THE RESPONSE OF SHEAR WALLS SUBJECTED

TO DYNAM~ C LOADS

by
VITELMO V. BERTERO

Ingeniero Civil
Universidad Nacional Del Litoral
R. Argentina
1947

S.M. in C.E.
Massachusetts Institute of Technology
1953

Submitted in partial fulfillment of the

requirements for the degree of

DOCTOR OF SCIENCE
at the

Massachusetts Institute of Technology

June 1957

Signature of Author: ,, , ----- --

Vitelmo V. Bertero
Department of Civil Engineering
June 1957
Signature of Professor
in Charge of Research: i
Professor Robert J. Hansen
Department of Civil Engineering
June 1957
Signature of Chairman
of Departmental Committee
on Graduate Theses: _

Professor Charles H. Norris

Department of Civil Engineering
June 1957
SS. INST. r

FEB 19 1958
IaIBRARI
 To my wife, whose patience and encouragement

have made this work possible.

 ABSTRACT

THE RESPONSE OF SHEAR WALLS SUBJECTED TO DYNAMIC LOADS

This report deals with the experimental and theoretical study of

the behavior of reinforced concrete shear walls being subjected to
dynamic loads. The first part includes a brief discussion of the dif-
ferent factors involved in the study of such structural members. An
analysis of the dynamic loading machine follows. This machine was
especially designed for this investigation.

Four specimens, identical in every respect except the strength of

the concrete, were tested, and attempts were made to simulate actual
structural and loading conditions. The specimens were approximately
1/h size replicas of a standard shear panel in a one-story building,
one bay wide. One wall out of these four was tested under static load-
ing. The results obtained are presented in the form of curves, tables,
and photographs.

The lattice analogy was used for the theoretical analysis. The
simplified lattice used for the elastic behavior during a previous in-
vestigation by Mr. Finerman, was modified for the cracked range. In
order to determine the properties of the modified equivalent lattice,
several assumptions were made. Making use of these assumptions, the pro-
gramming for the WHIRLWIND I and IBM 70h digital computers was developed.
Both programs, as well as the results they yielded, are included in the
theoretical discussion.

Thesis Supervisor: Robert J. Hansen

Title: Professor of Structural Engineering
 TABLE OF CONTENTS

CHAPTER I: INTRODUCTION
I. Statement of Problem
II. Previous Investigations
III. Objectives
IV. Summary of the Investigation
V. Acknowledgement

CHAPTER II: FACTORS INVOLVED IN THE PROBLEM OF A

SHEAR WALL UNDER DYNAMIC LOAD 9
I. Wall Materials and Their Properties
II. Type of Wall - Proportion of Wall Components
III. Construction Procedures
IV. Support Conditions
V. Type of Load and Method of Load Applications
VI. Scale Effect
VII. Instrumentation and Measurements
VIII. Miscellaneous Considerations

CHAPTER III: TEST SPECIMENS

I. Description of Test Specimens
II. Properties of the Materials
III. Casting and Curing of Test Specimens

CHAPTER IV: EQUIPMENT AND INSTRUMENTATION

I. Loading Equipment
II. Measuring Equipment or Instrumentation
III. Miscellaneous Equipment

CHAPTER V: TEST OF WALLS

I. Wall Preparation
II. Test Procedure
III. Test Results

CHAPTER VI: THEORETICAL INVESTIGATION

I. Introduction
II. Elastic Behavior
III. Cracked Behavior
IV. Theoretical Results

CHAPTER VII: DISCUSSION OF RESULTS, GENERAL

CONCLUSIONS, AND RECOMMENDATIONS FOR
FUTURE RESEARCH 72
I. Discussion of the Machine's Behavior
 II. Discussion of the Results Obtained from
Tests Performed on the Walls 7h
III. Discussion of the Theoretical Results 78
IV. Comparisons Between Theoretical and
Experimental Results 79
V. General Conclusions 80
VI. Recommendations for Future Research 82

REFERENCES 8h

FIGURES 86

APPENDIX A: GENERAL PROGRAM AND CODING FOR WHIRLWIND I 169

APPENDIX B: GENERAL PROGRAM AND CODING FOR IBM 70h 201

 CHAPTER I

INTRODUCTION

I. Statement of Problem

The type of framing selected for use in a particular structure

depends on the amount of protection desired against the forces which one

expects will act on the structure. It is obvious, therefore, that any

serious attempt to design future structures which will resist atomic

blast loads, wind gusts, or earthquakes, will lead to the selection of

structural forms particularly suited to resist a great amount of lateral

load. In this sense, the shear wall structure offers one of the most

satisfactory and economical methods of construction.

All buildings which derive their principal strength for resist-

ing lateral loads from structural walls (shear walls) capable of resist-

ing lateral loads acting in their own plane, are understood to be shear

wall structures. When forces caused by winds, earthquakes or blasts are

applied to these structures, the action of the structural components sub-

jected to lateral loads must be predicted, so that a proper design may

be accomplished. Since the shear wall is one of the most important com-

ponents, particular attention has to be paid to its behavior.

The shear wall is a very stiff member; that is, it has a high

resistance to deformation under load. It is essentially a plate loaded

in its own plane, but the analysis needed to predict its behavior is more

complicated than that of simple plates, for the following reasons. The

shear wall cannot act completely as an individual plate, but must deform

in a manner which is consistent with the bounding frame. Secondly, the

usual materials (reinforced concrete or brick) used in shear walls have

-1-
elastic and inelastic properties which are difficult to define, and which

make it necessary to predict three different phases of action for any

wall. Namely:

1. The elastic behavior.

2. Behavior under small deflections and partial cracking.

3. Behavior under large deflections.

In addition, the usual lateral loads produce dynamic effects which must

also be considered.

From the above considerations it becomes obvious that the

theoretical approach to the problem is not simple, due to the complexity

of the mathematics involved.

II. Previous Investigations

Serious earthquakes have periodically stimulated engineering

groups to set standards of design for structures subjected to lateral

loads. These standards are usually based on the inertia of the entire

structure considered. A portion of the total weight of a building is applied

to each story in the form of static lateral loads. The engineer has had

to rely on judgement to decide how much of the total lateral load will be

carried by the shear walls. Today, the possibility of enormous lateral

forces created by blasts, and the increasing desire to reduce the amount

of materials used, make the previous knowledge of a dependable design

criterion necessary.

Japan has suffered considerably from building failures caused by

earthquakes. For this reason, Japanese engineers have devoted a consider-

able amount of effort in the design of shear walls. A large part of their

work has been highly theoretical.

-2-
 Some experimental work has also been accomplished, yielding

various empirical formulas. A common practice in Japan consists of

* ( 1 ) (2 ) Another Japanese
replacing a wall by an equivalent rigid frame.

method entails a solution of the problem by classical energy theorems.

Dr. Naito , in an investigation on seismic walls assumed that:

1. The walled frame is completely fixed at the base, and when

subjected to the action of horizontal forces,

2. The walled frame would deform like a cantilever beam of

I-shape.

The Japanese have conducted a considerable amount of experi-

mental testing, but since their primary goal is earthquake study, most

of the walls have been tested by reversals of loading (5 ). For this

reason, it is difficult to interpret the results in the elastic range.

An experimental study which could be of some assistance is one concern-

ing ultimate strength. (6)

Very little research on reinforced concrete shear walls had

been done in the United States before 1949. At that time, an investiga-

tion was begun at the Massachusetts Institute of Technology, under the

direction of Prof. Robert J. Hansen. The main objective of this investi-

gation was to obtain experimental data that would help establish design

procedures for structural elements which would sustain impulsive loads.

Results were published in different reports.(7)( 8 )(9)(1O) Since 1951,

Stanford University has carried out extensive research programs.-- ex-

perimental and theoretical - to determine the behavior of shear walls.

As a result of these investigations, several reports have been

* Numbers in parenthesis refer to entries in the Bibliography, or

references.

-3-
 1 2 ) ( 1 3 ) ( 14 ) ( 15 )( 16 )
published which shed some light on the subject.(11)(
(17)(18)(19)(20) However, all these investigations were made considering

only the behavior of walls under static loads, and although the results

obtained throw some light on the present problem, it is well known that

the behavior of a structure under so called "static loads", or under very

slowly applied loads, is quite different from the behavior of the same

structure under rapidly-applied loads such as those which arise from wind

gusts, earthquake, or blasts from the explosion of bombs. Since much

attention has been paid lately to the effects of these loads, it has

become increasingly important to have available methods of predicting the

dynamic response of shear walls. The Massachusetts Institute of Tech-

nology, under contract to the Corps of Engineers, has undertaken recently

its theoretical and experimental investigation. The principal objective

of the experimental part of this investigation is to obtain sufficient

data to substantiate a rational means for predicting the behavior of

reinforced concrete shear walls under dynamic loads. From the theoretical

side, its principal aim is to find a method that may allow, in the future,

the elimination of a great part of experimental work, and act as a check

for more empirical solutions. Any recommended procedure for predicting

the behavior of such an element should also be based on principles which

are familiar to practicing engineers, and at the same time it should be

simple to apply with a fairly good degree of accuracy.

The first step of the theoretical investigation is described in

the report (2 1 ) submitted by Mr. Aaron Finerman, titled "The Theoretical

Elastic Response of Shear Walls Subjected to Dynamic Loads". The

first step in the experimental investigation was the construction of a

testing machine (22) capable of loading dynamically to destruction

-4 -
large-scale models of reinforced concrete shear walls.

Once the first step of the experimental program was completed,

it was decided that this preliminary investigation would be restricted

to the testing of several reinforced concrete shear walls, with the ob-

ject of analyzing the behavior of the machine, and at the same time ob-

tain some data on the parameters which can affect the dynamic resistance

of the element being considered. This preliminary investigation will

allow the development of a more extensive experimental research program.

The theoretical investigation may be considered as a continu-

ation of the work made by Finerman. As the title of his report(2 1 )

indicates, his work was restricted to the study of the dynamic response

of walls, within the elastic range of the member. Although it is true

that the designer must decide what point in the behavior of the structure

he wishes to consider as the maximum permissible response, it is obvious

that only rarely may he choose this point to be within the elastic range,

because of the extremely heavy construction this would require. There-

fore, it will be more economical to allow for a certain amount of crack-

ing, and in some cases -- especially for certain types of blast-resistant

structures -- it will be reasonable to even design for collapse, since

a factor of safety may be already included in the choice of the pressure

curve for which the design is to be made.

Furthermore, only under very favorable conditions may it be

assumed that the first loading on a shear wall will be resisted in part

by tension in the concrete. Repetition of the common live loads or un-

predictable volume changes can cause cracks to develop along any section

that is not constantly under compression.

-5-
 From the above considerations it is evident that it is

necessary to have a rational means for finding deflections and stresses

not only in the elastic, but also in the inelastic ranges.

III. Objectives

This work was made in order to supplement the above-mentioned

combined research program undertaken by M.I.T., and it is a theoretical

and an experimental investigation covering the following points:

A. An experimental investigation of shear wall behavior under

dynamic loads.

B. The mathematical prediction of the behavior of the tested

walls.

IV. Summary of the Investigation

The mathematical prediction of shear wall behavior under dy-

namic loads for the phase included between the first cracking load and

the ultimate load is accomplished by the use of a lattice analogy. The

method consists essentially of replacing the frame and panel by a pin-

connected lattice having the same resistance to deformation. The predic-

tion of the cracked action involves a number of assumptions which are

discussed in Chapter VI. The lattice analogy procedure is then used to

obtain theoretical load-deflection curves and the ultimate load for the

reinforced concrete shear wall.

Due to the limited time, the predictions and experimental

verifications reported here have been confined to shear walls of one

proportion. Four specimens were tested, and these were identical except

for the strength of the concrete. In two the compressive strength at

the day of the test was approximately 3,000 psi, and in the other two 5,100

-6-
and 6,300 psi respectively.

Attempts were made to simulate actual structural and loading

conditions, the specimens being approximately a 1/4 scale replica of a

standard shear panel in a one story building, one bay wide.

The model selected and the loading arrangement used in the

tests are shown in Figure 1.

One wall was loaded by static lateral load only. This test

served to check the operation of the equipment and to establish the sta-

tic properties of the walls which were to be tested dynamically later.

The applied load, reactions, horizontal and vertical deflections were

recorded against time, during each test, with the exception of the

static test. The experimental results of this investigation are pre-

sented in Chapter V, and the theoretical discussion is given in Chapter

VI.

V. Acknowledgment

The experimental work described in this report was performed

in the Structural Dynamics Laboratory of the Massachusetts Institute of

Technology, under Contract DA-49-129-Eng 325 with the Corps of Engineers.

The author wishes to thank his advisor, Dr. Robert J. Hansen,

Associate Professor of Structural Engineering, for his advice and direc-

tion of this investigation, and Howard Simpson, Associate Professor of

Structural Engineering for his immediate supervision and encouragement.

The author also wishes to acknowledge the assistance of Harry

Guzelimian, Research Assistant in Civil Engineering, in preparing and

conducting the tests. Also the work done by Shui Ho and Christopher Cal-

ladine, Research Assistants, in preparing the program for the high-speed

-7-
computers. The recording instruments were assembled and operated by

Arthur Casey, Senior Electrical Technician.

Sincere thanks are also expressed to Mr. Donald Gunn, who

offered much practical advice and did the machine work for some of the

apparatus, to Miss.S. Carlson for typing the thesis and to E. Lawlor,

A. Ostapenko, N. Durando, and K. Gentili for their help throughout the

investigation.

-8-
 CHAPTER II
FACTORS INVOLVED IN THE PROBLEM OF A SHEAR WALL UNDER
DYNAMIC LOAD

,In planning an experimental investigation of shear walls of

reinforced concrete, the following factors must be considered:

I. Wall materials and their properties.

II. Type of wall - Proportion of the wall components.

III. Construction procedures.

IV. Support conditions.

V. Method of load application and type of load.

VI.. Scale effect.

VII. Instrumentation and measurements.

VIII. Miscellaneous considerations.

For the prupose of a specific series of tests designed to

acquire information on a particular facet of the general problem, the

importance of certain of these factors may be negligible, or for prac-

tical purposes may be considered as such. For a full understanding of

the value of a series of tests and the relation of that series to the

overall problem, it is necessary to define the limits within which the

series falls, in terms of the above factors.

I. Wall Materials and Their Properties

Shear wall structures include articulated structural steel

frames with diagonal bracing and continuous frames with masonry or re-

inforced concrete panels. This investigation, however, has been limited

to monolithic reinforced concrete shear walls, which appear to be the

-9-
most satisfactory and economical choice for structures designed to

resist heavy blasts.

The properties of the material which are important in the case

of reinforced concrete are the tensile strength andmodulus of elasticity

of the concrete, and the yield point of the steel reinforcing.

II. Type of Wall - Proportion of the Wall Components

Although we are concerned with only one kind of shear wall --

the monolithic reinforced concrete shear walls -- many different factors

must be considered. In one case, the shear wall may belong to a one

story or to a multistory shear wall structure. However, since we are

essentially interested in the behavior of the structural elements, our

investigation will be limited solely to one-story walls. Once the dy-

namic response of this structural element is known, the behavior of

multistory shear wall structures may be predicted by modifying certain

parameters.

It is also recognized that if there exist openings in a shear

wall, the dynamic response of the wall can vary over a wide range,

depending on the relative size and location of the openings. However,

the investigations reported herein have been conducted on solid shear

walls only.

The important properties of a shear wall are:

A. Frame Effect

1. Column and beam concrete and steel areas relative to

thickness and length of wall panel.

2. The steel ratio of the columns and beams.

B. The steel ratio of the wall panel and its distribution

C. Length-height (L/H) ratio of wall panel.

- 10 -
A. Frame Effect: Ordinarily, the greatest gains in speed of construction,

cost and space usage in shear wall structures are obtained by simplify-

ing and standardizing the wall forms by elimination of all projections

and pilasters: (i.e.), they are to be constructed as monolithic, box-

shaped structures without conventional framing beams and columns around

shear walls. Since cross walls are ordinarily well tied to face walls

and to floor and roof slabs, it is obvious the latter will impart --

acting as a frame -- considerable strength to the shear walls, in addi-

tion to that inherent in the shear wall itself.

This integral action of slab and wall must be considered in

our investigation. The proportion of wall or slab that acts integrally

as flanges with the shear wall in resisting bending moment and shear

forces is highly debatable at this time. If one follows conventional T

beam practice, the allowance would be eight times the thickness of slab

or face wall on each side of the cross wall, and the maximum effective

width of flange should be taken no more than one-half of the distance

between shear walls. This would mean that a 6" wall or roof slab is

equivalent to a 24" square column or girder, although the percentage of

reinforcement would be low compared to the conventional member. It

should be noted that face walls and slabs enter into the frame action for

shear walls of any monolithic structure, whether or not there are framing

columns and beams. However, in the case where dynamic loads are applied,

the problem becomes more complex, due to the necessity of considering an

equivalent mass to that of the walls and slabs.

The frame effect is an important aspect of the shear wall

problem. While the general effect may be investigated by changing the

frame size for a given wall, this throws little light on the actual

- 11 -
effect of face walls and slabs in a structure. It is obvious that if

the slab develops cracks (but does not fail in flexure), the horizontal

load capacity of the structure may be reduced due to the redistribution

of the load which is caused by these cracks, as shovm in Figure 2. When

the stresses in the roof are not sufficient to cause cracking, the roof

slab will supply added resistance by acting as the flange of a T beam.

The experimental approach to the problem is difficult because

of the magnitude and type of loading required. Static tests on small

models were carried out at Stanford University (1 3 ) but the results should

be considered as qualitative, owing to considerable difficulties en-

countered during the tests. It is hoped that a full-scale structure test

program can be prepared, as it was prepared previously by the Office of

the Chief of Engineers, Department of the Army. (23)

1. The relation between column and beam areas and wall panel area

is important insofar as it greatly influences the value of the load that

will produce first cracking, by changing the stress pattern in the wall.

2. The steel ratio of the column and beam directly influence the

ultimate strength.

In general, the frame, and particularly the columns may give

rise to a different type of shear wall failure. If the column strength

is insufficient to cause cracking of the shear wall, a tension failure

along the base will result as shown in Figure 3. This requires a yield-

ing of the tension column steel. If the tension column steel does not

yield and produce a failure, the ultimate load can be associated either

with complete wall breakdown or a shear failure of the compression col-

umn at the foundation, as it is shown in Figure h. As far as can be

- 12 -
determined, reinforced concrete walls of practical proportions do not

fail at ultimate by sudden panel cracking.

According to the above reasoning the investigation was pro-

grammed using shear walls whose components have been designed so as to

induce progressive panel cracking, eliminating the possibility of a sud-

den failure.

B. The Steel Ratio of the Wall Panel and its Distribution: The design

of blast-resistant structures may require the use of shear walls designed

to produce as high a strength as possible. One of the basic parameters

involved is the panel steel ratio.

Although it may be expected that the wall rigidity and load

at first cracking would be virtually independent of the wall's steel

ratio, it is obvious that the placing of reinforcing in the panel changes

the mode of failure to some extent. If the percentage is small, the wall

will have an ultimate load derived from two sources -- the compression

column and the panel steel. As the panel strength increases with increas-

ing percentage of steel, it tends to become the dominant influence,

while the contribution of the column tends to become negligible. Studies

of the strengthening factors in the panel showed that the only important

parameters were the horizontal and vertical steel cross-sections. The

concrete is unimportant. Furthermore, a considerable portion of the panel

steel is stressed to its yield point at or before ultimate load. From

static tests it was found that the yield point of the panel steel was a

parameter. These same tests seem to indicate that there exists a prac-

tical limit in the neighborhood of one per cent panel steel, above which

the increase in ultimate load with further increases in panel steel

tends to become relatively smaller.

- 13 -
 Standard design procedures call for equal amounts of horizon-

tal and vertical panel steel. This steel is selected as a percentage of

the wall cross-sectional area, and is uniformly distributed throughout

the panel. The practice is reasonable when the steel ratio is low and

large amounts of steel are not involved. Howrever, blast design may re-

quire large amounts of steel in the panel. Some question then arises

as to the efficiency of the uniformly distributed design.

Although it may have been expected, according to the stress

pattern to be obtained, that the diaonal reinforcing might be more ef-

fective than that placed horizontally and vertically, the static test

results obtained at Stanford University (16 )(9)indicated that in all

cases the ultimate loads were smaller, in the case of a diagonal pattern

than for vertical and horizontal distributions. Both concentrated and

uniformly distributed patterns were studied, including an attempt to

provide local strengthening of critical areas. Concentrations of steel

on the diagonals and in the lower corners produced early shrinkage crack-

ing in these locations and noticeably decreased the strength.

1With uniformly distributed diagonal steel, the strength was

nearly equivalent to the rectangular design. However, Mr. Vhitney ( 2 3 )

after analyzing a limited number of tests conducted on concrete rein-

forced with h5-degree diagonal tension steel, reached the conclusion

that in general it appears that the above distribution of reinforcing

results in slightly stronger and stiffer shear panels, provided the walls

are not subjected to a transverse load at the same time. Further studies

under static loads show that uniform vertical steel is more effective

than horizontal steel. Vertical steel alone appears to be fully as ef-

fective as vertical plus horizontal steel. Complete omission of hori-

- 14 -
zontal steel would be undesirable from a practical standpoint, since a

certain amount of horizontal steel is necessary to hold the vertical

steel in place. The tests at Stanford ( 1 9 ) demonstrated that if it is

desired to reduce the panel steel, the horizontal wall steel in the up-

per three quarters of the wall can be reduced substantially, without

influencing the panel stiffness or strength.

The present study only takes into account walls with uniformly

distributed horizontal and vertical reinforcing steel, and a steel ratio

of 1/4 percent. However, it is obvious from the above considerations

that in the extensive experimental program, a large number of different

ratios must be taken into account.

C. Length-Height (L/H) Ratio of Wall Panel: The length-height (L/H)

ratio, is an important factor in wall behavior. Static tests at Stan-

ford University ( ), indicated that the ultimate strength in stress per

unit of shear area is essentially independent of L/H, whereas the first

cracking load increases greatly with the L/H ratio.

For walls with panel reinforcing, first cracking and ultimate

loads become essentially equal at an L/H value somewhat greater than

three.

However, these conclusions were obtained from tests on speci-

mens in which the bottom beam can be considered as fixed. The analyti-

cal study made by Finerman,(21 ) shows that for simple-support shear

walls, the first cracking load decreases after a certain L/H, instead

of increasing.

In this first series of dynamic tests, the L/H ratio was not

changed, but it is obvious that it would be necessary to do so in a gen-

eral investigation.

- 15 -
III. Construction Procedures

Although shear walls, except for pre-cast elements, would be

formed and poured in place in a vertical position, the models here

studied were poured flat on the casting floor.

It is possible that some small difference in strength may oc-

cur from this difference in method of casting, but it is believed that

this factor has no importance.

IV. Support Conditions

The support conditions of a shear wall are an important fac-

tor. Yield, sliding or other movements of the supports of a wall ex-

posed to a blast or to a motion of the footing that restrains the build-

ing frames from sliding or rotating will act to relieve the straining

parts, and must therefore be considered in the design.

The flexibility of the foundation materials and its resistance

to lateral deflection and to rotation are important in determining the

motion of a structure. When a structure is founded on spread footings

or on footings supported by piles, the relative degree of fixity of the

columns in the structure at their connection with the footings is a mat-

ter of considerable importance in determining the lateral resistance

and the strength of the structure being submitted to lateral forces.

In this investigation, the walls were supported as indicated --

in a general way - in Figure 5. Details of the support are shown in

Figures 6 and 7. The type of supports used may be considered as simple

supports, except for the fact that they are not entirely rigid, but have

some flexibility. The deflections of the supports were determined dur-

ing the first static test, and were found to be not negligible.

- 16 -
 In a shear wall, the loads are transmitted at the bottom of

the wall through a connecting shear wall, a floor slab, or the founda-

tion. It is evident that only in the case of single-story buildings,

(which are generally preferable to multistory constructions), may our

supports be considered similar to those encountered in an actual struc-

ture.

V. Type of Load and Method of Load Application

As it was stated previously, the main purpose of this work is

to study the behavior of shear walls which are submitted to loads pro-

duced by an atomic blast. Obviously then, attempts must be made to simu-

late this special kind of load.

Most of the energy released by atomic fission acts to heat the

surrounding air to extremely high temperatures. The rapid expansion of

this heated air exerts an enormous pressure on the surrounding atmosphere

which moves outward from the center of the explosion as a conventional

shock wave. This wave is characterized by a virtually instantaneous rise

to peak pressure which quickly decreases in intensity to atmospheric,

followed by a much less intense negative pressure. This intensity of

the peak pressure, the variation with time, and the duration of the

positive phase are of prime interest. These factors depend on the energy

release of the weapon, the ground zero distance, and the height of burst.

It is obvious that the exact loading to which any particular building

may be exposed will never be accurately known. The only thing that can

be done is to establish the energy level to be protected against, by the

selection of an appropriate bomb and ground zero distance. Once this is

done, curves of load vs. time for most types of rectangular structures

- 17 -
may be approximated by the methods given in , Chapter XI; and(23),

Appendix 1.

Evidently, the way in which impulsive loads are transmitted

in an actual structure depends largely on the type of construction and

the location of the wall. Figure 8 shows the typical loading on shear

wall structures.

In calculating the horizontal reactions of the exterior front

and rear walls, the roof and shear wall are usually assumed infinitely

rigid in their own planes. Then, the reactions of the exterior walls

rather than the direct blast pressures should be used in the design of

shear walls, because the dynamic strains in the stiff, quick-acting

shear wall are greatly affected by the rate at which the forces acting

on the front and rear walls and the roof are applied on it.

In structures where large plastic deformations of the exterior

blast-resistant walls are allowed, the reactions of the wall will have

an impulse curve which is very different from that for the pressures act-

ing on the exposed wall at any given time, as shown in Figure 9. This

difference in the shape of the impulse curve (elimination of the peak)

will, in most cases, reduce the strength needed for the structural ele-

ment which supports the exterior walls.

Usually, wall spacing would be large compared to story height

and the main face wall reinforcing would be in a vertical direction.

Hence, face wall pressures would be transmitted primarily to the floor

system and thence to the tops of the shear walls. If walls were so

closely spaced that face wall pressure were transmitted primarily in a

horizontal direction and distributed along vertical ends of the walls,

it is questionable whether the shear walls would be very heavily loaded.

- 18
 An exception would occur if large impulsive loads were applied

to very heavy face walls. However, the above conditions lead us to the

conclusion that the proper method of testing model shear walls is to

apply the load along the top of the wall. This conclusion introduces

another question: How should the load be distributed along the top of

a model 'wall? Four possibilities are shown in Figure 10.

Japanese investigators have found that the second and third

methods cause a different crack pattern than the first. The loading in-

dicated in Figure 10-D probably comes closest to the actual conditions,

since face walls and floors together act as an I-beam, the floor being

the web of the beam. It would seem reasonable to believe that shear

would be distributed to the cross wall by the floor in a manner consis-

tent with the parabolic distribution of shearing stress in an I-beam

web. However, this can only be true in the elastic range, because if

the slab develops cracks but does not fail in flexure, the load will be

redistributed according to the crack pattern on the roof, as mentioned

before and shown in Figure 2.

However, the truly few static tests carried out at Stanford

University (13), reach the following conclusion:

"The manner in which loading is applied to the shear

wall -- whether directly or through a roof -- has relatively
little influence on the shear wall behavior, providing the
structures are otherwise identical."

Another factor which will affect the behavior of a shear wall

is the presence of vertical forces. These forces may result from heavy

dead loads and the superimposed live load arising from vertical effects

of the blast or other dynamic forces. Impulsive roof loads of a high

order of magnitude will be present under ordinary conditions, and the

- 19 -
magnitude of the total vertical force will be of the order of 50 to l00

of the shear load applied to the top of the wall, depending on relative

dimensions and type of construction. It is evident that the added verti-

cal load may affect the response, because the moments of these loads

affect the deflections, which in turn affect the moments of the vertical

forces; but there is still another effect of the vertical load which is

potentially of great importance in a problem of this kind. This effect

concerns the change in the resistance properties of the cross sections

of the columns, which is brought about by the simultaneous presence of

direct forces and bending moments. Because the problem is a non-linear

one, and both the forces and the moments are independent functions of

time, enormous complexities would be introduced into any analytical con-

siderations. These effects may be approximated in the manner suggested

5)
by Nevwmark and Fraenkel.(24)(2

However, the tests performed under static loads at Stanford

University ( 1 3 ) , indicated that the shape or magnitude of the load deflec-

tion curve is not significantly influenced by vertical loads of the same

order of magnitude as the shear load, and the initial stress pattern and

initial cracks are influenced very slightly by the vertical loading.

The final crack pattern is similar to that of specimens subjected to

shear load only.

For the effects considered above, in the present series of

tests, a loading arrangement was used which subjected the panel to

stresses that would occur under cantilever loading with a concentrated

load at the upper corner. (Figure 6).

In a more extensive experimental program, some account of the

added vertical load may be taken by considering an increased dead load

- 20 -
equivalent to the added vertical force.

Special consideration must also be given to the effects of

various different combinations of load rise with time, peak intensity,

and duration of load.

VI. Scale Effect

If the various factors mentioned before are to be explored in

a satisfactory manner using full-scale shear wall models, the program

immediately assumes stupendous proportions. On the other hand, it is

difficult to properly simulate complex stress patterns with scale models,

particularly when cracking of concrete and yielding of reinforcing steel

are involved. Hence, some comparisons are needed.

For the static tests carried out at Stanford University (13)

(1 3 )

concrete models geometrically scaled to 1/8, 1/4, and 3/8, were used.

The results of these tests indicated that in general, there is no defi-

nite indication of scale effect.

It must be noted that if no scale effect exists -- in the case

of static loads -- the load will vary as the square of the scale factor,

and deflections vary directly as the scale factor, with like concrete

strength and modulus of elasticity.

However, these results were obtained considering that the

applied load and the internal forces are closely balanced at all times,

and that the acceleration is negligible.

In the case of rapidly or instantaneously applied loads, how-

ever, the internal forces resisting deflections will not be immediately

and continuously equal to the applied loads, and the resultant accelera-

tion of the member will not be negligible. The maximum stress and, more

- 21 -
important, the maximum displacement produced under blast loads, are a

function of the intensity of the applied force, the rate of application

of the force, the duration of the load, and the particular variation with

time of the applied force and the resistance. These stresses and dis-

placements are generally far different from those which would result from

a static load of equal magnitude, and the prediction of the way the load

and deflections will vary with the scale factor becomes very difficult.

As in any dynamic problem, the solution of this one becomes a question

of force, mass, stiffness, strength, and time. The scale factor between

the responses of model and prototype will not only depend on the dimen-

sional scale factor, but also on the rate of application of the force,

the duration of the load, and the natural vibratory frequency of the wall.

In the present work, a reinforced concrete panel of h5 1/8"

by 61" was considered to represent a 1/h-scale model of a prototype wall.

For the reasons given above, it is obvious that in order to determine

the scale effect with accuracy, it will be necessary to plan an extensive

experimental program including not only different scale concrete models,

but also varying the load-time function.

VII. Instrumentation and Measurements

In planning the experimental program for studying the dynamic

characteristics of the shear wall, several requirements have to be met.

One set of requirements is related to the design of the test-

ing machine, and the other set is related to the instrumentation neces-

sary to obtain the magnitude of load and reactions, strains and deflec-

tions, of the test specimen.

Electronic, photographic, and mechanical means are used. The

- 22 -
accuracy of the results to be obtained will depend on the errors intro-

duced by the different sources.

A discussion of some of these errors will be given in Chapter

IV.

VIII. Miscellaneous Considerations

There are other factors which may also be considered in the

case of a structure with shear wall construction, such as: horizontal

construction points, shrinkage cracks, expansion points, etc. However,

it is believed that the importance of these factors may be neglected in

this investigation.

- 23 -
 CHAPTER III
TEST SPECIMENS

I. Description of Test Specimens

The specimens used in this investigation were designed to rep-

resent shear walls in a single story building, one bay wide, which was

considered to represent a 1/4 scale model of a prototype wall.

The dimensions were chosen to duplicate as accurately as pos-

sible, the shear wall tested statically at Stanford University.(16)

Details of the shear wall configuration, stirrup-spacing, and

steel distribution are illustrated in Figure l1. The four walls cast

were identical except for the compressive strength of the concrete. For

two of these walls, the compressive strength was 3,000 psi, whereas in

the other two this strength was 5,100 and 6,300 psi respectively. Table

I contains the important characteristics of the walls.

As shown in Figure 11, the columns were extended to provide

sufficient bond for reinforcing, and to prevent local cracking failures

in the places where the load was applied. As it is indicated in the same

figure, special steel plates were used where the load and reactions are

applied.

II. Properties of the Materials

Information concerning the properties of the materials con-

stituting a specimen is of prime importance in an experimental investi-

gation.

Two types of concrete were used in the construction of the

specimens. The mixes used for the concrete had the following character-

istics:

- 24 -
 For wall #1 and #2

1 part by weight of high early strength Portland Cement

2.h6 part by weight of sand

2.82 part by weight of 1/4" - 1/2" gravel

6 gallons of water per sack of cement

Air entraining agent - Darex 28 cc per sack of cement

For wall #3 and #`

1 part by weight of high early strength Portland Cement

3.4 part by weight of sand

3.8 part by weight of 1/4" - 1/2" gravel

7.16 gallons of water per sack of cement

Air entraining agent - Darex 28 cc per sack of cement

As can be seen above, the types of concrete used were very

similar to those found in conventional practice.

Sufficient control specimens were made for each model, so that

the rate of gain in strength and the strength on the day of the test

could be ascertained. An attempt was made to test all specimens when

the strength was approximately that desired. This was possible in all

but the first two walls.

The values of the important characteristics of the concrete

used in the four walls are listed in Table II.

Reinforcing Steel

Due to the small amount of steel reinforcement used in the wall

panel, and in order to maintain the separation of the bars between prac-

tical limits, it was necessary to use undeformed bars for the panel re-

inforcing. Intermediate grade Inland Hi-Bond deformed bars were used

for the frame reinforcing.

- 25 -
 The important characteristics of the bars used in the four

walls are given in Table III.

III. Casting and Curiný of Test Specimens

All walls were cast on the floor, in a wood formwork especially

designed so that its part could be easily removed. Two walls were cast

simultaneously, The reinforcing cage and the steel plates -- where loads

and reactions of the supports acted -- were held in position by welding

the reinforcing of the frame to the plates, which were held in place by

an auxiliary steel frame. Figure 12 shows the fornmork and the rein-

forcement in place.

Special pins were inserted where the brackets for measuring

deflections had to be attached, and bolts were inserted for the subsequent

attachment of the lateral restraint mechanism.

All the concrete was mixed for three to five minutes in a mixer

of 6 cu.ft. capacityi Each wall was cast from two batches of concrete

of approximately the same proportions. Three 6 x 12" control cylinders

were also cast from each batch. The concrete was placed in the forms and

cylinder molds with the aid of a high-frequency internal vibrator.

The walls and control cylinders were stored under moist condi-

tions for five days, after which they were removed from the forms and

kept stored in the air of the laboratory until tested.

Figure 13 shows the walls and cylinders just after being cast.

- 26 -
 F-
o 0 0 0
0

I I I
Co CoCo
4
·r-l~ ·-F-

*H *H *

cc cc cx~
r-\ HH
I

0 0 O O
0 0 O0O

In I 1 tr\
I I

.~ I C Co
*H *.
COO CO
Co
0 0
F-. F--

E-4
bO
0 r- r-I H H H
0 O
O 1 ! I I
ho

o
00 .H--I
0 4 4-3
H
(D 0 H H0
4'. 4-.

tCj tC tCi
cl c MH
_ C\clJ
CN C
NC N Se -- r--
ma to~ us v
O o *H H H Hp H-
H 0 - 0 0
- 0
0
O
-) o -t o - -- --

Hr r- r- r-
. rtC
oc

E-i
0...
.

CO H c11
HCMz
t z
m .t

- 27 -
 M -tC co
blE-o co
(0

o ,H

a>
*H H<
V\ - CO O
r-I
-p
0
0)0 P-

H
CO

0 0 0 0
VI) *
0 OO C.)
M
\ ED 0 0
CO

0
4,
0H a,g
0 0 *(-
O O 0
+3 ·r

V\ u\ _O

H
H

4,
0)
H-

00)
0

0j CM CJ C,j

o,
C
0)

IH C'J
rn 4 % % :1

- 28 -
0
H

H
H
H

- 29 -
 CHAPTER IV
EQUIPMENT AND INSTRUMENTATION

The equipment and instrumentation necessary to apply the load

and to record the behavior of a test specimen are as important a part of

the test setup as the specimen itself. The reliability and consistency

of this equipment's behavior is of prime necessity for the success of the

test program. The fact that the equipment and instrumentation necessary

for a dynamic test are infinitely more complicated than the equipment for

the static testing of comparable specimens is perhaps not so obvious, but

this fact will be attested to in the following description.

The apparatus used is designed to provide a clear picture of the

behavior of a test specimen under dynamic loading, which will also serve

as an adequate basis for comparison with tests of similar specimens sub-

jected to static loading. The pulse of the loads applied should be control-

lable as to magnitude and duration, and the force-time shape should follow

any desired function. The apparatus should also be able to record reac-

tions, deflections and strains as functions of time.

The different parts of the apparatus may be grouped under the

following three principal items:

I. Loading equipment.

II.. Measuring equipment or instrumentation.

III. Miscellaneous equipment.

I. Loading Equipment

The loading equipment used in this investigation may be referred

to as "Dynamic Testing Machine". It was specially designed and constructed

- 30 -
for this experimental program, under contract with the Corps of Engineers.

This Dynamic Testing Machine consists of two basic parts:

A. The Basic Frame, and

B. The Loading Unit.

A. The Basic Frame

The supporting frame for the specimens and loading device con-

sists of many parts. Figure 5 is a schematical diagram of the principal

parts of the basic frame. Figure 1L is a side view of the heavy truss

and lateral foundation beam.

As shown in Figure 5, the specimen (A) to be tested is supported

vertically by two horizontal cantilever beams B and C. These beams may

be moved six inches at a time allowing the support of walls from 2 to 16

feet in length.

The horizontal reaction is offered by the fixed lateral horizon-

tal cantilever beam D, to which the reaction is transmitted by means of a

strut E, composed of two channels. The three support cantilever beams

transmit the forces to the two principal trusses I and II.

As it is also indicated in Figure 5, the main loading system F

is supported by a lateral loading cantilever beam G. This loading beam may

be moved up and down by increments of five inches allowing walls from 3ft

9 in., to 9 ft in height to be tested. This loading beam is bolted to the

basic trusses; therefore when the main loading device (hydraulic jack)

applies a load to the specimen, the reaction equal and opposite to the load

applied is transmitted to the basic trusses, and consequently, if the

applied load were static, the trusses would be under loads which equili-

brate themselves and the corresponding support reactions would be zero.

This is clearly shown in the sketch of Figure 15. However, since the loads

- 31 -
to be applied are dynamic ones, at any given time there may exist a con-

siderable resultant force on the truss's supports, which required the

addition of the support beams H and I. H is connected to the pile caps of

the foundation by means of high-strength bolts, and I is cast into a mass

of reinforced concrete J, which was also bolted to the pile caps. As it

is shown in Figure 5 and partially indicated in figure 14, this mass of

reinforced concrete consists of two lateral beams running parallel to the

trusses I and II - connected together by the heavy mass of concrete poured

on beam I. This special shape was required in order that the unbalanced

horizontal force may be transmitted to the foundation.

For more complete information on the desi;n, computations, and

details of the basic frame, refer to report (22).

B. The Loading Unit

The loading unit is the most important feature of the dynamic

testing machine. It was conceived by Professor Howard Simpson, and con-

structed under his direction. It is completely described in reference (22)

and Figure 16 shows diagrammatically its operation.

This loading device consists of several basic sections:

1. The hydraulic loading unit, which is the basic loading device.

2. The sequence control system for the automatic control of the

loading and unloading processes.

3. The pressurizing system.

1. The Hydraulic Loading Unit: This is actually a hydraulic jack

and consists, essentially, of the main cylinder, main piston, and main shaft.

Figure 17 shows a section of this hydraulic jack. Before a dynamic test,

equal forces are applied to both faces of the main piston by the introduc-

- 32 -
tion of compressed oil into the chambers on either side of the piston, and

pressures into the two compartments are adjusted according to the area of

the piston faces (73.631 sq. in. for the push side, and 65.97h sq. in. for

the pull side) and the desired value for the dynamic load, so that no load

is applied to the specimen through the main shaft.

At the time of loading, the oil in one face of the piston (pull

side) is permitted to escape, and the piston is acted upon by the pressure

of the remaining oil on the other face (push side). This pressure is kept

practically constant for small displacements of the piston, by means of

four five-gallon accumulators.

The unloading process involves the flow of oil from the ten-gallon

accumulator to the pull side, and as this oil is at a higher pressure than

that in the push side, the piston is forced to move back, away from the

specimen. The loading and unloading processes are controlled automatically

as it is explained below.

2. The Sequence Control System for the Automatic Control of the Loading

and Unloading Processes: The load command signal is obtained from

a mechanical function generator, which consists of a rotary wiper and fifty

contact points. Each contact has its own potentiometer voltage divider, so

that the voltage on the wiper arm can be made to follow any desired function.

Figure 18 is a schematic diagram showing the functional block diagram of

the electronic control system, and Figure 19 is a photograph of the function

generator control panel.

The voltage function produced by the function generator is at-

tenuated and sent to a summing circuit where it is combined with the signal

from the force transducer at the ram. The combined signal is the error or

the difference between the command and the load on the specimen. The error

- 33 -
signal is amplified and sent to a second summing circuit, where it is

combined with position feedback signals from the second and third stages

of a three-stage hydraulic servo valve. This combined signal controls

the first hydraulic stage. Thus, the error signal controls a feedback-

stabilized hydraulic system, or inner loop. The third stage is a large

three-way valve (dump valve), which controls the flow to and from the pull

side of the ten-inchldiameter hydraulic cylinder. The force on the speci-

men is produced by the difference between the system pressure at the rear

or push chamber of the cylinder, and the controllable bucking pressure at

the'front or pull chamber. Figure 20 shows the servo and dump valves

connected together.

3. Pressurizing System: The pressurizing system consists of two Dumps,

a manifold, h five-gallon accumulators, 1 ten-gallon accumulator, 1 one-

gallon accumulator, two one-pint accumulators, safety devices, filter units,

1 circulating unit, 1 transfer unit, a receiver unit or t-ank, the control

panel, and the necessary piping and tubing. Figure 16 shows the disposi-

tion of these different parts. A photograph of the pumps can be seen in

Figure 21.

The Vickers pump consists essentially of a variable-delivery

piston pump, which can deliver a maximum of h8 g.p.m., the maximum operat-

ing pressure being 3000 psi. The principal purpose of this pump is to

supply the oil to the servo valves. However, it is connected in such a

way that it may also deliver oil to the complete system. The other pump

is a Seco radial hydraulic piunp, whose principal characteristics are the

following:

Maximum pressure (spot rating) 10,000 psi

Maximum pressure (continuous rating) 6,000 psi
Delivery at max. pressure .60 gpm

- 34 -
 The principal purpose of this unit is to feed the accumulators,

and the hydraulic jack.

The manifold allows the delivery of oil from the pull side to the

receiver unit, or the flowing of oil from the ten-gallon accumulator to the

pull side. These two operations are made through the dump valves which are

directly connected to the manifold.

The principal function of the ten-gallon accumulator is to supply

oil, at a determined pressure, to the pull side of the hydraulic jack --

through the dump valve -- to reduce or remove the applied load when required.

The four five-gallon accumulators are connected directly to the

push side and their principal use is to act as a source of fluid power to

hold the pressure behind the face of the piston when it moves forward. The

one-gallon accumulator is connected to the servo valves to act as a leakage

compensation and to help supply the large flows required during the test.

The two one-pint accumulators are connected directly to the pull side with

the object of smoothing out pulsations, should this ever prove necessary.

Figure 22. shows some of the accumulators in place.

The safety devices consist of safety heads with rupture dia-

phragms, one of which is connected between the pull and push sides in order

to prevent the ramming of the piston against the front side of the cylinder

due to an unpredictable drop of pressure in the pull side and simultaneous

failure of the specimen. The other safety head with rupture diaphragm is

connected between the ten-gallon accumulator and the push side and it will

rupture if a sudden and unpredictable drop of pressure occurs in the push

side. This will permit a fast balance of pressure between push and pull

sides, thus preventing the ramming of the piston against the back of the

cylinder. The Barksdale valve located on the control panel may also be con-

- 35 -
sidered as a safety device, because if it is opened it will allow the rapid

draining of the push side. To open the valve it is only necessary to hit

down the lever.

The filter units consist of lO1 filters, located at different

points. One of these filters is connected to the Vickers unit, and filters

all the oil that this unit pumps into the system. Another 104 filter is

connected between the transfer unit and the Seco unit. Two 10 1 filters

are also placed in parallel just before the servo valves, with the purpose

of filtering all the oil that goes to these valves. In addition to these

filters, it was necessary to connect a unit which filters all the oil that

comes out of the system through a l1 filter, just before it is returned

to the tank of the Vickers unit. Figure 23 shows this unit.

The transfer unit is a small pump which transfers the oil from

the tank of the Vickers unit to that of the Seco unit. The receiver unit

is a three-gallon tank into which the oil from the servo valve and the pull

side of the cylinder is drained, before it is sent back to the tank of the

Vickers unit through the circulating unit.

Figure 2h is a photograph of the control panel, the schematic

diagram of which is shown in Figure 16.

Oil from any supply pump can be directed into the system through

the line interlock valves which are on the control panel.

As an example of the use of the panel in a dynamic test, we may

take the following description: One starts the Vickers pump with all valves

closed, except for valves 10, 2, and the Braksdale valve. All the pressure

gage line valves (6'; 4'; 1'; 5'; and 7), are just cracked so that the pres-

sure in any line will be indicated at any moment.. In this way, it is pos-

sible to bleed pressure into the servo system in an amount suffic'ient to

- 36 -
operate the dump valve -- approximately 2,500 psi. The test supervisor

operating the electronics maintains contact with the operator of the

pressurizing system control panel, and once he has stabilized the inner

loops and has moved the dump valve to a position that will connect the pull

side of the main cylinder and the ten-gallon accumulator, he gives the

order to supply pressure to the system. At this moment the control panel

oper'ator opens valve 1, and closes valve 2. This will allow the oil to

flow to any part of the system, once the line interlock valves 6,h,5,9, are

opened. One starts opening valves 6 and h, which will allow the supply of

pressure to the 10 gal. accumulator and pull side of the cylinder. If

the pressure required for the test is smaller than 2,500 psi, this is

obtained directly by operating the Vickers pump, but if the pressure is

larger than 2,500 psi, the Vickers pump is turned off at 2,500 psi pressure,

and closing valve 10, the Seco unit is started, which can then give a maxi-

mum pressure of 10,000 psi. Once the pressure in the 10 gal. accumulator

and the pull side of the cylinder is that desired for the test, valves 6

and h are closed, and then the Barksdale valve is closed at the same time

that valve h is opened. This operation will allow the supply of pressure

to the push side. Due to the fact that the diameter of the shaft is

greater in the pull side than in the push side, the area of the pull face

of the main piston is smaller than that of the push side. Consequently,

the pressure required in the push chamber to maintain equilibrium is

smaller than that required in the pull chamber by a factor of 0.896. Then,

one must take the precaution that the pressure in the push side does not

exceed a value equal to 0.896 times that of the pull side.

When the pressure in the push chamber has reached the desired

amount, determined by the desired magnitude of load, valve 5 is closed,

- 37 -
and valve 9 is then opened. If the pressure desired in the 10 gal. accumu-

lator and in the chambers of the cylinder required the use of the Seco pump,

this pump must now be turned off, and the Vickers unit is to be started once

more.

The operator of the control panel gives the signal of "ready for

testing" to the test supervisor, and the test can then be performed as soon

as the supervisor brings the ram in contact with the specimen. The test

is initiated by pressing the fire button of the function generator.

To bleed off the supply lines after the test, special precaution

has to be taken in extracting first the oil from the push chamber by open-

ing the Barksdale valve, then draining the complete system by opening valves

2, 6 and h. Servo pressure must be maintained until all push pressure has

disappeared, to avoid the accidental opening of the dump valve.

II. Measuring Equipment or Instrumentation

For the complete definition of the behavior of a test specimen

under dynamic loading, the measuring equipment had to be designed to provide

a finished record of the applied load, reactions, deflections, and strains,

all as functions of time. The instruments also had to be capable of record-

ing the detailed shape of the mentioned quantities for force-time curves of

any shape, and pulses of different magnitude and duration. However, as it

was pointed out in Chapter I, the principal objective of this preliminary

test program was not only the evaluation of the properties of the shear

wall, but also the evaluation of the testing machine. Since one of the most

important properties of the machine is the speed of the load build up and

the shape of the load-time curve during build up, it was decided to run this

first series of tests with a load-time variation as sketched in Figure 27,

- 38 -
Where the load-build-up time Tr was expected to be of the order of 20 to

30 milliseconds, and the total duration of testing T wvas planned to be

about hO to 60 milliseconds. This fact simplified considerably the design

of the necessary instrumnents, because, as the time T was planned to be

approximately equa wo to three times the time T , it was only necessary

r
to design a high-speed recording system.

It must be pointed out that, as only four Cathode Ray Oscillo-

scopes were readily available, and this would only allow the recording of

eight simultaneous signals -- without the use of special equipment -- it

was decided to measure only load, reactions and deflections in this first

series of tests.

The different electronic, photographic and mechanical means used,

are grouped and described below, under the following items:

A. Measurement of Load and Reactions

B. Measurements of Deflections

C. Recording Equipment

D. Timing System

E. Calibration

The functional block diagram for the instrumentation is shoT•

in Figure 28 and Figure 29 is a view of the instruments.

A. Measurement of Load and Reactions

The measurement of load and reactions may actually be regarded

as a measurement of strains, because the measurements were obtained

through the use of standard SR-h gages and strain measuring equipment. The

gages were attached to Load Cells and calibrated in terms of kips of load.

1i.Load: For purposes of analysis and proper functioning of the con-

- 39 -
trol of loading and unloading processes, it is important to know the

magnitude of the load applied by the machine as a function of time. Of

the various methods possible for the determination of this force-time

curve, that of using sensing elements such as Baldwin SR-h strain gages

seemed to be most practical and reliable. These sensing elements have to

be attached to an elastically strained member (load cell) near the point

of application of load to the wall, and the change in electrical proper-

ties, such as resistance due to the strain, is converted to force by

calibration with known applied load.

The design of the load cell for this particular investigation

depended upon the mounting problem, and the sensitivity required. The

simple column (hollow or solid), seemed to be the most advantageous and

simplest load cell shape for use in this case. As the load cell had to be

connected to the ram (or main shaft) of the hydraulic jack and to the ele-

ment (Ram Head) which will be in direct contact with the specimen to be

tested, and also had to be interchangeable (for sensitivity reasons that

will be explained below), the setup shown in detail in Figure 30 was

adopted solving in this way the mounting problem.

In order to satisfy the sensitivity requirements, the following

steps were taken:

With the object of obtaining the maximum strain possible for a

given stress intensity, aluminum alloy was selected.

Logically, the simplest method of obtaining a constant output

from the load cell as the maximum load decreases, is through the use of

tube with reduction of the wall thickness. With this in mind, and since

it was expected for this first series of tests, that the ultimate resist-

ance of the shear wall would not be greater than 150 kips, a drawn tube

- ho -
of aluminum alloy 6062-T6, with a h in. outside diameter, and a wall thick-

ness of 1/2 in., was selected.

Not only the stress in the load cell was checked, but also special

precautions were taken in checking that the natural frequency (which is a

direct function of the area in the form fn ), does not go below certain

limits imposed by the frequency of the wall to be tested.

The following are the equations used in order to satisfy the

requirements of the shear wall to be tested.

To obtain the output of a load cell for a given force, the follow-

ing expression was used:

GF x 1V)
o 2 2(1)
Er(r -r )

V = output voltage
o

GF" = gage factor of the sensing element

P = applied load

V = applied voltage

r = outside diameter of the cylinder

rl = inside diameter of the cylinder

And for the frequency:

1= ETE(r 2-r 2)
n 2-T 1(M + 0.33m)

M = lumped mass at the end of the load cell

m = mass of the cylinder gage section

E = elasticity modulus of 6062-T6

1 = length of the load cell.

- l1 -
 Past work has shovwn positively that SR-h gages have little, if

any, frequency restrictions below 50,000 cycles per second. Expression (1)

imposes a limitation on the load cell sensitivity. Type C-10 or CB-10 are

the best for high sensitivity load cells for dynamic tests. However, as

they were not readily available, C-7 models with a gage factor of 3.3 were

used.instead. Four C-7 strain gages were mounted on the outer surface of

the aluminum tube in an alternating pattern. Two of the gages were placed

horizontally and parallel to the axis of the tube, and two are circumferen-

tial (vertical). As it is shown by Figure 31(a), a Wheatstone bridge cir-

cuit is formed with the horizontal gages in opposite legs. This arrange-

ment eliminates the effect of eccentric loading, if present, and multiplies

the average strain output of the horizontal leg by approximately 2.6. Con-

sequently, a maximum voltage of 22.5 volts can be apolied to this circuit.

Substituting our data into equation (1), the output of the load cell is seen

to be 1.3 millivolts per kip of load, which can be considered as a satis-

factory sensitivity. However, the results of the preliminary tests revealed

that in order to obtain better control of the loading and unloading processes

it would be better to have large output voltage, for which reason it is

planned - for the successive tests - to use the arrangement shown in Figure

31(b).

Equation (2) yields a natural frequency of approximately 3,000

cycles per second. If it is now considered that the wall to be tested was

expected to have a maximum natural frequency of 200 cycles per second, it

is possible to expect that the unit designed would probably adequately

measure the load applied in the tests.

The calibration of the load cell was obtained previously in a

hydraulic testing machine, as it is shown in Figure 32.

- h2 -
 2. Reactions: The two vertical, and the horizontal reactions of a

shear wall specimen are measured by special load cells. The shape of this

load cell was determined by mounting requirements, and the necessity of

possessing great flexibility in the plane of the wall, since they also act

as supports. As it is shown in detail in Figure 33 these load cells con-

sist of three seven and a half inch pieces of 8 WF 48 steel. Actually the

design was made for an aluminum H-beam, with a depth of 8 in. and web thick-

ness of 0.5 in. But as this standard structural shape was not readily

available, it was decided to use three welded plates. Experience showed,

however, that it was too difficult in practice to obtain welds of sufficient

strength and it became necessary to use steel. In order to obtain the sen-

sitivity required, eight SR-h, type C-7 strain gages are mounted as shown

in Figure 33. This arrangement eliminates the effect of any eccentricity

of load and results in a signal output from the bridge equal to 2.6 times

the average of the vertical gage. These load cells have been calibrated

statically under an axial load of 70 kips compression and tension.

B. Measurement of Deflections

Deflections of the shear wall specimen are measured at two points

-- as shown in Figure 27 -- by means of electric-inductance gages of the

moving-core solenoid type, conventionally termed as Linear Variable Differ-

ential Transformer (L.V.D.T.) and manufactured by the Schaevitz Engineering

Company. Since previous static tests indicated that the maximum horizontal

deflection before failure would not exceed 1 in., a Schaevitz gage with a

maximum range of 1" was used. The gage used for measuring vertical deflec-

tions at the loading point had a maximum range of + 3/8", and for the one

used at the support for the vertical displacement, the range was + 1/4".

For the horizontal displacement of the support, one with a maximum range of

- 43 -
+ 1" was used. Due to the fact that the point at which the vertical deflec-

tion is to be measured will also move considerably in the horizontal direc-

tion, a difficult alignment problem was posed. This problem was solved by

using the device that is shown in Figure 34 and which essentially consists

of the use of two universal joints.

C. Recording Equipment:

The signals from the load, reactions, and deflections are detected

on four Du~ont type 322, Dual Beam Cathode-Ray oscilloscopes, and the perma-

nent record of the traces from the screen of the Cathode-Ray oscilloscopes

are obtained by using four DuMont Type 297 oscillograph record cameras;

which operating on the Polaroid-Land principle, produce a finished print

(oscillogram) 60 seconds after exposure.

D. Timing System

The number of events taking place and the short duration of the

loading period made it mandatory to adopt a triggering device such that the

starting of the complete instrumentation system could occur simultaneously

with the initiation of the test. This is obtained by pressing the fire

button of the function generator, thus sending -- 2 milliseconds prior to

the function - a voltage pulse to the external synchronization of all the

Dual Beam Cathode-Ray oscilloscopes.

The plotting of load, reactions and deflections versus time, was

made possible by energizing the Schaevitz with a 500-cycles voltage. This

was obtained by sending a carrier wave to the Schaevitz with a Hewlett-

Packard 200A audio oscillator, thus obtaining in this way a modulated wave

for the deflection signal, with a time interval equal to 2 ms. between wave

peaks.

- 44 -
 As there are four deflections and four forces (one load and three

reactions) to be measured, one deflection and one force signal were recorded

in each oscilloscope; and since the deflection is calibrated versus time,

it is possible to obtain the calibration of force vs. time.

E. Calibration

The load cells used to measure load and reactions were calibrated

under static loading, using a Baldwin 200,000 lbs. hydraulic testing mach-

ine. During this calibration, shunting resistances of a precision wire

wound resistance were placed across one leg of the bridge, so that their

equivalent values in terms of load could be determined. In order to obtain

the calibration trace on the screen of the oscilloscope, the zero reference

line is established first, and then the bridge is unbalanced with the

calibrated resistance.

The calibration of the Schaevitz is obtained from a linear wire

wound calibrated potentiometer, whose value was determined during initial

calibration of the Schaevitz, in terms of mechanical movement of an Ames

Dial Gage.

III. Miscellaneous Equipment

Among the various parts of the mechanical and electrical equip-

ment used on this experimental work, there are several small devices which

are indispensable to the smooth performance of a wall test, but do not merit

detailed discussion. These include:

A. The Lateral Restraint System

B. Deflection Gage Mount

C. Indicator of the Ram Head Position

D. Guide for the Ram and

E. Restraining Strut.

- 45 -
 A brief description of each of these auxiliary equipment follows.

A. The Lateral Restraint System

In order to prevent any lateral movement of the wall during the

test, a lateral restraint system was devised for these tests. Figure 35

is an overall view of this systme, connected to the specimen, and Figure

36 shows in detail the different parts of this device.

B. Deflection Gage Mount

The deflection gages were supported on an auxiliary frame, com-

posed of two horizontal bars bolted to two poles which were fastened between

floor and ceiling.

C. Indicator of the Ram Head Position

The function of this device is to indicate the position of the

ram head position with respect to the specimen. It is illustrated in Figure

37, and consists essentially in a slide-wire deflection gage, whose output

can be seen on the screen of an oscilloscope.

D. Guide for the Ram

Illustrated in Figure 38 is the guide for the ram, which is a

device that prevents the transmission of any large beniding moment to the

main shaft of the hydraulic jack. This bending moment can be produced by

friction between the ram head and the plate connected to the specimen.

Figure 39 gives the details of this device.

E. Restraining Strut

In order to protect the load cells, used for reactions, and the

ram from damage caused by a specimen overriding the stroke of the ram, a

special strut of steel was placed behind the wall, limiting the total hor-

izontal deflection of the wall to two inches. Figure hO shows this strut

in place.
- 46 -
 CHAPTER V
TESTS OF WALLS

Although the long time static resistance of shear walls is not

of major concern in this study, it was considered desirable to make an

initial static test on a wall similar in properties to that to be tested

dynamically. Such a test was considered to be valuable, not only because

it would provide a way of comparing the static resistance of the structural

element to its dynamic behavior, as well as a comparison with the static

tests made previously (16); but also because it would afford an opportunity

to check the correct operation of all the equipment at a loading rate slow

enough to permit -- as a double-check -- the simultaneous use of Ames dial

gages and Linear Variable Differential Transformer gages. This previous

test would also make possible thestudy of the importance of the support

displacements, measured by placing at the supports special Ames dial gages.

I. Wall Preparation

The preparation of the wall for testing is the same whether the

test is to be made dynamically (load duration 20-100 milliseconds) or

statically (load gradually applied over a period of 5-10 minutes), except

that more deflection brackets and supports for Ames dial gages were neces-

sary for the static test. Figure 42 shows schematically the location of

the gages for the static test, and Figure 43 for the dynamic tests.

Three days after the wall is cast, it is removed from the form-

work and stored in the laboratory until the tests on the corresponding cy-

linders give a compressive resistance close to that required. Then the

-0 -
wall is moved into position on the reaction-measuring supports, which are

already bolted in position on the supporting beams, as shown in Figure 4h.

The wall is then bolted by means of six 7/8" Allen Socket-Head Cap Screws

per support. Then the lateral restraint mechanism is bolted on the already

cast bolts, and connected with the part attached to the frame, as shown in

Figure 35. The wall is then white washed in order to improve the detection

of the cracks. Next, the brackets for deflections are attached to the wall

on the already cast bolts. Then the frame for supporting the gages is put

in place and the last step .is the attachment of the gages and the making

of all the electrical connections required for recording and calibrating

the various measuring devices. A wall ready for testing is shown in Figure

S5. It must be noted that the preparation of the wall for testing requires

at least the work of two men, for a period of twelve hours.

II. Test Procedure

In a static test, after the zero value of each measuring device

is read, the calibrating traces for each electronic measuring device are

recorded according to the procedures for calibrating, described in Chapter

IV, and the gain of each amplifier is first set so that the calibrating

step representing the greatest trace deflection -- which in turn represents

a value of load, reaction or deflection greater than that expected in the

test -- will remain on the record.

The load is applied by allowing the gradual flow of oil into the

chamber behind the main piston (push side), and readings are taken for every

50 psi of oil pressure increment, which is equivalent to a load increment

of 3680 lbs.
In a preliminary step, the load is increased until an oil pressure

of 300 psi is reached, and then released, with the object of checking if

- 48 -
all the measuring devices are working. The load is then increased until

the maximum resistance of the wall has been overcome and its travel is

stopped by the restraining strut placed behind the Wall.

In a dynamic test, after the calibration traces have been

recorded and the precharge pressure of the nitrogen in the different ac-

cumulators arranged according to the maximum pressure of oil to be applied,

oil is pumped into the servo mechanism until a pressure of 2,000 psi is

reached, with the object of activating the dump valve. Then the dump valve

is set in neutral position by means of the balance control, and the two

inner-loop gains (one corresponding to the servo, and the other to the dump

valve) are set in order to obtain the stability of these two inner loops.

Next, the dump valve is moved to the closed position, which wrill allow the

pumping of oil to the pull side, and also to the ten-gallon accumulators,

because they are now interconnected through the manifold and the dump

valve. Once the pressure on the pull side and ten-gallon accumulators

reaches a value high enough to give more force on the pull side than that

which will exist on the push side under the maximum required oil pressure

for the test to be performed, the control valves to the pull side and ten-

gallon accumulators are closed, and oil is pumped to the push side of the.

loading piston. This is done until the pressure reaches the value required

for the test.

Once the oil pressures in the different parts of the system are

those required for testing, the man who is operating the hydraulic system

gives the "ready" order to the man in charge of the electronic equipment;

who then, by means of the balance control moves the dump valve to the center

position (neutral), and then opens it slightly in order to move the ram

towards the wall. When the ram-head position indicator indicates that the

- 49 -
head of the ram is close to the wall, the shutters of all the oscillograph

record cameras are opened, and when the signal of the ram-head indicator

shows that the ram is practically touching the wall, the man in charge of

the function generator presses the fire button, which simultaneously re-

leases the trigger on all the oscilloscopes and sends the signal to the

summing point, where the output of the load-cell-ram-bridge is connected.

This activates the whole system, producing the load desired. This load

is established beforehand by setting the amplitude of the function genera-

tor's signal in correspondence to the magnitude of the load desired.

In these preliminary dynamic tests, it was decided to apply to

each wall dynamic loads of increasing magnitude, starting from a value

which is smaller than that at which cracking should be expected -- accord-

ing to the static test - and then increasing the magnitude of the peak

load, until a value that produces the collapse of the wall is reached.

III. Test Results

A. Static Test

1. Specimen #1: Specimen #1 was tested under lateral static load, and

the results of this test are presented in the form of curves and photographs.

Figure 46 shows the complete picture of the deflections that the wall and

its supports suffered, while Figure 47 shows the lateral movement of the

top of the shear wall with respect to the foundation (supports), which is

of primary importance.

The analysis of the curves of Figure 46 shows that although the

vertical displacement of support C given by curve 8 is negligible - which

was expected due to the special precautions .taken, such as the welding of

the supporting beam C to the principal truss, the vertical displacement of

support B, given by curve 4, was considerable, and the same can be said

- 5o -
with respect to the horizontal displacement given by curve 3. In spite of

the special precautions taken with the connection of the supporting beam

B with the trusses, such as the introduction of shims between the gusset

plate of the supporting beam and the web of the two channels that form the

bottom chord of the two trusses, and that the nuts of the high-strength

bolts were driven in with power wrenches in order to provide a good clamp-
ing force, the large vertical displacement seems to indicate that the fric-
tion produced between the joined members was not sufficiently high to pre-
vent vertical slipping.

The considerable horizontal movement can also be explained by horizon-

tal slipping at the junction of the horizontal support with the web of the

two channels that compose the strut E (see Figure 5).

Thus, if agreement between experimental and theoretical results is

expected, it is not only necessary to measure the displacement of the

support in the dynamic tests, but also to include in the theoretical anal-
ysis the effect of the rotation and sliding of the shear wall as a con-

sequence of non-rigid supports.

The curve (a) of Figure 47, obtained by subtracting from the total dis-

placement measured at the loading point the value obtained for the horizon-

tal movement of the support B, plus the horizontal movement due to the ro-

tation caused by the vertical movement of the supports; shows that the first
crack in the wall occurred at a load of approximately 25 Kips. This is what
actually happened during the first test, when -- in order to check the ins-

trumentation - the oil pressure in the push side was increased up to 350

psi. The first crack was that marked with number 1 in the picture of
Figure 48. The complete test was actually carried out on a cracked wall.

The failure of the wall happened at a load of 73.6 kips, due to

- 51 -
failure of the welded connection of the reinforcement of the tension column

with the supporting plate; as shown in Figure 48.

The maximum relative deflection recorded at the start of yielding was

0.12 in. With respect to these results, the following comparison should be
made:

a. The tests carried out at Stanford University (16) on an analogous

specimen -- where the only difference lay in the strength of the concrete,

(3 vs. 5) x 103 psi, and in the modulus of elasticity, (3 vs. 4.5) x 106 psi
-- gave the following results:

First crack load 24 kips Deflection at ul-

timate load 0.20 in.
Ultimate load 71 kips

The load-deflection curve obtained in this test is plotted in Figure

47, curve (b). It should be noted that the ultimate load was associated
by shear failure at the base of the compression column.

b. The computation of the ultimate load -- assuming that the stress-

strain relation for the steel reinforcement of the frame is the ideal plas-

tic relation, (i.e.) neglecting the increase of stress during strain hard-

ening - gives an ultimate value of only 55 kips.

When the results obtained are compared with those of the Stanford

University investigation, we can see that there is some difference, es-

pecially in the cracked region. Perhaps this discrepancy can be explained

by the difference in concrete properties, and by the different method of

holding down the tension column of the shear wall at the support. It seems
that for the test performed at Stanford University, the support of the ten-

sion column was considerably stiffer than that used in this test; and fur-

thermore, special reinforcements were introduced at the bottom of the tension

column, which avoid the yielding of the column at this point.

- 52 -
 The discrepancy between the experimental ultimate load obtained dur-
ing the test, and that obtained analytically, can be explained by the pres-

ence of a vertical forceat the loading point as a consequence of the

considerable rotation of the wall produced by the large movement of sup-

port B.

B. Dynamic Tests

A total of three specimens (numbers 2, 3, and h) were tested dyna-

mically. Continuous photographic recording of the oscilloscope's trace

for the applied load, reactions and deflections at four points as functions

of time was obtained during each dynamic test by means of oscilloscope-

record cameras. An example of this photographic recording is shown in

Figure 49. It is the complete data for the test number six, performed on

wall #4.
The results of the dynamic test on each specimen are presented in

the form of curves, tables, and photographs, and a brief description of

these results for each specimen is given as follows:

i. Specimen #2: This specimen was subjected to several load pulses.

The results obtained in four of these tests are presented in Figures 50

through 57. From these figures it can be seen that the load pulse applied

was not completely consistent in shape.

The sharp peaks and depressions in the load-time curves were caused

by an improper gain and vibrations in the system. However, if these peaks

and depressions are averaged and a curve is drawn through the average points,
a smooth load pulse can be obtained; however, its shape varies from one test

to another. The rise time varies from 10.0 ms. to 30.0 ms., and the total
duration from 35.0 to 75.0 msa.

- 53 -
 Figures 50, 52, 54, and 56 give the value of the load and reactions

versus time, and Figures 51, 53, 55, and 57 give, respectively, the value
of the deflections vs. time obtained in these tests. As it is shown in
this last group of figures, only the data corresponding to the horizontal
deflection at the loading point, and at the support was obtained, due to
the fact that at the time the tests were started, the complete set of de-

vices necessary for holding and attaching the Shaevitz gages was not avail-
able. Consequently, it was not possible to obtain the component of the

horizontal displacement at the loading point caused by the vertical move-

ment of the support.

The maximum value of the applied dynamic load, and the time required

to reach this peak load, the lateral movement of the loading point with
respect to the foundation (supports) at the time at which peak load is

reached, and the maximum value of this relative lateral movement with the

time at which it happens are shown in Table V-l, for each test. Also shown
in this table is the static deflection obtained on specimen #1 correspond-

ing to the value of the peak load of each dynamic test.

A comparison of the static deflection with the dynamic one, reveals

that:
a. Test No. 1 performed on an uncracked wall gives practically the

same deflection as that in the static test.

b. For the following tests, when the peak load is smaller than 30 kips,

the dynamic deflections are larger than the static.

c. For the first eight tests, when the peak load is larger than 30 kips,
. the dynamic deflection is smaller than the static, the difference
 being quite considerable in the case that the load pulse has a
short rise time and a definite and sharp peak load (see Test No. 8).

d. After test No. 8 the dynamic deflections are larger than the static

ones; the explanation of this result can be found in the fact that
the wall was considerably damaged (cracked) after test No. 8.

It must also be noted that the first cracking occurred during test
No. 5, at a load of 40 kips.
Because of an error when trying to precharge the ten-gallon accum-

ulator to a higher pressure, after test No. 12, the pressure in the pull

side was released without releasing that in the push side first, resulting

in an unbalanced force that produced the failure of the specimen. The

failure occurred by tension failure of the welding of the reinforcement of

the tension column with the support plate. Figure 53 shows the cracks on
the wall after test No. 6, which had a peak load of 31.5 kips. However,

the cracks shown in the picture were produced by a load of 40 kips, ap-

plied during test No. 5. Figure 59 shows the wall after failure.
2. Specimen #3: Specimen #3 was submitted to several load pulses of

varying magnitude and duration.

Figures 60 through 71 show the values of the load, reactions and
deflections vs. time, obtained in some of these tests. From these figures

it can be seen that the rise time varies from 12.5 to 25.5 milliseconds
and the total duration varies from 35 to 62 milliseconds. It may also be

seen that during the first tests the load did not come back to zero. The

force-time shape varies considerably from one test to another and only
that corresponding to test 13 can be considered as triangular in shape, if

the small peaks and depressions are smoothed with an average curve. -=

- 55 -
 Table V-2 gives a summary of the test results on specimen #3.
Analyzing these results and the corresponding figures, it can be seen that

when there exists a definite peak load, the maximum horizontal deflection

takes place almost simultaneously, the delay being only of a few millisec-

onds in certain cases. On the other hand, when there is.no definite peak
in the pulse, the maximum deflection may occur considerably before or
after the instant at which the peak load is reached, and more or less co-
inciding with some of the other peaks.

In this case, first cracking occurred during test No. 5, at a load

of 42.5 kips.
If the values of the maximum lateral deflection obtained on the tests

performed on this specimen are compared with those obtained on a specimen

being subjected to a similar static load, it can be seen that considerably

higher deflections resulted on specimen #3. This difference can be only

partially explained by the difference in the moduli of elasticity of the

concrete (4.70 x 106 psi for the specimen under static loading vs. 3.80 x 106

psi for specimen #3.)

Because of an error during test 16, the vertical reaction at the fixed

support and the vertical displacement at the same support were not obtain-
ed.
Figure 72 shows wall #3 after failure produced by a dynamic load with

a peak value of approximately 68 kips. This failure was due to the defect-
ive welding of the reinforcement of the tension column to the support plate.

3. Specimen #4: In Figures 73 through 84, the applied load, reactions

and deflections occurring on some of the several tests carried out on

specimen #4 are plotted as functions of time. It can be seen from thes6

figures that:

- 56 -
 a. The shape ;of the pulse load is not consistent, even if a smooth
curve is drawn through average points of the peaks and depressions.

b. The rise time of the pulse varies from 3 to 33.5 milliseconds.

c. The total duration of the pulse varies from 21 to 65 milliseconds.

d. The maximum lateral displacement of the loading point with respect
to the supports for the case when the pulse has a definite peak load

is reached almost simultaneously with the peak load.

Table V-3 gives a summary of all the data obtained on this specimen.

The first crack occurred at a load of 37.0 kips.

Comparing the values given in Tables V-2 and V-3, it can be seen that
the maximum value for the horizontal displacement obtained in the tests

performed on specimen #4 are larger than those produced by similar loads

on specimen #3.

Wall #4 failed under a dynamic load whose peak value was equal to 74
kips. Figure 85 shows the crack pattern of this wall after failure, which

was produced also by tension failure of the reinforcement of the tension

column with the support plate.

The cause for the failure produced in the four walls tested was un -
expected, (i.e.) the defective welding of the connection between reinforce-
ment of the tension column and the support plate. In an actual shear
wall construction, this kind of failure cannot happen, and consequently

the question of how much more load can the specimen withstand if this

premature failure is avoided, arises. In order to find a quick answer to

this question it was decided to repair the connection of wall #4, chipping

out the concrete at the bottom of the tension column and welding again the
reinforcement to the support plate, taking special precautions to avoid -

failure in the welding section.

- 57 -
 The repaired wall was then submitted to a new series of tests,

starting with a peak load of 30 kips in order to check the instrumentation.

Once the instrumentation was in order, two pulses with a peak load of

73 and 76 kips respectively, were applied. These peak loads were approx-
imately equal to those at which the failure of the connection occurred.

This time the wall endured the load, and when a new dynamic load with a

peak value equal to 80 kips was applied, the wall also withstood it.
The failure of the wall was produced by a load of 95 kips, with a

rise time of 21 ms. This time the failure was associated with a shearing
off of the compression column at its junction with the beam foundation.
This is the kind of failure that was expected according to the results
obtained in the static tests carried out on similar walls at Stanford
University. (16)
Figure 86 shows the wall after collapse. It is interesting to note
the local crushing of the concrete in the panel around the corner at which
failure occurred.

- 58 -
 in

(1U
%,-0
co - 0 0 0 0- c 01 - 0

b o 000000000000 a4
0
a,v
H
E-1
0 v \0 co o\
m .- -tHH -t o
0 ,
H co~ H o e H ca cH om\
4
0 0

DO 0. 0
H O ·C- '

H
.H
Cd
4'

'
CO
'0 1o
ca 1-4

C~j

H 0
H j \0.o\ \ uj\ -- 54-
0 I-4

HO
to 4' C
oH o r-
U)
Ho o) Ho- HHoHHo o o oAHc H -0 cr\ 0
0 CD
F-4
Ef O E--4 *
v

0 4'
0
*1-

0
U)
U)
rd
o o o - o o 0 o oc 0. t"
r-I M
v 000
0 Cr
to)
0HOH0 0r- M-
000000000000 O,\,
IA O,
H r
OH H...
,H HH
,, H r-4
0)0 -C- J±

H
v Hoo
ýo
v
A .rd

*rI
IO
U- 0 O
H1 0\ e- tO CN
HM Ctri'r -Et- o C- mc % -

V
rl

r-t
E-., v

- 59 -
 $4
a,

CiE44
C
C41240
·i
pj
Si

H 0oI!\ o 0 0C 01
O\ O O M
H>>cO~'j 0

0
E-4

0'Lr\o U
z;t -~t O
oz 00 0000 00000
0000
0 ir

Cd
edr
-q
o00 0 o~)\ ON'\

0
oL 66
00 000
0'68-
00
00 680666
00000

kWd
OS

*ri
• • 0 00 0 0 0 00 0 0 0 0 0

E-4 Nm C C'J
jA C% 0
O\ CNOt q lO \--IA
HH H N HA --- HH C\j
CHNNCM
E- -H
•dI -
'-•.,

C'j
b\ 0o 0nYU~ 0 b- r-ac 0 cc)
(0U j z .- t - cOc\O. r-
m'J At " m c C\j CUCU ---rIeO \O

E-4
zH .HH HAq

- 60 -
 0)
*0

o
)d )
0)1 i-

(1) o
oz 000000O0O
0 0 0 0 04 06A 00
0 0
,O
r-
8.. A muH.H I I.J
0f
H
C cc) C~j a C'0 s cJ 0
C.j
*w-l

NI
H
E-j E 0 'lAO
0 0 0
00O
0
o
0
O
0 ^
Vi
0
*0
01 rS
-OH *0
Cl,
0 0

*8
.0

C--

co 000 0 0 0 CC)0 0 0 C\
0 0CY00\
\0
0
I
E-4
iM

'0d
0
V1
0 0 OO 00 m 0 0
*rl ,1
0
r•
O 1 O O O O1 O 0
itv
14 0 0 00 0 0 0 0 0 0o
4 rda

Cd H v o
91
.*H 4
20

ed
0
*) 0H
*H
*Pc m

CV 0 . 0A 00 I00 I0
t> *0
-
E-4

E-O OH eCm
%~d
V b
-t' LA
% V
\0 r- co a\ ,•

- 61 -
 CHAPTER VI

THEORETICAL INVESTIGATION

1. Introduction

As stated in Chapter I, the basic problem being considered is

the prediction of the resistance and behavior of reinforced concrete
shear walls which are subjected to dynamic loading caused by air blasts

due to the explosion of nuclear weapons.

Elementary approaches to the shear wall problem assume that
the wall is a beam. The beam is considered to be cantilevered from the

foundation. In the case of static loads, wall deflections and stresses

are then found from elementary strength of materials formulas. Results
from such analysis are often quite inaccurate, and completely erroneous

conclusions may be drawn. Therefore, the analysis of a shear wall must

be based on a more exact solution.

The shear wall may be reduced to the problem of a plate loaded

in its own plane. This is a problem in plane stress analysis. Precise
mathematical solutions based on the theory of elasticity are available for
only a small number of special problems. The analysis made in Chapter II
of the different factors involved in the study of this specific problem,

shows that its solution is complicated not only by the nature of the
shear wall itself, which can take many forms, but also by the necessity

to predict three phases of action for any wall, namely:

1. The elastic behavior

2. The behavior when partially cracked and subjected to small

deflection, and

3. The large-deflection behavior.

- 62 -
If to this is added that the problem to be investigated is a dynamic one,
it is obvious that an exact mathematical solution is impossible and some
approximate methods must be developed to provide solutions of engineering
accuracy in a reasonable amount of time.

II. Elastic Behavior

This phase was investigated by Mr. Aaron Finerman (21). The
procedure followed in his studies may be summarized as follows:

Dynamic Analogy: The actual continuous wall was replaced by a con-

centrated mass system inter-connected by weightless springs as shown in

Figure 87. The stiffness and properties of the springs were obtained by
the lattice analogy method.
The fundamental concept involved in the lattice analogy is that

of replacing a plate by a pin-connected lattice having the same deforma-

tions as the plate, under similar stress conditions. In this manner, the

equivalent bar area of the lattice may be determined.

Equation of Motion: The equation of motion of each mass point can
be developed either from considerations of the requirement of dynamic
equilibrium or from the Lagrangian equations.
Numerical Integration: The set of simultaneous equations of motion

was solved by numerical integration.

To obtain the numerical solution, an open type formula was used
to predict the value of displacement in the succeeding time interval.

This predicted value was then corrected by the use of a closed type
formula which required a number of iterations.

Principal Tensile Stress Determination: After a consistent set of

displacements has been obtained, it is necessary to compute the principal

tensile stress at the center of each segment. The stresses are then com-

- 63 -
pared to the cracking stress of concrete, to determine where and when
the first crack appears. Cracking may also take place in the tensile column.

The analyses were carried out until the principal tensile stress
at the center of a segment exceeded the cracking stress of concrete.

III. Cracked Behavior

As was established before, one of the principal objectives of
this work was to predict mathematically the cracked behavior.

The prediction of cracked action involves a number of assump-

tions, and considerable time was spent trying to find a method of approxi-
mate analysis that is simple to apply and yet offers a fairly high degree
of accuracy. In this case, the effort made by Mr. Calladine in order to
draw an analysis of the cracked behavior from the theory of elasticity

must be mentioned. However, the proposed treatment was not a complete

solution by the mathematical theory of elasticity. Judgement had been

applied in the consideration of the stresses and slipping of the steel in

the neighborhood of the crack. Difficulties in the programming of this

method, in the short time.available, made it impractical.

Finally, it was decided to adapt for the cracked range the

simplified lattice used by Mr. Finerman for the elastic behavior. Where
cracks occur in the wall panel, or in the tension column, the behavior
of the wall depends on the location of the cracks, changing of modulus

of elasticity, the value of Poison's ratio, etc., and consequently it is

obvious that the areas determined for the equivalent lattice bars - for
the elastic phase - are no longer valid. Some further assumptions are

required to determine the properties of an equivalent lattice.

In order to simplify the problem, and considering the limited com-
puter time available and storage capacity of the high speed digital computer
Whirlwind I, the following assumptions have been made.
1) Poisson's ratio does not change due to cracking of wall.

2) Elastic modulus of the materials remains constant.

3) The crack extends through the whole segment.

4) Due to the fact that a careful study of the different crack
patterns obtained in the static tests of specimens similar to
those being studied showed that the cracks at the wall panel

occur practically at an angle of 45 degrees, it was decided to

neglect the effect of variation in the angle at what cracking oc-
curs in a segment, and to assume that all cracks occur at that
angle.
5) In order to satisfy the deflection of the segment once a crack occurs

under assumption 4, the stiffness of all the lattice bars changes

for that segment, except for the diagonal compression bar, which

retains the stiffness derived under the assumption of elastic be-

havior.

Assumption 5 may be justified by the following reasoning: The

reinforced concrete wall is considered as a laminated plate of three lami-

nas, as shown in Figure 88, the central lamina or core being a steel

lamina representing the steel reinforcement, and the other two laminas or
faces being the concrete layer.
As long as the stresses acting in the plane of this laminated

plate are smaller than those which will produce cracks in the concrete,
the elastic isotropy of the laminated plate permits the substitution of a

plate of concrete of equivalent thickness (transformed thickness) for the

real plate, and then this substituted by the pin truss according to the
lattice analogy.
When the stresses reach the value that will crack the concrete at
45 degrees (i.e. pure shear stresses), the collaboration of the concrete
laminas in absorbing stresses disappear completely, except for the case of
stresses parallel to the crack. This means that, if for a moment we

neglect this unique possibility of concrete collaboration, as cracks occur,

the effectiveness of the laminated plate is reduced to that of the steel
lamina alone, and consequently the pin truss of the lattice analogy will

be similar to that obtained for the case before cracking. The only dif-
ference will be that the area of the bars, instead of corresponding to

the total transformed thickness will now be proportional to the area of

the steel lamina alone. Now, taking into account the fact that for stresses

parallel to the crack, the cracked laminated plate behaves as if it were

not cracked, the diagonal bar of the corresponding lattice is assumed to

maintain its area proportional to the total transformed thickness.

It is recognized that the assumption of neglecting completely

the collaboration of the concrete with the steel due to the bond between

cracks in members under tension, will result in lattice members of smal-

ler stiffness than actual. However, if a very fine lattice is used,

this assumption will not introduce appreciable error and according to the
analysis of the results obtained by Galletly (10) for the case of static
loads, it is expected that,' even if a coarse lattice network is used, this

assumption is justifiable. Later, in the discussion of the stiffness of

the lattice members, a method for taking into account the collaboration
of concrete is proposed.
It is evident from the above considerations that the effect of the
fineness of the lattice pattern, (i.e.) how fine the lattice must be in or-

der to obtain a fair agreement between experimental and analytical results,

becomes important enough to require a more careful consideration. With this

-66 -
purpose, lattices of two different sizes were investigated. (More com-
plete information is given in appendices A and B)
6. When the strain in a bar of the lattice network has reached the
value corresponding to the yield point stress, the force in that
bar should then be kept constant for any increase in strain.

This assumption is also an approximation, because -- if we neglect

the collaboration of the concrete -- when any unit of the continuous plate

starts yielding under pure shear stresses, it would be necessary to assume

in the analogous lattice -- if the requirement that deformations be equal
is to be satisfied -- that not only the diagonal bar in tension is yield-

ing, but also all the other bars, with the exception of the diagonal bar
in compression; although the horizontal and vertical bars have not yet

reached the yielding point. However, an exact analysis of the deformation

of the unit-continuous plate under stresses that will produce yielding is

not possible and it is believed that the assumption made above will not

introduce any large error.

7. If at any moment the strain in a bar which is in yielding state

starts to decrease, the force in the bar will decrease following
Hooke's Law. If the strain continues to decrease to zero, the
cracks disappear and the concrete starts to collaborate again.

This is shown clearly in Figure 89.

8. The computation is carried out until the deflection at any point
reaches the value for the deflection at which - in the static
test - the load starts to decrease.

This is of course an arbitrary value, and does not mean in any way

that the wall will collapse.

- 67 -
 It is evident that a better point for stopping the computations
would be the moment at which enough bars have yielded to make the struc-
ture a mechanism. However, the requirements for such a method were con-
sidered prohibitive due to their complexity.

According to the above assumptions, the stiffness of the bars is

given by the following equations:

1. Elastic Range:
A
Ki i
Eo (1)
which is the value obtained by Finerman (21); where Ati is the transformed
area of the bar, 1i is the length of the bar, and Ec is the modulus of
elasticity of concrete. This stiffness determines a curve of force vs.

deflection like that shown in Figure 89 by line A' 0 A.

2. Cracked Range:

Once the segment cracks, the stiffness of the bars is given by:
A
s
i
Ki = Es (2)

where A is the steel area of the bar in consideration, and E is the

si s
modulus of elasticity of steel. In Figure 89, the curve F vs. - is
1
given by line OB.

Equation (2) is valid until the strain 1 reaches the value

for which the stress in the steel is equal to the yielding limit. After
that the curve F vs.Al/l in Figure 89 becomes the horizontal line BC.

As discussed before, the stiffnesses obtained from equation (2)

are smaller than the actual due to the assumption that the collaboration
of the concrete between the cracks could be neglected. Different

investigators and authors, (26) (27) (28), have proposed different

-68 -
methods in order to take into account the collaboration of the concrete,
after concrete has cracked. Following Brice (28), whose study seems to

be the most complete, the stiffness after crack can be obtained from:

tf li1 F c

where ftc is the tension stress in the concrete as given by:

i

Fi
A + nA
ci si

Aci is the concrete area.

p is the percentage of steel.

It must be noted that ft /2.75p is the factor that represents

ci
the collaboration of the concrete. According to this, the stiffness would
be:
As
K =
i
[ -
2.75(1
2 + np) ] E
8
This equation is valid until ftc reaches the maximum allowable

value in tension; from then on, the equation that governs the deflections
is given by:
Al.
1 F
fcmax
i s a

and consequently, the F is given by:

1i ft
F. =A E _. +
i si si li 2 max Aci

This equation is valid until the strain A li/li is such that the

stress of the steel on the cracked cross-section reaches the yield

limit, in other words, until:

- 69 -
 F
s y

From then on, the force Fi will remain constant with increasing Ali/l
i.
Figure 90 shows schematically the curve F vs. deflection for the different

phases considered above.

If in any moment - after cracking occurs -- the deformation in

a bar starts to decrease, the force vs. deflection curve follows a line

parallel to the line OB, as indicated by line CD in the figure; until the
total deformation becomes zero -- point D. From then on, the F vs. /1i

curve becomes parallel to the OA line -- line DE in Figure 89.

With these assumptions and incorporating the program developed
by Mr. Finerman (21) for the elastic range, the writing of the general

program for the use of the digital computer Whirlwind I was undertaken by
Mr. Ho, while Mr. Calladine undertook the writing for the new IBM com-
puter. A complete information on the coding for these two high-speed

computers, is given in appendices A and B.

IV. Theoretical Results

A. Data for the Analysis

1. Applied Load. The applied load curve is shown in Figure 91.

The actual loading duration is 75 milliseconds; however, the dynamic effect

to the wall is assumed to be continued until 25 seconds after the applied

load is over.
2. Number of Segments. Detail of the wall is shown in Figure 1.
For the Whirlwind I the wall was equally divided into six segments, each
of 22" square. The dynamic model in the form used for computer solution

is shown in Figure 92. As seen from the figure, two extra dummy columns
and rows are added to the model for the convenience of programming.

- 70 -
Equivalent wall properties for the dynamic model are shown in Appendix A,
item V. 0.10 milliseconds was used as time interval in this computation.

For the IBM 704 computer, the wall was divided into a grid system 7 squares

by 5, each square being 9.2". The time interval was equal to 0.0375
milliseconds.

B. Results
1. From the Whirlwind I program. The results of the analysis are

presented on Figure 93, 94, 95 and 96.

The nomenclature used in these figures is

P = applied Load (which is equal to the static horizontal reaction)

H = horizontal reaction

VL= vertical reaction at left support

VR= vertical reaction at right support
=
VS= vertical reaction for statically applied load P(L)

XL= horizontal displacement of loaded mass point

XR= horizontal displacement of extreme right hand mass point in top row.
YL= vertical displacement of loaded mass point

YR= vertical displacement of extreme right hand mass point in top row.

2. From the IBM 704. The more significant information is given

in Figure 97 and 98. The nomenclature used in these figures is the same
as that used above.

- 71 -
 CHAPTER VII
DISCUSSION OF RESULTS, GENERAL CONCLUSIONS, AND
RECOMMENDATIONS FOR FUTURE RESEARCH

I. Discussion of the Machine's Behavior

As stated in Chapter I, one of the principal objectives of this

preliminary investigation was the analysis of the testing machine's

behavior. In this respect, it is interesting to point out that the

results obtained demonstrated the following:

A. Although in the function generator the rise time of the load was
set at 15 ms., this time varied considerably from test to test. For
small loads, the rise time oscillated around 12 ms., except for some of

the tests carried out on wall #4. In this case, the rise time was con-
siderably less - perhaps this can be attributed to some impact effect.

As the peak value of the load increased, the rise time also increased;
oscillating between 24 and 33.5 ms. at loads greater than 50 kips.

It is felt that these variations in rise time are due to the satura-
tion of the only dump valve used in the system, and it is hoped that with

the introduction of a new dump valve and the use of higher pressure, it

will be possible to obtain a rise time close to the desired value.

B. The shape of the load-time pulse varies considerably from that
required and set in the function generator. From the shape of the load-
time curves given in tables V-1, V-2 and V-3 it can be seen that -- if

the small depressions and peaks are smoothed with an average curve - when
the load is small the shape is completely irregular, showing big oscilla-

tions. As the load increases, and for loads of approximately 30 kips,

- 72 -
the curve presents two definite peaks. When the peak load is larger

than 30 kips, the shape of the average curve tends to be triangular.

It can be also observed that in a few tests, the load did not

return to zero after the peak value was reached. A series of tests per-
formed on a reinforced concrete column after the four tests of the pre-

liminary investigation were completed, showed that these considerable

deviations of the load-time curve were caused by the use of an improper
pressure in the accumulators. It was discovered that the pressure of

nitrogen in the ten-gallon accumulator was too high in comparison to the

low oil pressure used for the push and pull sides, and most of the oscil-

lations were due to the use of high gain.

Considerable improvement in the shape of the load-time curve was
already obtained during the series of tests performed on the reinforced

concrete column. It is believed that all difficulties will be eliminated

when the appropriate nitrogen and oil pressures, and gain are determined

by trial and error.

C. The duration of the load pulse varies from 35 to 75 ms. although

the function generator was set for 50 ms.

The cause for this irregularity seems to lie in the same factors
discussed above under 1 and 2. Above all, the saturation of the dump

valve seems to be the principal reason for the greater duration of the load
pulse.

D. The movement of the supports for the specimen are considerably

larger than expected. This motion not only confuses the experimental

results, not allowing a good comparison between theoretical and experiment-

al results; but also has considerable influence in the behavior of the

- 73 -
loading unit, because a larger displacement of the ram is required, and

consequently a greater flow of oil.

It is planned to stiffen the supporting beam and to make the

connections more rigid, in order to minimize the displacement of the sup-
ports.

II. Discussion of the Results Obtained from Tests Performed on the Walls
It is known that the dynamic response of a member being subjected

to a dynamic impulsive load is not only a function of the intensity of

the applied force as in the static case, but also of the rate of applica-

tion of the force, the duration of the load, and the particular variation
with time of load and resistance. Then, it is believed that in order to
get an idea of the shear wall's dynamic behavior from the data obtained,
given in Chapter V, the following comparisons have to be made.
A. The shape of the reaction vs. time and displacement vs. time

curves must be compared with the load-time curve. When this analysis is

completed it can be seen that in all cases the reaction and displacement
curves resemble the load curve. In fact, if the small peaks and depres-
sions of the load-time curve are smoothed out by an average curve; and if --
in some cases -- some time shifting which is probably due to defective
synchronization and calibration is accounted for; the agreement in shape

is very good. It is possible to observe that any significant irregulari-

ties in the load vs. time curve were reproduced in the reaction and dis-

placement curves.

The resemblance discussed in the preceding paragraph is important,

because it reveals an almost complete absence of dynamic effect and indi-
cates that the response of the wall is practically a static one.

- 74 -
 The conclusions arrived at can be justified by a quick analysis

of the results obtained in the theoretical computations. In effect,

according to the figures obtained by the IBM 704 computer for the dynam-
ic horizontal reaction (see Fig. 97), it can be seen that in the elastic

region this reaction ("H") vibrates about the static reaction, which is
equal to the applied load. However, the amplitude of this vibration is

very small, and its period has a value of approximately 5 milliseconds.

The cause for this oscillation may be considered to be the fundamental

mode of the multi-mass system. Considering that the rise time for the

load was always greater than 13 ms., (i.e.) that the rise time/natural
period of the fundamental mode, ratio was always larger than 2.6, it

follows that no appreciable dynamic effect can be expected. Consequently

the amplitude of vibration and displacement has to be very small --

unobtainable with the sensibility of the intrumentation used in this

series of tests.
B. When the values of the displacements obtained on specimen #2

are compared with those obtained on specimen #1 - assuming that the

same vertical movement of the supports occurs during the dynamic test
rather than during the static test -- it can be seen that as long as

wall #2 was not badly cracked by previous loads, the dynamic displace-
ment that the specimen undergoes practically coincides with that obtained

for similar static loads on wall #1. Considering that the modulus of

elasticity for the concrete in both specimens was approximately the same,

the reasons for this similarity follows from the former conclusion (i.e.)

that there is practically no dynamic effect.

While this discussion was being written, a static test was con-

- 75 -
ducted on a specimen built with concrete whose characteristics were
similar to those of the concrete used in specimens number three and

four. A quick analysis of the results seems to also indicate a good

agreement between the displacements under static and dynamic loads. It

must be pointed out that in order to make a comparison of these results

it is necessary to subtract from the horizontal displacement obtained

in the test, the horizontal displacement due to the rotation of the

wall. This rotation is caused by the considerable vertical displacement

of the supports.
C. When the type of failure is analyzed it can be seen, as pointed

out in Chapter V, that the collapse of all four specimens was produced
by a defective welding between the reinforcement of the tension column

and the support plate. However, the failure of repaired wall #4 was

produced by a shearing off of the compression column at its junction with

the beam. This type of failure agrees with that obtained in the static

test conducted on a similar specimen at Stanford University.

D. Analysis of the Ultimate Resistances If no appreciable dynamic
effect exists, it is logical to expect a larger ultimate resistance
under dynamic than under static load. Unfortunately, due to the pre-
mature failure of the specimens for reasons given above, only the results

obtained during the test conducted on the repaired wall #t, may be

analyzed in an attempt to shed some light on this fact of the problem.

The failure of this specimen took place under a load of 95 kips. Although
it is true that no report of a static test conducted on a wall similar

to #4 is included here, such a static test was just completed, as men-

tioned before. The wall in this case failed under a load of 80 kips --

- 76 -
a value which agrees closely with the results obtained at Stanford
University; where a similar specimen failed under a load of 71 kips.

It is then significant that the repaired wall #4, although extensively

cracked by numerous blows applied prior to the ultimate load, was still

able to resist a load higher than that endured by a similar wall being

loaded statically. If the value of 80 kips is taken as a basis for

comparison, the increase in ultimate load is of the order of 20%. If

the Stanford tests are taken as a basis, the increase is approximately

equal to 33%i

To.justify this increase in ultimate load, two factors must be

considered:
1. An increase in the strength of the member due to the increase

of the yield point of the material under high rate of loading.

2. The ductility of the member -- which may be defined as the

ratio of the maximum deflection (just before collapse) to the yield deflec-
tion -- is an important influence in determining the resistance of a
member to dynamic loads.

Ductile members -- those which can maintain their peak or near

peak ultimate strength for large plastic strains -- are capable of ab-

sorbing a greater amount of energy without failure, and are less liable

to fail or collapse suddenly.

The ductility of a shear wall depends on the type of failure.

In the specimen being studied, failure occurred due to a shearing off of

the compression column, which is a sudden type of failure. The specimen

may be considered as a brittle structure member.

- 77 -
 Consequently, the increased strength of the member under dynamic

load finds its justification in the influence of the speed of loading

on the yield point of the materials. What this influence is precisely,

on members made of concrete and steel is difficult to evaluate, but

factors of the order to 20 to 35% greater than the static strength seem
reasonable.

III. Discussion of the Theoretical Results

If the results obtained from the computations carried out by the

WHIRLWIND I and IBM 704 computers are compared, the following facts

become evident:

A. There is a very good agreement in the values obtained for the

horizontal displacement of the loading point, up to a load of 25 kips,

(i.e.) while the wall behaves elastically.

B. According to the IBM 704 program, the first crack occurredat

a load of approximately 25 kips; whereas, according to the results ob-

tained from WHIRLWIND I's program, the first crack should occur at a load

of approximately 43 kips. This big discrepancy is the first indication

that there is something wrong in one of the two programs.

C. After cracking occurs; according to both programs, the lattice

started to oscillate and practically all the segment cracked. It was

felt that these oscillations and the widespread cracking which followed

were due to the abrupt weakening of the elements at cracking, under the

cracking hypothesis adopted. However, a second run was made by Mr. Calla-
dine, increasing the amount of steel in the wall and beams, retaining its

elastic properties but reducing its yield strength in such a way that the
ultimate strength per element was unchanged, and on the whole, this

- 78 -
second test gave results similar to those of the first.
A third test was also run by Mr. Calladine in order to check

whether an increased number of iterations would be helpful. Clearly,

the elastic solution would not be improved, but it was felt that the

abruptness of the cracking could be better dealt with by repeated

applications of the iterative procedure. However, the desired effect
was not achieved.

The trouble -- which is undoubtedly serious -- seems to be of a

mathematical nature rather than a physical one, and it is hoped that a

careful reappraisal of the programming will point out the error.

IV. Comparison Between Theoretical and Experimental Results

The theoretical computations run on IBM 704 yield what seem to be sa-
tisfactory results up to a load of 23 kips, and those run on WHIRLWIND I

give good results up to a load of 40 kips. It is evident that a comparison

of theoretical and experimental results can only encompass this region,

which may be considered as the elastic range of the wall.

It is interesting to point out, in the first place, that the first

cracks in specimens number three and four, occur under loads of 42.5
and 37 kips, respectively. These values are in very good agreement

with the results obtained for the program run on WHIRILIND I.

This seems to indicate that in the computations performed by IBM 704

the oscillations of the lattice obtained at a load of 23 kips are due

to a mathematical error, and not to cracking.

As it was mentioned before, the theoretical and experimental results

show clearly that the vibrational amplitude of reactions and displacements

- 79 -
is so small, that no dynamic effect is apparent.
If the theoretical and experimental values for the horizontal dis-

placement of the loading point are compared, it may be seen that the re-
sults are inconsistent. For example; at a load of 23 kips WHIRLWIND'S I

computations give a displacement of 0.0115 in., and IBM 704's computations

a displacement of 0.01 in., whereas, in the static test performed on
specimen #1, a value of 0.0017 was obtained for a similar load. The

difference can be explained by considering that the modulus of elasti-

city for the concrete in wall #1 was 4.5 x 106 p.s.i., while in the

theoretical computations, a value of 3 x 106 p.s.i. was used. However,

if the theoretical results are compared with the experimental values

obtained from test No. 3 on wall 3 and test No. 6 on wall 4, it can be

seen that the latter are of the order of 0.035 in., which is considerably
larger than the theoretical ones. The explanation for these inconsistent
results perhaps lies in the difficulties in obtaining a good evaluation

of the experimental value from the data, due to the complications in-
troduced by the displacement of the supports.

V. General Conclusions

The conclusions that can be drawn from the former discussion may be
summarized as follows:

A. Due perhaps to the use of inappLropriate gain in the amplifiers,

incorrect nitrogen and oil pressures, and saturation of the dump valve

of the system, aggravated by the extra oil flow required by the consider-
able displacement of the specimen's supports; the force-time shape of
the pulse, rise time of load, and duration of pulses given by the dynamic

- 80 -
loading machine differ from those set in and required by the function
generator.

B. Although the number of walls tested so far is too small to allow

the drawing of any general conclusion, the almost complete lack of vibra-
tions in the reactions and displacements obtained from the tests on

walls 2, 3, and 4 reveals that the applied load practically produces

no dynamic effect.

The values for the horizontal displacement of the loading point

during a dynamic test and those corresponding to a static test seem to

agree quite closely, in spite of the difficulties introduced in the

exact evaluation of this data by the necessity of taking into consider-

ation the displacement of the supports. The type of failure produced

on all four walls was an unexpected one--the defective welding of the

tension column's reinforcement to the support plate. In the wall

where the connection was repaired, failure occurred by a shearing off

of the compression column, which is the type of failure expected .

Although this wall was extensively cracked, it was still able to resist
several blows with peak values larger than those endured by a similar
wall under static loading.
C. The theoretical analysis performed on WHIRLWIND I yields satis-

factory results until first cracking occurs, at a load of 40 kips.

This value is in good agreement with the experimental data. After
cracking, however; the lattice started to vibrate and the results are
not consistent with the experimental data. Satisfactory results are

given by the analysis of the IBM 704 computer up to a load of 23 kips,

after which the lattice begins to vibrate. The cause of this oscillation
seems to be a mathematical error in the programming.

- 81 -
 D. The theoretical and experimental results agree in showing an

almost complete lack of vibration. However, when the values for the

horizontal displacement of the loading point are compared, they are

found to be inconsistent.

VI. Recommendations for Future Research

This preliminary series of tests demonstrated most convincingly thtd

considerably more experimental and theoretical work has to be done in

order to reach a good understanding of the shear wall problem. It

is suggested that further theoretical and experimental work be done

along the following lines;

A. Introducing a new set of dump and servo valves in the system in

order to investigate the possibility of decreasing the rise time and
obtaining better control of the duration of pulse. Performing a series

of tests in order to find the appropriate gain and the appropriate oil

and nitrogen pressures to be used for different values of the peak load.

Investigating the possibility of introducing modifications in the system

with the aim of obtaining a better control of the pulse.

B. Investigating the possibility of improving the instrumentation

used for the,recording of the load, reactions, and displacements.

C. Starting an extensive series of tests in order to determine

the influence of the following variables on the behavior of the shear

wall:

1. Strength of concrete

2. Effect of frame

3. Amount of panel reinforcement

4. Effect of rise time

- 82 -
 5. Variation of load pulse
6. Length/Height ratio of wall

7. Scale effect

D. From the theoretical point of view, a careful review of the pro-

grams already written is suggested, in order to find the cause for the
oscillation of the system.

An investigation of the proper time interval for the numerical

integration, and the effect of the grid sizes, with the aim of reducing

the comiputation time to the minimum possible, while still obtaining

a good approximation.

- 83 -
 REFERENCES

1. K. Muto, D. W. Butler and B. Osawa, "Solution of Frames by the

Consideration of Bending and Shear in the Rigid Region", Trans-
actions of the Architectural Institute of Japan (Japanese) N-45
(Tokyo, December 1952) p. 53.
2. K. Muto and B. Osawa, "Approximate Method of Calculation of
Forces in a Reinforced Concrete Wall with Openings", Transactions
of the Architectural Institute of Japan (Japanese) N-45 (Tokyo,
December 1952) p. 58.
3. Y. Otsuki, "Stress Analysis of a Wall with Rectangular Holes",
Transactions of the Architectural Institute of Japan (English)
N-41 (Tokyo, August 1953) p. 72.
4. "Massachusetts Institute of Technology", Lecture Notes on 1.559
"Structural Design for Dynamic Loads", p. 19.2.
5. K. Muto, and K. Kuromasa, "Experimental Study of Vibration-
Resistant Reinforced Concrete Walls", Study of Strength and Con-
struction of Reinforced Concrete Frames with Walls (Japanese)
(Tokyo: Japanese Architectural Engineering Research Fund 1952.)
6. Dr. Tadashi Taniguchi, "The Study of Vibration-Resistant Walls",
Transactions of the Architectural Institute of Japan (Japanese)
N-41 (Tokyo, August 1953) p. 62.
7. Hansen, R. J., "Behaviour of Structural Elements under Impulsive
Loads", Massachusetts Institute of Technology, April 1950, p. 97.
8. Hansen, R. J., "Behaviour of Structural Elements under Impulsive
Loads, II", Massachusetts Institute of Technology, November 1950,
p. 83.
9. Hansen, R. J., "Behaviour of Structural Elements under Impulsive
Loads, III", Massachusetts Institute of Technology, July 1951,
p. 89.
10. Galletly, Gerard D., "Behaviour of Reinforced Concrete Shear Walls
under Static Load", Massachusetts Institute of Technology, August
1952.
11. Williams, H. A., Benjamin, J. R., "Investigation of Shear Walls,
Concrete and Brick-Walled Bents under Static Shear Loading",
Technical Report No. 1, Part 1, Department of Civil Engineering,
Stanford University, Stanford, California, April 1952.
12. Benjamin, J. R., "Investigations of Shear Walls, Part 2 -
Prediction of Behavior of Plain Concrete and Brick Walled Bents
under Static Shear Loading by Lattice Analogy", Technical Report
No. 1i, Part 2, Stanford University, April 1952.
13. Williams, H. A., Benjamin, J. R., "Investigations of Shear Walls,
Part 3 - Experimental and Mathematical Studies of the Behavior
of Plain and Reinforced Concrete Walled Bents under Static Shear
Loadings", Technical Report No. 1i, Part 3, Stanford University,
July 1, 1953.

- 84 -
14. Williams, H. A., Benjamin, J. R., "Investigations of Shear Walls,
Experimental and Mathematical Studies of the Behavior of Brick
Walled Bents under Static Shear Loading", Technical Report No. 2,
Part 4, Stanford University, August 1, 1953.
15. Williams, H. and Benjamin, J., "Investigation of Shear Walls,
Part 5, Prediction of the Behaviour of Plain Concrete, Reinforced
Concrete, and Brick-Walled Bents under Static Loadings", Stanford
University 1953.
16. Williams, H., and Benjamin, J., "Investigation of Shear Walls"
Part 6, Continued Experimental and Mathematical Studies of Rein-
forced Concrete Walled Bents under Static Shear Loading, Stanford
University, 1954.
17. Walter, J. K. et.al., "Investigation of Shear Walls", Part 7,
Continued Experimental and Mathematical Studies of the Behaviour
of Brick Walled Bents under Static Shear Loading, Stanford Univer-
sity, 1954.
18. Stivers, R., Benjamin, J. and Williams, H., "Investigation of
Shear Walls", Part 8, Stresses and Deflections in Reinforced
Concrete Shear Walls Containing Rectangular Openings, Stanford
University, August 1954.
19. Benjamin, J. and Williams, H., "Investigation of Shear Walls,"
Part 9, Continued Experimental and Mathematical Studies of Rein-
forced Concrete Walled Bents under Static Shear Loading, Stanford
University, September 1, 1955.
20. Benjamin, J., Williams, H., Erickson, R. and Campbell, R., "In-
vestigation of Shear Walls", Part 10, Limited Study of Brick and
Concrete Block Shear Walls with Steel Frame, Stanford University,
September 1956.
21. Finerman, Aaron, "The Theoretical Elastic Response of Shear Walls
Subjected to Dynamic Loads", Massachusetts Institute of Tech-
nology, August 1956.
22. Simpson, Howard, "A Dynamic Loading Machine", Massachusetts Insti-
tute of Technology, 1957.
23. Proceedings of the American Concrete Institute, Year 1955, p. 600.
24. Proceedings of the Conference on Building in the Atomic Age,
Massachusetts Institute of Technology, June 1952.
25. Newmark, Nathan M., "An Engineering Approach to Blast-Resistant
Design", Transactions of ASCE, Vol. 121, 1956, pp. 45-6h.
26. Guerin, A., "Traite' de Beton Arme" Tome I, p. 238, Tome II, p. 54.
27. Telemaco van Langendouck, "Calculo de Concreto Armado", Associacas
Brasileira de Cimento Portland, Vol. 1 (2nd Edition) Sao Paulo,
1954.
28. Brice, Louis Pierre, "Etude des Conditions de Formation des Fis-
sures de Glissement et de Decohesion dans les Solides", Travant,
June 1954, p. 475-506.

- 85 -
 -j

,j I
~II

I-

86
8LAST -
-Rear Wall

SA
I
PLAN

crack caused by
BLAST ion of horizontal
c ked roof.

Lateral View A-A

Figure 2 - Shear Wall Crack Caused by Redistribution of

Horizontal Load in Cracked Roof

j.:~
 N

FJ88
- 8w
 'U

I-

½' -.
0
 I

I>

Cr

10

IN CH
0cd
U)
a,
b4

0
to
4,
0

.r)

K-

K.LA_
 4Y
7-

Figure 7 - Support Arrangement for the Shear Wall - Cross Section

 A - Lateral Load from
Horizontal End Shear
of Front Wall
B - Vertical Load from
Vertical end Shear of Roof
C - Lateral Load from Horizon-
tal End Shear of Rear Wall
D - Foundation Resistance
Below Shear Wall Footing
D'- Ultimate Plastic Soil
ResistanCe
E - Friction on Shear Wall
Footing
F - Lateral Load from Horizon-
tal Shear in Floor Slab
G - Dead Load
H - Vertical Load from Verti-
cal Shear in Front Wall
J - Vertical Load from
Vertical Shear in Rear
Wall
K - Lateral Load from Horizon-
PLAN tal Shear of Roof

Secti.on at Shear Wall

Figure 8 - T.yical Loadings on Shear Wall Structure

 LuI
b*

"0
H

r-4

0
Cd
0)

00 r,

o II
tt

*r4
4,
-d
0O
0l 0
i-1

H
0

rI
0\

3JlynSS3dId
I
7ý a;
a,
o

Cd
L- O
0
O

4o
E-

0
I H
0

1l
,.H

4.,-

N -t Cd
0

0
co

o0
H
r-t

0)

fi)

c-,
. ý11
 *08(
c:

1I

N. N

-- ·
Figure 12 - F.ormwork and Reinforcement in Place for the
Shear Wall Specimen

7 97
Figure 13 - Walls and Cylinder, after being Cast

Figure l4 - Side View of Heavy Truss and Lateral

Foundation Beam

98
 O

Q) Od

.H
-• 0) 0
-tP O G 0

4-Q3
-p
q; liw10(1)
0 0
-)0 to-I -C,
a~ 6-
) -0) 0
+3 o
'o4
ed
O mdO_C m
.H U)
LD
0 c0
cHi
+3
0

4.•- i1
-o I

-P
O

43
0
E 0
O I

Sco
catJo
,oe,

-4

99,
 L

c a
V. >
-r IL
ICL
~P; 0 a

0P 35
0 0
w0 3 0 1 OCL
_143
E 0 0
C)

U) 0 U 1-
":I "'
w
Cz

00 o .,-
o oo
cc
= -a c €
E

;, -
E > 0) 0

" 041 t u •,C,o ,.

Qu,
- , •

~0 CL

LL j0

r•• ,• :0
i
I • I,_
.. (j
i,7_° ...
-u I II ,I -
3•,"
', I II.,.

,. I - Ico _

--
4-'-- .....
v0,1'91
_ _ -_-_, - _ I
0 13

r.4.

•,OO•
4, I~
Figure 19 - View of ýontroi ranel of the Function Generator

Figure 20 - View of the Servo and Dump Valves

reO3~
 Figure 21 - General View of Pump Setup

Figure 22 - View of Accumulators in Place

1'04 'a
 Figure 23 - General View of the Filter Unit

Figure 24 - View of Control Panel of the Pressurizing System

LOAD -P

T7ME

Figure 27 - Load-Time Function

O

I-C

418
toi

Ac Q siz
LL R
K
Figure 29 - General View of Instruments

Figure 32 - Calibration of Ran Load Cel 1

 X
0
Ck:
apt:
ki p Ct co

I II - 11I
r-7 k

i~LiJ

Lu

41U-r Lu

& (0

U)e
t-. Z
~------c
CL
C)
0u N 0
U,
'U

-----------
-------'
'-----'

"I

LL 4 1j
1-
CL
a.~
S
LUL"
-5ý0

N,
i,
izz
Lu
I,, (,j

& I

Ltj

LIJ
LU

(LU
,,•L

/ 1 f -1-40 (
 I

f-l I

JA

'~..-,
i
Y

Figure 31-(a)

H4 H2
V
H3

I
H4 SH3 V H2

I I
I I

N...~-'
Figure 31-(b)

Figure 31 - Gage arrangement and circuits for ram load cell

"140
EJO

EJ0

1111
Figure 34 - View of the Device used to Connect the
L.V.D.T. with the Specimen

Figure 35 - View of the Lateral Restraint System

i iý2
 I
01

,·
a
a
Q

--

LAJ

I'M~c
Figure 37 - View of Indicator of the Ram Head Position

Figure 38 - View of the Guide for the Ram

Id4
 O
LLi

Lu
-i

CI)

5t

Lu~
U)

__
// 3 7! Fi
---
Figure 40 - View of the Restraining Strut
Figure h2 - Arrangement of Gages for LMeasuring
Deflections Under Static Load

S DIAL - DEFLECTION
SES
AR VARIABLE DIFFERENTIAL
4NSFORMER

Figure h3 - Arrangement of Gages for i'easuring

Deflections Under Dynamic Load

. /
Figure 44 - Setup of Load Cells for Reactions

Figure 45 - A Typical Wall Ready for Testing

 LOAD
(Kips)

73.60

66.24

58.88

51.52

44.16

36.80

29.44

22.08

/4.72

7.36

.02 .04 06 .08 .10 .12

DEFLECTIONS (In.)

Figure h6 - Load-Deflection Curves for Wall #1 (Static Test)

;I
U)

7.36

.02 .04 .06 .08 .10 .12 .14 .16 .18 .20 .22 .24 .26
LATERAL DEFLECTION (in.)

Figure 47 - Lateral Deflection of the Top of the Wall

with Respect to the Supports (Wall #1 -
Static Test)
Figure h8 - STATIC TEST -
Final Condition of Wall #1

i21
 A 8

C
Figure 49 - Example of a Complete Record of a Test
(Test 6 on Wall #4)
A- Load and vertical deflection C-Vertic al reaction at support
at loading point C an d vert. def I. at support 8
B- Horizontal reaction and hori- D-Vertical reaction at support
zontal deflec. at loading pt. 8 and horiz. deflec. at sup-
port 8
L2O2
 24

22

2C

18

016

10 20 30 40 50
o TIME (ms.)

- Load
-- Lc*#3
-- Lc#2
........ Lc# I
Figure 50 - Observed Load and Reactions vs Time
for Test No. 2 on Wall #2

S23
.030

.028

.026

.024

.022

.02C

I
//
.016
I
iI II
.014 IIA
II
.012

.010

.008

.006
I/-__ ___
__ _ _ I
_ /__
__

.004

.00;

TIME (milliseconds)
- HORIZONTAL DEFLECTION AT
LOADING POINT
-- HORIZONTAL DEFLECTION AT
SUPPORTS

Figure 51 - Observed Deflections vs Time for Test No. 2

on Wall #2

-24
 32

30

26

24

22

20

18

16
L&j

14

12

-0

-2

TIME (ms.)

Figure 52 - Observed Load and Reactions vs Time for

Test No. 6 on Wall #2

.w.
 .080

.076

.0 72

.068

.064

.060

.056

.052
U,

.048

.044

o .040

LU .036

.032

.028

.024

.020

.016

.012

.008

.004

.000
0

TIME (milliseconds)
- HORIZONTAL DEFLECTION AT LOADING
-- HORIZONTAL DEFLECTION AT SUPPORT

Figure 53 - Observed Deflections vs Time for Teqt No. 6

on Wall #2 .
 55

50

45

40

35

15
00

10 20 30 40 50 60 70 80 90 100
o TIME (ms.)

Load
---
-- Reactlon No. 3
.......... Reaction No. I
Figure 5h - Observed Load and Reactions vs Time for Test No. 11
on Wall #2

-27
0.

0.

0.

0.

0.

0A

0.

0.(

0 10 20 30 40 50 60 70 80
TIME (milliseconds)

Figure 55 - Observed Deflections vs Time for Test No. 11 on Wall #2

-2S
60

55

50

45

40

30

25

20

15

10

10 20 30 40 50 60 70 80 90 I00

TIME (ms.)

--- Load
- Reaction No. 3
........ Reaction No. I
Figure 56 - Observed Loads and Reactions for Test No. 12 on
Wall #2

`29
 0.24

0.22

0.20
0-- -
0.18

0.16

2
- Of

0.14
-t--
0.12

0.10 o-- -

0.08

0.06

0.04
-- /
--I /-
-I-!I
\N
- a

/
I
0.02

0 I o 30
a
4
TlI s 07
04E (ms o 6

-o0.02

Horiz. deflection at support

.----
-- Horiz. deflection at loading point

Figure 57 -.Observed Deflections vs Time for Test

No 12 on Wall #2

/:23 0
t!r3O
 Figure 58 - Crack Pattern of Wall #2 After a
Dynamic Load with a Peak Value
of 31,5 kips was Applied

Figure 59 - Final Condition of Specimen #2

'I)

0~

TIME (ms.)

Figure 60 - Observed Load and Reactions vs Time for Test No. I

on Wall #3

-0 02
0.17

0.16
SVRTICof
-- V"RTICAL DEF ECTIOV AT ýUPPOAT
0.15
- RIZONTAL

0.14

0.13

0.12

0.11

0.10

0.09

0.08

0.07
r-IF--- -x - - -

0.06

0.05 --
1-.1---

0.04 I .
i-k
0.03

0.02

0.01
_ I --.
_
-
_

' -. - _ ' r .-.--- Ž

2*--'
S1 409 50 60 7 a

TIME (ms.)

Figure 61 - Observed Deflections vs Time for Test No. i on Wall #3

433
 15

I0

10 20 30 40
0
TIME (ms.)

Figure 62 - Observed Load and Reactions vs Time for Test No. 7

on Wall #3

4/34
 0.090

0.080

0.070

0.060

0.050

Laj
_l
L4 Q040

0.030

0.020

0.010

0 /0 20 30 40 50 60
TIME (ms.)
- HORIZONTAL DEFLECTION AT LOADING POINT
VERTICAL
---- VERTICAL DEFLECTION AT SUPPORT
--- HORIZONTAL

Figure 63 - Observed Deflections vs Time for Test No. 7 on

Wall #3 4 1 t
U)

"p.q I .1 1 b-
%ft
yj SI f-I-I ?--1-I--tI--l-II I I

(3
Lu
La

TIME (ms.)

Figure 6h - Observed Load and Reactions vs

Time for Test No. 9 on Wall #3
 _·_
0.16

0.15

0.14

0.13

0.12

0.11

0. 10

L 0.09

0.08

,-J 0.07

0.06

0.05

0.04

0.03

0.02

0.01

TIME (ms.)

- HORIZ. DEFLECTION AT LOADING POINT

---- VERTICAL t I o
. . ...... VERTICAL DEFLECTION AT SUPPORT
" " "
--- HORIZ.

Figure 65 - Observed Deflections vs Time for

Test No. 9 on Wall #3

r!37
 (I)

(V)

LW I I f I ! i Ii I
Q

h
5J
rr I· * I
0-

TIME (ms.)

Figure 66 - Observed Load and Reactions vs Time for Test

No. 13 on Wall #3
.20
I_

.19

.18

.17

.16

.15

.14

.13

.12

.11

.09

.08

.07

.06

.05

.04

.03

.02

.01

TIME (ms.)
Figure 67 - Observed Deflections vs Time for Test No. 13 on Wall #3
i3/9
 __
68

64

60

56

52

48

36

32

g 28

" 24

20

16

12

o TIME (ms.)
---- Load
---- Lc No. 3
--- LcNo. 2
....-. LcNo. I
Figure 68 - Observed Load and Reactions vs Time for
Test No. 15 on Wall #3
 0..

0.O

0.1

0.1

0.;

0.1

O.

0.

0.

0.

0.

0.

0.(

0(

0.4

0.4

- 0.4

- 0.(

-0.
- HORIZ. DEFLECTION AT LOADING POINT
--- VERTICAL " i t
. . ...... VERTICAL DEFLECTION AT SUPPORT
--.-- HORIZ.

Figure 69 - Observed Deflections vs Time for Test No. 1 r

on Wall #3.
141k
 LOAD
.0-

ye'
- Load
--- Lc No. 3
.......... Lc No. I
Figure 70 - Observed Load and Reactions vs
Time for Test No. 16 on Wall #3

2
14
 _ __
0.42

0.40

0.38

0.36

0.34

0.32

0.30

0.28

0.26

0.24

0.22

0.20

S0.18

0.16

L&J 0.14

S0.12

0.10

0.08

0.06

0.04

0.02

0 10 20 30 40 50
TIME (m s.) - HORIZ. DEFLEC. AT LOADING POINT
-----VERT. " I o
--- HORIZ. DEFLEC. AT SUPPORT

Figure 71 - Observed Deflections vs Time for

Test No. 16 on Wall #3

?1 43
 . _|-

rr

7" A

Figure 72 - Wall #3 After Failure

 (1

E-4
0

,o
OO

Cd
0

*r8

(X)
C)

L
0=
0)
rrx-
Fr

- Load
Lc No. 3
---- Lc No. 2
- I .......... Lc No. I

-2

-3

-4

45
 0.12

0.11

o.IQ

0.09

0.08

0.07

0.06

0.05

O
•, 0.04

0.03
LL&

0.02

0.01

- 0.01

- 0.02

- 0.03

- nn

Horizontal deflection at loading point

Vertical
Vertical deflection at support
Horizontal " "

Figure 74 - Observed Def lections vs time for

Test No. 2 o n Wall #4
 28

26

24

22

20

/B

16

'
L14

c 12

00
8

-2
TIME (ms.)

Figure 75 - Observed Load and Reactions vs Time for Test No. L on

Wall #4

.97?
1-47
 0.19

0.18

0.17

0.16

0./15

0.14

0.13

0.12

0.11

0.10

0.09

I 0.08

0.07

0.06

0.05

0.04

0.03

0.02

0.01

- 0.01

-n fl

Figure 76 - Observed Deflections vs Time for Test No. h on Wall #h

 __
40

38

36

34

32

30

28

26

24

22

20

18

16

12

10

- 2

- 4

Figure 77 - Observed Load and Reactions vs Time for Test No. 5

on Wall #4

i141'
 O.l9

0.18

0.17

0.16

0.15

0.14

0.13

0.12

0.11

O. 10

0.09

0.08

Lai
S0.06

0.05

0.04

0.03

0.02

0.01

TIME (ms.)
- HORIZ. DEFLECTION AT LOADING POINT
........ VERTICAL DEFLECTION AT LOADING POINT
HORIZ. DEFLECTION AT SUPPORT
--- VERTICAL DEFLECTION AT SUPPORT

Figure 78 - Observed Deflections vs Time for Test No. 5 on Wall #4

-"` 150
 50

47.5

45

42.5

40

375

35

32.5

30

25

22.5

L15
20

Q S17.5
15

IO

75

2.5

- 2.5

-5

Figure 79 - Observed Load and Reactions vs Time for Test No. 6 on

Wall #4
L-.

Ql
Lt

TIME (ms.)
- HORIZ. DEFLECTION AT LOADING POINT
........ VERTICAL It to ot
•-•- HORIZ. DEFLECTION AT SUPPORT
--- VERTICAL

Figure 80 - Observed Deflections vs Time for Test No. 6 on Wall-#4

°° .
 60

56

52

48

44

40

36

32

28

24

20

16

12

-4

-8

Figure 81 - Observed Load and Reactions vs Time for Test No. 7 on Wall #4

/ -3
All
0. 3
0.31

0.31,

0.3(

0.21

0.24

0.2,

0.2,

0.2(

O.11
O. IE
0.10

0.I'

0.11

0.1(

0.0ý
0.01

0.04

0.0'

10 20 30 40 50 60 80
TIME (ms.)
- HORIZ. DEFLECTION AT LOADING POINT
........ VERTICAL
-- HORIZ. DEFLECTION AT SUPPORT
---- VERTICAL

Figure 82 - Observed Deflections vs Time for Test No. 7

on Wall #4

-54
 __
elu

76

72

68

64

60

56

52

48

44

40

S36

Q 32

Q 28

-J 24

20

16

12

-4

-8

- 10

Figure 83 - Observed Load and Reactions vs Time for Test No. 9 on

W':all #h .
 - HORIZ. DEFLECTION AT LOADING POINT
........ VERTICAL t
--.- HORIZ. DEFLECTION AT SUPPORT
--- VERTICAL

O 10 20 30
TIME (ms.)

Figure 8. - Observed Deflections vs Time for Test No. 9

on Wall #h
-J'·f
Figure 85 - Crack Pattern of Specimen #4 After Failure
of the Connection Between the Reinforcement of
the Tension Column and the Support Plate

Figure 86 - Crack Pattern (After Failure) of the

Repaired Specimen #4
/i 7
 77 r1F

1 L

3
Figure 7(a) - Actual Wall

H L
1I
!

L---------

Figure 87(b) - Dynamic Model

 at3m
4Zk
&IZt
CL)
q)(
1
j
qas~L~

K
I I

S59
a,
a,
(o
0

-d
0

a,
 D)

0cd 0
00oH

o0

H
o)l O C
z 0)
0
OH,

Q: 0)
*H r-t -)
o0
O
* COS
(A- k

*Hjc

* ." /
 P (Kips)

53

t {ms.)
O0

Figure 91 Applied Load Curve

I lx\
/ / \
I
\/ /\LV \I/
i X\
\/
7 /V• I

Figure 92 Dynamic Model

""-7.
 0

.H

q) 04,
O

0
-Hl

0 tO
0 0 0 0
(001)) dPUO H3
(sd.I) A c A
 U)

('u!) 'X pUD 7X

6
;'16
 Vj

*...

(u!) 'A puo'

IGG•
 46

44

42

40

38

36

34

32

0 3C

224

I(

I
I1

o TIME (each interval =.0375 ms.)

Figure 97 - Horizontal Reaction vs Time

167
340

320

300

280

260

240

220

200

180

160

140

120

100

80

60

40

20

0
TIME (each interval = .0375 ms.)

Figure 98 - Horizontal Displacements vs Time

(I
 APPENDIX A

GENERAL PROGRAM AND CODING FOR

WHIRLWIND I

As it was established in Chapter VI, the writing of the general

program and coding for Whirlwind I was undertaken by Mr. Shui-Ho, and the

following description is a summary of his work.

NOMENCLATURE

In this appendix the following nomenclature is employed.

a = side of the square grid

A2 = transformed net area of bouding beam of column.
E = modulus of elasticity of concrete.
c

h time interval from tn- 1 to .tn, or tn to t n+, etc.

Kd = spring stiffness of diagonal bar.

Ke spring stiffness of exterior bar.

K. = spring stiffness of interior bar.

1
m mass of the particle considered.

M = mass of the mass point.

P = the component force of the mass point in the q direction.

Pn = value of the applied load on the mass point at time tn in that

direction, etc.

q = the independent displacement coordinate of the mass point.

qn = displacement at time tn .

qn+l =
displacement at time tn+ '
1
qn-1 =
displacement at time tn-l'

- 169 -
4n+1 velocity of the mass point at time tn+1 in that direction, etc.
qn+l = acceleration of the mass point at time tn+1 in that direction, etc.

R = the component of resisting force on the mass point in the q-q

q
direction generated by the displacement of all mass points.
R = value of developed resistance on the mass point at time t in
n+l n+l
that direction, etc.

t transformed wall thickness.

tn-1 tn, tn+ 1 three successive values of independent time variable.

I. LATTICE ANALOGY FOR ELASTIC BEHAVIOR OF REINFORCED CONCRETE SHEAR WALLS

A. Dynamic Analogy
According to the lattice analogy method, the wall is replaced

by a pin connected lattice (truss). In order for the lattice unit in

Figure A-l(b) to be equivalent to the plate unit in Figure A-l(a) the fol-

lowing conditions must be satisfiedi

Figure A-l(a) Figure A-l(b)

1) The deformations of the units must be the same under direct uniform
stress.

2) The deformation must be the same under the Poisson's ratio effect.

3) The deformation must be the same under a pure shear loading condition.
4) The deformation must be the same under a pure moment loading condition.

As the wall is subjected to dynamic loads it is necessary to replace the

equivalent lattice with massless springs and masses. A simplified model

- 17 -
was shown in Figure 87(b). The continuous distribution of the actual

system is approximated by a number of concentrated masses, and these

masses are coupled by interconnecting weightless springs. For a square
grid assuming that the Poisson's ratio equal to one-third and neglecting
the effect of pure moment, as shown in Mr. Finerman's report (21) the

spring stiffnesses can be found as:

Ki = 3/4-t. Ec (A-1)
,e
K (3/8 t +-
a
) Ec (A-2)
Kd = 3/8 t. Ec (A-3)

B. Equation of Motion

The equation of motion for each mass point can be developed from
the dynamic equilibrium as:

P + R - m(' (A-3)

If the x and y directions are chosen as the two independent coordinates

to specify the location of the mass point at any instant, the resisting

bar forces at the mass point can be derived as follows and according to
Figure A-2. +X
I0

tS
y Figure A-2 Typical Interior Mass.

- i 71 -
 = K1
x - x d] (A-4)

= K2 (A-5)
-K 3 (xi - xa) + (Yi - Ya)] (A-6)

K4
[(x - xf) + (Yf - Yl (A-7)
= K5
= K (x - xi ) + (Yl - Yc) (A-8)

[(xh -xi) + (h - Yi)] (A-9)

= K7
[Yi - Yb] (A-10)
= K8
[Yg -Yj] (A-11)

In the above equations, effective diagonal spring stiffness (equal to

half the actual stiffness) is implied and only components of force rather

than total force are considered. The two equations defining the horizon-
tal and vertical motion of the interior mass point are:

Px + Rx -m(c) = 0 (A-12)

P + R -m() = 0 (A-13)
Y Y
where

R = - Fl + F2 - F3 - F4 + + F6 (A-14)
R = - F + F - F + F4 - + F6
y 7 8 3 h

C. Numerical Integration
The open type formula used to predict the value in the succeed-

ing time interval was the Acceleration Impulse Equation

Qn+1 2qn - -1 + (h)

2" (A-16)
and the closed type formula used to correct the predicted value was the

Linear Acceleration Equation

- 172 -
 2

n+ - + 4(h) +F (2"'n n+1

) (A-17)
where if tn1 tn+1
tn, Sn+l three successive values of independent time
variable

= displacement at time t

displacement at time t
n
= displacement at time t
n-1

velocity at time tn+1

n+l = acceleration at time tn+1

h = time interval from tn+1 to tn, or tn to tn+l, etc.

n+l n n+l'
D. Principal Tensile Stress

The cracking stress of concrete varies between 0.06(fc) and

c
0.10 (fc) for statically applied loads. With f' equal to 3000 psi, a
criterion stress of 400 psi is adopted as the dynamic cracking stress.

The principal tensile stress at the segment center can be

obtained directly from the displacement as
0 +O - cr 2 1
0m x2 y 2Y
2X T
m 2 2 (-S
For a Poisson's ratio equal to one-third, the corresponding stresses can

be expressed as
3E
rx 8- (3Ex + C ) (A-19)
3E
yl -- (3 E + 6x )
(A-20)
3E
(A-21)

where acc ording to Figure A-3

-173--
 0~ rb
a b

C
r+y
Figure A-3. Typical Plate Segment

x 2a S(axb 1 - ) +
a) · (x-~
(xd X )(A-22)

y = 2a (Y Ya) (YdYb (A-23)

[(x
X -x) + (X- Xb) + (yb Ya

+ (Y - Yc)] (A-2h)

In general the first crack may occur at the center of any segment or in

the most highly stressed tension column. The stresses at the tensile column

can be found from the equation

E
)
St c (Yd - Yb (A-25)

II. LATTICE ANALOGY FOR CRACKED REINFORCED CONCRETE WALLS

According to what was established in Chapter VI, after crack-

ing has occurred, assuming 450 crack, all bars except the diagonal which

is parallel to the crack will be modified based on the reinforcing steel

of the wall unless the bar is subjected to compressive force. For the

diagonal which is parallel to the crack, bar area will be reduced only

if the tensile stress exceeds the allowable cracking value.

III. PLANNING FOR THE WHIRLWIND I PROGRAM

Before coding a program, it is essential to lay out the major

routines of the complete program and all the equations involved.

- 174 -
 The major operations of the program are as follows.
A). "Compute load P for use in acceleration impulse equation"
n

As the load curve may not be represented by formulas, it is

decided to store the value of load for each time interval in the computer
memories. Instead of computing the load from formulas, the computer will

pick up the proper value of the applied load from the memories at each
time interval.

B). "Predict the deflection qn+1 for each mass point from the accelera-

tion impulse equation"

2
qn+ = 2 qn - qn-1 + (h) "2

where
an
Pn Rn
n M M

The values of qn+l can be obtained directly from this equation.

For the loaded mass point, Pn can be picked up from the computer memories
while for all other mass points Pn will be equal to zero. At the first

time interval, use P 1 instead of P 0 to balance the error

introduced by assuming qn-1 = 0 in the equation. After computing the
value qn+l for each mass point in the x direction, repeat the operations and

compute the value qn+l for each mass point in the y direction.
C). "From the predicted values of n+l calculate the resistances Rn+1 for

each mass point"

The spring forces between the masses are calculated from the

deflections according equations (A-h) to (A-11) and the resistances Rn+ 1

for each mass point can be obtained from equations (A-14) and (A-15).

- 175 -
 The analysis involved is rather tedious after cracking has
occurred. For convenience in programming, the analysis is broken down

into several portions.

As discussed in Chapter VI after stress in center of a segment

exceeds the cracking stress of concrete, all bar areas of the lattice

segment except the diagonal which is parallel to the crack will be

reduced. It is not likely that compression failure will appear in the

shear wall; in order to simplify the analysis, straight line relationship
of force against elongation is assumed for each individual spring as shown
in Figure A.h where
ab - represents the force-elongation curve of the bar before cracking.

The bar area is based on the transformed area of the plate

segment.

bc - represents the state of cracking. When the stress at center of

segment exceeds the cracking stress of concrete, the bar area

is reduced and the bar force drops down from point b to c.

cd - represents the force-elongation curve of the cracked bar.

de - represents the state of plastic elongation of the cracked bar.

ef - represents the force-elongation curve of the cracked bar after

the plastic elongation.

fg - represents the force-shortening curve of the bar. When the gap

of the crack is less than zero (i.e. e = 0), the bar force is
in compression and the cracked bar will behave as an uncracked
element.

gh - represents a state of plastic shortening of the bar.

hi - represents the force-shortening curve of the bar after the

plastic shortening.

-. 176 -
 FORCE

LONGATION

Figure A.h - Typical Force-Elongation Curve I

Aflwtr

I
I
I
I
I
I b'
IA
ELONGATION

Figure A.5 - Typical Force-Elongation Curve II

- 177 -
 ij - represents the force-elongation curve of the cracked bar. The

bar force is in tension and the bar will behave as a crack

element again.

If plastic shortening has occurred before cracking starts, the force elon-

gation curves will have the same shape except that the origin "a" will
move to point fal" as shown in Figure A-5. For the diagonal which is

parallel to the crack, the cracking state (bc) in Figure A-h) will depend

on the tensile force at the bar instead of the stress at the center of

the segment.

ist Resistance Routine for Horizontal Lattice Bar

As shown in Figure A-6 the horizontal lattice bar between the

masses consists of two independent springs. The upper spring represents
the equivalent resistant force in the upper segment and the lower spring

represents the equivalent force in the lower segment.

For lattice bar at the edge, the outer spring will represent

the equivalent resistance force of the bounding beam. For each lattice
bar, the following storages are required from the computer memories.

a - storage to store the value of "kink" (compression) for the whole

lattice bar.

b - storage to store the value of "kink" (tension) for each spring.

c - storage to store the index for each spring indicating whether

the spring is in plastic or elastic condition.

d - storage to store the index for the edge spring indicating whether

the bounding beam is cracked or not. This storage is required

for the lattice bar at the edges only.

- 178 -
 In the following discussion tensile forces are taken as posi-

tive bar forces; subsequently the "kink" (tension) will be a positive

value. The operations involved in this portion of the routine are:

a) Assuming no crack has occurred compute bar force between the two
horizontal masses from the equation

F = KL xb - a - "Kink" (compression)

where KL is the spring stiffness of the whole lattice bar and

kink (compression) is a negative value.

b) If the bar force is in compression, compare this value with the

allowable compressive force for the bar:
1) If still remains in elastic condition, no correction is

required.
2) If newly plastic, reduce the bar force to the allowable value.

3) If already plastic since the last time interval, reduce the

bar force to the allowable value and compute the value of

"kink" (compression).

c) If the bar force is in tension, check each spring independently.

1) If no crack occurred in the corresponding segment (or bound-

ing beam) except for the edge spring, no correction is neces-

sary. For edge spring, compare the spring force with the

cracking tensile force for the bounding beam; no modification

if cracking has not taken place. Otherwise reduce the spring

area (stiffness) and recompute the spring force.

2) If crack has occurred in the corresponding segment (or bound-

ing beam), reduce the spring area (stiffness) and re-compute

the spring force. Compare this value with the allowable

tensile force for the modified spring.

i) if still remains in elastic condition, no forther correc-

- 179-
 tion is required.

ii) if newly plastic, reduce the spring force to the allow-

able value.
iii) if already plastic since the last time interval, reduce

the spring force to the allowable value and compute the

value of "kink" (tension) for the spring.

3) Correct the bar force with the modified spring forces.

d) A counting system set up for the above operations to compute force

at each horizontal lattice bar.

2nd Resistance R6utine for Vertical Lattice Bar

This portion of routine is about the same as for the horizontal

lattice bar except that the system concerned is in the vertical direction.

3rd Resistance Routine for Diagonal Lattice Bar

For each diagonal bar only one spring is implied and the

storages required from the computer memories are

a) Storage to store the value of "kink" (compression) for the bar.

b) Storage to store the value of "kink" (tension) for the bar.

c) Storage to store the index indicating whether the bar is in plastic

or elastic condition.

d) Storage to store the index indicating whether the bar is cracked

or not.

The following operations are involved in this portion of the routine:

a) Assuming no crack has occurred compute the diagonal bar force from
the equation (see Figure A-6)

- 1 80- -
 U-

Figure A.6 - Typical Interior Lattice Segment

Figure A.6 - Typical Exterior Lattice Segment

- 181 -
 F [xd - Xa) (y -a) -"kink" (compression)

b) If the bar force is in compression, compare this value with the

allowable compressive force of the bar.

i) If still remains in elastic condition, no correction is

required.

ii) If newly plastic, reduct the bar force to the allowable value.

iii) If already plastic since the last time interval, reduce the

bar force to the allowable value and compute the value of

"kink" (compression).

c) If the bar force is in tension and no crack has occurred in the

segment, no correction isrequired.

d) If the bar force is in tension and crack has occurred in the seg-

ment, assuming that the direction of the crack is at 45 degrees:

i) If the crack is parallel to this diagonal bar, compare the

bar force with the cracking tensile force for the bar.

(1) If the bar force is greater than the cracking tensile

force, no correction is required.

(2) If the bar force is greater than the cracking tensile

force, the diagonal will crack and only the reinforc-

ing steel contributes resisting force to the system.

Reduce the bar area (spring stiffness) and re-compute

the bar force.

ii) If the crack is perpendicular to this diagonal bar, reduce

the bar area as the diagonal is cracked. Re-compute the bar

force and compare this value with the allowable tensile

force for the cracked bar:

- 182--
 (1) If still remains in elastic condition, no further

correction is required.

(2) If newly plastic, reduce the bar force to the allowable

value.

(3) If already plastic since the last time interval, reduce

the bar force to the allowable value and compute the
value of "kink" (tension) for the bar.

e) A counting system setup for the above operations to all diagonal

lattice bars.

D). "Compute load Pn+l for use in linear acceleration equations."

In this special case instead of computing the load from formulas,

the computor will pick up the proper value of the applied load from the

memories at time interval tn+l .

n+l -and Pn+l, calculate the corrected values

E). "With the values of R

qn+1 for each mass point from the linear acceleration equation."
2
qn+l = n + b (2
+ +" +l)

where 4n l + h/2 +n-

+)
1

P R
n + n and
Pn+1nl Rn+1
qn+1 M M
F). "Calculate the new values of resistance Rn+ 1 for each mass point from

the corrected values , and compute the corrected values qn+l

again from the linear acceleration equation. Repeat this operation
for three times."

G). "To store the values qn+l' +1'' n+l for use in the next time interval."

- 183--
H). "Determine the stresses at the center of all segments: Compare these

values with the cracking stress of concrete. If cracking has taken

place compute direction of the crack."

The maximum principal stress at the segment is given by Eq. (A-18).

If cracking has taken place, compute and store the direction of

the crack. The direction of crack is

1 t -1 xy where + 90
2 an 0o- 0
y x

and counterclockwise is considered as positive direction.

I). "Compute the stresses at the tensile column and compare these values
with the cracking stress of concrete".

This is obtained from Eq. (A-25)

J). 'With the values of resistance R+ 1 and load P predict the deflec-
tion qn+2 for each mass point from the acceleration impulse equation."

i.e., start a new cycle of the above operations for the next

time interval.

K). "Repeat the above operations until computer stops at certain specified

deflection or after the dynamic load is over."

L). "Print out results"

1) Time

2) Applied Load

3) Horizontal support reaction

4) Vertical support reaction
5) Deflection in the x direction for each mass point
6) Deflection in the y direction for each mass point

7) The principal stresses at the center of the plate segments.

- 184-.
The direction and location of the cracks are also printed out at the last
time interval to give a general picture of the cracked wall.

IV. CODING FOR THE WHIRLWIND I PROGRAM

A. General
Two systems of instructions are involved in the present high
speed digital computer, Whirlwind I.
1) Whirlwind System. This system consists of the fundamental
instructions to the computer. It is generally used in the counting system
and used as index (e.g. indicating crack or plastic condition, etc.)

2) Comprehensive System. This system consists of the approved

instructions for the computer. It is generally used as the arithmetical

instruction in programming.

B. Storage Arrangement

The storage system of this program is shown in Figure A-7 and

A-8). Contents of the storage are:

Comprehensive System
Registers xl Horizontal displacements x
yl Vertical displacements y
R
rl Horizontal resistances _x
R
r2 Vertical resistances -Z

x2 Horizontal displacements at previous time

y2 Vertical displacements at previous time
zl Horizontal and vertical acceleration i and y

z2 Horizontal and vertical velocities times h,

i(h) and *(h)
 Principal tensile stresses
Direction of cracks

Horizontal bar forces

Vertical bar forces

Diagonal bar forces (8 = -45*)

Diagonal bar forces (.6 = +46')

al "kink" (tension) for lower springs

"kink" (tension) for upper springs

"kink"t (compression) for horizontal bars
"kink" (tension) for right springs

"kink" (tension) for left springs

"11kink" (compression) for vertical bars

Registers dl "kink" (tension) for diagonals (6 = -450 )

d2 "kink" (tension) for diagonals (E= +45' )

d3 )
"kink" (compression) for diagonals (8 = -645

d4 "kink" (compression) for diagonals (e = +450 )

q3 Applied loads

Whirlwind System
Registers e4 Crack indexes for segments and bounding beams
a4 Plastic indexes for lower springs

a5 Plastic indexes for upper springs

b4 Plastic indexes for right springs

b5 Plastic indexes for left springs

d5 Plastic indexes for diagonals ( 8 = -45*)

d6 Plastic indexes for diagonals ( e = +45)

- 186-
O 2 4 0 2
24 26 28 30

l I'l A 10 12 14

1]2 22
32 34 36 38

18 20 22
rl, r2, x1,x2, yl, y2 40 42 zl,z2 44 46

6 8 /0

12 14 16
e2, e3 e5, al, a2, a3

4 8 12

6 /0 14

e6, bl, b2, b3 e7, dl, d3

0 2 41 M
CCj
1"c
It
6 B 0/0

d2, d4
n-2 =I h columns

Figure A.7 - Details of the Storage Arrangement. Comprehensive System

(Each Number Occupies Two Registers)

- 187 -
0 I 3 4
6 7 8 9
1L3 4 5
t0 II 12 13 14
15 16 17 18 19
6 7 8
e4 ah, a5

2 4 6

3 5 7

bh, b5 d6, d8

\2
, d7

3 K4 5

d~, d7

Figure A.8 - Details of the Storage Arrangement. Whirlwind Sytem:

(Each Number Occupies One Register)

- 188 -
 d7 Crack indexes for diagonals (0 = -45c)
d8 Crack indexes for diagonals ( 8 = +459 )

C. Field Arrangement
The computer core memory consists of six fields each of 1024

registers. The Whirlwind I computer is designed to operate with a full

complement of 2048 registers of storage and combination of any two

fields can be engaged at a time. In addition, huge memory called Aux-

iliary Drum is also available as extra storage.
In this program, all instructions and data are first stored in

the Auxiliary Drum and will be distributed into six fields in the follow-
ing order:

1) Group A (registers 0 - 1024):

Field 0 The main program except the resistance routine.

Field 2 Resistance routine, horizontal bar force computation.

Field 3 Resistance routine, vertical bar force computation.

Field 4 Resistance routine, diagonal bar force computation.

Field 5 Resistance routine, resistance computation for mass

points.

2) Group B (registers 1024 - 2048):

Field 1 Constant and storage.

D. Overall Program (see Figure A-9)

Step A

Load Computation. Pick up Pn for use in acceleration impulse.

Proceed to Step B.

- 189- .
 START

STOP

Figure A.9 - Block Diagram of Overall Program

- 190 -
Step B
Integration Routine. Predict qn+l for each mass point from the
acceleration impulse equation. Proceed to Step C.

step C
Resistance Routine. Calculate Rn+1 from the predicted values

qn+l for each mass point. Proceed to Step D.

Load Computation. Pick up Pn+l for use in linear acceleration

equations. Proceed to .Step E.
Step E

Integration Routine. Calculate the corrected values qn+l for

each mass point from the linear acceleration equation. Proceed to Step
C to recorrect qn+l" After performing this complete operation four times

proceed to Step F.

Load Computation. Pick up Pn+l for use in acceleration impulse

equations at next time interval. Proceed to step H. At every fourth time

interval, before proceeding to Step H, enter Step G.

Print Out Subroutine. Print out the results of this time inter-
val on the oscilloscope. Return to Step F.

StepH
Tensile Stress Computation. Compute the stresses at center of

all segments. If cracking has taken place compute direction of the crack.
Proceed to Step I.

- ,191- -
Step I

Stopping Device. After the loading is over or if the wall fails,

stop the computer and print out results. Otherwise, proceed to Step B.

E. Explanation of the Program

1) Load Computation
P
ml Compute - for next time interval. Print out results and stop the

computer at a specified time interval, if the loading is over.

m2 Compute
4 P
m3 Compute P ( = -).
1 Used in the first time interval only.

2) Integration Routine

a) Acceleration Impuls e Equation

nl Compute qn+l from the equation.

n3 A simplified routine for the first time interval to skip the un-

necessary operations.

nh Counter for the mass point, check column.

n5 Reset for next column and next row.

n6 Counter for the mass point, check row.

n8 Reset support deflections = 0.

n7 Change field and proceed to the resistance routine.

b) Linear Acceleration Equation
n2 Compute qn+l from the equation.

n1l Counter for the number of iterations (recorrection of qn+l ) .

nll Counter for the mass point, check column.

n12 Reset for next column and row.

nl3 Counter for the mass point, check row.

-192 -
nl Reset n2.
n16 Obtain resistance from the resistance routine.

n17 Reset certain instructions for the fourth iteration.

n18 Used in the fourth iteration only. Store n+1 (h) and "+1I
n19 Used in the fourth iteration only. Additional instructions for

n12 when resetting n18 for next mass point.

n20 After four iterations, reset the whole operation.

3) Resistance Routine
a) Horizontal Bar

fl Subroutine consists of all operations for a typical spring.

f2 Subroutine of checking tensile stress for the edge spring.

f3 Subroutine of checking compressive stress for the whole bar.

a8 Compute bar force.

a9 Counter for the bar, check column and row.

alO Reset for next column.

all Reset for next interior row.

a12 Reset for the bottom row.

al3 Reset the whole operations. Change field and proceed to compute
forces for the vertical bars.

b) Vertical Bar

f4 Subroutine consists of all operations for a typical spring.

f5 Subroutine of checking tensile stress for the edge spring.
f6 Subroutine of checking compressive stress for the whole bar.

b8 Compute bar force.

b9 Counter for the bar, check column and row.

b10O Reset for next row.

- 193 --
bll Reset for next interior column.

b12 Reset for the right column

bl3 Reset the whole operation. Change field and proceed to compute forces

for the diagonal bars.

c) Diagonal Bar

f7 Subroutine consists of all operations for a typical bar.

f8 Subroutine of checking tensile stress for the bar.

f9 Subroutine of checking compressive stress for the bar.

dll Compute bar force.

dl2 Counter for the bar, check column and row. After completing

computation for diagonals in one direction, reset the whole operation

for diagonals in other direction.

dl3 Reset for next column.

dl4 Reset for next row.

dl5 Reset the whole operation. Change field and proceed to compute
resistances.

d) Resistance Computation
R R
k2 Compute x and - at left upper corner.
M M
k3 Compute __ and a for interior mass point at upper edge.
M M
kh Counter for mass point at upper edge.
R
k5 Compute Rx and at right upper corner.
M M
k6 Reset k3 for next mass point.

k7 Counter for the mass point, check row.

R R
k8 Compute - and - for mass point at left column.

uM M

- 194--
k10 Counter for the typical interior mass point, check column.

kll Reset k9 for next mass point.

R R
k12 Compute - and - for mass point at right column.
iF M
k13 Counter for the mass point, check and reset for next row.
R R
klh Compute x and Y at left lower corner.
R R
kl5 Compute X and for interior mass point at lower edge.

k16 Counter for mass point at lower edge.

R R
k17 Compute x and- at right lower corner.

kl8 Reset k16 for next mass point.

kl9 Reset the whole operation. Change field and leave the resistance

routine.

4) Tensile Stress Computation

gl Compute principal tensile stress.

g2 LSR FU8 RAPID SQUARE ROOT SUBROUTINE.

g3 Counter for the segment, check column.

gh Reset for next column.

g5 Counter for the-segment, check row.

g6 Reset for next row.

g8 Compute direction of cracks.

g9 LSR FU7 ARC TANGENT SUBROUTINE.

hl Reset the whole operation. Check time interval, proceed to the

print out routine at every fourth time interval.

S) Print Out Routine

h2 Compute support reactions.

- 195--
h3 Reset the whole routine.
pl LSR OS4 SCOPE MRA DECIMAL FOMAT SUBROUTINE.

p2 Counter for number of values to be printed.

P3 Print out one value.

ph Reset for next value.

P5 Set up the oscilloscope

6) Stopping Device

ml Stop the computer at the specified time interval when loading is

over.

glO Stop the computer when deflection exceeds the specified value.

F. Preparation of Tapes

The program is introduced to the machine by means of punched

tapes. For the details of the tape the reader is referred to the report

of Mr. Shui Ho, "The Theoretical Analysis for the Dynamic Behavior of

Shear Walls", Department of Civil and Sanitary Engineering of the Massa-

chusetts Institute of Technology, May 20, 1957.

- 196--
V. EQUIVALENT WALL PROPERTIES FOR THE DYNAMIC MODEL

Wall size: 66" x 44"

Grid size: 22" x 22"

Areas: Top edge and columns = 32.5 in2

Bottom edge = 80.25 in2

Columns: Interior segment = 22 x 22 x 2 = 968 in3

Exterior segment = 1/2 x 968 + 22 x 32.5 = 1199 in3

Bottom segment = 1/2 x 968 + 22 x 80.25 = 2249.5 in3

Bottom corner = 1/4 x 968 + 22 x 1/2 (32.5 + 80.25) 1482.25 in3

Upper corner = 1/4 x 968 + 22 x 32.5 = 957 in3

Masses: Interior segment = 968 x .150 x 1

12 x 12 x 12 32.2 x 12

.2174632 x 10-3 k-sec 2 /in

0=

Exterior segment .150 1

12 x 12 x 12 32.2 x 12

= 0.2693578 x 10- 3 k-sec 2 /in

Bottom segment = 2249.5 x 12 x.150

12 x 12
X150 1
1
32.2 x 12

= 0.5053548 x 10- 3 k-sec 2 /in

Bottom corner = 1482.25 x 12 x.150

12 x 12 32.2 x 12

= 0.3329905 x 10 - 3 k-sec 2 /in

Upper corner S957x.15o

12 12 x 12 x 32.2 x1 12

= 0.2149920 x 10-3 k-sec 2 /in

Transformed thickness of wall: #2 at 9.8" c-c

As 0.05 (one direction) 0.00255102%
p = -8
bd 9.8x2

- 197--
E = 3000 k.s.i.

n = 10
tw = t 1 + (n-1) Pw = 2 1 + 9 x 0.00255102 = 2.0459184 in

Transformed thickness of edge: h#h

h x 0.2 =
pW = 4 0.02133333%

te = 7.5 (1 + 9 x 0.02133333) = 8.939999775 in

Transformed thickness of Bottom edge: 4#4

x 0.2=0.008
4xO.2
Pw = 12 x 7.5 0.008
tb = 7.5 (1 + 9 x 0.008) = 8.099999993 in

Bar areas:

A2 (edge) = 8.939999775 x 5 - 2.0459184 x 2.5 = 39.585203 in2

A2 ' (bottom edge) = 8.099999993 x 12 - 2.0459184 x 4.875 = 87.2261477 in2

Al (interior) = 3/8 x a x tw
Interior bar = 2A1 = 3/4 x a x tw

Exterior bar = A1 + A2 = 3/8 x a x tw + A2

Bottom bar = Al + A2'
2 3/8 xax tw +A2'
Diagonal bar = Al 2 = 3/8 x a x tw x 2
F = AE
-
Stiffnesses (concrete): E
e L
Interior spring = 3/4 tw Be = 3/h x 2.0459184 x 3000 = 4601.316 k/in
Exterior spring = (3/8 tw + A2 /a) B = (3/8 x 2.0459184 + 39.585203 x

1/22) x 3000 = 7699.6404 k/in

Bottom spring = (3/8 tw + A2'/a) Bc

= (3/8 x 2.0459184 + 87.2261477 x 1/22) x 3000

= 14196.9 k/in
Diagonal spring = 3/8 tw Ec = 3/8 x 2.0459184 x 3000 = 2300.2582 k/in

- 19 0-1.
Effective stiffnesses (concrete):

Interior spring = 4601.3160 k/in

Exterior spring = 7699.6406 k/in
Bottom spring = 14196.1329 k/in

Diagonal spring = 1150.6291 k/in

Stiffnesses (steel):
= 1090.9091 k/in
Edge spring 22 x 4 x 0.2 x 30000
-=

Typical interior spring = 1/2 (0.05 x )x 153.0612 k/in

Effective stiffnesses (steel):

Edge spring = 1090.9091 k/in

Typical interior spring = 76.5306 k/in

Bottom spring = 1090.9091 k/in

Diagonal spring = 76.5306 k/in

= 4 k.s.i.
Allowable compressive forces (concrete(: fc' -- 000 p.s.i.

Interior bar = 4 (2 x 3/8 x a x tw)

= 3 x 22 x 2.0559184

= 135.0306144 k
Exterior bar = 67.5153072 + 39.585203 x L
= 225.8561192 k

Bottom bar = 67.5153072 + 87.2261477 x 4

= 416.419898 k

Effective diagonal bar = 1/2 x 135.0306144

= 67.5153072 k

= 0.4 k.s.i.
Allowable tensile forces (concrete): ft'

Top edge and columns = 39.585203 x 0.4

= 15.8340812 k

- 199--
Bottom edge - 87.2261477 x 0.4

= 34.8904591 k
Effective diagonal bar = (3/8 x a x tw) x 0.4
= 6.7515307 k
= 39-3 x 1.33 = 52.229 k.s.i.
Allowable tensile forces (steel): Panel - f'
Y
Frame - f' = 47 x 1.33 = 62.51 k.s.i.
y
Edge spring = 4 x 0.2 x 62.51
= 5o.008 k
Bottom spring = 50.008 k

Typical interior spring = 1/2 (0.05 x .) x 52.229

= 2.9312194 k
= 2.9312194 k
Effective diagonal spring
 APPENDIX B

GENERAL PROGRAM AND CODING FOR IEB 704

The general programming and coding for IBM 704 was undertaken

by Mr. Christopher Calladine and the following description is a summary

of his work.

In its general structure, the program is similar to that

described in Appendix A, and this discussion will only include points

which may be considered as modification of the program described

previously.
NOMENCLATURE

In this appendix the following nomenclature is used.

= /2 o thne slae o0 tne square grid.

= side bar area per element = 3/4 at.

= diagonal bar area per element = 3diagonal

per
bar
area
---- at.
element
at.
AE = edge bar area = Ab + (n-l) Pb

E = Ec [1 + (n-l) p is the transformed modulus of elasticity of

the bars.

m = s/Ic = Density of Steel/Density of Concrete.

n Es
E
= Young Modulus for Steel/ Y.M. for Concrete.
c

p = proportional area of steel in wall each way.

t = wall thickness

Ab' Pb, As, b, d are defined in Figure B-1.

MN Mass of Central Element = 10c 4 a2 t [1 + 2p(m-l)]
MB = Mass of Edge Element = [0
c [Ab + at] 2a + (m-.1) 2aY
[AbPb + at . 2p

- 201 -
 bAS

Figure B-1

S- (b1 d 1 + b2r hdat

M = Mass of Corner Element = + b2)a
d2 - -

d2at dld2t 2
2
-- + (A+ As2) a (m-

I. COMPUTATION OF THE ACCELERATIONS OF THE MASS-POINTS

A. Uncracked Wall

In order to compute the resistance R for each mass point

from the predicted values of qn+1 (see item II, part C of Appendix A)
the operator technique was used. This technique basically consists in
the following:

Consider the system shown in Figure B-2, whose members are

unstrained and whose mass point may be considered to be undeflected.

If the eight external mass-points are held fixed and the

central point is moved a small distance in the x direction, the following

forces will act upon the fixed mass points.

Point 1 : x9 /ha Ad E in direction 1-9

3: x9/ha Ad E in direction 9-3

4 : x9 /a Al E in direction 9-4

5: x9 /ha Ad E in direction 9-5

- 209 -
 9--
I
I
.@ y

8
X X x

2a

7
X
m M
X b '1*

Figure B-2

Ad E in direction 7-9
xg9 /a
x 9 /a
Al E in direction 8-9

A E+ 9 - in direction 9-8
fJa1 Ad E

Substituting for Al and Ad the values given in the nomenclature, we

have the following components of force.

Point Horizontal Force Vertical Fo:rce

1 + 3/16 x Et 9 3/16 x 9 Et - - X9 B
3 3/16 x Et = + x9 + 3/16 x Et SX9+
4 3/4 x9 Et = + 2x a
3/1 6 X9 Et = + x 9 - 3/16 x 9 Et S- x9
5
3/16 x 9 Et = + x9 + 3/16 x9 Et
7
8 3/h x 9 Et = + 2x9 a

9 -9/h x9 Et = x9
where a "3/8 Et, PB 3/16 Et, (a+ P)
X= .- - 9/16 Et.

- 203 -
A similar set of reactions can be found for a displacement of point (9)
in the y-direction.
The whole wall system may now be analyzed by superimposing

solutions of this type for each mass point. If a means of storing

accumulated totals of forces is available, then the resulting forces

at each mass point can be built up by the use of the above operation

for each mass point.

At mass points on the boundaries of the grid system the effects

will necessarily be modified. The constants concerned may in this case

be evaluated by the use of a method similar to that described above.

Figure B-3 and B-4 show the complete operations for displace-
ments in the x and y directions. The forces computed are divided by
the corresponding masses to give the desired accelerations.

In the program all the operator constants are divided by the

mass of the central element MN, thus giving the accelerations directly.

Edge accelerations are found by multiplying the computed values by

MN
the mass of the cor-
responding element

B. Cracked Wall
0 crack across an element has
The effect of a 45 already been
discussed in Chapter VI and Appendix A.

In the computation of accelerations from deflections of the

mass points, the system is first analyzed as if there were no cracks, and

the solution is then modified by replacing the forces as originally

computed in the bars by the forces resulting from an analysis of the bars
that are assumed to be cracked.

A flow diagram of the subroutine which deals with cracked bars

- 204 -
S=O +d 6 AEE

Figure B-3 Operators for the Contributions to Forces in

the x and y directions resulting from unit deflec-
tion in the x direction. The upper members in the
square refer to x-multipliers, the lower, -o y.
- 205 -
Figure B-h Operators for the contributions to forces in the x and
y-directions resulting from unit deflection in the y
direction. The upper members in the squares refer to
x-multipliers, the lower to y.

- 206 -
 d =bar
dp =plo,
dy =defi
de = d-
p = elo1
del

(contraction) (extension)

(no plastic
set)

No (some
plasticprevious
set)

(no more
yield: bar yielding)
elastic)

yield)

COMPUTE
i --crs x dp

ADD TO COMPUTED
ELASTIC FORCE

OUT

Figure B-5 Flow diagram for the programme which computes the
difference in force due to cracking
- 207 -
is shown in Figure B-5 and the following is a brief explanation of the

different possibilities that it was necessary to consider for a given

bar deflection (6).

1. 6 is Negative. In this case, we have contraction, and the following

two cases must be considered:

a) If the plastic set 6 p was negative just before this 6 was

considered, it means that no previous tension plastic set occurred

and we have the situation given in Figure B-6

.vigure Bt-o

b) If (p was positive, we have the situation illustrated in

Figure B-7 by point F.

From this figure it can be seen that if we had computed the

force assuming the bar was elastic, the value obtained would be that

shown by point P. However, the actual value is given by Pe and in order

to obtain this it is necessary to add to the elastic force P the value

PPe which is equal to OP'. This value is directly as x p

- 208 -
 F~IC(S

EFLECTION

Figure B-7

2. c is Positive. In this case we have extension and depending on

whether[l 5el - dy is positive or negative, the following two
cases must be considered:

a) If [el - y] is negative we have the case illustrated in

Figure B-8, where we can see that the exact value of the force

is P . The value computed assuming the bar is uncracked was P.

Therefore, in order to obtain the exact value Pe e x as, it
isneessary
value
toadd
tothecomputed
Pthe
value
is necessary to add to the computed value P the value [c on - P
I.

Figure B-8

- 209--
b)If[ ce - y] ispositive we have to consider two new
possibilities, depending on whether ce is positive or negative.
i) If 6( is negative, we have the situation shown in Figure

B-9 where in order to obtain the exact value of the force

Pe = (-a
s
x 6 y ) it is necessary to add to the computed

elastic value P the value [-a y -P]

In order to obtain the new plastic set 6c it is necessary

to subtract from the former 86 the value (16de - C6y)

Id I- c3

Figure B-9

ii) If 6e is positive we have the situation shown in Figure

B-10 and the exact value Pe may be obtained by adding to

the computed force P the value ( y x a s - P) and in order

to obtain the value of the plastic set for the next step

- 210 -
 it is necessary to add to cp the value [IeI - Sy]

Figure B-10

II. THE ITERATIVE SOLUTION OF THE EQUATIONS OF MOTION OF THE POINT MASSES

As already explained in Appendix A, three equations are used.'

Acceleration Impulse Equation.

q+i = 2% - %- + hi n (B-1)
Linear Acceleration Equation.
2

= +'qn+1) (B-2)
qn+l n n+ h + (2 B

Velocity Equation.

S n=-1 + h/2 1 "n)

+% (B-3)

First, with "inq

nQn_ 4n-i
n n , known, n is computed from (B-3).
Equation (B-1) is then used to obtain an estimate of qn+1l This estimate

- 21- -
is then used to compute the accelerations qn+l by use of the processes

described above. Substitution of this value, and the value of in in(2)

gives a second estimate of qn+l. Again iqn+ is computed, and a third

estimate of *'n+1 is obtained. These iterations are continued until

an adequate convergence is obtained. First trial was made with 3

iterations. Figure B-11 is a "Flow Diagram" showing how the equations

are used.

Compute
h/2 "-1n+ + n- n
(B-3)

Compute

2qn 1 n-1 + h2
n n+l
(B-i)

Compute
h2
3T 4 +4n h +qn
h2 ..
%+L 6 %+i =b

AZ I
Jrom Deflections
qn+1 compute acceler-
3 times
ations q'n+l

Compute
b + h2 /6 n+l =n+ out

Figure B-11

- 212 -
III. PROGRAM FOR THE IBM 704 BASED ON SCHEMATIC FLOW DIAGRAM
A. Key to the Overall Flow Diagr

The following is a description necessary in order to interpret

the flow diagram of Figure B-i.
Wall : Starts program off. Reserves space for final Stop, and for

Dump Stop.
Num 1 : Reserves all storage space used: the size of the matrices being

determined by the input data.

Orig.: Fills in with -0 the blocks which will contain the amounts of

plastic yield in the bars.

Fill : Sets up Reference Grid.

CAl : Computes all constants subsequently used - bar stiffnesses,

element masses, etc. Points out column headings and results at

zero time. Applies equation 2 for the top left-hand corner for

the first time interval.

P03 : Sets counter for r iterations: i.e. put CHECK = r + 1

Rout : Computes n+l from qn, assuming the wall uncracked and of uni-

form section (i.e. no heavier edge beams). Sets corner accelera-

tions = 0 at points of rigid support.

CF10 : Checks the indicators for cracking of the wall anywhere. If

there is any cracking, control is transferred to CRAX; if not,

control is transferred to BOUND.

CF1 : Checks for the last iteration of the current time interval. If
the iterative procedure is over, control is transferred to CF2.

Otherwise control is transferred to BOUND.

CF2 : Stores the corner accelerations before they are set to zero by ROUT

Crax : Modifies 'n+l wherever there is cracking.

- 213--
Bound : Modifies qn+l to take account of the additional mass of the

edge beams. Applies equation 2: i.e. makes estimates of qn+1'

Stres : Computes the principal tensile stress in each panel of the wall
and in the edge beams. If the breaking stress is reached, a

record of the time is made in appropriate registers. For wall

cracks, the angle of the crack in each segment is also recorded.

Mast : Master control. Sets the force for the next time interval.

Increases TIM, the time-interval counter, by 1. If the end of

the computation is reached control is transferred to PRO.

PRO : Prints out the sequence of cracking and the angles of the wall

cracks.

OUTP : Collects required deflections, and assembles together with the

corner reactions for printing. Prints out the results at certain

predetermined times.

PI : Applies equations 1 and 3, and finds three terms of equation 2

(See BOUND)

IV. CONTROL REGISTERS

CHECK Counts the number of iterations per time interval.

CATCH Current results are printed out when CATCH = 1. Any new crack

puts CATCH = 1.

CUT Results are printed out when CUT = 0. Each successive time in-

terval CUT is reduced by 1. CUT is set to S-1 after a "CUT = 0"

point and to q-1 after a "CATCH = 1" point.

TEST +1 or -1. When positive, the routines deal with forces, velo-
cities and accelerations in the x-direction; when negative, in

the y-direction.

SPLIT Originally set to O, it is set to +1 on the occurrance of any

crack, and x acts as an overall cracking indicator.

TIM Counts the time-intervals. Originally 0, it contains +1 during

the computations of the first time interval, +2 during the second, etc.
TCO +1 or -1. When TIM is odd, TCO is +1

When TIM is even, TCO is -1.

As qn during one time-interval becomes qn-1 during the next, etc,

it is necessary to have an indicator to decide what information

is to be used in the application of each equation. This control is

achieved by using TCO.

Parameters r = number of iterations per interval

s = number of time intervals elapsing between normal points.

q = number of time intervals after the occurrance of a crack

at which a print is made.

- 215--