % of the strength,
The 150 by 300-mm (6 by 12-in.) cylinder is the specimen the shape of the stress-strain curve at high stresses is of signifi-
size, a commonly used specimen geometry for the determina- cance in determining the ultimate load-carrying capacity of a
tion of the modulus of elasticity in compression; however, it concrete member [7–9]. When tested under load control, con-
should be noted that 100 by 200-mm (4 by 8-in.) cylinders may crete cylinders fail suddenly, shortly after the maximum load
be common for concretes with smaller aggregates. In order to has been attained.
compensate for the effect of eccentric loading or nonuniform Several different approaches have been used to assess the
response by the specimen, strains should be measured along complete stress strain response of concrete. Hognestad et al.
the axis of the specimen or along two or more gage lines uni- [9] utilized U-shaped specimens where the central portion of
formly spaced around the cylinder. The selection of the gage the specimen was loaded eccentrically to back calculate the
length is important. It must be large in comparison with the stress-strain response. Shah et al. [10] tested steel in parallel
maximum aggregate size so that local strain discontinuities do with a concrete cylinder and subtracted the linear response of
not unduly influence the results, and it must be large enough the steel to obtain the complete stress-strain response of con-

Fig. 4—Compressometer testing details: (a) compressometer testing apparatus, and (b) geometric
relation for calculating strain [3].
198 TESTS AND PROPERTIES OF CONCRETE

crete. While the testing of two materials in parallel provides a

simple approach for obtaining the stress strain response, it can
be subject to errors in the later portions of the post-peak stress-
strain response since the large contribution of the steel is
being subtracted from the large composite response. To over-
come these limitations a closed-loop testing control method
was developed in which the expansion of the cylinder in the
circumferential deformation was controlled at a constant rate
[11]. This approach takes advantage of the fact that as damage
develops in the specimen (at high stress levels and in the post-
peak region) the circumferential (lateral) deformation in-
creases much more rapidly than the lateral deformation. The
circumferential control was used to illustrate the stress-strain
curves of concrete of various strengths (Fig. 5) [12]. It can be
observed that as the strength of the concrete increases the pre-
peak behavior is linear up to a higher level of stress. The
higher strength materials show a steeper response in the post
peak region that corresponds with a more brittle material. It is
important to understand how the end conditions of the cylin-
der influence the results. For example, capping compounds
may demonstrate a disproportionate deformation and friction
at the plattens may result in a confining effect. One common Fig. 6—Influence of specimen size on the measured
method that is used to overcome end confinement or excessive compressive stress-strain response.
deformation is grinding the specimen ends and using friction
reducing substances or brush plattens [13]. Figure 6 can also Modulus of Elasticity in Tension and Flexure
be used to show the influence of testing specimens of different Substantially less work has been done to determine the elastic
length [14–16]. In the pre-peak region the stress-strain response modulus when concrete is tested in tension [17–19]. The test is
is relatively similar irrespective of specimen length. After the complicated by the problems associated with gripping the
peak is reached, however, the larger specimens demonstrate a specimens in tension, low strains at failure, and the need to
more brittle response. This occurs since the zone of damage avoid eccentricity. Several different approaches have been ad-
appears to be constant irrespective of specimen size. Newer vocated over the last three to four decades mainly aimed at de-
testing procedures are being developed in which closed-loop veloping procedures for improving the stress distributions at
testing enables the stress strain diagram to be measured on the specimen ends for different grips and removing difficulties
specimens of different size [15]. associated with eccentricity [20,21]. Much of the research has
not focused specifically at only assessing the elastic modulus,
but rather much of the work has focused on assessing the soft-
ening response that occurs when cracking begins to develop
[22–24].
Since a principal use of concrete is in flexural members,
several investigators have determined the elastic modulus on
specimens loaded as beams. An obvious approach is to meas-
ure deflections caused by known loads and to calculate the
modulus of elasticity from well-known beam deflection for-
mulas. It should be noted, however, that the span-to-depth ra-
tios of concrete beams normally used for such tests are so
large that shear deflection comprises a significant part of the
total deflection. In applying shear corrections, certain other
corrections must be made to take care of discontinuities in
the shear deflection curves at load points. For center-point
loading, Seewald [25] gives the following deflection formula:

P l3
    
2 3
h h
  1
(2.4
1.5
)   0.84 
48EI l l
While the deflection for third-point loading can be computed
using the following expression from ASTM C 1018.

23P l3 216h 2(1

)

   1

1296EI 115l2 
where

  maximum deflection
Fig. 5—Typical stress versus axial strain plots for a normal P  applied central load
strength, medium strength, and high strength concrete. l  distance between supports
 WEISS ON ELASTIC PROPERTIES, CREEP, AND RELAXATION 199

E  modulus of elasticity
I  moment of inertia of the section with respect to the
centroidal section

 Poisson’s ratio
h  depth of the beam

The portion of the expression outside the brackets is the

simple beam formula without considering the effects of shear.
It should be noted that the deflections used in determining the
elastic modulus should be carefully measured on the flexural
specimens. Researchers have illustrated that substantial errors
can occur if deformations are not measured correctly. This is
illustrated in Fig. 7 in which the deflection of the machine (i.e.,
stroke or ram deflection) is compared with the deflection
measured directly on the specimen [26]. It can be seen that the
measured machine deflection can be much higher than the de-
flections measured on the specimen. This can be primarily
attributed to crushing of the specimen at the load points and Fig. 8—Rate of material property development as a
deflections of the testing fixtures and machine itself. Deflection function of degree of hydration.
is commonly measured at the center point of the beam using a
“Japanese Yoke,” which is a frame that is attached to the neu-
tral axis of the beams directly over the supports. The frame is
designed so that rotations are permitted at one end while rota- the influence of aggregate fracturing in strength related prop-
tion and translation is permitted at the other end. The deflec- erties while this does not occur for modulus measurements
tion frame provides a reference location which does not move due to the low level of stress that is applied.
during the test and to which the measurement device is at-
tached so that it can react off of a smaller element that is at- Elastic Modulus from Ultrasonic Measurements
tached to the beam. In addition to measuring elastic properties by applying a
A large number of results have been reported in the litera- mechanical load, elastic properties can be determined nonde-
ture. The range of results has been from 7 to 40 GPa (1 to 6  structively. Of all the nondestructive methods, ultrasonic meth-
106 psi) [27]. It should be noted that the elastic modulus in ten- ods offer a distinct advantage in that they can be conducted
sion or flexure does not appear to be substantially different with a relatively low cost and without causing any new damage.
from the elastic modulus in compression at low stress levels. The main premise of the ultransonic pulse velocity test is re-
It is also interesting to note that Olken and Rostasy [28] lated to the concept that the velocity with which a compression
presented relationships for the development of various me- wave travels through an elastic body is proportional to the
chanical properties (Fig. 8 shows this relationship as a function square root of the elastic modulus. ASTM Standard Test
of the degree of hydration). It can be seen that the elastic Method for Pulse Velocity Through Concrete (C 597) can be
modulus develops at a much faster rate than either the tensile used for the determination of the compressional wave speeds.
or compressive strength. While all of the reasons for the A schematic illustration of the testing equipment is illustrated
differences in the rate are not completely understood [29], it in Fig. 9. This testing procedure relies on the development of
appears that this may be related to the composite nature of compression wave pulses that are generated by exciting a
concrete and the fact that the aggregate, interfacial transition piezo-electric crystal inside the transmitting transducer with a
zone, and paste influence each of these properties differently. high voltage pulse. The transmitting transducer is held in con-
Barde et al. [30] recently demonstrated that differences in flex- tact with one end of a specimen (usually a coupling agent is
ural strength and elastic modulus development may be due to applied between the specimen and the transducer). The second
transducer is held on the opposite side and used to record
the time that it takes this wave to reach the second transducer.
Using information obtained from the test, the wave speed
(or pulse velocity) can be determined using the following
equation:

L
V  
t

Fig. 7—Conceptual illustration of the importance of Fig. 9—An illustration of the ultrasonic pulse velocity
measuring deflections on the specimen. measurement procedure, ASTM C 597.
200 TESTS AND PROPERTIES OF CONCRETE

where Poisson’s Ratio

Static determinations of Poisson’s ratio are made by adding a
V  the compressional wave speed (i.e., the pulse velocity) third yoke and second dial gage to a compressometer so that a
L  the distance between the transducers magnified transverse strain may be measured, or by mounting
t  is the transit time (i.e., the time it takes the wave to travel strain gages on the surface of a specimen perpendicular to the
between the transducers) direction of loading. The same considerations apply to gage
length for lateral strain measurement as for longitudinal strain
The elastic modulus can be determined using the velocity measurement. Procedures for determination of Poisson’s ratio
of the compression wave through concrete if a value for Pois- are included in ASTM C 469. Poisson’s ratio is also commonly
son’s ratio is assumed (typically 0.22–0.24). The elastic modu- computed from results of the elastic modulus and shear mod-
lus can be determined using the pulse velocity from the ulus determined dynamically. The static value at stresses below
following expression: 40 % of the ultimate strength is essentially constant; for most
concretes the values fall between 0.15 and 0.20. The dynamic
[(1

)(1  2
)]
E  V 2  (1 
)
values are usually in the vicinity of 0.20–0.25.
It should be noted, however, that at high stresses or under
where conditions of rapidly alternating loads, the measured value of
Poisson’s ratio can change dramatically. When the applied
 density stress is below 50 % of the peak strength there is a decrease in
V the pulse velocity volume of the body as a compressive load is applied. However,

 the dynamic Poisson’s ratio at higher loads cracking develops which results in an increase
E the dynamic modulus of elasticity in the volume of concrete and an increase in Poisons ratio [33].

Typical values of the ultrasonic wave speed can range Property Specification and Estimation
from 3500 to 5500 m/s depending on the strength of the con- of Elastic Properties
crete or age at which the concrete is tested. To improve the Frequently designers do not specify elastic modulus but rather
accuracy of the results the wave speed of a known material is they rely on approximations using other properties (namely
typically measured to enable the testing apparatus to be prop- compressive strength) to estimate these properties for their de-
erly calibrated before a test. The modulus of elasticity deter- sign. It should be noted however that in some structures where
mined from the ultrasonic test (typically referred to as the dy- deflections need to be minimized aggregates may be restricted
namic modulus) can be up to 25 % higher than the static to those that can economically produce low elastic deforma-
modulus. This occurs for two reasons. First, the ultrasonic test tion and low creep. Other structures may specify the use com-
is conducted at low stress levels and as such the test results posite sections (of concrete and steel or concrete and fiber re-
more closely resemble an initial tangent modulus (Fig. 3). Sec- inforced composites) to increase the stiffness of the overall
ond, the elastic modulus is dependent on the rate at which load structure.
is applied. Loads applied at a higher rate result in a higher elas- Although code specifications are primarily associated with
tic modulus. It should be noted that the elastic modulus in sat- concrete strength, information on the elastic modulus is re-
urated concretes may be 5 % higher than that in dry concrete quired for many aspects of civil engineering design. The ACI
[31]. Further information on the method for measuring pulse Building Code [34] permits the modulus of elasticity to be
velocity can be found in ASTM C 597 or in the committee re- taken for normal weight concrete as
port from ACI 228.1R [32]. It should be noted that alternative
ASTM test methods can be used for measuring the compres- E  57000 ƒ
c (in psi)  473 ƒ
c (in MPa)
sional wave speed (P-wave) in concrete such as ASTM Standard
Test Method for Measuring the P-Wave Speed and the Thick- where
ness of Concrete Plates Using the Impact-Echo Method (C
1383). Results from this test can provide an alternative method E  modulus of elasticity
to measure the wave speed which can be used in the preceed- ƒc  specified comprehensive strength
ing equation to estimate the elastic properties.
An alternative nondestructive test is based on the concept For concrete with a hardened unit weight between 90 and 155
that the natural frequency of vibration of an elastic body is pro- lb/ft3 (1440 to 2885 kg/m3) the modulus can be taken as
portional to the square root of either the elastic modulus or the
shear modulus, depending on the mode of vibration. ASTM E  33w1.5
c ƒc (in psi and lb/ft3)
Standard Test Method for Fundamental Transverse, Longitudi- c ƒ
 0.043w1.5 c (in MPa and kg/m3)
nal, and Torsional Resonant Frequencies of Concrete Speci-
mens (C 215) was developed for determining the fundamental where
transverse, longitudinal, or torsional resonant frequencies of
concrete specimens. This test is generally conducted either by E  modulus of elasticity;
forcing the specimen into resonant vibration using an electro- wc  unit weight; and
mechanical driving unit or using a small impactor to generate ƒc  specified comprehensive strength.
a vibration that is recorded by an accelerometer. This test
method is commonly used in the lab for assessing freeze-thaw It should be noted that while this equation is useful, the elas-
damage in prisms. Further details on this testing procedure tic modulus is highly sensitive to the modulus of the aggre-
and its use for determining elastic properties are available in gate and as such the measured modulus may be expected to
ASTM C 215. vary by approximately 20 % of the computed value. It has
 WEISS ON ELASTIC PROPERTIES, CREEP, AND RELAXATION 201

been also been noted by ACI 363 that the aforementioned ex- to an improved understanding of creep for prestressed beams,
pressions overestimate the elastic modulus for higher plates and shells, and long-span structures. In the 1970s inter-
strength concretes. As such, it has been suggested that the fol- est in creep once again increased due to applications in nu-
lowing empirical relationship can be used for concretes with clear structures. In the 1980s concerns with creep in higher
compressive strengths between 3000 psi (21 MPa) and 12 000 strength concretes emerged since these materials were being
psi (83 MPa) [35]: used in offshore oil structures and high-rise buildings. In the
1990s interest in creep once again increased due to problems
w1c .5
E  23w1.5
c ƒ
 
c
1106  (in psi and lb/ft3)
145
associated with early age cracking and the desire to understand
how stress relaxation influences the behavior of concrete at
very early ages.
w1c .5
E  0.030w1.5
c ƒ
 
c
6895  (in MPa and kg/m3)
2325
First it should be noted that creep is a property of the
paste. The cement paste exhibits creep due to its porous struc-
where ture with a large internal surface area (nearly 500 m2/cm3) that
is sensitive to water movements. It appears reasonable to con-
E  modulus of elasticity clude that the movement of water in the paste structure is re-
wc  unit weight sponsible in large part for creep in concrete elements. In fact,
ƒc  specified comprehensive strength Mullen and Dolch [37] found no creep when pastes were oven
dried. The fact that creep is associated primarily with the
Poisson’s ratio is also commonly assumed to be 0.18–0.20 for cement paste adds a particular complexity to the problem of
static measurements while values are usually assumed to be trying to describe creep that does not appear in many other
0.20–0.22 for dynamic measurements or rapid loading materials. This complexity is called aging which generally
conditions. refers to the fact that cement pastes continue hydration which
means that the pore structure and elastic properties are chang-
Rheologic Properties: Creep and Relaxation ing over time or with age. This frequently results in problems,
however, when information is desired at early ages since the
Creep is defined in ASTM E 6 as “the time-dependent increase cement is hydrating relatively rapidly. Although creep is a paste
in strain in a solid resulting from force.” Nearly all materials property it is important to note that this does not mean that the
undergo creep under some conditions of loading. Unlike aggregates play no role. On the contrary, aggregates (especially
other materials, however, the creep of concrete is unique stiff aggregates) substantially reduce the creep of a material. In
since it is observed under normal service conditions at all some structures, where deflections need to be minimized, ag-
stress levels. Furthermore, creep of concrete is approximately gregates may be restricted to those that can economically pro-
a linear function of stress up to 50 % of its strength (on ma- duce low elastic deformation and low creep.
ture concrete in compression) and it appears to increase at At this point it should also be noted that the names applied
higher stresses presumably due to the cumulative effects of to the rheological response (often, as done in this document,
creep and microcracking. creep is generally used to describe all aspects of the rheologi-
The creep of concrete appears to have been first described cal response) of concrete are frequently less than precise.
in the United States in 1907 by Hatt [36]. Since that time over Strictly speaking, creep describes the deformation that may oc-
1000 papers have been written on various aspects of creep. It cur under a constant stress. Creep is illustrated in Fig. 10a. It
should be noted, however, that the interest in creep has been can be seen that initially the specimen is unloaded (a). At some
high at various times over the last century due to various ap- time (to) the specimen is loaded with a stress (o) and the speci-
plications. In the 1930s the rise in dam construction was driv- men exhibits an initial elastic deformation. Over time this -
ing research in creep, and this gave way over the next 20 years deformation increases (c) due to the effect of creep. If the load

Fig. 10—A conceptual illustration: a) creep, and b) relaxation.

202 TESTS AND PROPERTIES OF CONCRETE

is removed at some time (d) the specimen unloads elastically counted for by shrinkage “drying creep” [38]. Thus, the total
and continues to unload over some time (creep recovery) shortening at any time may be considered the sum of elastic
though it should be noted that only two thirds of the original strain, basic creep, drying creep, and shrinkage.
creep deformation is recovered. On the other hand stress re-
laxation describes the reduction in stress that occurs when a Effect of Specimen Size
specimen is deformed and this deformation is maintained con- It has been demonstrated [39,40] that creep of sealed speci-
stant. While stress relaxation is related to creep because both mens is independent of specimen size. This observation plus
occur due to the movement of water in the microstructure un- the observation concerning mass concrete in the preceding
der an applied stress, stress relaxation is illustrated in Fig. 10b. paragraph indicate that the techniques and specimens of
An initial specimen can be considered (a) that is deformed elas- ASTM C 512 are applicable to all types of concrete sealed to
tically at time (to) to a strain of (o). This deformation causes prevent loss of moisture. For unsealed specimens exposed to a
the specimen to develop an initial elastic stress (b); however, drying atmosphere, it is evident that there must be a size effect
over time if this deformation is maintained constant the stress associated with the moisture gradients within the specimen.
will decrease (c) due to relaxation. If the applied deformation It should be noted that the creep of a structure may be
is then released (d) the stress in the specimen will be reduced only a fraction of that in a test specimen. Hansen and Mattock
to zero while some permanent deformation may remain in the [41], in an investigation of both size and shape of specimens,
material. found that shrinkage and creep were dependent only on the ra-
Creep and relaxation properties are not frequently de- tio of surface to volume. Information of this sort may make it
fined in specifications. Rather designers often use very possible to apply correction factors to the data obtained from
approximate calculation procedures or apply larger safety ASTM C 512 to determine the creep in any size and shape of
factors to account for them. The reason for the approximate structure.
nature of many of the calculations may be traced to the fact
that the composition and size of the cement and composition Measurement of Tensile Creep or Relaxation
and size of the aggregate play such a large role in determin- Early-age cracking sensitivity of concrete recently has been a
ing the amount of creep that can be expected. Further, these topic receiving much attention. Toward this end early-age
materials vary from location to location making it very diffi- creep and relaxation properties have been heavily investigated.
cult to “predict the effects of creep” with a high level of Although no standardized testing procedures have been devel-
accuracy without having testing information on the local ma- oped, the testing procedures fall into a few distinct categories.
terials. Finally creep tests are frequently labor intensive, re- The first category of tests consists of a uniaxial tensile creep
quire a conditioned space to perform the tests, and take a test that relies on the application of load through a dead
substantial amount of time to perform. weight. Umehara et al. [42] and Bissonette and Pigeon [43]
have conducted uniaxial tensile creep tests where a specimen
Measurement of Creep in Compression is loaded using a dead-weight that is attached to a lever arm.
The age at which creep tests begin and the stress level to The second category of tests consists of a pressurized cylindri-
which specimens are loaded are usually dependent on how cal specimen that applies a constant pressure on the inner sur-
the data will be used. A test procedure has been standardized face of a hollow cylinder [3,44]. The third category of tests con-
in ASTM Test Method for Creep of Concrete in Compression sists of using an electric or hydraulic mechanical testing device
(C 512). The method stipulates loading moist-cured speci- to apply either a constant load or constant displacement to a
mens at an age of 28 days to a stress not exceeding 40 % of single specimen or series of specimens [45–47]. The final cate-
the strength of the concrete at the time of loading, although gory of testing devices consists of horizontal testing frames
provision is made for other storage conditions or other ages that use a closed-loop control to rapidly adjust the force on a
of loading. The stress is restricted to the range throughout specimen to maintain a specified displacement. The closed
which creep has been found to be proportional to stress. loop test provides the total stress history of a specimen and is
Limitations on gage lengths similar to those in the test for quite useful [48–53]. It should be noted that in each of these
modulus of elasticity apply. The method is intended to com- tests the entire testing frame or specimen is generally placed in
pare the creep potential of various concretes. Testing at a sin- a controlled environment.
gle age of loading is satisfactory for this purpose. It is re-
quired in the test method that the stress remain constant Property Specification and Estimation of Creep
throughout the one-year duration of the test within close tol- and Relaxation Parameters
erances. The load may be applied by a controlled hydraulic As previously mentioned, the time under loading influences
system or by springs, provided in the latter case the load is the corresponding deformation of the concrete. The ratio of
measured and adjusted frequently. long-term strain to immediate strain can be as high as 3.0. The
ASTM C 512 requires companion unloaded specimens. amount of creep exhibited is generally proportional to the
Length changes of these specimens are measured and sub- stress level (at least to 50 % of the peak strength), to the age at
tracted from the length changes of the loaded specimens to loading with materials loaded at an earlier age showing more
determine creep due to load. This correction is intended to creep, to the duration of loading with more creep in materials
eliminate the effects of shrinkage and other autogenous vol- under a longer duration of loading, and to the strength of the
ume change. While this correction is qualitatively correct and material being tested with higher strength materials showing
yields usable results, most modern theories deny the inde- less creep.
pendence of shrinkage and creep and thus indicate that the A simple method for computing the effects of creep at
two effects are not additive as assumed in the test. It is now various times under loading was defined by ACI-209R-92 using
common to label creep which occurs in the absence of drying a creep coefficient. This creep coefficient (CCU) can be thought
“basic creep” and to label the additional deformation not ac- of as simply the ratio of the long-term (ultimate) strain, which
 WEISS ON ELASTIC PROPERTIES, CREEP, AND RELAXATION 203

includes both elastic and creep effects (CU), to the initial elas- Relatively little creep testing is directed to predicting deflec-
tic strain (CI): tions of specific structures; however, predictive equations and
approximations are frequently used.
t0.6
CCT (t)   CCU
d
t0.6 Loss of Prestressing Force
where In contrast to the lack of precision needed for deflection meas-
urements, an accurate knowledge of the early-age rheological
CCT  coefficient at any time properties of concrete is valuable to the prestressed concrete
t time in days industry. After the prestressing force is applied, there is a loss
d is a constant (typically assumed to be 10) in days of prestressing force resulting from creep (and shrinkage) of
CCU  is the long-term on creep coefficient (typically assumed the concrete and relaxation of the steel. Since the initial pre-
to be 2.35) stressing force is limited by the strength of the steel and the
load-carrying capacity of the member is limited by the residual
Typical values for CCU range from 1.3 to 4.15 but the recom- prestressing force, a knowledge of the factors governing loss of
mended value is 2.35. It should be noted that ACI-209 provides prestressing force has important economic implications.
an approach to correct CCU to account for moist or steam cur-
ing, duration of moist curing, relative humidity, member size, Residual Stress Calculations
and surface to volume ratio. Recently it has become increasingly common to see concrete
Over the last three decades several models have been pro- structures developing cracks due to thermal, drying, or autoge-
posed to overcome the shortcomings of the creep coefficient nous shrinkage. Although numerous factors influence whether
approach to predict the response of concrete under sustained a concrete will crack [61], it can be simply stated that cracking
loads. These models are commonly referred to as the Gardner- will occur if the residual stresses that develop exceed the tensile
Lockman Model (GL2000) [54], the Bazant Models (BP or B3) strength of the material. Figure 11a illustrates how one can
[55,56], or the CEB model (CEB-90) [57]. It should be noted compare the time dependent strength development with the
that there is no universally accepted model for creep and even time dependent residual stresses that develop [62]. As a first
the most accurate models are commonly believed to be accu- point of analysis, it can be argued that if the residual stress de-
rate to only approximately 35 %. Due to space limitations the velopment exceeds the strength of the specimen the concrete
reader is referred to the original documents or summary doc- can be expected to crack. This is illustrated in Fig. 11a as the
uments for further information on predictive creep models point at which these two lines intersect. Similarly, it follows that
and their application [58–60]. if the strength of the concrete is always greater than the devel-
oped stresses, no cracking will occur.
Significance and Use of Elastic Properties, The residual stress that develops in concrete as a result of
Creep, and Relaxation restraint may sometimes be difficult to quantify. Shrinkage
strains can be converted to stresses with knowledge of the elas-
Deflection of Compressive and tic and creep (relaxation) behavior of concrete. This residual
Flexural Members stress cannot be computed directly by multiplying the free
Concrete members undergo deflection upon application of shrinkage by the elastic modulus (i.e., Hooke’s Law) since
load and continue to deflect with the passage of time. This may stress relaxation (creep) can substantially reduce the stress by
be of interest for reinforced beams, girders, slabs, or columns. 30–70 %. This reduction can be described by Fig. 11b in which
It is not uncommon for a reinforced concrete flexural member a specimen of original length (i) is exposed to shrinkage. If the
eventually to reach a deflection three times as great as its ini- specimen were unrestrained, the applied shrinkage would
tial deflection, while a precise prediction of these deflections is cause the specimen to undergo a change in length (shrinkage)
possible only if the elastic and creep properties are known. of L
(ii). To maintain the condition of perfect restraint (i.e.,

Fig. 11—An illustration of the restrained shrinkage cracking problem: a) residual stress development,
and b) a schematic description of stress development.
204 TESTS AND PROPERTIES OF CONCRETE

no length change) a fictitious load can be envisioned to be ap- • Standardized practices are needed for conducting early
plied (iii). However, it should be noted that if the specimen age creep and relaxation tests. As these tests are developed
were free to displace under this fictitious loading the length of and standardized it is recommended that experimentalists
the specimen would increase (due to creep) by an amount L- consider recommendations of model developers that will
(iv). Again, to maintain perfect restraint (i.e., no length change) enable this data to be utilized in future model develop-
an opposing fictitious stress is applied (v) resulting in an over- ments [70].
all reduction in shrinkage stress (vi). This illustrates that creep • It currently appears that predictive models may be needed
(relaxation) can play a very significant role in determining the that are capable of meeting the needs of two distinct au-
magnitude of stresses that develop at early ages. As a result sub- diences. The first audience desires a model that includes a
stantial research has been conducted over the last decade to description of material behavior from first scientific prin-
better determine how stresses develop at early ages. For fur- ciples. It would be anticipated that these models would be
ther information on early age stress development the reader is incorporated by the computer modeling community since
referred to RILEM TC-181 [63]. It should also be noted that a complicated calculations can be performed in the models.
simple test method was recently added as an ASTM standard (C The second audience for these models are users that may
1581) to assess the behavior of restrained concrete elements. want to be able to perform approximate calculations very
This test consists of casting an annulus of concrete around a quickly using hand calculations.
steel ring. As the concrete dries (or experiences autogenous • It is recommended that future standard test procedures
shrinkage) it attempts to shrink; however, this shrinkage is re- consider the addition of an appendix that would define
stricted by the restraining steel. This places the concrete in ten- standard reporting procedures for reporting data from the
sion and compresses the steel. While the test is primarily used test and the development of a data bank for elastic modu-
to determine if sufficient tensile stresses develop in the con- lus, creep, early-age creep, and relaxation tests that is sim-
crete ring to cause cracking, additional research has described ilar to that RILEM Shrinkage and Compliance Data Bank.
how this test method could be used to determine the residual
stresses and effect of creep [64–66].

Ongoing Efforts and Future Needs References

[1] Puri, S., “Assessing the Development of Localized Damage in
Interest in more accurately assessing rheological properties of
Concrete Under Compressive Loading Using Acoustic Emission,”
hardened concrete is increasing due to concerns related to early MSCE Thesis, Purdue University, West Lafayette, IN, May 2003,
age cracking, prestressed concrete, behavior of thin elements, p. 111.
and behavior of high-strength low-water-to-cement ratio con- [2] Puri, S. and Weiss, W. J., “Assessment of Localized Damage in
cretes. The following list provides an outline of ongoing efforts Concrete under Compression Using Acoustic Emission,” Under
in the area of rheological properties as well as future needs: review by the ASCE Journal of Civil Engineering Materials.
• Standardized testing practices exist for determining the [3] Weiss, W. J., “Prediction of Early-Age Shrinkage Cracking in
elastic modulus; however, standardized tests do not exist Concrete Elements,” Ph.D. Dissertation, Northwestern Univer-
for determining the complete stress-strain response for sity, Evanston, 1999.
concrete. Other organizations have assembled substantial [4] Walker, S., “Modulus of Elasticity of Concrete,” Proceedings,
data for assessing the stress-strain response of concrete. In ASTM International, West Conshohocken, PA, Vol. 21, Part 2,
the absence of ASTM standards on the subject, it is sug- 1919, p. 510.
gested that readers review the practices advocated by the [5] Teller, L. W., “Digest of Tests in the United States for the De-
RILEM SSC (strain softening in concrete) committee [67]. termination of the Modulus of Elasticity of Portland Cement
Mortar and Concrete,” Proceedings, ASTM International, West
• Significant advancements are taking place in attempting to
Conshohocken, PA, Vol. 30, Part 1, 1930, p. 635.
describe the behavior of concrete at early ages. This fre-
[6] “Bibliographies on Modulus of Elasticity, Poisson’s Ratio, and
quently includes computation of deformations and stresses Volume Changes of Concrete,” Proceedings, ASTM Interna-
which require elastic properties at early ages when they tional, West Conshohocken, PA, Vol. 28, Part 1, 1928, p. 377.
may be changing dramatically due to hydration. Further re- [7] Whitney, C. S., discussion of a paper by V. P. Jensen, “The Plas-
search is needed to define how elastic properties develop ticity Ratio of Concrete and Its Effect on the Ultimate Strength
over time (i.e., different maturities). This includes both rate of Beams,” Journal, American Concrete Institute, Nov. 1943,
of development as well as “time-zero” and it is suggested Supplement, Proceedings, Vol. 39, pp. 584–2 to 584–6.
that research and standardized procedures be developed to [8] Ramaley, D. and McHenry, D., “Stress-Strain Curves for Con-
define how these properties should be defined [68,69]. Fur- crete Strained Beyond the Ultimate Load,” Laboratory Report
ther, it is suggested that engineers begin to discuss how No. SP-12, U.S. Bureau of Reclamation, Denver, CO, March 1947.
calorimetric measurements of property development can [9] Hognestad, E., Hanson, N. W., and McHenry, D., “Concrete
be compared to ultrasonic, electrical, and mechanical Stress Distribution in Ultimate Strength Design,” Journal,
measurements which may measure fundamentally differ- American Concrete Institute, Dec. 1955; Proceedings, Vol. 52,
pp. 455–479.
ent aspects of the system.
[10] Shah, S. P., Namaan, A. E., and Moreno, J., “Effect of Confine-
• Interesting work is being performed in the area of the in-
ment on the Ductility of Lightweight Concrete,” International
creased use of nondestructive testing procedures to assess Journal of Cement Composites and Lightweight Concrete, Vol.
the elastic properties of concrete, especially the age- 5, No. 1, Feb. 1983, pp. 15–25.
dependent elastic properties. Research linking ultrasonic, [11] Shah, S. P., Gokoz. U, and Anasari, F., “An Experimental Tech-
acoustic, and electrical measurements with rheological nique for Obtaining the Complete Stress Strain Curves for High
properties should be further developed and validated over Strength Concrete,” Cement, Concrete and Aggregates, Vol. 3,
a wide range of material properties. No. 1, Summer 1981, pp. 21–27.
 WEISS ON ELASTIC PROPERTIES, CREEP, AND RELAXATION 205

[12] Jansen, D. C., Shah, S. P., and Rossow, E. C., “Stress Strain Results [34] Building Code Requirements for Structural Concrete (ACI 318-
of Concrete from Circumferential Strain Feedback Control 02) and Commentary (ACI 318R-02), American Concrete Insti-
Testing,” ACI Materials Journal, Vol. 92, No. 4, July-Aug. 1995, tute, Farmington Hills, MI, 2002.
pp. 419–428. [35] Nilson, A. H. and Winter, A. H., Design of Concrete Structures –
[13] Van Mier, J. G. M., et al., “Strain Softening of Concrete in Uni- 11th Edition, McGraw Hill, Inc. © 1991.
axial Compression,” Materials and Structures, RILEM 148-SSC, [36] Hatt, W. K., “Notes on the Effect of the Time Element in Load-
Vol. 30, pp. 195–209. ing Reinforced Concrete Beams,” Proceedings, ASTM Interna-
[14] Van Mier, J. G. M., Fracture Processes in Concrete, CRC Press © tional, West Conshohocken, PA, No. 7, 1907, pp. 421–433.
1997. [37] Mullen, W. G. and Dolch, W. L., “Creep of Portland Cement
[15] Jansen, D. C. and Shah, S. P., “Effect of Length on Compressive Paste,” Proceedings, Vol. 64, 1964, pp. 1146–1171.
Strain Softening of Concrete,” ASCE Journal of Engineering [38] Neville, A. M., Properties of Concrete, 4th ed., Wiley, New York,
Mechanics, Vol. 123, No. 1, Jan. 1997, pp. 25–35. 1996.
[16] Weiss, W. J., Güler, K., and Shah, S. P., “Localization and the [39] Polivka, M., Pirtz, D., and Adams, R. F., “Studies of Creep in
Size-Dependent Response of Reinforced Concrete Beams,” ACI Mass Concrete,” Symposium on Mass Concrete, ACI SP-6, Amer-
Structures Journal, Vol. 98, No. 5, Sept.–Oct. 2001, pp. 868–695. ican Concrete Institute, Detroit, MI, 1963, pp. 257–283.
[17] Davis, R. E., Davis, H. E., and Brown, E. H., “Plastic Flow and Vol- [40] Karapetrin, K. S., “Influence of Size upon Creep and Shrinkage
ume Changes in Concrete,” Proceedings, ASTM International, of Concrete Test Specimens,” I. Tekhiniceskie Nauki, Armenian
West Conshohocken, PA, Vol. 37, Part 2, 1937, p. 317. Academy of Sciences (Yerevan), Akad. Nauk Armianskoi SSR,
[18] Johnson, A. N., “Tests of Concrete in Tension,” Public Roads, Fiziko—Mathematicheskie, Estestvennye, Vol. 9, No. 1, 1956,
Vol. 7, No. 4, June 1926. pp. 87–100 (in Russian).
[19] Johnson, J. W., “Relationship Between Strength and Elasticity [41] Hansen, T. C. and Mattock, A. H., “The Influence of Size and
of Concrete in Tension and in Compression,” Bulletin No. 90, Shape of Member on the Shrinkage and Creep of Concrete,”
Engineering Experiment Station, Ames, IA, 1928. Proceedings, American Concrete Institute, Vol. 63, Feb. 1966,
[20] U.S. Bureau of Reclamation, Procedures for Direct Tensile pp. 267–290.
Strength, Static Modulus of Elasticity, and Poisson’s Ratio of [42] Umehara, H., Uehara, T., Iisaka, T., and Sugiyama, A., “Effect of
Cylindrical Concrete Specimens in Tension (USBR 4914-92), Con- Creep in Concrete at Early Ages on Thermal Stress,” in RILEM
crete Manual, Part 2, 9th ed., U.S. Bureau of Reclamation, Den- Proceedings 25, Thermal Cracking in Concrete at Early Ages, R.
ver, 1992, pp. 726–731. Springenschmid, Ed., E&FN Spon, 1994, pp. 79–86.
[21] Gérard, B., Breysse, D., Ammouche, A., Houdusse, O., and Didry, [43] Bissonnette, B. and Pigeon, M., “Tensile Creep at Early Ages of
O., “Cracking and Permeability of Concrete under Tension,” Ordinary, Silica Fume and Fiber Reinforced Concretes,” Cement
Materials and Structures, Vol. 29, April 1996, pp. 141–151. and Concrete Research, Vol. 25, No. 5, 1995, pp. 1075–1085.
[22] Ueda, M., Hasebe, N., Sato, M., and Okuda, H., “Fracture Mech- [44] Ross, A. D., “Experiments on the Creep of Concrete Under Two-
anism of Plain Concrete under Uniaxial Tension”, “Concrete Dimensional Stressing,” Magazine of Concrete Research, June
Library of JSCE,” No. 24, Dec. 1994, pp. 31–45. 1954, pp. 3–10.
[23] Li, Z., Kulkarni, S. M., and Shah, S. P., “New Test Method for Ob- [45] Brooks, J. J. and Jiang, X., “The Influence of Chemical Admix-
taining Softening Response of Unnotched Concrete Specimen tures on Restrained Drying Shrinkage of Concrete,” ACI SP-173-
under Uniaxial Tension,” Experimental Mechanics, Vol. 33, No. Chemical Admixtures, 1994, pp. 249–265.
3, 1993, pp. 181–188. [46] Bernader, S. and Emborg, M., “Risk of Cracking in Massive Con-
[24] Li, Z. and Shah, S. P., “Localization of Microcracking in Concrete crete Structures—New Developments and Experiences,” Pro-
under Uniaxial Tension,” ACI Materials Journal, Jul.-Aug. 1994, ceedings of the International Symposium Thermal Cracking in
pp. 372–381. Concrete at Early Ages, Munich Germany, R. Springenschmidt,
[25] Seewald, F., “Abhandlungen,” Aerodynamischen Inst. an der Ed., 1995.
Technischen Hochschule, Aachen, Vol. 7, 1927, p. 3. [47] Parillere, A. M., Buil, M., and Serrano, J. J., “Effect of Fiber Ad-
[26] Gopalaratnam, V. S., et al., “Fracture Toughness of Fiber Rein- dition on the Autogenous Shrinkage of Silica Fume Concrete,”
forced Concrete,” American Concrete Institute Materials Jour- ACI Materials Journal, Vol. 86, No. 2, 1989, pp. 139–144.
nal, Vol. 88, No. 4, 1991, pp. 339–353. [48] Springenschmidt, R., Gierlinger, E., and Kernozycki, W., “Ther-
[27] State-of-the-Art Report on High Strength Concrete (ACI 363- mal Stress in Mass Concrete: A New Testing Method and the
R92), American Concrete Institute, Farmington Hills, MI, 2002. Influence of Different Cements,” Proceedings of the 15th Inter-
[28] Olken, P. and Rostasy, F. S., “A Practical Planning Tool for the national Congress For Large Dams, Lausanne, R4, 1985, pp.
Simulation of Thermal Stresses and for the Prediction of Early 57–72.
Thermal Cracks in Massive Concrete Structures,” in Thermal [49] Altoubat, S. A. and Lange, D. A., “Early-Age Creep and Shrinkage
Cracking in Concrete At Early-Ages, R. Springenschmid, Ed., of Fiber Reinforced Concrete for Airfield Pavement,” Aircraft
Proc. RILEM Symp., E&FN Spon London, 1994, pp. 289–296. Pavement Technology, F. V. Hermann, Ed., 1997, pp. 229–243.
[29] Graveen, C., “The Use of Nondestructive Test Methods to Assess [50] Bjøntegaard, Ø., Kanstad, T., Sellevold, E. J., and Hammer, T. A.,
Pavement Quality in a Performance-Related Specification,” Pur- Stress-Inducing Deformations and Mechanical Properties of
due University, M.S. Thesis, W. Lafayette IN, May 2001. HPC at Very Early Ages, Utilization of High Strength/High Per-
[30] Barde, A., Mazzotta, G., and Weiss, J., “Early Age Flexural formance Concrete, I. Holland and E. J. Sellevold, Eds., Sande-
Strength: The Role of Aggregates and Their Influence on Ma- fjord Norway, 1999.
turity Predictions,” Materials Science of Concrete VII, American [51] Bloom, R. and Bentur, A., “Free and Restrained Shrinkage for
Ceramic Society, 2004, pp. 247–264. Normal and High Strength Concretes,” ACI Materials Journal,
[31] Bungey, J. H., Testing of Concrete in Structures, Second Edition, Vol. 92, No. 2, 1995, pp. 211–217.
Chapman and Hall, 1989, p. 52. [52] Kovler, K., “Testing System for Determining the Mechanical
[32] “In-Place Methods to Estimate Concrete Strength” (ACI 228.1R), Behavior of Early Age Concrete under Restrained and Free Uni-
American Concrete Institute, Manual of Concrete Practice, Part axial Shrinkage,” Materials and Structures, RILEM, London,
II, Farmington Hills, MI. U.K., 27(170), 1994, pp. 324–330.
[33] Shah, S. P. and Chandra, S., “Critical Stress, Volume Change, and [53] Pigeon, M., et al., “Equipment for the Analysis of the Behavior
Microcracking of Concrete,” Journal of the American Concrete of Concrete under Restrained Shrinkage at Early Age,” Maga-
Institute, Vol. 65, September 1968, pp. 770–781. zine of Concrete Research, Vol. 52, No. 4, 2000, pp. 297–302.
206 TESTS AND PROPERTIES OF CONCRETE

[54] Gardner, N. J. and Lockman, M. J., “Design Provisions for Drying [64] Attiogbe, E. K., See, H. T., and Miltenberger, M. A., “Tensile
Shrinkage and Creep of Normal-Strength Concrete,” ACI Mate- Creep in Restrained Shrinkage,” Concreep 6: Creep, Shrinkage,
rials Journal, Vol. 98, March-April 2001, pp. 159–167. and Durability Mechanics of Concrete and Other Quasi-Brittle
[55] Bazant, Z. P. and Panula, L., “Practical Prediction of Time De- Materials, F. J. Ulm, Z. P. Bazant, and F. H. Wittman, Eds.,
pendent Deformations of Concrete, Parts I–IV,” Materials and Elsevier, Cambridge, MA, 2001, pp. 651–656.
Structures, Vol. 11, 1978, pp. 307–316, 317–378, 425–434; and [65] Weiss, W. J. and Ferguson, S., “Restrained Shrinkage Testing:
Vol. 12, 1979, pp. 169–183. The Impact of Specimen Geometry on Quality Control Testing
[56] Bazant, Z. P., “Creep and Shrinkage Prediction Model for Analy- for Material Performance Assessment,” Concreep 6: Creep,
sis and Design of Concrete Structures – Model B3,” Materials Shrinkage, and Curability Mechanics of Concrete and Other
and Structures, Vol. 28, 1995. Quasi-Brittle Materials, F. J. Ulm, Z. P. Bazant, and F. H. Wittman,
[57] Müller, H. S., “New Prediction Models for Creep and Shrinkage Eds., Elsevier, Cambridge, MA, 2001, pp. 645–651.
of Concrete,” ACI SP 135-1, 1992, pp. 1–19. [66] Hossain, A. B. and Weiss, W. J., “Assessing Residual Stress De-
[58] Neville, A. M., Dilger, W., and Brooks, J. J., “Creep of Plain and velopment and Stress Relaxation in Restrained Concrete Ring
Structural Concrete,” Construction Press, Longman Group, Lon- Specimens,” Journal of Cement and Concrete Composites, Vol.
don 1983. 26, 2004, pp. 531–540.
[59] Bazant, Z. P., “Theory of Creep and Shrinkage in Concrete Struc- [67] Van Mier, J. G. M., et al., “Strain-Softening of Concrete in Uni-
tures: A Precis of Recent Developments,” Mechanics Today, Vol. Axial Compression,” Materials and Structures, RILEM 148-SSC,
2, 1975, pp. 1–93. Vol. 30, 1997, pp. 195–209.
[60] “Fourth RILEM International Symposium on Creep and Shrink- [68] Bisschop, J., “Chapter 3.3 – Evolution of Solid Behavior, Early
age of Concrete: Mathematical Modeling,” Z. P. Bazant, Ed., Age Cracking In Cementitious Systems,” RILEM TC-181 EAS Early
Northwestern University, August 26–29, 1986. Age Shrinkage Induced Stresses and Cracking in Cementitious
[61] Shah, S. P., Weiss, W. J., and Yang, W., “Shrinkage Cracking— Systems State of the Art Report, A. Bentur, Ed., 2003, pp. 27–36.
Can It Be Prevented?” Concrete International, Vol. 20, No. 4, [69] Weiss, W. J., “Chapter 6.1 – Experimental Determination of the
1998, pp. 51–55. ‘Time-Zero,’ Early Age Cracking In Cementitious Systems,”
[62] Mehta, P. K. and Monterio, P. J. M., Concrete: Structure, Prop- RILEM TC-181 EAS Early Age Shrinkage Induced Stresses and
erties, and Materials, 2nd Edition, Prentice Hall, Englewood Cracking in Cementitious Systems State of the Art Report, A.
Cliffs, New Jersey, © 1993. Bentur, Ed., 2003, pp. 196–206.
[63] Early Age Cracking in Cementitious Systems, RILEM TC-181 [70] “Guidelines for Characterizing Concrete Creep and Shrinkage
EAS, Early Age Shrinkage Induced Stresses and Cracking in Ce- in Structural Design Codes or Recommendations,” RILEM TC
mentitious Systems State of the Art Report, A. Bentur, Ed., 107: Creep and Shrinkage Prediction Models: Principles of their
2003. Formulation, Materials and Structures, Vol. 28, 1995, pp. 52–55.
20
Petrographic Examination

Bernard Erlin1

Preface but his work was, perhaps, the first widespread enough so that
it truly reached out to engineers. His work was published in a
ALMOST 50 YEARS HAVE PASSED SINCE THE PUBLICA- six-part series from January through March 1915 [1].
tion of ASTM STP 169A, in which this chapter was first published. Subsequent and more recent informative documents in-
Katherine Mather, who wrote the first two versions of this chap- clude Refs 2–15. The International Cement Microscopy Associ-
ter in ASTM STP 169A and 169B, quoted, in her closing, from ation (ICMA) was founded in 1978. That organization, whose
St. Paul, “Things which are seen”—concrete and mortar—“were main emphasis originally was on portland cement, has ex-
not made of things which do appear.” But then, St. Paul was not panded and now includes work on many aspects of concrete.
a petrographer. St. Paul is not with us—neither is Katherine, who It also provides publications of interest to concrete petrogra-
died in 1991. Much of this paper is still her “quote” on petrog- phers and others. About three years ago the Society of Con-
raphy, but with changes and additions that update the subject. crete Petrographers (SCP) was founded and now provides a
home base to concrete petrographers.
Introduction In June 1989, the first symposium specifically directed to-
ward presenting information on petrography of concrete and
Almost 50 years have passed since the inception of ASTM STP concrete aggregates was sponsored by ASTM. The papers pre-
169A, and 27 years have passed since the updated ASTM STP sented at that symposium resulted in ASTM STP 1061 [16]. In
169B was published. During these times, there was little refer- 1999, ACI sponsored a symposium on determining the water-
ence material available on petrography of concrete. Today, cement ratio and durability aspects of concrete [17] that in-
there are literature references to petrographic examinations cluded a number of papers on petrographic methods. A book
available in which the focus is on petrography as a tool used to devoted entirely to details of concrete petrography and corre-
provide information about concrete instead of petrography lations to field concrete performance, and aptly titled Concrete
that is incidental to the main subject of a paper. Both ap- Petrography [18], was published in 1998. ASTM Practice for
proaches, however, are needed because doing petrography Petrographic Examination of Hardened Concrete (C 856) has
solely for the sake of providing information can be an exercise been modified throughout the years and now provides for
in futility, while doing petrography for the purpose of provid- more stringent qualification requirements for petrographers
ing links connecting “activity” within concrete to its behavior doing petrographic examinations and more detailed informa-
and performance is indeed a fulfillment of the science. tion on a number of aspects of petrographic examinations. The
Significant advances in petrography and petrographic scanning electron microscope (SEM) and its ancillary energy
methods have been published related to concrete. Once con- dispersive spectroscopical (EDS) method for elemental analy-
sidered by many to principally revolve around light optical mi- sis is used so frequently for petrographic work that ASTM Com-
croscopes, the science has greatly expanded to include new mittee C09.65 on petrography has been working to standardize
types of instrumentation and techniques that cover subjects that approach.
from light optical microscopy to specimen preparation, “wet” The closing remarks of Katherine Mather, who authored
chemical analyses to infrared spectroscopy, X-ray diffractome- the initial versions on petrographic examination of concrete in
try and spectroscopy, scanning electron microscopy with at- ASTM STP 169A and ASTM STP 169B, foretold the potential ad-
tendant elemental analysis, differential scanning calorimetry, vances in the science of petrography. Today, the imagination
and other analytical tools. extends even further, and it is gratifying that her visions are in
In the history of petrography of concrete as we know it to- use and being extended.
day, at least from published papers, was the work of Johnson
who described and related microscopical observations of the Petrography
composition and texture of deteriorated concrete to its per-
formance. He further applied what he saw toward a philosophy Within the realm of petrography, which includes the use of a
of what makes inferior concrete inferior, and said, “even with broad variety of analytical and physical methods, is the
the very best of materials, only concrete of inferior strength is scientific description of the composition and texture of mate-
commonly produced.” There were trailblazers before Johnson, rials, including the systematic classification of rocks. It also in-

1
Petrographer, The Erlin Company, Latrobe, PA 15650.

207
208 TESTS AND PROPERTIES OF CONCRETE

cludes almost anything that can be said about a concrete, from of the petrographer for obtaining the information necessary
its mineralogy to its strength and volume stability. to answer the question that prompted his need for the petro-
Petrographic analysis of concrete—a man-made rock—is its graphic study.
examination by analytical techniques that will identify proce- The petrographer should not expect his petrographic re-
dures and the sequence of its production, its composition and sults to be taken on faith. Standard procedures should usually
internal structure, and allow its classification as to type, origi- be used, or modifications of standard procedures, as may be
nal and existing conditions, and future serviceability. appropriate—the rationality of the techniques that are used
The physical and chemical properties of concrete, espe- should be demonstrable. Useful background information that
cially immediately and shortly after it is made, are a physical may help the petrographer direct his work and evaluate the pet-
wonder. Within a very short time, measurable in terms of days, rographic data may be from plans, specifications, materials
it becomes hard and strong and usually endures for long peri- tests, inspectors, contractors, concrete producers, suppliers of
ods. Its strength originates within itself by complex chemical the concrete-making components, engineers, architects, and
reactions. It is re-created “rock” akin to the rock conglomerate most anyone involved in the concrete. The petrographer should
that Mother Nature has made. examine concrete with all available means commensurate with
Like rocks and minerals, concrete is a “mirror” with a the needs and economic restrictions imposed by his client.
memory. Petrographic examinations allow us to interpret the In a recent article, Paster [19] comments that the water-
concrete’s past as it really was—to identify, beyond all of the ob- cement ratio (w/c) cannot be determined using petrographic
scurities, essential facts about its manufacture and performance. methods. However, in ACI SP 191 [17] and in dozens of other
Its makeup and past performance, in light of research and prac- papers, a number of methods for doing so are discussed. That
tical experiences, allow projections of its future serviceability. comment brings up the competence of the now relatively large
Petrography is used frequently to assist in forensic evalua- number of concrete petrographers and their limitations in ex-
tions, where it is vital for supplying factual information. That amining and interpreting what they find in concrete and their
information can relate to mixture proportions that include comfort level in estimating water-cement ratios. That comfort
coarse and fine aggregates, portland cement and other cemen- level does indeed vary between petrographers, depending
titious materials, water-cement ratios, air contents, mixing, upon their education, experience, and training. Because one
placing, bleeding, cement hydration, finishing, curing, crack- petrographer has a low comfort level of evaluation does not
ing, scaling, spalling, low strength, excessive wear, blistering, mean all petrographers are at that same low level, and it does
delamination, adverse chemical and physical reactions result- not speak highly of anyone that apparently does not have
ing from internal and external sources, and other features. To background information upon which to make that kind of
be effective, the petrographer must have a good understanding statement. A petrographer’s responsibilities include not only
of each concrete-making material, concrete manufacture, and petrographic examinations, reports of the petrographic data,
the influence of environmental exposure on its stability and and interpretations of the data, but also data that are qualified.
performance. ASTM C 856 establishes minimum requirements for concrete
This chapter provides insight into the use of petrographic petrographers. Recently, The Society of Concrete Petrogra-
examinations for evaluating hardened concrete, noting prob- phers2 was formed as a means of getting together hands-on
lems inherent in its applications, outlining what it involves, concrete petrographers. The Society recognizes the weak-
describing information it can produce, and showing how this nesses of some petrographers in areas of their work—one of the
information can be applied. goals of the Society is to provide a meeting ground for
General questions that petrography tries to answer are: (a) petrographers to discuss issues of mutual interest so that they
does the concrete conform to specification requirements? (b) can better perform and interpret their work.
how will it behave in use? (c) why did it behave the way it did?
and (d) what can be anticipated in the future? The most useful Methods—Standardization and Description
method for developing practical information from which to an-
swer these questions is to study the concrete in the laboratory ASTM C 856 was adopted in 1977. It turned out to be more
and correlate what is found to its field performance. complicated than was expected because there are different
Petrographic examination of hardened concrete is among purposes for examining concrete. The concrete also may rep-
the subjects in this volume because it provides direct observa- resent exposures to different environments for different peri-
tional information on what is in concrete and provides another ods of time, specimens from laboratory test programs, and
way of establishing the relationship of concrete composition to samples from newly cast laboratory and field structures. ASTM
concrete performance. C 856 does the following: (1) serves those who request a pet-
rographic examination and want more understanding of what
Responsibilities they are getting and why; (2) provides information to petrog-
raphers conducting petrographic examinations; and (3) serves
A petrographic examination usually begins with discussions as a reminder to petrographers of things they may have for-
between someone who requests the examination and a petro- gotten or neglected.
grapher. Usually the discussions include a summary of the ASTM Guide for Petrographic Examination of Aggregates
situation that prompted the need for the examinations. The for Concrete (C 295) was first published in 1952. ASTM Prac-
person requesting the examinations may not be familiar with tice for Examination and Sampling of Hardened Concrete in
techniques petrographers use or petrographic approaches Constructions (C 823) was adopted in 1975. It gives guidance
available. His reliance is on the background and experience on steps to be taken before laboratory examinations, and pre-

2
Derek Cong, Secretary ([email protected]); Bernard Erlin, President ([email protected]).
 ERLIN ON PETROGRAPHIC EXAMINATION 209

liminary investigations are outlined and include: (1) the desir- vide ancillary information needed to properly evaluate the
ability of assembling reports and legal documents concerning problem that prompted the studies. The transition from the
the construction, and (2) the usefulness of interviews with con- macro to the micro scale and the information from each unit
tractors and others connected with the construction and with of effort provides separate and complimentary data that can be
the owners, occupants, and users of the construction. Proce- used in final analyses and interpretations of the data.
dures for detailed investigations of the concrete in-place are de- Care is needed to ensure that proper sampling is done for
scribed. Sampling hardened concrete is discussed along with each part of the study. Field samples may include examples of
the preparation of appropriate sampling plans and selection of different outwardly appearing concretes, with cracks, scales,
the number and size of samples. Information needed to ac- spalls, efflorescence, and anything else that may have con-
company samples is described. ASTM C 823 was prepared to tributed to or be part of a problem. Specimens for the differ-
be useful to petrographers, engineers, and others who have ent analyses should be taken to represent inherent parts of the
reasons for a need to examine concrete. concrete needed to understand its original makeup and subse-
ASTM Quantitative Determination of Phases in Portland quent physical or chemical alteration, or items that may have
Cement Clinker by Microscopical Point-Count Procedure (C contributed to its current condition. The number of samples
1356) provides a standard systematic procedure used to iden- needed is dependent upon the specific problem—and should be
tify the volumetric properties of portland cement clinker representative of the various features of the concrete. The sam-
phases. It also provides information that assists in an evaluation pling should include unaltered concrete for comparison.
and interpretation of the manufacturing history of clinker. In Sorting out major from minor causes or secondary effects
1976, Transportation Research Circular No. 176 [20] was pub- of deterioration requires knowledge of many aspects of con-
lished. It provides help in (1) identifying alite and belite residues crete, from individual concrete components to its manufacture
in paste, and (2) examining paste by X-ray diffraction, differen- to the things that affect cement hydration, and effects of the
tial thermal analysis, and infrared absorption spectroscopy. chemical and physical environment on concrete performance.
Another publication for help in petrographic examination of Analytical techniques offer a means for understanding in-
hardened concrete is Highway Research Board Special Report ternal chemical reactions in detail so that it is possible to char-
No. 127 [21]. Other publications of interest are in volumes of acterize cement hydration and hydration products. Thus, the
the Fifth International Symposium on the Chemistry of Cement chemical reactions of normal hydration or of abnormal
[22] and those of the Sixth International Symposium on the deterioration and alteration to concrete can be detected with
Chemistry of Cement [23]. A stimulating publication originating reasonable clarity so that detailed information is available to
in 1971 [24] covers a wide field of cement and concrete topics decipher a concrete’s history. The relative roles of chemical
and has the advantage of being truly international because the attack and physical attack that dictate concrete performance
editorial board and contributing authors come from various are usually decipherable. The basic microscopical techniques
parts of the world where cement and concrete research is used for petrographic examinations of hardened concrete can
carried out. be augmented, but not replaced, by other techniques.
Today, sophisticated analytical instrumentation is bring-
ing about a breakthrough by orders of magnitude in our abil- Purpose and Approach
ity to decipher and interpret (and sometimes misinterpret) the
composition and performance of concrete. These methods can Purpose
bring with them the danger of losing touch with the primary The purpose of a petrographic examination is initially to under-
purposes of an examination due to great differences in scale. stand the original concrete makeup, which includes: (1) aggre-
Those differences include concrete elements with dimensions gates—their grading, composition, and distribution, and any
in hundreds of metres (kilometres in the case of pavements); physical or chemical alteration; (2) portland cement, supple-
core samples with dimensions in hundreds of millimetres that mentary cementing materials, pozzolans, and their hydration
can be examined with great advantage using low-power stereo- and other reactions; (3) air-void systems including the size and
microscopes; thin sections about 800 mm2 by 15- to 30-
m distribution of air voids; and (4) textural characteristics resulting
thick and immersion preparations even smaller examined us- from its placement, consolidation, and finishing. Once that
ing petrographic microscopes; X-ray diffraction samples that understanding is in hand, the appropriateness of each item and
may be milligrams of material hand-picked under the stereo- alteration to the original concrete can be identified. The ques-
microscope or a few grams of material concentrated by hand- tion of why the concrete malperformed can then be identified,
picking paste or aggregate from carefully broken concrete sur- if suitable, and appropriate analytical techniques can be used.
faces; scanning electron microscope specimens that may be
200 mm3 down to micro-picked specimens—and ancillary ele- Approach
mental analysis using EDX methods of material volumes The concrete problem must usually be defined in order to pro-
equivalent to one-trillionth of a cubic inch (1.64  1011 cm3); vide direction for the petrographic examinations. Sometimes
single crystals of a few angstroms examined using micro- the problem is purposely not defined because of a desire to not
probes; and nanometer-sized material examined using trans- prejudice the petrographer in his examination and interpreta-
mission electron microscopes. However, few cases will be tion of the analytical data or to test his skillfulness. The latter
found in which all of these will be needed. is unneeded because his prowess ultimately becomes known
The sequence of steps involved in a petrographic exami- when his report is finalized. A good petrographic examination
nation can include: (1) observing the concrete construction; (2) is the one that provides enough information to resolve the
evaluating samples using the naked eye and low-power stereo- problem. A better petrographic examination is one that re-
microscopy; followed by (3) the selection of microscopical solves the problem with maximum economy in minimum time.
methods that allow greater insight into the makeup of the con- In both cases, the end result is what is important. The good, or
crete; and (4) the selection of additional methods that will pro- better examination, varies depending on the problem and
210 TESTS AND PROPERTIES OF CONCRETE

particularly the skill and adeptness of the petrographer. The may also be needed. The petrographer should be familiar with
petrographer’s skills include deftness in obtaining data and in specimen preparation, the kind of information each of these can
interpreting the data. provide, and limitations of their use.

Texture and Composition Age of Concrete Under Study

Concrete is a very heterogeneous material because it is made The age of the concrete may be important for judging the sig-
up of a variety of coarse and fine materials in terms of its basic nificance of petrographic observations. For example, calcium
components, coarse and fine aggregates, paste, and air. The sulfoaluminate (ettringite) found in many voids as far as 130
processes of batching, mixing, transportation, discharging, mm (5 in.) from the outer surfaces of a concrete pavement of
placing, consolidating, the differential settlement of compo- unknown age and of high flexural and compressive strength
nents (among the results is bleeding), finishing, and curing are may be of importance relative to projected service. That ob-
independent factors that control its final makeup and texture. servation is of particular importance when in other 15-year-old,
Texture is the way in which concrete components are finally similarly exposed field concretes from the region calcium sul-
allied and aligned. On a macro scale, honeycombing, lift lines, foaluminate is not abundant. If the concrete of unknown age is
and stratification of components are examples of textural in fact five or seven years old and it differs conspicuously, the
features that are usually undesirable. What is most desirable difference probably justifies some concern about its future per-
is a uniform distribution of concrete components. On a micro formance; if it is 15 years old, it is peculiar, but the peculiarity
scale, texture can be reflected by the uniformity of distribu- is probably of less practical importance.
tion of, for example, calcium hydroxide, a main hydration Recently, Neville [25] reviewed dating the age of cracks us-
component of cement hydration that can be uniformly or ing depths of carbonation as signposts. Although relative ages
nonuniformly distributed throughout the paste, at aggregate- could be identified, specific times of cracking could not unless
paste interfaces, or preferably located along the bottom side the cracks occurred very early (e.g., plastic shrinkage cracks).
of aggregate particles.
Petrographic examinations are useful for identifying tex- Sources of Concrete
tural and compositional characteristics, cracks, and crack pat-
terns. Some examples are: low strength resulting because the Knowing the concrete background and its geographic location
physical structure (texture) is inadequate can be caused by is sometimes of importance because aggregates, cements, and
poor mixing or inadequate consolidation; cement hydration environmental conditions vary both locally and regionally. For
may have been curtailed because of a number of factors; con- example, regional geology—as it determines quantity and uni-
taminants may be present; air-void systems may be inadequate formity of aggregates—may be of importance. However, today,
to provide resistance against damage by cyclic freezing; poor with longer-range shipment of concrete-making materials,
finishing may have created incipient delaminations in the sur- changes to cement manufacture, use of supplemental cemen-
face region; there may have been undesirable chemical reac- titious materials, and due to material shortages, there is a
tions of the aggregates; and aggressive chemicals may have broader overlap of different materials so that new problems
been introduced from the environment. may arise in areas previously thought immune to some prob-
Each concrete and each part of a concrete is unique in lems. For example, whereas low alkali cements may have once
terms of composition, texture, history, and exposure. Deci- been common to a regional area because of local cement pro-
phering the texture and composition of concrete is where pet- duction, increased shipment of higher-alkali cements into an
rography shines. area can change the previous “immunity” of some siliceous ag-
gregates to alkali-silica reactivity, or significantly affect the ef-
Observations ficacy of air-entraining and other admixtures.

Petrography is the science of observation, whether on a macro Reconstruction of History of Field Concrete
or micro scale. In a sense, everyone is a petrographer on a
macro scale. The petrographic effort becomes valuable on a To progress from consideration of simple petrographic exami-
micro scale, which requires patience, skill, and knowledge—not nation to the petrographic examination of concrete that has
only about microscopes, as most people recognize, but also aged and perhaps deteriorated in service introduces two im-
about ancillary techniques to which attention is directed based portant new unknowns—time and the precise environment of
upon microscopical evaluations. the concrete. The effects of the passage of relatively short peri-
The cupboard of a petrographic laboratory, from an instru- ods of time on the constituents present in several cement pastes
mentation viewpoint, must include: (1) a series of light-optical of known w/c stored under laboratory conditions have been in-
instruments that includes a low-power stereomicroscope, petro- vestigated. However, anomalies remain in the results, even
graphic microscope, and may also include a metallurgical though compositions of pastes and nature of the environments
microscope; (2) specimen preparation equipment that includes were known and controlled far more thoroughly than the com-
saws, grinding and lapping equipment; and (3) accessory equip- position and environment of any field concrete. Today, where
ment and supplies needed to make each of the preceding effec- there is extensive use of fly ash, silica fume, and ground granu-
tive (see ASTM C 856). Supplementary equipment may include lated blast-furnace slag, there are even more anomalies.
a scanning electron microscope and an ancillary elemental
probe. A variety of analytical instrumental methods that include Composition
X-ray diffractometry, differential scanning calorimetry, differ-
ential thermal analysis, infrared spectroscopy, a variety of in- Petrographic examinations provide the most direct and more
struments for elemental analysis, and wet chemical methods information about concrete than any other technique because
 ERLIN ON PETROGRAPHIC EXAMINATION 211

it can be used for identifying its physical and chemical makeup, to which the data will be used; and (5) the needs of those who
and its proportions and factors leading to an understanding of either want to accept the estimates or debunk the estimates—for
its behavior. Things that can be deciphered include: (1) the types whatever reason.
of cementitious materials present, the products that have ASTM Practice for Use of the Terms Precision and Bias in
formed as a consequence of their chemical reactions, and ASTM Test Methods (E 177) says, in so many words, that from
estimates of amounts present; (2) aggregates, their rock and a statistical viewpoint it is virtually impossible to have an ac-
mineralogical types, grading, uniformity of distribution, and in- tual, true reference to base accuracy on. As a result, in ASTM
formation about their chemical and physical stability or insta- test documents where numerical values are obtained, there is a
bility; (3) estimates of water-cementitious materials ratios and mandate that precision and bias statements, suitably qualified,
their uniformity throughout a concrete; (4) air-void character- be used to express accuracy.
istics, whether non-air-entrained or air-entrained, distribution Here is an example of how deceptive the w/c expressed as
of the voids and parameters of air-void systems; and (4) features an average or mean value of a number of estimates can be.
related to consolidation, finishing, curing, premature carbona- Jakobsen et al. [29], in a paper on the fluorescence method for
tion, surface dusting, aggregate coatings, mineral surface estimating the w/c, used round robin tests to provide informa-
hardeners, depth of carbonation, and cracks. tion on the accuracy of the method. Their “E” Round Robin
One of the many items of interest and concern about con- Test Series has a mean w/c value of 0.47 versus the mix design
crete is the water-cementitious materials ratio. Petrographers w/c of 0.44, a difference of 0.03. That doesn’t sound too bad.
have been using a variety of techniques for estimating the w/c However, the six w/c values reported are from 0.40 to 0.55 (a
and water-cementitious materials ratio (w/cm) of hardened difference of 0.15), which means that the variation from the
concrete. All of them are subjective and rely upon the petrog- mean is from 0.07 to 0.08. Although average or mean val-
rapher and his or her experience and expertise in interpreting ues may look good, the range of values is hidden. The reported
the data, and they are generally accepted by the concrete standard deviation of 0.52 for those values is more realistic for
industry. expressing the variance of values of a method than the mean
There are four general methods among petrographers or average.
that are popular today: (1) a water-droplet absorption tech- ASTM Committee C09.65 on petrography of hardened con-
nique [26]; (2) scratch hardness [26] and indentation hardness crete is currently assessing different methods petrographers
[27] techniques; (3) a combination of 12 or so microscopical use for estimating the w/c of hardened concrete. The methods
and physical observations of the paste technique [28,29]; and and techniques used for estimating the elusive w/c and w/cm
(4) methods where thin sections of concrete are impregnated of hardened concrete may never be developed to the satisfac-
with dyed epoxy and are compared to thin reference sections tion of all, but perhaps when that assessment is completed and
made from concrete having known w/c’s. The latter includes some verification studies are done, the distance to the estimate
two techniques in which the dye is either blue [26] or fluores- may well be narrowed.
cent [29]. Sometimes combinations of methods are used.
Unfortunately, there are few specific details in the litera- Environment
ture of how to translate what is observed to assessing the w/c
using these methods, except for the fluorescence method. That Singularly, the environment to which concrete is exposed has
method originated in Europe and has been evolved by a hand- the greatest influence on its performance. If the petrographic
ful of people over which there has been controversy about its examination is of laboratory-made concrete from an environ-
claimed accuracy of 0.02 when estimated using a manual mentally controlled exposure, many key factors controlling its
technique, and an accuracy of 0.01 when estimated using a performance and consequent alteration can be more easily de-
semi-automatic technique. There is nothing in the literature ciphered than if the examination were of field concrete, in
about the accuracy of the scratch hardness, water droplet which many variables are unknown. Usually, for samples of
absorption, and blue tone methods. The w/c estimate using the field concrete, there is a decrease of available background in-
combined optical and physical paste properties technique is formation compared to examinations of new concrete or labo-
reported to cover a 0.02 range. ratory test specimens of concrete.
The degree of interpretation of data from each method Sometimes an answer to the specific questions posed is
is based upon the comfort of the petrographer in extending not interpretable from the data initially uncovered even though
his or her expertise to provide that estimate. That comfort the petrographer has recovered evidence not accessible by
zone is from no interpretation (no comfort) to 0.01 and other approaches. Yet, the results of a petrographic study can
0.02 (a relatively high degree of comfort) to 0.03 to 0.06 be used to eliminate certain factors or direct attention to oth-
(a range of intermediate comfort levels) to 0.10 (a low com- ers. Complicating the interpretation is the fact that the deterio-
fort level)—and some petrographers even decline to provide rated field concrete may have performed abnormally because
an estimate. The deftness, skill, and experience of the petrog- of more than one cause.
rapher will usually dictate his or her comfort zone and, Field concrete examined by a petrographer is concrete
hence, the degree of interpretation of the items used to that has worried a responsible person enough to put forth the
estimate the w/c. However, sometimes confounding that in- effort and expense of sampling and testing. Usually concrete
terpretation is the existing concrete condition, depending, for sent for petrographic examination is “controversial” concrete.
example, on the varieties of chemical and physical alteration In practice, this usually means that sampling was from the poor
that the concrete has undergone. concrete, and the results are frequently extrapolated to the en-
The acceptability of the petrographic estimates depends tire area of concrete in a project. Unless samples representing
on: (1) the competency of the petrographer; (2) the validity of all field and other conditions and exposures are included, there
the techniques involved; (3) the ability of the petrographer to is a built-in bias in the sampling process—good concrete is
qualify the techniques used for the estimates; (4) the purpose rarely controversial.
212 TESTS AND PROPERTIES OF CONCRETE

Although, for example, it can be deduced from petro- recognizable peculiarities in cases of unsatisfactory service with
graphic examinations of old outdoor-exposed concrete that the possibly expensive consequences. Alkali-carbonate and alkali-
w/c was high or low, the cement content was low, medium, or silica reactions are described in other papers in this publication.
high, fly ash or other supplementary cementing materials were Because of varying specimen orientation and variable
or were not present, general quality of workmanship was good exposure conditions, features symptomatic of specific types of
or inadequate, finishing was improper, etc., the alkali content distress may not always be similar.
of the cementitious materials cannot usually be reconstructed
although it may have helped direct the performance of the Petrographic Assistance in Other Types of
concrete. Examinations
Finally, deteriorated field concrete usually shows superim-
posed traces of several processes, with at least one in an ad- The identification of compositional properties of portland
vanced stage. The most advanced process may conceal evidence cement-based products analyzed using other methods and tech-
of others that had a more important effect. niques is sometimes essential for proper interpretation of the
Evaluations of data from laboratory test exposure samples analytical data. For example, ASTM Standard Test Method for
make the interpretative process easier because the materials Analysis of Hardened Masonry Mortar (C 1324) (also applicable
and mixture design are known compared to natural exposures to portland cement-based plaster and stucco) has an initial de-
where many factors are unknown and environmental condi- tailed section that deals exclusively with petrography and that
tions are variable. Consequently, laboratory-induced proce- supplements the second section on chemical analysis and al-
dures often result in different diagnostic “symptoms” than lows suitable interpretation of the chemical data because these
evidenced by field-exposed concrete. For example, samples of materials are made using portland cement and other calcare-
field concrete, when examined using light optical microscopy, ous and siliceous materials that skew the chemical data. ASTM
are frequently found to contain secondary calcium carbonate Standard Test Method for Portland-Cement Content of Hard-
near outer surfaces, along old cracks, and sometimes in the in- ened Hydraulic-Cement Concrete (C 1084) provides for petro-
terior. That calcium carbonate, when examined by optical graphic examinations as a means of understanding chemical in-
methods, is generally found to be calcite, rarely aragonite, and fluences of constituent materials, other than portland cement,
almost never vaterite (the birefringent spherulitic form of cal- on the analysis. These materials include aggregates, pozzolans,
cite containing interstitial water). Vaterite, however, was found supplementary cementing materials, and mineral admixtures.
by optical methods to be common in mortar bars that had been Petrographic analysis of post-test specimens, such as specimens
tested according to ASTM Test Method for Potential Alkali Re- from a variety of ASTM tests involving alkali-silica or alkali-car-
activity of Cement-Aggregate Combinations (Mortar-Bar bonate reactivity, is mentioned in those test standards as a
Method) (C 227) and had been found in concrete specimens means of providing supplemental assessment of damage.
tested for resistance to freezing and thawing according to
ASTM Test Method for Resistance of Concrete to Rapid Freez- Conclusion
ing and Thawing (C 666).
The use of X-ray diffraction to examine cement paste con- The kinds of information about hardened concrete that pet-
centrates from field concrete has revealed that vaterite (not rographic examinations reveal can be used to identify a great
easily identified using optical methods) is frequently a major variety of data about its physical and chemical makeup. That
constituent of secondary calcium carbonate [30], especially information, in turn, can be used to identify unsound aspects
on samples from seawater exposures or from other wet envi- of portland cement that have led to bulk concrete expansion
ronments. Vaterite is known to persist for several months in [34,35]; improprieties of aggregates that can cause low
laboratory specimens stored at room conditions; however, concrete strength [36]; stray naturally occurring organic
conversion of aragonite and the poorly crystallized vaterite to chemicals in aggregates that can cause variable concrete air
well-crystallized calcite in pastes and mortars has been contents [8]; the causes of concrete failure to set; rapid slump
clarified based upon work in Ref 31 that identified naturally loss; poor proportioning and batching; factors resulting in
occurring vaterite [32]. large concrete shrinkage; early and late concrete instability
Concrete exposed to accelerated freezing and thawing in due to a variety of causes; conformance to specifications;
water using Procedure A of ASTM C 666 can produce a charac- malperformance due to chemical reactions of aggregate;
teristic loss of surface mortar that is unlike the condition of surface distress due to improper finishing, curing, and inade-
specimens exposed, for example, on the mean-tide rack at East- quate air entrainment; and poor performance due to compo-
port, Maine [33]. Field concrete that is not air-entrained and has sition, manufacture, and exposure to aggressive environments
deteriorated because of natural freezing and thawing usually for which it is not designed.
develops sets of sub-parallel cracks normal to the direction of Before the petrographic trail was blazed and during its
the freezing front (because of directional freezing) that may not early adventuresome period, the factors that led to good or
be reproduced in accelerated freezing and thawing in water. poor concrete performance were unclear; even today, with
Field concrete that has deteriorated because of alkali-silica new and exciting methods and techniques being constantly
reactions usually has much more advanced and conspicuous developed, we still have some ignorance concerning these
internal symptoms of this reaction than are found in mortar factors. However, there is no need to view our analytical level
bars of expansive combinations examined after they are tested of understanding as insignificant.
according to ASTM C 227. On the other hand, some field It makes no difference if the concrete is a 10 ton cherry
concretes regarded as undeteriorated have shown a range of pie 14 ft (4.2 m) in diameter and 2 ft (0.6 m) deep constructed
evidence of alkali-silica reaction. in Charlevoix, Michigan to commemorate the birthday of
Alkali-carbonate and alkali-silica reactions sometimes exist George Washington, or a dam, nuclear reactor facility, bridge,
concurrently in varying degrees of development in some con- foundation, wall, pavement, sidewalk, concrete electrical insu-
cretes, and inconspicuous degrees of reaction may be the only lator, the liner of a bank safe, or coffin vault located anywhere
 ERLIN ON PETROGRAPHIC EXAMINATION 213

in the world; the basic constituents are similar at any age. Pet- [10] Idorn, G. M., Durability of Concrete Structures in Denmark,
rographic methods are ideal for evaluating how concrete has Technical University of Denmark, Copenhagen, Jan. 1967.
been put together and changes that have occurred to it since it [11] Klemm, W. A., Skalny, J., Hawkins, P., and Copeland, L. E.,
was installed, and provides an understanding of its behavior “Cement Research: Boon or Boondoggle,” Rock Products,
and performance. It is invaluable in assisting engineers and Cement International, Vol. 80, No. 4, 1977, pp. 156–170.
others in understanding why it failed to function as expected [12] Porvar, T. O. and Hammersley, G. P., “Practical Concrete Petrog-
raphy,” Journal, Concrete Society, England, Vol. 12, No. 8, 1978,
and, less often, why it functioned as expected.
pp. 27–31.
Petrography of concrete is flourishing. Like in other
[13] Bennitt, J. G., “Petrography: New Man on the Construction
sciences, the development of new methods and techniques of Team,” Concrete Construction, 1981, pp. 585–587.
examination, more sophisticated equipment, more qualified [14] Jensen, A. D., Eriksen, M., Chatterji, S., Thaulow, N., and Brandt,
petrographers, more laboratories available to complete work, I., “Petrographic Analysis of Concrete,” Danish Building Export
and better understanding of the physical and chemical makeup Council, Ltd., Denmark, 1984.
of concrete now make possible better interpretations of the [15] Pitts, J., “The Role of Petrography in the Investigation of
causes of concrete performance and malperformance. Concrete and Its Constituents,” Journal, Concrete Society, July
Along with improvements in analytical methods and 1987, pp. 5–7.
equipment comes a new problem—the development of petrog- [16] Petrography Applied to Concrete and Concrete Aggregates,
raphers who are able to obtain the “right” analytical data and ASTM STP 1061, B. Erlin and D. Stark, Eds., ASTM International,
also have the background to interpret the data. There is a need West Conshohocken, PA, 1990.
for hands-on appreciation of mixtures of portland cements; [17] “Water-Cement Ratio and other Durability Parameters,”
blended cements; mineral and chemical admixtures; water; American Concrete Institute, SP-191, 2000.
varieties of coarse and fine aggregates; methods of mixing, [18] St. John, D. A., Poole, A. W., and Sims, I., Concrete Petrography,
placing, and consolidation; the physics and chemistry of what John Wiley & Sons, New York, 1998.
happens to and within plastic and hardened concrete; and the [19] Paster, T. R., “Petrography for Contractors,” Concrete Construc-
tion, Addison IL, March 2003.
effects of time and environmental exposure on performance.
[20] “Evaluation of Methods of Identifying Phases of Cement
Too frequently, the analyst becomes a specialist who operates
Paste,” Transportation Research Circular No. 176, W. L. Dolch,
in a narrow walkway of self- or organization-imposed semi- Ed., Transportation Research Board, National Research Council,
isolationism. Washington, DC, June 1976.
In this day of specialization, the techniques and particu- [21] “Guide to Compounds of Interest in Cement and Concrete
larly the analytical equipment are very expensive. The cost of Research,” Special Report No. 127, Highway Research Board,
getting petrographic information dramatically increases as the National Research Council, Washington, DC, 1972.
size and scale of what is analyzed progressively decreases. [22] Proceedings, Fifth International Symposium on the Chemistry of
However, there is still a lot of information obtainable at Cement, Tokyo, 1968, Vols. I–IV, 1969.
moderate and justifiable costs. [23] Proceedings, Sixth International Symposium on the Chemistry
of Cement, 1974; Moscow Stroyizdat 1976; in part, available in
English preprints.
[24] Cement and Concrete Research, an International Journal
References (bimonthly), D. M. Roy, Ed., Pergamon Press, New York, 1971.
[1] Johnson, N. C., “Application of the Microscope to the Study of [25] Neville, A., “Can We Determine the Age of Cracks by Measuring
Concrete,” Engineering Record, 23 Jan., 6 Feb., 13 Feb., 27 Feb., Carbonation,” Concrete International, Part 1, December 2003,
6 March, and 13 March 1915. Part 2, January 2004.
[2] Brown, L. S., “Petrography of Cement and Concrete,” Journal, [26] Liu, J. J. and Kahn, M. S., “Comparison of Known and
Portland Cement Association, Research and Development Determined Water-Cement Ratios Using Petrography,” Water-
Laboratories, Vol. 1, No. 3, Sept. 1959, pp. 23–24. Cement Ratio and Other Durability Parameters, SP-191, ACI
[3] Mielenz, R. C., “Petrography Applied to Portland-Cement International, Farmington Hills, MI, 2000.
Concrete,” Reviews in Engineering Geology, Significance of [27] Erlin, B. and Campbell, R. C., “Paste Microhardness—Promising
Tests and Properties of Concrete and Concrete-Making Technique for Estimating Water-Cement Ratio,” Water-Cement
Materials, Geological Society of America, Boulder, CO, Vol. 1, Ratio and Other Durability Parameters, SP-191, ACI Interna-
1962, pp. 1–38. tional, Farmington Hills, MI, 2000.
[4] Mielenz, R. C., “Diagnosing Concrete Failures,” Stanton Walker [28] Erlin, B., “Other Viewpoints on Petrographic Reports,”
Lecture Series on the Materials Sciences, University of Mary- Reader Response, Concrete Construction, Addison IL, February
land, College Park, MD, Nov. 1964. 2000.
[5] Mielenz, R. C., “Why Do Some Concretes Fail,” Concrete Con- [29] Jakobsen, U. H., Laugesen, P., and Thaulow, N., “Determination
struction, Sept. 1972. of Water-Cement Ratio in Hardened Concrete by Optical
[6] Mather, K., “Petrographic Examination,” Significance of Tests Fluorescence Microscopy,” Water-Cement Ratio and Other
and Properties of Concrete and Concrete-Making Materials, Durability Parameters, SP-191, ACI International, Farmington
ASTM STP 169A, ASTM International, West Conshohocken, PA, Hills, MI, 2000.
1966, pp. 125–143. [30] Mather, K. and Mielenz, R. C., “Cooperative Examination of
[7] Dreizler, I., “The Microscopy of Concrete,” Zement-Kalk-Gips, Cores from the McPherson Test Road,” Proceedings, Highway
No. 5, 1966, pp. 216–222. Research Board, Washington, DC, 1960, pp. 205–216.
[8] Erlin, B., “Methods Used in Petrographic Studies of Concrete,” [31] Cole, W. F. and Kroone, B., “Carbonate Minerals in Hydrated
Analytical Techniques for Hydraulic Cement and Concrete, Portland Cement,” Nature, Vol. 184, 1959, p. B.A.57.
ASTM STP 395, ASTM International, West Conshohocken, PA, [32] McConnell, J. D. C., “Vaterite from Ballycraigy, Larne, Northern
1966, pp. 3–17. Ireland,” Mineralogical Magazine, Vol. 32, No. 250, Sept. 1960,
[9] Erlin, B., “Analytical Techniques,” Observations of the Perfor- pp. 535–544.
mance of Concrete in Service, Special Report 106, Highway [33] Kennedy, T. B. and Mather, K., “Correlation Between Labora-
Research Board, Washington, DC, 1970, pp. 29–37. tory Accelerated Freezing and Thawing and Weathering at
214 TESTS AND PROPERTIES OF CONCRETE

Treat Island, Maine,” Proceedings, American Concrete Institute, [35] Erlin, B. and Jana, D., “Forces of Hydration That Can Cause
Vol. 50, 1953, pp. 141–172. Havoc in Concrete,” Concrete International, Vol. 25, No. 11,
[34] Erlin, B., “The Magic of Investigative Petrography: The Practical November 2003.
Basis for Resolving Concrete Problems,” Petrography Applied [36] Davis, R. E., Mielenz, R. C., and Polivka, M., “Importance of
to Concrete and Concrete Aggregates, ASTM STP 1061, B. Erlin Petrographic Analysis and Special Tests Not Usually Required in
and D. Stark, Eds., ASTM International, West Conshohocken, Judging Quality of Concrete Sand,” Journal of Materials, Vol. 2,
PA, 1990, pp. 171–181. No. 3, Sept. 1967, pp. 461–486.
21
Volume Change

Fred Goodwin1

Preface are discussed elsewhere in this publication and therefore are

not covered by this chapter.
IN THE PREPARATION OF THIS CHAPTER, THE There are other types of volume change that occur as a
contents of the 4th edition of this series, ASTM STP 169C, result of concrete’s response to environmental effects such as
were drawn upon. The author acknowledges the author of wetting, drying, heating, cooling, freezing, and thawing. Ther-
the volume change chapter in the 4th edition, P. Kumar mal volume changes are discussed in the chapter by Tatro, and
Mehta [1], as well as those in the previous editions. In ASTM moisture effects are covered in the chapter by Hearn et al. Due
STP 169 and ASTM STP 169A, the chapter on volume to the relatively low tensile strength of concrete, nonuniform
changes in hardened concrete was authored by George W. restrained expansions or contractions can easily exceed the
Washa [2,3]. In ASTM STP 169B, the chapter was authored cohesive tensile strain capacity of the concrete resulting in the
by James L. Sawyer [4]. In the interest of consistency, the es- formation of cracks.
sential format as well as many of the fundamental aspects of Ingress of sulfate ions from the environment or an excess
the earlier publications are retained in this paper. The cur- of sulfate intermixed within the concrete causes an expansive
rent edition will review and update the topics as addressed by reaction known as sulfate attack resulting in the formation of
the previous authors, introduce new technology that has been ettringite (calcium sulfo-aluminate). This expansive mecha-
developed, and include up-to-date references. nism is discussed in the chapters by Forster and Meininger.
Concrete’s reaction with carbon dioxide from the environ-
Introduction ment (carbonation) lowers the pH (thereby lowering the cor-
rosion threshold of imbedded reinforcing steel) as well as
Concrete is subject to several types of volume changes during causes shrinkage due to chemical reaction with the cement hy-
its service life. The magnitude of volume change is generally dration product. Carbonation that occurs during the first few
stated in linear rather than volumetric units, since test data hours following concrete placement causes a soft, dusty sur-
are generally evaluated in these terms. In engineering prac- face on the area that is exposed to the carbon dioxide [5]. Cor-
tice, the volume change is generally expressed as “strain,” rosion of imbedded steel, while not a volume change of the
which is defined as the change in length per unit length and concrete constituents, still results in cracking and spalling of
therefore is a dimensionless quantity. The term “microstrain” the concrete due to the expansive nature of the formation of
is frequently used to quantify the magnitude of shrinkage. steel oxidation products within the rigid concrete matrix.
One microstrain is 1 millionth of a unit length change per Uncombined crystalline calcium oxide (CaO) and magne-
unit length change (i.e., 1 millionth of a metre per metre or sium oxide (MgO) present in the cement or intermixed within
1
106 inch per inch). the cement cause another expansive mechanism as the
Freshly mixed concrete remains plastic for a relatively hydroxides of these minerals form during hydration.
short period of time. During this period, volume changes can The reaction of alkalis (sodium and potassium) with
occur due to cement hydration, water evaporation, thermal certain types of aggregates as described in ASTM C 295 and
change, absorption, and bleeding. These types of volume other references produces expansive reaction products
changes, commonly called “plastic shrinkage,” are discussed causing cracking and potential disintegration of the concrete.
elsewhere in this book. This topic is covered in more detail in the chapters on
The instantaneous strain on loading that is fully re- petrographic examination by Erlin and Wong, respectively.
versible and also is proportional to the applied stress is called Volume changes that are not covered elsewhere include car-
“elastic strain.” On sustained loading, especially when the ap- bonation shrinkage, drying shrinkage, and expansion due to hy-
plied stress level is 50 % or more of the ultimate strength of dration of free CaO and MgO in hardened concrete. Also dis-
concrete, the material shows nonlinear stress-strain behavior. cussed in this chapter is the autogenous shrinkage associated
The gradual increase in strain with time under a constant with cement hydration, which can occur in sealed and unloaded
load, which is not completely reversible on unloading, is specimens. The last section of this chapter contains a brief
termed “creep strain” or, more commonly, “creep.” The description of volume change in expansive cements that are
strains resulting from concrete’s response to applied stress used in the production of shrinkage-compensating concrete.

1
Principal Scientist, Degussa Construction Systems, Cleveland, OH 44122.

215
216 TESTS AND PROPERTIES OF CONCRETE

Under conditions of restraint, the volume changes involving less absolute volume than the volumes of the reactants, resulting
shrinkage commonly produce cracking through the generated in the densification of the solid phases during the hydration
tensile stress from shrinkage exceeding the tensile strain capac- process [8]. Chemical shrinkage has been determined by
ity of the concrete. Significant work is underway to relate shrink- dilatometry and by the change in weight of a hydrating sample
age, creep, tensile strength, and modulus of elasticity; however, suspended in water [10]. The factors affecting chemical shrink-
standards are only now beginning to be developed [6,7]. age of hydrating cement systems include the mineralogical com-
Concrete typically is under sufficient restraint due to connecting position of the cement, the particle size distribution of the ce-
members, foundations, subgrade friction, steel reinforcement, ment [11], the kinetics of the hydration reaction, and the use of
etc. to exceed the tensile strain capacity. While cracking may not supplemental cementing materials and other admixtures
be serious enough to jeopardize the structural integrity of an [12,13]. The magnitude of chemical volume change in concrete
element, it is still undesirable for a variety of reasons. for a given consistency increases as the fineness of cement and
Cracks create a plane of weakness within the formerly the cement content are increased. Ultimate contraction appears
homogeneous structure. Continuing movement of cracks to be somewhat greater for high C-S-H forming cements
frequently causes problems with coatings, toppings, or over- (namely, ASTM Type IV cement) than for normal portland ce-
lays installed as barriers to protect the underlying concrete. ment (Type I). The addition of pozzolans such as silica fume and
Cracks in structures subjected to traffic can rapidly deteriorate metakaolin, which increase the amount of CSH formation, can
the adjacent concrete due to point loading of the crack edges therefore also increase the amount of autogenous shrinkage.
as well as D-cracking in freezing environments. Cracks also Chemical shrinkage is the dominant factor in autogenous
look bad and are the most frequent subject of complaints shrinkage during the time prior to setting or the formation of a
about concrete, particularly architectural concrete. More self-supporting skeleton of hydration products. Self-desiccation
importantly, large external cracks, interlinking with internal becomes dominant after setting [14,15].
voids and micro cracks that are always present in concrete, Occasionally, a period of swelling has been observed be-
make it possible for water, harmful chemicals, and gases to ginning a few hours after initial mixing and lasting up to 20 h.
penetrate with relative ease into the interior of the concrete This appears to be due to the formation of ettringite as some
mass. This phenomenon is likely the most common cause of correlation with C3A, C4AF, and gypsum levels has been found
numerous concrete durability problems with field structures. [16–18].
Unrestrained, expansive volume changes also cause Self-desiccation results from the consumption of the in-
cracking and are therefore harmful. Frequently, restrained ternal water due to the hydration of the cement. The formation
controlled volume changes are used to shrinkage-compensate of a rigid structure of decreasing permeability as hydration
concrete and mortars in an attempt to produce expansion at progresses decreases the migration of water to the hydrating
rates and magnitudes approximately equal to the shrinkage. regions near the cement particles. Air fills the pore structure of
Many of these mechanisms rely on attempts to control the concrete resulting in a decrease of internal humidity within
otherwise deleterious reactions such as ettringite formation, the system. Hydration can cease to occur when the relative hu-
uncombined calcium oxide, and metal oxidation. If condi- midity of the pore structure falls below 75–80 % [19,20]. The
tions exist in which insufficient restraint or inhomogeneous development of high-strength systems with very low water-
uncontrolled expansion occurs, internal fractures can develop cement ratios has brought the phenomenon of autogenous
within the concrete causing failure. shrinkage to the attention of researchers again [21,22]. Using
cement pastes made with a 0.23 or 0.30 water-cement ratio, a
Autogenous Volume Changes high early-strength portland cement (430 m2/kg, Blaine), and a
superplasticizer, Tazawa and Miyazawa [23] reported that the
Autogenous volume changes are associated with cement hy- autogenous shrinkage of sealed specimens at an age of 70 days
dration alone and do not include environmental effects due to was of the order of 1000 microstrain. This autogenous shrink-
variation in moisture and temperature. Autogenous deforma- age value increased to almost double when the cement was
tion combines the mechanisms of chemical shrinkage and self- replaced with 10 or 20 % condensed silica fume by weight.
desiccation [8]. The autogeneous volume change with ordinary
portland cement concrete is usually small, that is, less than Carbonation Shrinkage
0.010 % or 100-microstrain expansion or shrinkage. The mag-
nitude is dependent on the overall effect of two opposing phe- Carbonation shrinkage occurs as a result of chemical interac-
nomena: (a) the increase in the disjoining pressure in poorly tion between atmospheric carbon dioxide (CO2) and hydration
crystalline C-S-H2 and ettringite due to water adsorption and products of cement. Carbonation shrinkage is probably caused
(b) the reduction in the disjoining pressure due to removal of by dissolving crystals of Ca(OH)2 while under compressive
adsorbed water by desiccation. According to Washa [2], in stress imposed by drying shrinkage and the subsequent depo-
most instances the initial expansions obtained during the first sition of CaCO3 in spaces free from stress [24]. Hydration prod-
few months do not exceed 0.003 %, while the ultimate shrink- ucts near the pore dissolve continuously to furnish Ca ions
age obtained after several years usually does not exceed 0.010 to the liquid phase from which the ions migrate to the CaCO3
%. Differences between volume and length change measure- nucleating site, where the crystal growth takes place.
ments of autogenous deformation have been reported by Since it takes place concurrently with drying shrinkage,
Barcelo et al. [9]. most reported data do not distinguish between the two and
Chemical shrinkage results from the chemical reaction of designate both as drying shrinkage. As the cement hydrate is car-
the cement with water, which produces hydration products of bonated, the mass of the sample increases. Drying shrinkage is

2
Cement chemistry abbreviations are used: C  CaO, S  SiO2, A  Al2O3, F  Fe2O3, S  SO3, and H  H2O.
 GOODWIN ON VOLUME CHANGE 217

accompanied by a loss of water resulting in a mass loss. This in- method. It was found that superplasticized concrete had a lower
verse relationship can prove misleading when the specimen depth of carbonation than the corresponding normal concrete.
mass is used to determine the equilibrium point of drying [25]. Malhotra et al. [31] reported that a superplasticized, high-
The rate of carbonation is dependent on several factors volume fly ash concrete containing 56 % fly ash by weight of the
such as porosity of concrete, size of the member, relative hu- cementitious material, with 30 MPa compressive strength at 28
midity, temperature, CO2 concentration, time of exposure, days, showed only 3 mm carbonation in one year and 7 mm in
method of curing, and the sequence of drying and carbonation. five years at 50  5 % relative humidity. Paillere et al. [32] found
Carbonation proceeds slowly and usually produces little that portland cements containing up to 29 % Class F fly ash or
shrinkage at relative humidities below 25 % or near saturation. blast furnace slag showed similar behavior to carbonation as
At low humidities, there is insufficient water in the pores for neat portland cements. At higher levels of cement replacement,
the CO2 to form carbonic acid and react with the alkaline ce- the carbonation of concrete increased.
ment hydrates. As the pore humidity approaches saturation, Oye and Justness [33] investigated the carbonation resist-
the diffusion rate of CO2 decreases. With permeable concrete, ance of latex-modified cement mortars and reported that, con-
the magnitude of carbonation shrinkage may approach the trary to a general improvement in other properties, the carbona-
magnitude of drying shrinkage in a CO2-rich environment at tion resistance of latex- and epoxy-modified mortars showed no
50–65 % relative humidity [26]. Concrete that has been sub- significant improvement when compared to unmodified mor-
jected earlier to carbonation shrinkage will still shrink or swell tars. Aimin and Chandra [34] investigated the effect of acrylic
with changes in relative humidity; however, the magnitude of polymer additions on the rate of carbonation of portland cement
these volume changes will be smaller than before carbonation. paste. Increasing the polymer content reduced the porosity and
Carbonation reactions, such as the one shown here, tend water absorption of the cement paste, but the rate of carbonation
to release moisture reached maximum at 10–15 % polymer loading. Sakuta et al. [35]
found that the addition of amino alcohol and glycol ether deriv-
Ca(OH)2  CO2 → CaCO3  H2O atives was effective in controlling carbonation. Amino alcohol
derivatives that are used for desulfurization and deoxidation pos-
All of the constituents of hydrated portland cement, sess the ability to absorb atmospheric CO2. The use of a water-in-
namely, C-S-H, CH, ettringite, and monosulfate hydrate, are soluble glycol ether derivative as an anti-foaming agent reduced
eventually subject to carbonation. The porosity of the cement the size of air bubbles in concrete thereby decreasing the per-
paste and the specimen size are controlling factors in carbon- meability. Paillere et al. [32] investigated the effect of freezing
ation because they determine the diffusion rates of CO2 ingress and thawing (F-T) cycles on air-entrained concrete containing fly
and the egress of moisture released by CO2 reacting. While au- ash or slag. With increased F-T cycles, carbonation depths were
toclaved blocks are relatively free from carbonation, concrete increased. Whereas the presence of entrained air improved the
blocks cured in steam at atmospheric pressure show maximum resistance to freezing and thawing cycles, it did not help against
carbonation shrinkage near 50 % relative humidity [26]. Also, the penetration of CO2; Massazza and Oberti [36] reported that
it is reported that precarbonation improves the volume stabil- carbonation depths of pozzolanic cements were similar to those
ity of blocks cured in steam at atmospheric pressure. of portland cements provided that concretes having similar
Ying-yu and Qui-dong [27] investigated the mechanism of strengths were compared. The carbonation behavior of blended
carbonation of cement mortars and the dependence of car- cements was inferior to portland cement only when the moist-
bonation on pore size. Since the gaseous-phase carbonation curing period was too short. With prolonged moist curing, there
process is a chemical reaction controlled by diffusion, the au- was usually an improvement in carbonation resistance of
thors suggest “prediction of the carbonation coefficient” by the blended cements.
equation Sakai et al. [37] compared the carbonation characteristics
of a Type K expansive cement concrete with ordinary portland
a1(2C1/kP)1/2 cement concrete from a 22-year-old building in Nigata Prefec-
ture, Japan. Core samples with 58 MPa compressive strength
where C1 is the partial pressure of CO2, k is a constant, and P and 0.57 water-to-cement ratio were found to have carbonation
is porosity. The authors found that pores with radii above 32 depths of about 12 mm. Mineralogical analysis of the carbon-
nm increased the carbonation coefficient greatly. ated material confirmed that C-S-H was decomposed to calcite
Factors influencing the depth of carbonation in different and silica gel, whereas the ettringite phase in the expansive ce-
concrete types were discussed in several studies. Schubert [28] ment concrete decomposed to calcite, gypsum, and Al(OH)3 gel.
reported the carbonation behavior of mortars and concretes The authors observed that the decomposition rate of ettringite
containing different types of cements and fly ashes. The author by carbonation was somewhat slower than that of C-S-H.
found an inverse relationship, (1/
fc  ), between the carbona-
tion rate and the 28-day compressive strength. Cements with Drying Shrinkage
unusually low lime contents (that is, 70 % slag cement) showed
increases in carbonation rates. Carbonation depths of mortar When plain, normal-weight, hardened concrete is dried from a
specimens moist-cured for 28 days were much less than the 7- saturated condition to a state of equilibrium with air at 50 %
day moist-cured specimens. For fly ash concretes with seven relative humidity, the shrinkage associated with moisture loss is
days of standard moist curing, Ho and Lewis [29] found that in the range of 0.04–0.08 % (400 to 800 microstrain). The change
carbonation depth depended mainly on the water-cement, not in volume of drying concrete is not equal to the volume of wa-
the water-binder, ratio regardless of the mix constituents used ter removed. The initial loss of water causes little or no shrink-
in their study. age since the most easily evaporable water resides in large pores
Dhir et al. [30] studied the effect of 28-day concrete strength of the concrete. Further evaporation of adsorbed water is
on depth of carbonation in 20 years, using an accelerated test reversible upon resorbtion of additional moisture [38]. One
218 TESTS AND PROPERTIES OF CONCRETE

source of irreversible drying shrinkage in concrete is the loss of values. Portland cements with high-C3A and high-alkali
water held in small capillary pores of the hydrated cement paste contents tend to give high drying shrinkage, which can be
[39]. It has been suggested that the evaporation of capillary wa- controlled by using optimum gypsum content via ettringite for-
ter causes a disjoining pressure when confined to narrow spaces mation and the associated shrinkage-compensating effect (see
between two solid surfaces. The removal of the capillary water the last section of this chapter that deals with testing methods).
reduces the disjoining pressure and brings about the shrinkage Admixtures that increase the water requirement of con-
of hydrated cement paste on exposure to drying conditions. In crete for a given consistency (namely, diatomaceous earth or
regard to capillary water, it has been suggested that the water calcined clays) generally increase drying shrinkage but many
meniscus in small capillaries (5–50 nm) exerts hydrostatic water-reducing admixtures that reduce the water content do
tension, and removal of this water tends to induce a compressive not reduce drying shrinkage. Accelerating admixtures such as
stress on the walls of the capillary pores, thus contributing to the calcium chloride and triethanolamine tend to increase drying
overall contraction of the system. The introduction of shrinkage- shrinkage.
reducing admixtures has allowed a reduction in drying Atmospheric diffusion of the adsorbed water present within
shrinkage of up to 50 %. Most of these materials reportedly work cement paste and the water held by capillary tension takes place
by changing the capillary water surface tension [40]. over a long period of time and is accelerated by high environ-
Among the more important factors influencing the drying mental temperatures, high porosity, and low relative humidity.
shrinkage of concrete are the content of cement paste, the wa- For a wide range of concrete mixtures only 20–25 % of the 20-
ter-cement ratio, the degree of hydration, the elastic modulus year drying shrinkage was realized in two weeks, 50–60 % in
of aggregate, the characteristics and amounts of admixtures three months, and about 75 % in one year. Almost twice as much
used, the time and the relative humidity of exposure, the size drying shrinkage is obtained at 45 % relative humidity as
and shape of the concrete mass, and the amount and distribu- compared to 80 % relative humidity exposure.
tion of internal reinforcement (such as steel and fibers [41]). The size and shape of the concrete mass have a consider-
Drying shrinkage of concrete and the factors influencing it are able effect on the rate and total amount of drying shrinkage.
discussed in numerous reports, such as those authored by The rate and ultimate shrinkage of a large mass of concrete are
Ytterberg [42] and Meininger [43]. A brief discussion of these smaller than the values for small-size concrete elements, al-
factors follows. though the shrinkage process continues over a longer period
Concrete is made up of two constituents: cement paste and for the large mass [46]. In full-sized concrete members, orien-
aggregate. The former shrinks and the latter restrains shrink- tation and edge effects complicate this relationship. There ap-
age. The effectiveness of aggregate to restrain the shrinkage of pears to be a linear correlation between the volume-to-surface
cement paste is related to the elastic modulus of the aggregate. ratio and the logarithm of shrinkage, as well as to the loga-
From the data obtained by an experimental study of concrete rithm of time required for half of the ultimate shrinkage to be
mixtures with water-cement ratios of 0.35 or 0.50, Powers [44] achieved [47]. In realistic terms, it seems that slower drying
developed the following expression showing how the drying conditions yield lower ultimate shrinkage values, but theoreti-
shrinkage of concrete (Sc) is related to the drying shrinkage cally the ultimate shrinkage may be independent of size
of cement paste (S), the volume fraction of the cement paste [48,49]. Also, in massive concrete members differential drying
(1  g), and a constant (n) that is dependent on the elastic conditions produce larger shrinkage at and near the surface.
modulus of aggregate: This gives rise to tensile stresses at and near the surface while
compressive stresses are developed in the interior. Conse-
Sc  Sp(1  g)n quently, surface cracking may occur if the tensile stresses ex-
ceed the tensile strength of the material and the stress relax-
Thus, for a given aggregate and cement paste the drying ation provided by creep. These cracks then increase the
shrinkage may be doubled when the aggregate volume fraction available surface area for evaporation of water, as well as cause
is decreased from 80 to 50 %. Similarly, long-term studies on redistribution of internal stresses.
the effect of aggregate type on drying shrinkage [45] showed Shrinkage of reinforced unrestrained structures pro-
that with a given concrete mixture the 23-year shrinkage was duces tension in concrete and compression in steel. By in-
more than twice for the low-elastic-modulus aggregates than creasing the amount of reinforcement, the shrinkage can be
for the high-elastic-modulus aggregates. Concrete made with correspondingly reduced but it will increase the tensile
dirty sand or with unwashed aggregates containing silt and stresses in concrete such that it is likely to crack when exces-
clay shrinks significantly more than concrete made with clean sive reinforcement is used. Reinforced concrete elements
aggregates. To the extent that the water requirement for a with normal amounts of reinforcing steel may show drying
given consistency of concrete is influenced by the maximum shrinkages of the order of 0.02–0.03 %. An increase in steel
size, grading, shape, and surface texture of the aggregate, these reinforcement will cause cracks at a closer spacing and also
factors also affect drying shrinkage. reduce the crack widths. Fibers added to concrete also redis-
The quality of the cement paste is primarily a function of tribute the stresses resulting in a similar reduction in crack
the water content for a given cement content in concrete width and closer spacing of the cracks [41].
(water-cement ratio) and the degree of hydration, which in According to Washa [2], drying shrinkage values for
turn depends on the fineness and chemical composition of structural lightweight concrete may vary between 0.04–0.15
cement as well as the curing time, temperature, and humidity %. Higher shrinkage values generally result from concretes
conditions. Since unhydrated cement particles can offer containing aggregates that have high rates of absorption,
restraint to shrinkage of hydrated cement paste in the same low modulus, a high proportion of fineness, and/or require
manner as aggregate particles, the factors leading to a greater high cement contents to obtain the specified strength.
degree of hydration generally contribute to a higher drying Moist-cured, cellular (foamed) products made with neat
shrinkage. Thus, finer cements usually show greater shrinkage cement and weighing 10–20 lb/ft3 (0.22–0.44 kg/m3) may
 GOODWIN ON VOLUME CHANGE 219

have drying shrinkage as high as 0.3–0.6 %. Autoclaved In regard to crystalline MgO or periclase, some early port-
concrete products containing fine siliceous additives may land cements, which were made at much lower temperatures
weigh 40 lb/ft3 (0.88 kg/m3) and have drying shrinkage in than used today, contained large amounts of this compound.
the range of 0.02–0.1 %. Conversely, internal curing occurs The expansion and cracking of cement pastes and concretes
when saturated lightweight aggregate is used, resulting in a containing these cements were attributed to periclase. It is now
decrease in shrinkage [50]. accepted that the periclase formed in modern portland cement
Silica fume is being used increasingly as a supplemen- clinker at a kiln temperature of 1400–1500°C is either inert to
tary cementing material for concrete elements. Data from moisture or too slow to hydrate under ambient temperature
ACI 234R indicate that the drying shrinkage of silica fume conditions. Only under autoclaving conditions, in accelerated
concrete is generally comparable to that of a control concrete laboratory tests such as ASTM Test Method for Autoclave Ex-
after 28 days of moist curing regardless of the water-to- pansion of Portland Cement (C 151), is periclase known to hy-
cementitious materials ratio (w/c  m) [51]. The silica fume drate and cause considerable expansion in unrestrained pastes
quantity and curing regimen prior to drying were found to be (see the section on test methods and specification for addi-
important factors. Concrete shrinkage is influenced little by tional discussion). Since portland cements meeting ASTM
silica fume content up to 10 % by mass of cement [52]. Early Specification for Portland Cement (C 150) [59] are required to
drying increases shrinkage for lean silica fume mixtures comply with the maximum limits of expansion in the autoclave
(w/cm greater than 0.60) and for high silica fume contents test, it is one of the chemical requirements of ASTM C 150 that
(greater than 10 % by mass of cement) because early drying the total MgO content of portland cement shall not exceed 6 %.
inhibits the pozzolanic reaction. Fly ash and ground granu- Note that approximately 2–5 % MgO goes into the solid solu-
lated blast furnace slag in the concrete mixture are reported tion of portland cement clinker compounds (C3S, C2S, C3A, and
to both increase drying shrinkage (probably due to the in- Fss),3 and the remainder shows up in the form of periclase.
creased gel formation of the pozzolanic reaction products) as
well as to have minimal effect [53–55]. Incorporation of silica Volume Changes in Expansive Cements
fume leads to reduced pore sizes in the cement paste, thus
increasing the surface tension in small capillary pores and ASTM Specification for Expansive Hydraulic Cement (C 845)
therefore the autogenous shrinkage. [60] covers hydraulic cements that expand during the early
hardening period after setting. Large expansion occurring in
Expansion Due to Hydration of Free CaO an unrestrained cement paste can cause cracking. However, if
and MgO the expansion is adequately restrained, its magnitude will be re-
duced but a self-stress will develop. When the magnitude of ex-
It has been often reported in the published literature that pansion is sufficiently small such that the prestress developed
hydration of crystalline MgO (periclase) or CaO, when present in concrete is on the order of 25–100 psi (0.2–0.7 MPa), the ce-
in significant amounts in a portland cement, can cause expan- ment is known as shrinkage-compensating. This amount of ex-
sion and cracking. The crystalline size, the thermal history of pansion is generally enough to offset the effect of tensile stress
the material, and the solubility of the material within the other due to subsequent drying shrinkage. During the last 20 years,
clinker constituents determine the reactivity of both CaO and the cements of this type have been commercially used for mak-
MgO in cement. Very fine particles react rapidly after mixing ing crack-free industrial floors, airport taxiways, water-storage
with water and likely are consumed in the plastic state of the tanks, and post-tensioned concrete members for parking
cement. The higher the calcining temperature of both CaO and garages.
MgO, the lower the reactivity, as the crystalline structure The formation of ettringite and hydration of crystalline
becomes denser. CaO and MgO are both chemically combined CaO are two mechanisms known in concrete technology that
within the clinker phases. Low burning temperatures can leave are capable of causing disruptive expansion. Both phenom-
unreacted calcium oxide present in the cement, as well as ena have been harnessed to produce shrinkage-stress-
clinker composition containing calcium levels above the compensating concretes. The expansive additive in U.S. prac-
stochiometric requirement for combination with silica, tice is a modified portland-cement clinker that contains
alumina, and ferrite (Lime saturation factor  1) [56]. significant amounts of anhydrous calcium aluminosulfate
Laboratory tests on early portland cements showed that (4CaO3Al2O3SO3) and calcium sulfate in addition to the
the cement pastes made with low-MgO cements containing cementitious compounds of portland cement, 3CaOSiO2,
over 3 % free (crystalline) CaO gave considerable expansion, 2CaOSiO2, and 4CaOAl2O3Fe2O3. The cement produced by
which caused cracking of the unrestrained paste [57]. The blending this additive with normal portland cement is called
phenomenon is virtually unknown in modern portland Type K expansive cement.
cements due to better manufacturing controls, a result of In Fig. 1, a graphical representation is given of the con-
which is that the content of uncombined or crystalline CaO cept showing how a Type K shrinkage-stress-compensating ce-
seldom exceeds 1 %. In addition to analytical verification of the ment, when compared to normal portland cement, works to
free CaO content, physical verification using the ASTM C 151 reduce the risk of drying shrinkage cracking [61]. Immedi-
[58] autoclave expansion test is now also conducted. Expansive ately following the start of hydration of Type K cement, it is
additives for cement containing a large amount of crystalline observed that large amounts of ettringite start forming. After
CaO are being used in Japan to obtain controlled expansions initial set, the concrete will bond to the reinforcing steel and
under restraint in shrinkage-compensating concretes, which any expansion will be restrained by the steel. Under these
are described in the next section of this chapter. conditions, the steel will go into tension and concrete into

3
Fss stands for the calcium aluminoferrite solid solution phase.