STATE-OF-THE-ART REPORT

A Review of Strand Development

Length for Pretensioned
Concrete Members
In October 1988, the Federal Highway Administration
(FHWA) issued a memorandum that placed restrictions on
the use of seven-wire strands for pretensioned concrete
members in highway bridge applications. As a result of this
memorandum, research projects on strand transfer and
development lengths were initiated at many institutions
across the United States and Canada, often resulting in
conflicting design recommendations for transfer and
development lengths of pretensioned strands. In an attempt
C. Dale Buckner
to reconcile some of the differences in the design
Ph.D., P.E. recommendations, the FHWA conducted an independent
Professor review of the recent research. This paper represents a
Department of Civil and summary of that effort. The specific objectives of the study
Environmental Engineering were : (1) .c onduct a review of literature related to strand
Virginia Military Institute
Lexington , Virgin ia transfer and development length research; (2) analyze data
from recent studies and rationalize discrepancies among
conclusions drawn from these stuqies; and (3) recommend
equations for strand transfer and development lengths
consistent with the current state-of-knowledge.

he development length of pre- pretensio n tran sfers in to a co ncrete

T stressi ng strand is the minimum

embedment needed to prevent
slip when the strand reaches its design
member
2. T he flexura l bo nd le ngth, Lb,
where bond stresses equi librate the
stress, fp s· T his length is meas ured difference between the design stress
from the point of maximum stress to and the effective prestress,fse
the free end of a strand, as shown in An equation for strand development
Fig. 1. length is contained in the American
The development length consists of Concrete Institute (ACI) Building Code
two segments: and the American Association of State
1. T he tran sfe r le ngth, L 1, where Highway and Tran sportation Officials

84 PCI JOURNAL
 For checking shear strength, the trans-
Embedment ~ L.t fer length is further simplified to 50db.

~
Fully Bonded Strand
r: ~. BACKGROUND
TO THE CURRENT
ACI/AASHTO EQUATION
Eq. (1) was derived by ACI Com-
De bond
Embedment ~ L.t Zone mittee 323 (now 423) for inclusion
1~ in the 1963 ACI Building Code.
t 'I AASHTO adopted the expression in
c. ., 1973. The equation is based largely on
A,..f,. • .... J data from tests conducted at the Port-

Oebonded Strand f land Cement Association (PCA) by

Hanson and Kaar,3 as explained in a
recent paper.•
Fig. 1. Development length of fully bonded and debonded pretensioned strand. Hanson and Kaar reasoned that after
flexural cracking in a pretensioned
concrete member, a bond stress wave
(AASHTO) design specifications: "2 progresses from the point of maximum
Ld = ~e db +(ips - fse )db (2) stress toward the transfer zone. Gen-
Ld =(ips -~ f se )db (1) eral bond slip occurs if the wave
The term f sedb/3 represents the reaches the end of the transfer zone.
Eq. (1) is expanded in the ACI transfer length and the second term They deduced the shape of the bond
Building Code commentary to read: represents the flexural bond length . stress wave from measurement~ on 47

Increase In
Steel Stress,
Cfsb • '••>
1
and
(f su • 'sa>•
kill

40 • 'su ·fse. Increase in Staat Stntss to Ultimate Bond Slip

o 1m ·lse, tncreasaln Steel Stntas to Ganarm Bond Slip

20

20 40 60 80 100 120 140 160

Fig. 2. Flexural bond length recommended by ACI Committee 323 (from Ref. 4).

March- Apri l1995 85

pretensioned rectangular beams. Using in a member. In subsequent ACI Tests conducted in 1986 at North
a lower limiting curve of local bond codes, allowable bond stresses were Carolina State University (NCSU)
stresses, Hanson and Kaar calculated replaced by development length crite- found development lengths for un-
the embedment length necessary to de- ria. In the current ACI Code, develop- coated pretensioned strands that were
velop the breaking strength of Grade ment of reinforcement is treated in significantly longer than predicted by
250 strands of 1/4, 3/8, and 1h in. (6.4, Chapter 12, "Development and Splices Eq. (l).s
9.5, and 12.7 mm) diameters. of Reinforcement." Pretensioning strand in use when
Based on a reappraisal of Hanson A statement from the Commentary the Hanson and Kaar tests were per-
and Kaar's data, ACI Committee 323 to Section 12.0 of the ACI Building formed was stress-relieved Grade 250
chose a less conservative expression Code is worth noting: "The strength strand with a specified tensile strength
for the flexural bond length. Data con- reduction factor 1/J is not used in of 250 ksi (1.7 GPa). In current prac-
sidered by the committee, plotted in this chapter. The basic development tice, Grade 270 strand with a higher
Fig. 2, show the increase in strand lengths !db already include an al- tensile strength, 270 ksi (1.86 GPa),
stress from the effective prestress, fse, lowance for understrength" (see Ref. and larger cross-sectional area is used.
to failure stress, fps• in terms of strand 1, p. R-171). This statement is true for Low-relaxation strand with higher
embedment and diameter. Open cir- reinforcing bars, for which the code yield stress has replaced stress-
cles represent the increase in strand equations give development lengths relieved strand for new construction.
stress at first slip (general bond fail- about 15 percent longer than predicted These improvements allow higher
ure) and solid circles represent the in- from experimental results. A similar pretensioning stresses and larger
crease in stress at ultimate. General margin of safety is not provided for strand sizes than those tested by
bond slip corresponds to the first mea- prestressing strand development Hanson and Kaar.
surable slip at the free, unstressed end lengths. The longer development lengths
of the strand. After general bond slip, measured in the NCSU study and the
the helical shape of the wires provides higher pretensioning stresses in cur-
Concerns with the rent practice prompted the Federal
mechanical resistance, which permits
an increase in strand stress with addi- Current Expression Highway Administration (FHW A) to
tional strand slip. At ultimate bond With the exception of cantilevers question the current ACIIAASHTO
stress, the strand slips with no further and short span members, strand devel- expression for development length. In
increase in stress. An equation to rep- opment seldom governs the design of October 1988, the FHW A issued a
resent a "reasonable mean for points pretensioned concrete members. Nev- memorandum that imposed the fol-
representing general bond slip," and ertheless, several bond-related failures lowing restrictions on seven-wire
deemed "not overly conservative at of pretensioned members have been strands in bridge applications: 9
larger embedments," was recom- reported since adoption of the current 1. The use of 0.6 in. (15.2 mm) di-
mended by the committee.' criteria. Martin and Scott describe the ameter strand in a pretensioned appli-
Though the committee's intent was failure of a pretensioned member that cation shall not be allowed.
to represent the mean to points corre- collapsed under construction loads. 5 A 2. Minimum strand spacing (center-
sponding to general bond failure, it is similar member was load tested and to-center) will be four times the nomi-
apparent from Fig. 2 that the expres- failed in bond at 85 percent of its ex- nal strand diameter.
sion is unconservative for long embed- pected capacity. Martin and Scott 3. Development length for all strand
ments (i.e., Lb/db greater than about called attention to the lack of conser- sizes up to and including 9h6 in. (14.3
80). Seven of the 10 specimens with vatism in the current criteria and rec- mm) shall be determined as 1.6 times
Lb/db greater than 80 experienced gen- ommended adoption of criteria to AASHTO Eq. (9-32).
eral bond slip at lower stresses than match the Hanson and Kaar data. They 4. Where strand is debonded (blan-
calculated by the committee's equa- derived curves of best fit for 1/4, 3/8, keted) at the end of a member and ten-
tion. For one-half of the specimens in and 1/z in. (6.3, 9.5 and 12.7 mm) sion at service load is allowed in the
this region, the expression is uncon- strands consistent with Hanson and precompressed tensile zone, the devel-
servative with respect to ultimate bond Kaar's results. opment length shall be determined as
failure. Most pretensioned concrete Zia and Mostafa also reported a 2.0 times AASHTO Eq. (9-32), as cur-
members designed today fall into this bond-related failure in a pretensioned rently required by AASHTO Article
region [i.e., stress increases more than concrete beam. 6 In this case, the beam 9.27.3.
80 ksi (552 MPa) from effective to de- did not collapse, but sagged exces-
sign stress]. sively with about 3/8 in. (9.5 mm)
strand slip. They recommended the Recent Research
Eq. (1) was adopted for the 1963
ACI Building Code, which introduced following more conservative equation The FHW A memorandum created
strength design as an alternative to for strand development length: 7 problems for the prestressed concrete
working stress design for structural industry. In some cases, designs
concrete. At that time, bond and an- Ld = 1. 5 ~: db - 4. 6 + 1.25(!ps - fse )db needed revisions to conform to the in-
chorage of reinforcement were treated Jet terim criteria, stock precast elements
by limiting calculated bond stresses (3) became obsolete, pile caps had to be

86 PCI JOURNAL
Table 1. Proposed equations for strand transfer length.
Source (Reference) Expression Remarks
,• e
f---------- --- --
Current AASHTO/AC I (Refs. I and 2)

-
l L
'
=
/ ,.db
3
~ 50dh I
-~

Zi a and Mostafa (Ref. 7) L1 = [ I .5(/,;f/;;)db] - 4 .6

--- ~

I
Cousins et al. (Ref. 8) L1 =O.S(UrfB ) + J, Apsi( TrdbU1) U1 and B are empirical bond coe ffi cients.
- - ~-
---- - -
Shahawy et al. (Ref. 12)
1 L - f ,,db
I -

Eurocode CEB -FIP. C90

3

l_ - --
Coo""" a,, a,, ""' aw "=""' fm
-
I
L' = a sa"a'o ( AP,
- )( -!,, ) sudden or gradua l release , stress condtti on
(Ref. 20)
Jrd ,, !,,,}(, around strands and strand surface co~dttton ~
-- -- --- -- --

Russe ll and Burns (Ref. 16) / , db

L, = 2
- -- ~-
.- "
!,; d b 3
Mitchell et al. (Ref. 19) L = -

I
3 ~ J;;
11

thickened to accommodate longer em- 3. A project was conducted by the crete compressive strength and strand
bedments needed for pretensioned FHW A to study tiansfer and develop- size on development length. '9
piles, and so forth. More ominous for ment lengths of both epoxy coated
the future is that 0.6 in. (15.2 nun) di- and uncoated strands in small-sized
Objectives of Present Study
ameter strands at 2 in. (51 mm) spac- elements. " This study considered the
ing are crucial to attaining the benefits effect of grit-impregnated epoxy As these projects neared comple-
of high strength concrete [i.e., con- coating, strand size, and age of speci- tion, it became apparent that research
crete with a compressive strength of men. A second phase of this project is was leading to myriad conflicting rec-
10,000 psi (69 MPa) or higher] in underway that includes full-scale ommendations. To resolve some of
long-span pretensioned concrete struc- specimens and will examine the ef- these conflicts, the FHW A conducted
tures. '0 Many research projects were fects of strand spacing and concrete an independent review of literature re-
initiated to find better guidelines for strength. lating to strand transfer and develop-
strand development length. Among 4. A comprehensive research pro- ment lengths. This review inc luded an
these projects were the following: gram was conducted at the University analysis of data from recent projects
1. The University of Tennessee at of Texas at Austin (UT A) .' 6 Testing and formulation of design equations.
Knoxville (UTK) conducted a pro- included measurements of transfer This paper presents a summary of that
ject, co-sponsored by the Precast/ lengths and development lengths for effort. In the sections that follow, the
Prestressed Concrete Institute (PCI), both 'hand 0.6 in. (12.7 and 15.2 nun) research projects cited above are dis-
involving small prisms and full-sized diameter strands, and the behavior of cussed in more detail , along with other
AASHTO Type I girders. " This pro- beams made with debonded strands. relevant studies. The review focuses
ject considered both uncoated and Variations in specimen size and shape, on seven-wire, uncoated, low-relaxation
epoxy-coated strands and examined strand spacing and strand confinement strands in normal concrete. Research
several important variables including were also examined. currently underway should provide
strand size and spacing, strand surface 5. Tests of fu ll-sized AASHTO guidelines for transfer and develop-
condition and stress level at release. Type II girders with cast-in-place ment lengths for epoxy-coated strands
2. Three projects were conducted composite slab were conducted at Pur- and for high strength concrete.
by the Structures Research Laborato- due University . 17 These tests were part Specific objectives of the present
ries of the Florida Department of of a study of pretensioned girders with study were as follows:
Transportation (FDOT). One project debonded strands made continuous for 1. Conduct a review of literature re-
involved tests of full-sized AASHTO live loads. lated to strand transfer and develop-
Type II girders with cast-in-place 6. Tests were conducted jointly by ment length research.
composite slabs . 12 Another project Louisiana State University (LSU) and 2. Analyze data from recent studies
studied strand development length in Auburn University to study the effect and rationalize discrepancies among
solid and voided pretensioned slab ele- of epoxy coating, strand spacing and concl usions drawn from these studies.
ments. " A third project examined the cover requirements.' 8 3. Recommend design criteria for
special case of a pretensioned pile em- 7. McGill University researchers strand transfer and development
bedded within a cast-in-place concrete conducted tests on small-sized speci- lengths consistent with the current
cap or footing .•• mens to determine the effect of con- state-of-knowledge.

March-April 1995 87
 TRANSFER OF tions is given in Table 1. More com- lease of prestress, for which the stress
PRESTRESS plex expressions have been derived fsi is needed. Thus, expressing L 1 in
analytically. 20•2 '· 22 terms of fsi• rather than fse, is both ra-
The current ACI/AASHTO expres- Transfer length is influenced by tional and convenient from a design
sion for transfer length was derived many factors. Strand surface condition standpoint.
using a transfer bond stress of 400 psi (e.g., as-received vs. rusted) is known Considering the wide variation in
(2.76 MPa), which represents the aver- to be an important factor. Sudden re- measured transfer lengths, there are
age value from tests conducted at the lease of pretension (e.g., by flame cut- reasons to question whether a mean, or
Portland Cement Association (PCA) ting) results in longer transfer lengths upper- or lower-bound expression
for Grade 250 strand. This stress ap- than those required when stress is should be adopted. A mean-value ex-
plies to the actual perimeter of a gradually applied . These and other pression is consistent with ACI and
seven-wire strand, 4ndJ3. For equi- factors are discussed in greater detail AASHTO criteria that depend on
librium of a strand over the transfer in a recent study.' 6 transfer length. For checking stresses
length: Despite wide variation in measured immediately after transfer, the roughly
4
values, several points have been well 20 percent longer length computed by
L.Fx = 0.4oo( ~b )L1 - established by recent tests: Eq. (6) is slightly unconservative as
(4) 1. Transfer length increases directly compared to calculations made using
0. 725( ndt I 4 )fse =0 with strand diameter for diameters the current expression. Typically, the
ranging up to 0.6 in. (15.7 mm). calculated stresses at the end of the
where the bond stress and effective 2. Transfer length is established im- transfer length will differ by less than
prestress are in ksi units. The constant mediately after release and remains 5 percent using the two equations .
0. 725 is the ratio of the actual area of practically unchanged with time. This is acceptable considering the fac-
Grade 250 seven-wire strand to the 3. Strand spacings less than 4db tor of safety in the allowable stresses
area of a circle of the same nominal [specifically, 0.6 in. (15.7 mm) at 2 in. at transfer and the fact that overstress-
diameter (for Grade 270 strand, this (51 mm) centers and 1/z in. (12.7 mm) ing at transfer is not catastrophic.
constant is about 6 percent larger). at 1.75 in. (44 mm) centers] do not A longer transfer length is more
affect transfer lengths of uncoated conservative for checking member
Solving Eq. (4) yields: strands. strength under design loads and is
An analysis of recent test results warranted for checking shear strength
L = fsedb (5) shows that the transfer length for and strand development length. Cur-
I 3
seven-wire, low-relaxation strand in rent ACI and AASHTO specifications
For an effective prestress of 150 ksi normal weight concrete having 3500 approximate the transfer length as
(1.04 GPa), Eq. (5) results in a transfer psi (24.1 MPa) compressive strength, 50db for calculating the contribution of
length of 50db, which is the approxi- or higher, at release can be reasonably pretensioned strands to shear strength.
mation used in both ACI and approximated by the expression: 21 This value should be increased by 20
AASHTO shear design criteria. percent, to 60db, to account for the
In a strict sense, Eq. (5) became in- L = fs;db (6) longer transfer length of Grade 270
I 3 strands. The current ACI and
valid when the industry changed from
Grade 250 to Grade 270 seven-wire where /s; is the strand stress at the end AASHTO shear criteria were conser-
strand. This is because the cross- of the transfer zone immediately after vatively developed and a strength re-
sectional area of Grade 270 strand is release. This expression was recom- duction factor of rp =0.85 is applied to
about 6 percent larger than that of mended in the FDOT study 12 and rep- the calculated nominal shear strength
Grade 250 strand of the same nominal resents the mean value approximation of a member. Thus, allowance is made
diameter, hence, requiring (theoreti- to their measurements. for extreme values of transfer length
cally) about 6 percent longer transfer Eq . (6) differs from the current in the shear criteria and continued use
length. Also, newer low-relaxation ACI/ AASHTO transfer length expres- of a mean value is justified. The de-
strand generally has higher strand sion in that fsi replaces the term fse· gree of conservatism and suitability of
stress after transfer, requiring longer The use of the term /s; is more rational Eq. (6) for calculating strand develop-
transfer length than predicted by considering that the transfer length is ment length are discussed in the fol-
Eq. (5). established at release of prestress and lowing sections.
Recent studies have evaluated Eq. does not change significantly with
(5) for Grade 270, low-relaxation time. For usual designs, the transfer
strands in diameters ranging from length by Eq. (6) is about 20 percent
DEVELOPMENT LENGTH
3/s to 0.6 in. (9.5 to 15.2 mm) diame- longer than that computed by the cur- The FHWA memorandum requiring
ters. These studies have shown wide rent expression. a 60 percent increase in development
variation in measured values and led An important step in the design of length, among other restrictions, stim-
to recommended expressions that dif- pretensioned concrete members is to ulated many research projects. These
fer significantly from the current ex- check the top and bottom fiber stresses studies led to several new proposals
pression. A sampling of recommenda- at the end of the transfer zone at re- for pretensioned strand development.

88 PCI JOURNAL
Development lengths calculated by reduced the prestress from an average ranging from 10.6 to 27.4 kips (47.2 to
different proposals for the same pre- 186 ksi ( 1.28 GPa) to about 165 ksi 122 kN) at failure. 25 This analysis was
tensioned member vary widely (typi- ( 1.14 GPa) at the time development based on a theory that the concrete
cally, the highest computed value is length tests were performed. near the load point was strengthened
more than double the lowest, as shown No explanation is given in the UTK by confinement supplied by the load-
by the example in Appendix A). One report for this unusual behavior. ing plate. Confinement was assumed
study has recommended eliminating However, the researchers note in an to increase the compressive strength to
the strand development length concept early version of the project reporf• more than twice the measured uniaxial
and replacing it with crack prevention that, " ... reliability of the resistance strength of some specimens. Calcu-
criteria. '6 strain gauges during the construction lated strand stresses by this analysis
In the following sections, several re- phase was questionable. At best, no were generally larger than the guaran-
cent proposals for strand development more than half of the installed gauges teed ultimate strength and consider-
are discussed and an attempt is made provided test results after transfer. ably larger than values tabulated in the
to reconcile some of the differences Also, correlating the output strains final report.
among the proposals. A recommenda- with theoretical strains at the failing Given the uncertainties of the actual
tion is made to adopt a more conserva- load provided questionable results" strand stresses in the UTK specimens,
tive expression for development (see Ref. 24, p. 53). a modification to the development
length to reflect current code philoso- Erroneously large effective prestress equation based on the UTK data is
phy and probable member behavior. in the specimens Jed to the conclusion unjustified.
that average flexural bond stresses
UTK Proposal for were lower than predicted by the cur- FOOT Proposal for
Development Length rent expression, hence, requiring a 50 Development Length
percent longer flexural bond length.
Based on tests of 20 full-sized The FDOT conducted three separate
(Note that the analysis on which the
AASHTO Type II girders, researchers studies of strand development. These
final UTK recommendation was made
at UTK proposed the following devel- included tests of 17 AASHTO Type II
fixed this increase more precisely as
opment length equation:" girders with composite slabs, 12 seven
42 percent) . 24 Assuming a design
solid and voided slabs, 13 and a series
stress of 250 ksi (1.73 GPa), which is
simulating piles embedded in a cast-
close to the average reported for speci-
in-place concrete cap.' 4 Based on these
mens that failed in flexure and to the
tests, FDOT submitted a proposal to
The UTK study targeted many im- value obtained by strain compatibility
AASHTO Committee T-1026 to revise
portant variables related to strand de- analysis, the consequence of erro-
the development length expression to:
velopment. Unfortunately, there are neously large effective prestress can
reasons to question whether the UTK be shown:
data justify the recommended 50 per-
cent increase in the flexural bond .
Rat1o=
(ips - fse L.elY
length. This recommendation is based
on the average bond strength at failure (ips - fse t eported (8} where llave equals 250 psi (1.73 MPa)
for the 20 specimens, which is a func- =250 - 165 =1. 5 and kb is a dimensionless constant de-
tion of the design stress and effective 250-193 fined as: kb = 8 for piles embedded in
prestress at the time of test. Effective a concrete footing or pier cap; kb = 4
prestresses reported in the UTK study Thus, to justify a 50 percent increase for slabs and slender members; and kb
were determined from strain gauge in flexural bond length requires that ef- = 2 if the computed development
readings and are generally unrealistic. fective prestress increases with time, length (using kb =4) to member depth
The average effective prestress im- which is unlikely. Assuming effective ratio is less than or equal to 3. For
mediately after transfer reported for prestress diminished to about 165 ksi "slender members," kb = 2, the FDOT
the UTK specimens is 186 ksi (1.28 (1.14 GPa) when development length proposal gives about the same result
GPa), which is reasonable considering tests were performed implies that the as the current ACI/AASHTO expres-
that strands were pretensioned to 203 current ACIIAASHTO equation is ade- sion. But the FDOT criteria abruptly
ksi (1.4 GPa) and elastic shortening quate for the UTK specimens. doubles the current value for deep
losses average about 15 ksi (104 MPa) The UTK development length tests members and halves the current value
for the specimens. But the average ef- were performed with the specimens for embedded piles.
fective prestress for the same speci- supported on concrete-filled structural The basis for the FDOT recommen-
mens at the time development length steel tubes. Friction at these supports dation regarding embedded piles is
tests were performed (i.e., several created an arching action that en- founded on the belief that shrinkage of
months later) was reported to be 193 hanced resistance to the applied loads. cast-in-place concrete around a precast
ksi (1.33 GPa}, an increase from the A precise analysis of this effect is im- concrete pile creates a clamping force
stress immediately after transfer. More possible. An analysis by the UTK re- that reliably improves the flexural
likely, creep, shrinkage and relaxation searchers gave horizontal reactions bond of the pretensioned strand. Tests

March-April 1995 89
conducted by FOOT for this applica- pression fs;db/3 for transfer length has
tion involved testing segments of a merit. However, abruptly doubling the
salvaged 14 x 14 in. (356 x 356 mm) development length based on the span-
(10)
pretensioned pile. The segments were to-depth ratio or halving the develop-
loaded to failure after clamping them ment length for embedded piles is not
with a 200 kip (889 kN) force in- justified by the FDOT research.
tended to simulate the effect of shrink-
This expression agrees well with
age in surrounding concrete. There are
Purdue University Tests data from the McGill tests but does
several questionable assumptions re-
not compare well with the data from
garding this test program: Tests performed at Purdue Univer-
other recent studies. 23 Some of the dif-
1. The clamping force was deter- sity involved full-sized AASHTO
ference may result from the gradual
mined from strains measured within a bridge girders and box beams with
release method employed in the
42 x 54 x 48 in. (l.l x 1.4 x 1.2 m) con- both debonded and fully bonded strand
McGill tests, as compared to sudden
crete mass surrounding a precast con- patterns. 17 Girders were made compos-
release in most other studies. Also, the
crete section. Large strains (more than ite with a cast-in-place slab. The fully
strand surface condition in the McGill
300 microstrains) were recorded after bonded specimens were loaded to fail-
study was described as slightly rusted,
one day near the middle of the mass ure with embedments exceeding the
which is known to improve bond.
concrete and initially attributed by the development length calculated by the
The McGill tests provide clear evi-
researchers to drying shrinkage in the current ACI/ AASHTO equation. For
dence that strand bond is better in
concrete. These strains are orders of embedments of about 1.2 times the cal-
high strength concrete members than
magnitude larger than can realistically culated development length, the speci-
in similar members of normal con-
be attributed to concrete shrinkage. mens failed at two-thirds to three-
crete. Research in progress will pro-
More recently, the researchers have ac- quarters of their predicted capacities.
vide a better data base to evaluate the
knowledged that the initial strain read- Only with an embedment of 1.8 times
McGill expression. Until more data
ings were caused by temperature rise.27 the calculated development length was
are available, the McGill University
But they maintain that after two weeks, the full capacity achieved. This led to
equation is not recommended for de-
the concrete temperature returned to the conclusion:
sign purposes.
ambient and that subsequent strains "The flexural and shear design of
were caused by shrinkage. Considering both bonded and debonded pre-
the disparity between tensile and com- tensioned 1-beams, where the Strand Development
pressive strains associated with this ex- flexural capacity controls, based by UTA Criteria
planation, it appears unlikely that such on current ACIIAASHTO design
large strains were caused by drying The UTA study concluded that
provisions would be adequate
shrinkage. seven-wire pretensioned strands could
provided that the fully bonded
2. A stress distribution within the be developed by preventing cracks
strands in the member have an-
mass concrete was formulated by sim- within the transfer zone. Instead of
chorage length of at least 1. 7Ld.
ply multiplying principal strains by an checking strand embedment against a
This recommendation is based on
assumed concrete modulus of 3600 ksi calculated development length, design
the results of the fully bonded
(24.8 GPa). This overlooks the fact guidelines were developed to prevent
beam in Specimen Set 3 as a
that creep significantly alleviates in- concrete cracking in the debond/trans-
lower bound. This value of 1. 7Ld
ternal stresses caused by confined fer zones of girders.16 For this purpose,
is also in agreement with the find-
shrinkage in concrete. The resulting the UTA study recommended an ex-
ing in the FDOT study." (see Ref.
stress distribution bears no resem- pression for transfer length, f se db/2,
17, p. 65.)
blance to a distribution caused by dry- which gives transfer lengths that are
ing shrinkage. In reality, restrained 50 percent longer than those calcu-
drying shrinkage in mass concrete re- McGill University Proposal lated by the current ACIIAASHTO ex-
sults in tensile stresses near the sur- pression. The UTA strand develop-
Based on tests of 22 single-strand
face, not compressive clamping stress, ment approach is summarized in the
rectangular specimens, researchers at
as used in the FOOT tests. following general conclusion:
McGill University expressed strand
3. Specimens cut from an existing development length as a function of " ... to prevent anchorage failures,
pile section are likely to have signifi- concrete compressive strength. 19 The beams should be designed so that
cantly shorter transfer lengths than a expression is similar in format to the no concrete cracks will propagate
typical pretensioned pile. The transfer current ACIIAASHTO expression ex- through the transfer zone of a pre-
length in a pretensioned pile is more cept that the transfer length and flexu- tensioned strand. This observation
than one-half the FDOT' s proposed ral bond length are multiplied by the is comprehensive for all sizes of
development length, leaving only terms involving concrete compressive pretensioned strand, for all pre-
about a 24 in. (610 mm) flexural bond strength, and fs; replaces fse in the tensioned applications, and for
length. transfer length. This results in the both fully bonded and debonded
FDOT's proposal to adopt the ex- equation: strands." (see Ref. 16, p. 209)

90 PCI JOURNAL
 The best documented evidence found
100 0 .1 to explain the interaction between shear
II Web Sheer. CliO !no and bond is Fig. 3, which gives a strong
80 I and End Slip
OA 'E'
c;.
indication that general bond slip oc-

~h
curred prior to sudden shear failure. No
JBon ~Falkn I <
matter whether the failure mode is
g

~
80

40 I
/

/
0.3

0.2
Ii shear/bond or bond/shear, the failure is
sudden and undesirable and should be
(/) prevented by the adoption of conserva-

I - Deflection ! tive criteria.

20

0
v
0 0 .6
v
Deflection (In)
1.1
_._ EnciSUp

2
0.1

0
The behavior of several rectangular
specimens in the UTA study also cre-
ates doubt as to whether this crack
prevention approach is a practical way
to ensure strand development. Speci-
Fig. 3. Load and end slip vs . deflection for a UTA speci men (from Ref. 28). mens FR350-l and -2, containing 112
in. (12.7 mm) diameter strands, both
failed with sig nificant amounts of
For members containing debonded depicted in Fig. 3, however, casts strand slip. Though transfer lengths
strands, the importance of preventing doubt on whether web-shear cracking were not measured for these speci-
cracks within the debond/transfer zone initiated strand slip or vice-versa. This mens, they were estimated to be about
under design loading was well estab- figure shows load-deflection and twice as long as usual [52 to 64 in .
lished by the UTA tests. But the gen- strand end slip measurements from a (1.32 to 1.6 m) based on end slips at
erality of this criterion for all preten- UTA test. 28 transfer] .
sioned applications is questionable. For this specimen, initial strand slip On this basis, the failures were ra-
That it is not generally applicable is occurred at an applied load of about tionalized to satisfy the UTA failure
shown by several specimens from the 70 kips (311 kN) [i.e., 0.7 in. (18 rnm) criterion. But a designer must rely on
Hanson and Kaar tests 3 that failed in midspan deflection] but web-shear calculations and the UTA criterion
bond with ultimate moments less than cracking was not observed until about predicts that the strands in these spec-
predicted and without cracks within 77 kips (343 kN) [i.e., 0.9 in. (23 rnm) imens fully develop, even at embed-
their transfer lengths. deflection]. The initial recorded end ments as short as 60 in . (1.5 m ).
As one example, Specimen 3-10 slip is about O.D15 in. (0.4 mm) . If this When te s ted, Specimen FR350-l
from the Hanson and Kaar study can slip occurred over a distance of 40 in. achieved only about two-thirds its
be cited. This specimen was reported (1.0 m) , which is only hypothetical, full nominal moment strength with
to have failed in general bond at only and resulted in a uniform reduction in embedments of 60 and 72 in. (1.5 and
77 percent of its full nominal moment bond stress over this length, the corre- 1.8 m). Specimen FR350-2 achieved
strength and to have attained only 90 sponding loss of prestress would be 90 percent capacity with an embed-
percent of its nominal strength at ulti- about 20 ksi (138 MPa). This sudden ment of 84 in. (2 .1 m) and reached
mate. The specimen had a reported loss of prestress causes a sudden in- essentially full capacity with 96 in .
transfer length of 26 in. (660 rnm) and crease in principal tensile stresses in (2.4 m) embedment. But even at this
a cracking moment of 467 kip-in. (5.3 the web and this could lead to subse- embedment the strands slipped signif-
MN-m) . The closest probable crack quent web-shear cracking. icantly at ultimate.
was about 5 ft (1.5 m) from the mem- Other recent studies offer conflict- The UTA report attributes the likely
ber end, which is roughly twice the re- ing opinions about whether bond slip cause of the poorer-than-expected
ported transfer length. Several other results in a shear failure or vice-versa. bond in Specimens FR350-1 and -2 to
exceptions to the UTA general conclu- The UTK report, for example, states, accidental strand contamination,
sion can be found among the speci- "Some slight slippage of the strands which led to exceptionally long trans-
mens tested by Hanson and Kaar. during static testing did not signifi- fer lengths. Extensive test data show,
The UTA program included devel- cantly reduce the beam flexural however, that transfer lengths vary
opment length tests for 19 I -shaped strength, but it did make the beams widely, with standard deviations in
beams resembling AASHTO-type more susceptible to shear failure ... " most test programs on the order of
composite girders. These specimens (see Ref. 24, p. 67) . Conversely, the ±10 in. (±254 mm) . Variation is well
frequently showed initial bond slip oc- FDOT study for composite AASHTO- illustrated by several full-scale speci-
curring coincident with web-shear type girders concludes , "The results mens in the UTA study, which had
cracking. The UTA researchers ob- indicate a direct interaction between significantly longer transfer lengths
serve that, " ... this and other research shear and bond with the initial slip oc- [up to 74 in. (1.9 m)] than laboratory-
consistently demonstrate web shear curring immediately or shortly after cast specimens. Thus, for reasonable
cracking to precipitate anchorage fail- the appearance of the first shear assurance that flexural cracks do not
ures" (see Ref. 16, p. 51) . Behavior crack." (see Ref. 26, p. 89). intersect the transfer zone, an ex-

March-April 1995 91
 can be reliably developed with embed-
ment lengths as short as implied by the
UTA criteria.
Data;_ Though the UTA report makes a
Wu= 12 kif
b= 16" h =16" strong case for adopting transfer zone
dp= 14" crack prevention criteria, which have
0 =
f' 5000 psi important implications for shear de-
Section c . 2 sign and for de bonded strands, UTA's
Ap?0.30~1n
general conclusion should not replace
fse= 155 ks1 the current concept of strand develop-
Mcf 64.4 kft ment length.
Elevation
L =5" ctM if 80.9 kft
b
Test Procedure in
Recent Studies

fS8= 155 ksi Recent tests to establish strand de-

velopment length have followed a
=
ub,ave 1500 psi similar approach. Simply supported
pretensioned concrete beams were
Flexural Bond Length tested with varying strand embed-
ments. If a flexural failure occurred,
Note: 1 1n • 25.4 mm; 1 kip • 1000 lb • 6.9 kN
the test embedment length was taken
as an upper bound to the development
Fig. 4. Example of a short canti lever that satisfies UTA development length criteri a. length. If bond failure occurred, a
lower bound to the development
length was assumed.
tremely conservative expression for On this basis, a uniformly dis- In this way, researchers attempted to
transfer length is required. To reiterate tributed design load of 12.0 kips per verify the current ACI/AASHTO ex-
an important point made by the UTA linear ft (175 kN/m) can be supported. pression by finding the embedment
researchers: At the end of the transfer zone, 39 in. length delineating the change in fail-
" .. .wide variation exists in trans- (991 mm) by the UTA criteria, the ure modes from bond to flexural.
bending moment is 63.4 kip-ft (86.0 Aside from the difficulties caused by
fer length from one specimen to
another. This possibility of varia- kN-m), which is less than the cracking scatter in test results, there are reasons
tion should be reflected in the de- moment, 64.4 kip-ft (87.4 kN-m), as to question whether this approach es-
calculated using a modulus of rupture tablishes the development length for a
sign and fabrication of preten-
7.5 {1: . Shear is not a problem; thus, pretensioned strand.
sioned structures." (see Ref. 16,
p. 39)
the UTA crack-prevention criteria are One problem with the approach is
satisfied. But this leaves a flexural that it does not account properly for
The expression for transfer length bond length of only 44 - 39 = 5 in. the interaction and variation among
recommended in the UTA study, (127 mm) over which the difference important variables; namely: transfer
though more conservative than the between the design stress , 263 ksi length, effective prestress and design
current ACI/AASHTO expression, (1.86 GPa), and the effective prestress, stress. To illustrate this point, consider
does not make adequate allowance 155 ksi (1.07 GPa), must be equili- a hypothetical case in which testing
for probable extremes, as the perfor- brated by bond stresses. leads to the conclusion that 80 in.
mance of Specimens FR350-1 and -2 This requires an average flexural (2.0 m) embedment is adequate to de-
shows. bond stress of 1500 psi (10.4 MPa), velop a 112 in . (12 .7 mm) diameter
Equally important, the UTA general which is roughly four times the peak strand.
conclusion is not proven applicable for bond stress reported by Hanson and For simplicity, the stress immedi-
all pretensioned applications. Con- Kaar. 3 Bond stresses of this magnitude ately after transfer is assumed to be
sider, for example, an extremely short, cannot be consistently developed. This 180 ksi ( 1.24 GPa) and the design
uniformly loaded, pretensioned can- is an extreme case, but it demonstrates stress as 270 ksi (1.86 GPa). The aver-
tilever beam as shown in Fig. 4. Using that the UTA criteria are readily satis- age transfer length for 1h in. (12.7 mm)
the approximate Eq. (18-3) of the ACI fied in cases that are highly suscepti- diameter strand under these conditions
Building Code and assuming fully de- ble to bond failures (i.e. , short can- is about 30 in. (760 mm). But there is a
veloped strands, the design stress for tilevers). There are enough documented wide variation in transfer length mea-
the beam is 263 ksi (1.8 GPa), which exceptions, including Specimens FR- surements, with standard deviations on
gives a design moment strength of 80.9 350-1 and FR-350-2 from the UTA the order of ±10 in. (±254 mm) com-
kip-ft (110 kN-m), slightly greater than study, to question whether 1h in. (12.7 mon in recent tests. Thus, there are
the code minimum of l.2Mcr mm) diameter pretensioned strands distinct probabilities that the actual

92 PCI JOURNAL
transfer length for a specimen might Table 2. Hypothetical results for specimens with 112 in. (12.7 mm) diameter strands
be as short as 15 in. (380 mm) or as tested with 80 in. (2032 mm) embedment.
long as 45 in. (1.14 m) . A few possible Ratio
outcomes for test results are summa-
L1 (in.)
I ifps- fse) "t,ave ub,ave
I
"t,ave
rized in Table 2. Case Lb (in.) fse (ksi) (ksi) (psi) (psi) ub,ave

On the first line of Table 2, which is I. Good bond 15 65 175 95 880 110 8.0
labeled "Good Bond," are data for a 2. Average
specimen with an extremely short
30 50 ISO I 120 440 180 2.4
3. Poor bond 45
transfer length, taken here as 15 in.
35 135 I 135 290 I 280
I
1.0
Note: I in. = 25.4 mm; I ksi = 1000 psi = 6.9 MPa.
(380 mm). The effective prestress at
time of test is taken as relatively large
at 175 ksi (1.2 GPa)(e.g., a high qual-
ity concrete specimen tested at an Variation in Local Bond Stress Over the Flexural Bond Length
early age). The average transfer bond
stress for this case , acting on the
actual strand perimeter of 2.09 in.
(74 mm), is:
-400 .
(11) !
Ill
:g 300
180x0.153 iE Constant Flexural Bond Stress
15x2.09 -g Implicit in Current ACUAASHTO
dl200 ..
= 0.88 ksi (6.1 MPa) Expression (About 145 psi)
iii
For a 15 in. (380 mm) transfer
g
..J 100
length, the complementary flexu-
ral bond length is (80 - 15) = 65 in.
(1.65 m). The average flexural bond 0~--------------------------------~
stress that must be developed for this 0 5 10 15 20 25 30 35 40
case is: Distance From End of Transfer Zone (inches)

Note: 1 In • 25.4 mm; 1 ksi • 6.9 Mpa

(12)
Fig. 5. Bond stress vari ation over the flexural bond length.

- (270 -175)0.153
65x2.09 ral failure under these "Average" con- failure , under extremely favorable
= 0.106 ksi (0. 7 MPa) ditions does not provide an upper conditions, 27 does not prove that
bound to development length. strands can develop reliably in similar
Thus, average transfer bond stress in Design criteria must allow for ex- applications with an embedment
this case is roughly eight times the av- tremes. Effective prestress might be as length that is roughly one-half the
erage flexural bond stress and bond low as 0.5fpu• which is 135 ksi (930 length calculated by current criteria.
failure is highly unlikely. But consider GPa) in this case. The transfer length A more rational approach to estab-
the sensitivity of the average bond for a particular member with 112 in. lish the development length is that fol-
stresses to transfer length and effective (12.7 mm) diameter strand might be as lowed by Hanson and Kaar, in which
prestress. Results for an "Average" large as 45 in. (1.14 m). Such a "Poor limiting bond stresses are found and
case are listed in Line 2 of Table 2. Bond" condition is illustrated on used to calculate the required length of
Here, a transfer length of 30 in. (762 the third line of Table 2. In this case, embedment. 3
mm) implies poorer bond than existed relatively poor bond coupled with
for the "Good Bond" case. Further, a high average flexural bond stress,
lower effective prestress of 150 ksi
Variation in
U 1,ave l ub,ave = 1.0, indicates probable
(1.04 GPa) requires that a larger force
Flexural Bond Stress
failure by bond.
be equilibrated over a shorter flexural Verifying the minimum embedment The current AASHTO/ACI develop-
bond length, 50 in. (1.26 m), resulting length to achieve flexural failure for ment length equation (in essence) as-
in a much higher flexural bond stress any number of specimens does not en- sumes a constant flexural bond stress
than in the "Good Bond" case. For this sure that similar members will achieve and obtains a linear relationship be-
case, the ratio of average transfer bond flexural fail ure with the same embed- tween the flexural bond length and the
to average flexural bond stresses drops ment. In particular, testing preten - change in strand stress over this
to one-third the ratio of the "Good sioned pile sections with 48 in. (1.2 m) length. For equilibrium of the flexural
Bond" case. But even attaining flexu- embedment and achieving flexural bond length:

Ma rch-April 1995 93
Table 3. Flexural bond length
multiplier, A., in equation Strain at Failure for Two Ultimate Stresses
Ld =L, + Mfps - fs,)db. Typical 270K Low Relaxation Strand with 150 ksi Effective Prestress

Source Multiplier Stress-strain Curve for Grade 270 Seven-wire Strand

(Reference) A- 400r-----~----------------------------·
Current ACIIAASHTO
= =
M.- 268 - 150 113 ksi
(Refs. I and 2)
1.0 lle = 0.035 - 0.006 = 0.029
300 . .... .
Zia and Mostafa
1.25
(Ref. 7)
FHW A Interi m
1.6
(Ref. 9)
UTK
1.5
(Ref. II )
Purdue
1.7
(Ref. 17)

McGill ru
-
OL---~----~--~~--~---~--~--~
0 O.Ql 0.02 0.03 0.04 0.05 0.06 0.07
(Ref. 19)
~ !/ Strain (in/in)
0.5, 1.0 or 2.0
FOOT Note: 1 In • 25.4 mm; 1 ksl • 6.9 Mpa
depending on
(Ref. 26)
parameter kb
Fig . 6. Stress-strain relationship for a seven-wire strand.
Proposed 1.0 $ (0.6 + 40t:ps)
(Ref. 23) $2.0
posals to revise the development 150 ksi (1.05 GPa) to, say, 250 ksi
length equation apply modifiers to the (1.73 GPa), the strand experiences an
flexural bond length, which, in effect, increase in strain of roughly 0.005.
(13) alters the assumed average flexural However, if the stress had been re-
bond stress. Table 3 contains proposed quired to reach, say, 268 ksi (1.86
flexural bond length multipliers from GPa) (about 7 percent greater stress)
several recent studies. the increase in strain would be 0.029,
where the term 4rrdtf3 is the perimeter which is 500 percent greater. Consider-
Reduction in local bond stress from
of a seven-wire strand. For Grade 270 its peak value of 400 psi (2.8 MPa) to ing the bond mechanism, stretching
strands, strand area can be approxi-
the limiting 50 psi (345 kPA) appears and contracting the strand six times as
mated as 0.6ldb>, and Eq. (13) can be
reasonable considering the generally much in the latter case should logically
solved to obtain an expression for the
accepted bond mechanism for seven- result in a lower average bond stress.
flexural bond length similar to the cur-
wire strand. 29 After a flexural crack The reduction in average bond stress
rent ACI/AASHTO expression:
forms, adhesion between the steel and with increasing strain, in terms of the
concrete is broken and a strand slips index pfp/f: = mP' is evident in Han-
Lb =(+ps _+Jse )(3x0.61}b
J (14) and twists along the helical path son and Kaar' s data and has been
4 rrub,ave formed in the concrete. Slip is initially noted in several reappraisals of their
resisted by the slight mismatch in cross. dataY
In the AASHTO/ACI expression, sections that occurs when one cross Hanson and Kaar noted, " ...for a
the term (3 x 0.6l)/(4rrub,ave) is unity, section is displaced a short distance given embedment length, the average
implying a constant value of bond away. As the stress increases, the bond stress at flexural failure will be
stress, ub,ave == 145 psi (1 MPa). strand stretches and contracts laterally, less in a beam with high steel percent-
The local bond stress wave deduced reducing the amount of misfit and, age than a beam with low steel per-
by Hanson and Kaar is reproduced in hence, reducing the resistance to slip. centage. The risk of exceeding the av-
Fig. 5. Bond stresses vary from a peak In view of this bond mechanism and erage bond stress at which general
value of 400 psi (2.76 MPa) at the end considering the nonlinear stress-strain bond slip occurs for a particular em-
of the transfer zone to a limiting value relationship for seven-wire strand, it bedment is therefore greater with a
of 50 psi (345 kPa) beyond 30 in. (762 seems illogical to assume that flexural beam having a low percentage of
mm). Hanson and Kaar established bond resistance is constant regardless steel." (see Ref. 3, p. 795) The fact
this curve as a lower bound to the of stress level and that the develop- that average bond stress is higher
bond stresses at general bond slip. ment length is a linear function of /p 5 over segments nearer the transfer
The AASHTO/ACI expression devi- for a given strand diameter and effec- zone than over the entire flexural
ates from the Hanson and Kaar criteria tive prestress. Consider, for example, bond length is the basis for applying
in applying a constant bond stress, an idealized stress-strain curve for the development length equation in
about 145 psi (1 MPa), regardless of Grade 270 strand as shown in Fig. 6. current specifications.
embedment length. Most recent pro- As stress increases from an effective Both ACI and AASHTO permit the

94 PCI JOURNAL
 investigation of strand development to
be limited to the critical sections nearest Comparison of Calculated Strand Strains at Nominal Moment Strength
to each end of a member. For some Hypothetical Case of UTK Specimen with Composite Slab
members, the available length of em-
bedment is less than that required to de-
velop the strand to its ultimate stress, Composite Section Neutral Axis

/pu- Because the average flexural bond

stress is higher for shorter lengths of
embedment, it is conservative to use the
development length equation to solve Neutral Axis

for the maximum strand stress that can

be developed for the given embedment
length. This approach is used, for exam-
ple, in AASHTO Eq. (9-19) to calculate
the limiting strand stress for precast,
A) As Tested (Precast Only) B) If Tested as Composite Section ·
prestressed deck panels.2
Fig. 7. Ultimate strand strains in composite and noncomposite girders.

Failure Strains in
Recent Test Specimens Calculated Strand Strain in Test Specimens at Nominal Moment Strength
Based on Strain Compatibility; ACI Principles; PCI Stress-Strain Curves
Apparently to simplify testing, most
development length specimens have Streaa-atraln Curve for 270Kal Seven-wire Strand
been proportioned to fail at relatively 400r----------------------------------,
low strains. The exceptions have been
girders with cast-in-place composite
slabs (Purdue, FDOT and a few oth- 300
ers) . The typical effect of adding a
composite slab is illustrated in Fig. 7, Composite Test Specimens with
using a test specimen from the UTK · · ·aonci s·n"P &i "F&iiui&:· ·· ·· · ·· ··· · ·· · ·
study as an example. Girders in the Purdue Type II (12L.; e,.• 0.036)
FOOT Type II (1 .7l.; e,. • 0.027)
UTK study were tested without a slab.
The neutral axis in these noncompos-
ite sections at failure was about 10 in.
(254 mm) below the compression edge,
resulting in calculated strand strains by
0.01 0.02 0.03 0.04 0.05 0.06 0.07
strain compatibility of about 0.0 ll .
Strain (in/in)
Had these sections been made compos-
ite with a cast-in-place slab- as most Note: 1 in • 25.4 mm; 1 ksl • 6.9 Mpa
actual bridge applications would have
been - the neutral axis at failure Fig. 8. Calculated strand strains at flexural failure in recent test specimens.
would be above the precast section and
near the compression edge. This would tested AASHTO Type II girders with Fig. 8 compares calculated strains at
result in much larger strand strains that composite cast-in-place slabs and found failure for specimens from several test
are typically close to the specified min- that these members generally required programs. Generally, those specimens
imum elongation of 0.035. up to 1.7 times the ACIIAASHTO pre- with calculated failure strains greater
Experimental results from most test dicted development length. 13 One possi- than 0.025 experienced strand slippage
programs suggest that average bond ble reason for this discrepancy is the rel- with greater embedments than that re-
strength is lower in specimens with large ative difference in strain levels between quired by current ACIIAASHTO crite-
strand strains at failure (e.g., strains near the two groups of specimens . Using ria. In addition to the FDOT girders
the guaranteed minimum elongation of strain compatibility analysis, the calcu- discussed above, five composite
0.0350) as compared to specimens that lated strains at failure for the FDOT slab AASHTO Type I girders with fully
failed with strains near the yield strain specimens are about 0.011 (by ACI 318 bonded strands were tested at Purdue
(i.e., 0.010). For example, FDOT con- principles). The FDOT composite gird- University. 17 Four of the specimens
ducted tests on solid and voided slab ers have calculated strains at failure of were tested with embedments of 1.1
sections. The conclusion from this series about 0.027. Assuming the effective to 1.2 times the ACI/AASHTO devel-
was that the current ACIIAASHTO de- strain at decompression was about 0.006 opment length and failed with strand
velopment length expression was con- in all specimens, the additional girder slippage at moments significantly
servative for these sections. strains would be about four times as below their expected capacities. Only
In a separate test program, FDOT much as in the slab specimens. with an embedment of 1.8 times the

March-April 1995 95
 ing percentages of prestressed rein-
Comparison of Calculated Development Lengths forcement. Also shown in the figure
rc .. 5000 psi; rei .. 4000 psi; fsi .. 180 ksi; fse "'
180 ksi are the development lengths for this
12- by 32-in beam with varying number of 1/2-in strands member calculated by several other
criteria. Calculations for the example
_200.---------------------· are included in Appendix A. Eq. (15)
§. agrees with the current ACI!AASHTO
-1969MSHTO
expression for the case in which fp s =
242 ksi (1.7GPa), where OJP = 0.36{3 1•
+ Current ACI/MSHTO
The equation yields about 1.6 times
+ McGill Univ.
the current ACII AASHTO develop-
... FOOT Proposal
ment length for f ps = 263 ksi ( 1. 8
-UTK GPa), where OJP = 0.072.
-+- Zla-Moustafa Strand development is often critical
-g 50 . ... ··· ·· · · ·· · · ·· · · ·· ·· · ... · · · ·· ··· ·· · ·· · · - Martin-Scott in pretensioned cantilevers , where
'Iii +Purdue strands typically have several feet of
'3
u •Proposed concrete cast beneath them . While
iii
0
0~--------------------~
none of the studies reviewed in this re-
80 85 90 95 100 105 110 115 120 port have examined the effect of
Stress Change, (fps- fse), (ksl) strand position on development length,
Note: 1 In • 25.4 mm; 1 kll • 8.9 Mpa there is evidence that settlement of
freshly placed concrete and the accu-
Fig. 9. Comparison of development length by various criteria. mulation of bleed water around
strands can weaken bond. Studies at
the University of Illinois found that
current development length did one length from its current value at cps = the bond strength of strands with
specimen reach full capacity without 0.010 to 1.7 times the current value at 10 in. (254 mm) or more of concrete
bond failure. cps = 0.035, consistent with the recom- cast beneath them was only about 75
mendation from the study at Purdue percent of the strength for strands in
University. This is accomplished by similar specimens having only 2 in .
Recommended Development (51 mm) of concrete below. 29
the expression:
Length Equation Also, comparative end slips for top
To ensure both strength and ductil- strands at transfer reported in the UTK
ity in pretensioned concrete mem- study were, on average, .about 2.6
bers, a more conservative develop- times the corresponding slips in the
ment length expression is required for where, for general applications, A is bottom strands of the same specimen.24
members that must attain large design taken as (0.6 + 40cps) and cps is the Tests have shown a correlation be-
strains. Test specimens proportioned strain corresponding to fps· tween large strand slip at transfer and
to have strains at failure near yield For many applications, design stress poor strand development. 30
(0.010) have usually achieved their is approximated using Eq. (18-3) of The ACI and AASHTO specifica-
predicted moment strengths at strand the ACI Building Code. For these tions both require that the basic devel-
embedments calculated by the current cases, strain is expressed in terms of opment length of reinforcing bars be
ACI/AASHTO expression . Speci- OJP = Pp/'p.ff: and the equivalent ex- increased by a factor of 1.3 when hori-
mens proportioned to achieve strains pression for A is (0.72 + 0.102{3 1/wP) . zontal reinforcing bars have 12 in.
at failure of about 0.035 have typi- Although the calculated strain in pre- (305 mm) or more of concrete cast be-
cally failed to reach their expected tensioned concrete members fre - neath. For these reasons, it is recom-
moment capacities due to premature quently exceeds 0.035, the ASTM mended that Eq. (18) be regarded as a
bond failures . specified minimum elongation is satis- basic development length and that it
To provide a conservative expres- fied at 0.035, which corresponds to be multiplied by a factor of 1.3 when
sion, it is recommended that the cur- A = 2.0. It is reasonable then to set an 12 in. (305 mm) or more of fresh con-
rent expression for transfer length, upper limit of 2.0 for A. A lower crete is placed below either a horizon-
fsedb/3, be replaced by the expression bound of A = 1.0 applies for the un- tal or draped strand that ends within
f sidb/3 . Further, the constant bond usual cases in which design strains are the upper one-third region of a mem-
stress applied to the flexural bond below yield, 0.01 . Thus: ber's overall depth. For strands ending
length in the current expression should in other regions, the 1.3 multiplier
be replaced by a variable bond stress, Lo~ [;., =(0. 6+40cps )]~2.0 (16) need not be applied.
similar to the approach followed by Several example problems to illus-
Hanson and Kaar. The variable bond Eq. (15) is plotted in Fig. 9 for an trate the proposed equation are in-
stress should increase the development example rectangular beam with vary- cluded in Appendix B.

96 PCI JOU RNAL

 SUMMARY AND of 5000 psi (34.5 GPa) or higher. governed by flexure was to provide a
RECOMMENDATIONS The transfer length calculated by strand embedment of at least 1. 7 times
Eq. (17) is about 20 percent longer the development length as computed
Recent studies provide a wealth of than that computed by the present by the current ACI/AASHTO expres-
information relating to transfer and de- ACI/AASHTO expression. In the cur- sion. This recommendation was noted
velopment of pretensioned strand. rent ACI and AASHTO shear criteria, to be in agreement with the earlier
This contributes to new applications, the transfer length is approximated as finding in the FDOT tests of AASHTO
especially for pretensioned beams 50db. This value should be increased composite beams.
made with high strength concrete. 3 '· 32 by 20 percent, to 60db, to account for Tests conducted at UTK, FDOT,
Unfortunately, conclusions relating to the longer transfer length of Grade UTA and Purdue have shown an im-
strand transfer and development 270 strands. portant relationship between strand
lengths have differed significantly
bond and shear strength. The tests
from one study to another.
demonstrate the potential for a web
The objectives of the present study Development Length shear crack to propagate through the
were to conduct a review of literature
Design recommendations from re- strand transfer zone and cause general
relating to transfer and development of
cent studies give wide-ranging values bond failure. The consequent destruc-
seven-wire pretensioning strand, to ra-
of development length when applied tion of tension anchorage leads to a
tionalize discrepancies among conclu-
to the same design situation (typically, failure that is frequently sudden and
sions drawn from various studies, and
the highest computed value is more without warning. This behavior indi-
to recommend equations for strand
than double the lowest value). One cates a deficiency in the current
transfer and development lengths. At
study has recommended abandoning ACI/AASHTO shear design criteria
present, there are several research pro-
the concept of strand development for pretensioned members that should
jects in progress related to strand de-
length in favor of crack prevention cri- be remedied. Design recommenda-
velopment length. Thus, recommenda-
teria within the transfer zone. Most of tions for pretensioned members gov-
tions made in this paper will need
the studies, however, have recom- erned by web-shear cracking are avail-
re-evaluation as additional data be- able in the UTA report. 16
mended expressions similar to the cur-
come available.
rent ACI/AASHTO expression, with The UTA study also established the
modifications to either or both of the importance of preventing cracks in the
Transfer Length transfer and flexural bond lengths. debonded and transfer zones of mem-
The current ACI/AASHTO transfer A study conducted at UTK con- bers containing debonded strands. The
length equation is based on data that cluded that the current ACI/AASHTO UTA criterion provides a rational ap-
are inappropriate for current practice. expression should be modified by proach for development of debonded
This expression underestimates the multiplying the flexural bond length strands, provided that in addition to
mean transfer length for the Grade 270 by a factor 1.5. Unfortunately, there preventing cracks in the debond/
low-relaxation strands used for most are inconsistencies in the reported data transfer zone, there is strand embed-
contemporary pretensioned concrete from the UTK study that cast doubt on ment from the critical section to the
members. The expression recom- the basis for this recommendation. end of the debond zone equal to or in
mended in the present study is the The FDOT has recommended an ex- excess of their development length.
same as recommended in an earlier pression that abruptly modifies both These criteria should replace the cur-
study by FDOT. The recommended the transfer and flexural bond lengths rent code requirement of doubling the
expression is: depending on the span-to-depth ratio development length for debonded
and member type. The rationale for strands that terminate in regions of
L _ fsidb ( 17 ) such abrupt changes is questionable, flexural tension under service loads.
t- 3
and it is recommended that the FDOT The general conclusion drawn in the
For strands, either straight or criteria not be adopted. UTA study was that seven-wire pre-
draped, that end in the upper one-third McGill University researchers have tensioned strands could be fully devel-
of member depth and have 12 in. (305 recommended an expression that mod- oped by preventing cracks within the
mm) or more of concrete cast beneath, ifies both the transfer and flexural transfer zone. The UTA researchers
it is recommended that the transfer bond length depending on concrete recommend replacing the strand de-
length be taken as 1.3 times the value compressive strength. The recom- velopment length expression in cur-
calculated by Eq. (17). Strands ending mended equation is consistent with rent codes with crack prevention crite-
in other regions need not be multiplied data from the McGill tests, but differs ria. This conclusion is questioned in
by 1.3. Recommendations apply to significantly from results from other the present study from the standpoint
Grade 270, seven-wire, low-relaxation, recent studies. Better understanding of that it is generally unconservative and
uncoated strands used in pretensioned the influence of concrete strength on unproven for all types of pretensioned
members with normal weight concrete bond performance should be provided concrete applications.
having a compressive strength at re- by tests that are in progress. Expressing development length as
lease of 3500 psi (24.2 GPa) or higher, The recommendation from the Pur- the sum of transfer and flexural bond
and a specified compressive strength due University study for members lengths is fundamentally sound and

March-April 1995 97
conceptually simple. Adopting the Eq. (18) be increased by a factor 1.3. 3. Hanson, N. W., and Kaar, P. H.,
UTA crack prevention criteria would For strands ending in other regions, the "Flexural Bond Tests of Pretensioned
require a set of equations specific to 1.3 multiplier need not be applied. Prestressed Beams," ACI Journal,
the support conditions and loadings on V. 55, No.7, 1959, pp. 783-803.
the pretensioned concrete member. 4. Tabatabai, H., and Dickson, T. J.,
ACKNOWLEDGMENTS "The History of the Pretensioning
The current ACI/AASHTO approach
This paper stems from a study the Strand Development Length Equa-
is consistent with the code treatment
author conducted as a visiting research tion," PCI JOURNAL, V. 38, No. 6,
of "attainable average bond stress over November-December 1993, pp. 64-75.
the length of embedment" (see Ref. 1, engineer at the Turner-Fairbanks
5. Martin, L. D., and Scott, N. L., "De-
p. R-181) that has been successfully Highway Research Center (TFHRC)
velopment of Prestressing Strand in
applied to nonprestressed reinforce- of the Federal Highway Administra-
Pretensioned Members," ACI Jour-
ment for many years. There is merit to tion (FHWA) while on leave from the nal, V. 73, No. 8, August 1976, pp.
retaining a concept that has been a part Virginia Military Institute (VMI). Fi- 453-456.
of the ACI Code for 30 years. Modifi- nancial support provided by the 6. Anderson, Arthur R., Anderson,
cations to both the transfer and flexu- Wachtmeister Sabbatical Leave Pro- Richard G., Holmberg, Ake, Martin,
ral bond lengths are needed, but the gram at VMI and by FHWA is grate- Leslie D., Scott, Norman L., Ni-
basic concept should be preserved. fully acknowledged. Opinions ex- jhawan, Jagdish C., Sanderson, K. J.,
The current ACI!AASHTO develop- pressed in the paper are those of the Zia, P., and Mostafa, T., Discussion
author and do not necessarily reflect of Ref. 7, "Development Length of
ment length expression is generally un-
those of FHW A. Prestressing Strands," PCI JOUR-
conservative. Fortunately, strand devel- NAL, V. 23, No. 4, July-August
opment rarely governs the design of The cooperation and help provided
1978, pp. 97-106.
pretensioned concrete members. The by the Precast/Prestressed Concrete
7. Zia, P., and Mostafa, T., "Develop-
relatively few documented bond related Institute (PCI) is gratefully appreci- ment Length of Prestressing Strands,"
failures in pretensioned members is ated. Paul Johal, PCI's Research Di- PCI JOURNAL, V. 22, No. 5,
probably due to the infrequent occur- rector, visited TFHRC three times dur- September-October 1977, pp. 54-65.
rence of critical conditions, instead of ing the course of the project and 8. Cousins, T. T., Johnston, D. W., and
adequate safety in the design criteria. contributed significantly by providing Zia, P., "Transfer and Development
Pretensioned concrete beams failing documents, reviewing report drafts Length of Epoxy-Coated and Un-
in general bond and followed by ulti- and offering insight into related coated Prestressing Strand," PCI
mate flexural failure typically show research. JOURNAL, V. 35, No. 4, July-August
The author appreciates the time and 1990, pp. 92-103.
gradual, ductile failures. However,
effort from many individuals who re- 9. FHWA, "Prestressing Strand for Pre-
many bond related failures in test tension Applications - Development
specimens have shown a relationship viewed report drafts and offered sug-
Length Revisited," Memorandum,
with shear strength, and generally fail gestions for improvement. In many
Chief, Bridge Division, Federal High-
in a sudden mode without adequate cases, the reviewer's opinions differ way Administration, Washington,
warning. The possibility of such fail- significantly from those of the author D.C., October 1988.
ures demands conservative develop- and are not necessarily reflected in the 10. Russell, Bruce W., "Impact of High
ment length criteria, consistent with final report. Comments and sugges- Strength Concrete on the Design and
the strength design philosophy of cur- tions were provided by the following: Construction of Pretensioned Con-
rent codes. A conservative expression Susan Lane, Lloyd Cayes and Jim crete Bridges," PCI JOURNAL, V. 39,
for strand development recommended Hoblitzell of FHWA; John Dick, Paul No.4, July-August 1994, pp. 76-89.
in this study is the following: Johal and George Nasser of PCI; 11. Deatherage, J. H., Burdette, E. G., and
Basile Rabbat of PCA; Edwin Bur- Chew, C. K., "Development Length
dette of UTK; Ned Burns of UTA; and Lateral Spacing Requirements of
(18) Prestressing Strand for Prestressed
Julio Ramirez of Purdue University;
Concrete Bridge Girders," PCI JOUR-
Bruce Russell of the University of Ok-
where for general applications, the NAL, V. 39, No. 1, January-February
lahoma; and Mohsen Shahawy of 1994, pp. 70-83.
multiplier A is taken as (0.6 + 40ep,).
FDOT Structures Research Center. 12. Shahawy, M. A., Issa, M., and Batch-
For applications in which design
stress is calculated by the approximate elor, B., "Strand Transfer Lengths in
Eq. (18-3) of the ACI Code, an equiva- Full Scale AASHTO Prestressed Con-
REFERENCES crete Girders," PCI JOURNAL, V. 37,
lent expression (0.72 + 0.102{3 1/wp) is
I. ACI Committee 318, "Building Code No.3, May-June 1992, pp. 84-96.
required to compute A. In either case, A
Requirements for Reinforced Con- 13. Shahawy, M., and Batchelor, B.,
shall be taken greater than or equal to
crete (ACI 318-89)," American Con- "Bond and Shear Behavior of Pre-
1.0 and less than or equal to 2.0. For stressed AASHTO Type II Beams,"
crete Institute, Detroit, MI, 1989.
strands, either straight or draped, end- 2. AASHTO, Standard Specifications Progress Report No. 1, Structural Re-
ing in the upper one-third of member for Highway Bridges, Fifteenth Edi- search Center, Florida Department of
depth and having 12 in. (305 mm) or tion, American Association of State Transportation, February 1991.
more of concrete cast beneath, it is rec- Highway and Transportation Offi- 14. Shahawy, M.A., and Issa, M., "Effect
ommended that the length calculated by cials, Washington, D.C., 1992. of Pile Embedment on the Develop-

98 PCI JOURNAL
 ment Length of Prestressing Strands," chorage Zone of Prestressed Concrete 27. Buckner, C. D., Shahawy, M. A., and
PCI JOURNAL, V. 37, No.6, Novem- Members," Draft Bulletin of the CEB, Issa, M., Discussion of "Effect of Pile
ber-December 1992, pp. 44-59. Delft University of Technology, Delft, Embedment on the Development
15. Lane, S. N., "Development Length of The Netherlands. Length of Prestressing Strands," PCI
Prestressing Strand," Public Roads- 21. Balazs, G. L., "Transfer Lengths of JOURNAL, V. 38, No. 5, September-
A Journal of Highway Research and Prestressing Strand as a Function of October 1993, pp. 86-88.
Development, Federal Highway Ad- Draw-in and Initial Prestress," PCI 28. Russell, B. W., and Burns, N. H.,
ministration, V. 54, No. 2, September JOURNAL, V. 38, No. 2, March- "Development Length and Flexural
1990,pp.200-205. April1993, pp. 86-93. Bond Behavior of AASHTO-Type
16. Russell, B. W., and Burns, N. A., 22. Loov, R., "A General Equation for the Girders With Fully Bonded and Blan-
"Design Guidelines for Transfer, De- Steel Stress for Bonded Prestressed keted Strands," Technical Memo
velopment and Bonding of Large Di- Concrete Members," PCI JOURNAL, 1210-2, Center for Transportation Re-
ameter Seven-Wire Strands in Preten- V. 33, No. 6, November-December search, University of Texas at Austin,
sioned Concrete Girders," Research 1988, pp. 108-148. Austin, TX, March 1991, 20 pp.
Report No. 1210-SF, Center for 23. Buckner, C. D., "An Analysis of 29. Stocker, M. F., and Sozen, M.A., "In-
Transportation Research, University Transfer and Development Lengths vestigation of Prestressed Concrete
of Texas at Austin, Austin, TX, 1993, for Pretensioned Concrete Structures," for Highway Bridges, Part VI: Bond
286 pp. Report No. FHWA-RD-94-049, Fed- Characteristics of Prestressing
17. Abdalla, 0. A., Ramirez, J. A., and Lee, eral Highway Administration, Wash- Strand," Bulletin 503, University of
R. H., "Strand Debonding in Preten- ington, D.C., December 1994, 108 pp. Illinois, Urbana, Illinois, 1970.
sioned Beams - Precast Prestressed 24. Deatherage, J. J., and Burdette, E. G., 30. Anderson, A. R., and Anderson, R. G.,
Concrete Bridges With Debonded "Development Length and Lateral "An Assurance Criterion for Flexural
Strands - Simply Supported Tests," Spacing Requirements of Prestressing Bond in Pretensioned Hollow Core
Part 2, Final Report FHW AI Strand for Prestressed Concrete Units," AC/ Journal, V. 73, No. 8,
INDOT/JHRP-92, 1993,228 pp. Bridge Products," PCI Report, Trans- August 1976, pp. 457-464.
18. Cousins, T. E., Francis, L. H., portation Center, University of Ten- 31. Russell, B. W., and Burns, N. H.,
Stallings, J. M, and Gopu, V. K. A., nessee, Knoxville, September 1991, "Static and Fatigue Behavior of Pre-
"Spacing and Cover Requirements of 127 pp. tensioned Composite Bridge Girders
Epoxy-Coated Prestressing Strands in 25. Deatherage, J. H., Chew, C. K., and Made With High Strength Concrete,"
Unconfined Sections," PCI JOURNAL, Burdette, E. G., "Behavior of Pre- PCI JOURNAL, V. 38, No. 3, May-
V. 38, No. 5, September-October stressed AASHTO Girders Under June 1993,pp.116-128.
1993, pp. 76-84. Static Loading," Preprint No. 0870, 32. Ralls, M. A., Ebonies, L., and
19. Mitchell, D., Cook, W. D., Khan, A. Transportation Research Board, Na- Panache, J. J., "The New Texas U-
A., and Tham, T., "Influence of High tional Academy of Sciences, Wash- Beam Bridges: An Aesthetic and Eco-
Strength Concrete on Transfer and ington, D.C., January 1993, 23 pp. nomical Design Solution," PCI JOUR-
Development Length of Pretensioning 26. Shahawy, M., "An Investigation of NAL, V. 38, No. 5, September-
Strand," PCI JOURNAL, V. 38, No.3, Shear Strength of Pretensioned Con- October 1993, pp. 20-29.
May-June 1993, pp. 52-66. crete AASHTO Type II Girders," 33. PC/ Design Handbook: Precast and
20. den Uijl, J. A., "Background CEB-FIP Structures Research Center, Florida Prestressed Concrete, Fourth Edition,
MC90 Clauses on Anchorage and Department of Transportation, May Precast/Prestressed Concrete Institute,
Transverse Tensile Actions in the An- 1993, 132 pp. Chicago, IL, 1992.

March-April 1995 99
 APPENDIX A- STRAND DEVELOPMENT LENGTH
BY DIFFERENT CRITERIA
A problem example is useful in comparing development AASHTO Specification Prior to 1973
lengths calculated by various criteria. Consider a rectangu- In 1973, AASHTO adopted the development length equa-
lar pretensioned concrete beam 12 in. (305 mm) in width, tion that was included in the 1963 ACI Code. Before adop-
32 in. (813 mm) in overall depth, and 28 in. (711 mm) in tion, AASHTO simply used the minimum embedment re-
effective depth. Data for the example are: effective pre- quired to develop the breaking strength of Grade 250 strand,
stress of 160 ksi (1.1 GPa); strand stress immediately after which was calculated as 134 in. (3.4 m) for 1iz in. (12.7 mm)
transfer of 180 ksi (1.2 GPa); concrete strength at release of diameter strand by Hanson and Kaar.
4000 psi (27.6 MPa); and specified compressive strength of
5000 psi (34.5 GPa).
The minimum number of 1/z in. (12.7 mm) diameter Current ACI/AASHTO Method
strands that furnish a moment strength in excess of 1.2Mcr is The development length equation proposed by ACI Com-
three. The maximum number of 1iz in. (12.7 mm) diameter mittee 323 is currently included in both the ACI 318 and
strands that provide ductile failure is 12. The corresponding AASHTO Specifications. From this equation:
strand stress at failure ranges from 242 ksi (1.7 GPa) with 12
strands to 263 ksi (1.8 GPa) for three strands, as approxi-
mated by Eq. (18-3) of the ACI Code. Sample calculations
Ld =(ips-~ ise )db (A3)

below are for the case using three strands. Results for other = [263 -0.67(160) ]0.5
cases are shown in Fig. 9.
= 78 in. (2.0 m)
In the sections to follow, the embedment length required
to develop the strand stress of 263 ksi (1.8 GPa) is calcu-
lated by current and proposed development length criteria. Martin and Scott Equation
The equations proposed by Martin and Scott are curves of
Hanson and Kaar Criteria best fit to the Hanson and Kaar data. Using the equation for
1
The Hanson and Kaar criteria were developed from tests iz in. (12.7 mm) diameter strand gives:
on Grade 250 strand. For comparison purposes, their local
bond stresses will be assumed applicable to the Grade 270 Ld = (di0.39)[fps- 1351db 116] (A4)
strands of the example. = (0.510.39)[263 - 13510.5<116)]
Transfer length is based on a bond stress of 400 psi = 143 in. (3.63 m)
(2.8 MPa), which applies to the actual strand perimeter [i.e.,
2.09 in. (53.3 mm) for a 1iz in. (12.7 mm) diameter strand]. Zia and Mostafa Equation
Thus, to transfer an effective prestress of 160 ksi (1.1 GPa)
in the strands requires: The development length equation proposed by Zia and
Mostafa is more conservative than the current ACII
AASHTO equation. This equation gives:
(Al)
(A5)
= 160 X 0.1531(0.4 X 2.09)
=29 in. (737 mm) = (1.5 X 180 I 4.0)0.5- 4.6+ 1.25(263 -160)0.5
Hanson and Kaar provide local bond stresses in a curve in = 94 in. (2.4 m)
Fig. A-1 of their paper. This figure is reproduced as Fig. 5
of this paper. Using linear interpolation, the average bond
stress over the first 25 in. (635 mm) from the end of the Proposed UTK Equation
transfer zone is approximated for 5 in. (127 mm) incre- The expression proposed in the UTK final report gives:
ments. This results in successive values of 225, 125, 80, 65,
and 55 psi (1550, 863, 552, 449, and 380 kPa). The average
bond stress beyond 25 in. (635 mm) is approximated as
Ld = ~e db + 1. 42( ips -he )db (A6)
50 psi (345 kPa). On this basis, the development length is: = (160 I 3)0.5 + 1.42(263-160)0.5
= 100 in. (2.54 m)
Ld = 4 + [(ips- fse )Aps -Suo;!( ub,ave ); ] I (0. 05uo )+ 25
(A2) FOOT Proposal
= 29 + [(263- 160)0.153- (5)(2.09)(0.250 + 0.080 The FDOT proposal provides significantly different results
+ 0.065 + 0.055)]1(0.050 X 2.09) + 25 for piles, slabs and girder sections. For the beam of the ex-
= 146 in. (3.7 m) ample problem, the data are assumed such that kb = 2; thus:

100 PCI JOURNAL

 (A9)
L = fsedb
(A7) t 2
=160x0.5/2
= [(180 I 3)0.5 + (263 -160)0.5]/ (2 X 0.250) = 40 in. (1.0 m)
=163in. (4.lm)
The cracking moment at the end of the transfer zone is:

McGill University Equation Mer= l/(h/2)lfbot +!,.) (AlO)

= (32768/16)(0.62 + 0.53)/12
The equation proposed by McGill University researchers
= 196 kip-ft (270 kN-m)
expresses development length in terms of concrete strength.
This gives: If the design bending moment 40 in. (1.01 m) from the
member end is less than 196 kip-ft (270 kN-m), and web
L _ ( fsidb ) / 3 + (f _f, )d /4.5 (AS) shear cracking is not a problem, the UTA criterion predicts
d - 3 IJ fci ps se b V J;
that strands can fully develop regardless of strand stress.
180
=[ I 3
3 lj 4.0
45
+(263-16o)o.5 / · ]
\ 5
Proposed Development Length Equation
= 75 in. (1.9 m)
Using the development length equation proposed in the
current study involves for this case:
UTA Criterion

coP=(~~)(;; J
The UTA study recommended a different approach to
strand development than calculating a development length. (All)
The UTA approach is summarized:
" ... to prevent anchorage failures, beams should be de- = (3x0.153)(263)
12x28 5
signed so that no concrete cracks will propagate
through the transfer zone of a pretensioned strand. This =0.072
observation is comprehensive for all sizes of preten-
sioned strand, for all pretensioned applications, and for Thus, Eq. (18) gives:
both fully bonded and debonded strands." (see Ref.l6,
page 209)
Ld = fsidb +[0.72+ 0.102/31 )Vps- fse)db (A12)
Thus, for the example problem, the UTA criterion re- 3 coP
quires that the calculated flexural stress within the transfer
zone not exceed the modulus of rupture plus effective pre- = 180x0.5 +(0. 72 + 0.102x0.8)( 263 _ 160 )0 . 5
stress at the point of consideration. The transfer length rec- 3 0.072
ommended by UTA is: = 125 in. (3.2 m)

March-April1995 101
 APPENDIX 8 - EXAMPLE PROBLEMS

~ 121n ~

.5
~
A
TJ..

ExamPle 1: Inverted Tee

r b.= 381n 1

\:1 - A,. • 0.459 ln

2
~
1:1 u

Example 2: Double Itt

.. . 181n

I
-tl ~ 21n

~ ~...··-a_strands
, 1_/2_1_n_·_. (typ)
• •
Example 3: Squara Pile Example 4: AASHTO
Section ComPOsite Girder
Note: 1 In·• 25.4 rom

Fig. 8 1. Cross sections of pretensioned beams in example problems.

To illustrate the proposed development length equation,

fps =Jp{1- ;~ (Pp;; Jl
(B1)
consider the typical pretensioned concrete members shown
in Fig. B 1. Data common to each member include: normal
weight concrete with a specified compressive strength of
5000 psi (34.5 MPa); 1/z in. (12.7 mm) diameter Grade 270 = 270 { 1 - 0. 28 [( 10 X 0.153 )( 270 )]}
0.8 12x21 5
low-relaxation strands; strand stress immediately after re-
lease of 180 ksi (1.24 GPa); and an effective strand stress of = 239 ksi (1.65 GPa)
160 ksi (1.1 GPa).
The corresponding steel index is:
Example 1: Inverted Tee
As a first example, Eq. (18) is used to calculate the devel-
opment length for the inverted tee section shown in Fig. B 1. wP = : ; ( ; ; J (B2)
The width of compression face is 12 in. (305 mm) and the
prestressed reinforcement consists of ten 1/z in. (12.7 mm) = (10 x 0.153) 239
diameter strands positioned such that dP = 21 in. (533 mm). 12 X 21 5
Eq. (18-3) of ACI 318 is applicable and gives: = 0.29"" 0.36/3,

102 PCI JOURNAL

 Thus, from Eq. (18): Thus, by Eq. (18):

(o. 72 + 0.1()2/31 ]Vps - fse )db

Ld = fsidb + (B3) Ld = ~~~!z_ +A {!ps - fse )db (B8)
3 (J)p
18
= 0xO.S +(2.0)(266-160)0.5
x · + (o. 72 + Q.l ~><9~~)(239 -I60)o.s
180 0 5 0 3
=
3 0.29 = 136 in. (3.45 m)
= 69 in. (I. 75 m)
This represents an extreme case in which the strand strain
For this example, the calculated strand stress is close to at ultimate exceeds the minimum guaranteed elongation of
the yield stress and Eq. (18) gives a development length that 3.5 percent. In this case, the development length is roughly
is practically the same as the value computed by the current 1.7 times the value that would be computed using the cur-
ACIJAASHTO equation. rent ACIJAASHTO expression. (Note that the factor of 2.0
applies only to the flexural bond length and not to the trans-
Example 2: Double Tee fer length.)
For the double tee shown in Fig. B1, the effective width
for each stem is taken as 38 in. (965 mm), the flange thick- Example 3: Square Pile Section
ness is 2 in. (51 mm) and each stem contains three 1h in. Consider the 18 x 18 in. (457 x 457 mm) pile section pre-
( 12.7 mm) diameter strands positioned such that dP = 17.5 in. stressed with eight symmetrically placed 1/z in. (12.7 mm)
(445 mm). Assuming, initially, that the depth of compres- strands, as shown in Fig. B I. In this case, part of the pre-
sion zone is less than the flange thickness, the ultimate stressed reinforcement is in the compression zone and the
strand stress can be approximated using Eq. (18-3) of ACI approximate Eq. (18-3) in the ACI Code does not apply.
318. For this case: The strand stresses must be determined by a strain compati-
bility analysis. If the axial force on the section is negligible,
(B4) the location of the neutral axis at flexural failure can be
found by trial-and-error to be 4.1 in. (104 mm) below the
compression face. This value was found using the principles
= 270{ 1 - 0.28 [( 3 x0.153)(~?9)]} of ACI 318 and an idealized stress-strain curve for Grade
0.8 38x17.5 5 270 low-relaxation strand. 33 For this location:
= 266 ksi (1.84 GPa)
(B9)
Checking the depth of compression zone: = (--1_()2__) + ( 8 X 0.153 X 160) + ( 16- 4.1)0.003
28500 4030xl8x18 4.1
Apsfps
a=~---'-- (B5) = 0.0145
0.85J;b
266x3x0.153 Thus,
0.85x38x5
0.04
= 0. 76 in. (1. 93 m) (BIO)
ips= fpu- cps -0.007
The calculated depth of compression zone is less than the
= 270- ___O_.Q±__
flange thickness; therefore, Eq. (18) is satisfactory. The pre- cps -0.007
stressed steel index is:
= 265 ksi (1.83 GPa)
= Aps (fps)
OJ
p bd t: (B6) The strain compatibility analysis provides the strand
strain, cps• so the flexural bond length multiplier is:
= (3x0.1~~)(?66)
38x17.5 5 A= 0.6 + 40cps (Bll)
= 0.6 + 40 X 0.0145
=0.037
= 1.18
In this case, the flexural bond length multiplier is:
From which:

A=( 0.72+Q}~~pj_) (B7)

Ld = fs~d!J_ + A(!ps - fse )db (B12)

=(0.?2+ 0.102x0.8) = 180xO.S +(1.18)(265-160)0.5

0.037 3
= 2.9; use 2.0 = 92 in. (2.3 m)

March-April 1995 103

 Most piles carry significant axial forces in addition to fpsAps (B14)
bending moments. These forces must be included in a strain a = __._____.___
0.85J:b
compatibility analysis. For example, if the pile in this ex-
ample had 100 kips (445 kN) axial compression force act- 264x22x0.153
ing simultaneously with the moment, the neutral axis at fail- 0.85x3x96
ure would be 5.5 in. (140 mm) below the compression face; = 3.63 in. (92 mm)
the strand strain on the tension side would be 0.011 at fail-
ure; the controlling strand stress would be 261 ksi (1.8 GPa) The depth of compression zone is less than the slab thick-
and would require a development length of 84 in. (2.1 m) ness; therefore, AASHTO Eq. (9-17) is satisfactory. The
by Eq. (18). prestressed steel index is:
Because piles are horizontally cast, the strands in the top
row will terminate in the upper one-third of member depth
and contain more that 12 in. (305 mm) of concrete cast
wP = ~; (;;) (Bl5)

below. Thus, the development length should be increased by

=(22x0.153)264
a 1.3 multiplier for these strands.
96x46 3
=0.067
Example 4: Composite AASHTO Girder
The flexural bond length multiplier is:
For the composite AASHTO girder shown in Fig. B 1 com-
pression is resisted by the 3000 psi (20.7 MPa) cast-in-place
deck slab. The effective width of slab is 96 in. (2.43 m) and
the twenty-two 1/z in. (12.7 mm) diameter strands are posi-
A= [ 0. 72 + 0.1~~/J, l (B16)

tioned such that dP = 46 in. (1.17 m). Using Eq. (9-17) of

=(0. 72 + 0.102x0.85)
AASHTO gives: 0.067

!,, = t:[ 1- ;, kjj)l (B13)

=2.01; use 2.0

Thus, Eq. (18) gives:

= 270 { 1 - 0.28 [( 22 X 0.153)(270)]} Ld = fs~db + A-(!ps - fse )db (Bl7)

0.85 96x46 3
180 0 5
= 264 ksi (1.82 GPa) = X · + (2.0)(264 -160)0.5
3
Checking the depth of compression zone: = 134 in. (3.4 m)

104 PCI JOURNAL

 APPENDIX C - NOTATION

ACI AASHTO
Symbol Notation Notation Description
--

t~
A - - Cross-sectional area
--~
--- ~

Af'S Af>I As * Area of prestressed re inforcement in tension zone

~

'-
b b b Width of compression face of member (

~
b, - - Effective width of compression edge in composite beam
B - - Bond coefficient (see Table I)
"'
·-
m
db db D Nominal diameter of prestressing strand

fo ,£ - ~ l-~=--
I'
-- '
Effective depth from compression face to center of gravity
dl' dl' d
of prestressed reinforcement in tension zone
Concrete stress in bottom stress fiber due to effective prestress
--
Concrete compressive strength immediatel y after prestress (see Table I)
);<"

-
f~
f,M
f ci
;c
f~ _
fci
f~
f ci _i
Specified compressive strength of concrete
Compressive strength of concrete at time of initial prestress
fps _l J;rs , _f....,* Stress in prestressed reinforcement at nominal strength
J; ..
~)/I J,m ... Specified te nsi le strength of prestressing strands

=r
};. !, !, Modulu s of rupture of concrete
f,_ f oe f ,_
... Effective stress in prestressed reinforcement after all losses
-
1

- -
" Stress in prestressed reinforcement at time of initial prestress
f ,i
! (immediately after release in a preten sion ed member)
t·c
~

F,
h
-

h
-

h
-~-
Force directed along longitudinal axis
Overall thickness of member -
--
I I I Moment of inertia of section
---
--
kb - - Constant used in FOOT development length expression
- ~ --
L -

+- Span of member
-

Lb - - Flexural bond length

-
Ld ld ld Development length
L, - - Transfer length
·-c
Mer M er M er * "' Cracking moment
M, M, M,
-~

Nominal moment strength ~

-
M., M, M,
~

Factored moment at section

--
- ~

llu I - - Perimeter of strand , 4rrdb/3 for a seven-wire strand

fl b.m ·e - - _ Average flexural bond stress per unit area
u,_m·e __,
- - Average transfer bond stress per unit area
u, -
-I
- Bond coefficient (see Table I)
..
- Factored load per unit length
WI/ lt'/1 wu -

as a9 aJO Coefficients in expression for transfer length (see Table I)

f-
/31 /31 /31 Ratio of depth of equivalent rectangular stress block to depth of neutral axis
)j, r;, )j,* Constant for strand type (0.28 for Grade 270 low-relaxation strand)
-
tlJ;,s - - Change in stress in prestressed reinforcement
de -+ ~-
- -
~
Change in strain in prestressed reinforcement
I Strain in concrete at center of gravity of prestressed reinforcement
Ece - -
due to effective prestress
---C~

f ps - - Strain in prestressed reinforcement at nominal strength

-
- - Strain in prestressed reinforcement after all losses
-
€se
- -
Strain induced in prestressed reinforcement from decompression
E_w
... to nominal strength __
,,__,
'A - - Multiplying factor applied to flexural bond length
-- -
'
~

J.l m•e - - Average bond stress in FOOT equation for development length
7r - - Constant= 3.14 I59
¢ ¢
·- ¢
-
Strength reduction factor
---
_,
~r-

Ratio of prestressed reinforcement to effective depth

Pp Pp p*
times width at compression face
~

WI' WI' p*fw*lf~ Reinforcement index, p1J;,.JJ:

March-April 1995 105