Académique Documents
Professionnel Documents
Culture Documents
INTRODUCTION
In ACI-318-08,1 the allowable compressive stress at
prestress transfer was increased from 0.60fci′ to 0.70fci′ near the Fig. 1—Prestressing strand profiles and potential locations
ends of prestressed concrete members. At other critical locations, of maximum release stress.
such as midspan or the hold-down points, the allowable stress is
still 0.60fci′. In the AASHTO Load and Resistance Factor
Design (LRFD) Bridge Design Specifications,2 the allowable member. Also, little information has been published that
compressive stress at release is 0.60fci′ at all sections. quantifies potential benefits of increasing the compressive
The maximum compressive stress at release can occur in stress at release. The information from this study will aid
the end regions, near midspan, or at any other section of the code developers in appropriately weighing the benefits and
pretensioned beam depending primarily on the profile or the weaknesses of increasing this allowable stress.
debonding of the prestressing strands. If only straight, fully-
bonded strands are used, the maximum compressive stress at HISTORICAL BACKGROUND
release will always be at the transfer section near the ends of The first U.S. code provisions for prestressed concrete
the beam (Fig. 1(a)). If some of the straight strands are members were adopted by the American Association for
debonded, the location of maximum stress can shift to the State Highway and Transportation Officials (AASHTO) in
section where all strands are bonded. If harped strands are 1961 and by the American Concrete Institute (ACI) in
used in combination with straight strands, then the compressive 1963.3,4 The code provisions in each standard were based on
stress at the hold-down point and near the ends of the beam can the following two documents: the “Criteria for Prestressed
be very similar (Fig. 1(b)). For post-tensioning applications Concrete Bridges” published by the Bureau of Public Roads
in which strands are draped, the maximum compressive in 1954 and the “Tentative Recommendations for
stress at release can occur at practically any section along the Prestressed Concrete” published by Joint ACI-ASCE
length of the member (Fig. 1(c)). Committee 323 in 1958 (later, ACI Committee 423).5,6 In
In this paper, benefits of increasing the release stress in both documents, the recommended allowable release stress
compression where the maximum stress occurs anywhere in compression was 0.60f ci′ for pretensioned members and
along the member, are discussed. Afterward, the results of an 0.55f ci′ for post-tensioned members. While the origin of
experimental research study are analyzed, in which a negative these values was not directly stated, two references allude to
outcome of increasing the release stress at midspan was observed. their development. In a paper by E. L. Erickson, in which the
“Criteria for Prestressed Concrete Bridges” specification
was described, the opinions of several “authorities” on the
RESEARCH SIGNIFICANCE
Over the last decade, a substantial amount of research and
discussion have been published related to increasing the ACI Structural Journal, V. 107, No. 1, January-February 2010.
MS No. S-2008-184 received June 6, 2008, and reviewed under Institute publication
allowable release stress in compression. None of the policies. Copyright © 2010, American Concrete Institute. All rights reserved, including the
previous investigations focused on the effect of increasing making of copies unless permission is obtained from the copyright proprietors. Pertinent
discussion including author’s closure, if any, will be published in the November-December
the allowable stress on the flexural performance of the 2010 ACI Structural Journal if the discussion is received by July 1, 2010.
always applicable. The release stress in compression must prestress loss procedures present in the literature. The
govern the design of the girder. For a TxDOT Type C girder, comparison of the measured and estimated cracking loads
this requirement was restrictive for short-to-moderate span was used to evaluate the impact of increasing the allowable
lengths when the tensile stress limit at release governed the compressive stress at release near midspan on the live-load
design. Also, it should be emphasized that production benefits performance of pretensioned girders.
and design benefits can contradict one another. Production
benefits of increasing the allowable release stress are maximized Materials
when the section design and the span length are unchanged. It Five concrete mixture designs were used to fabricate the
seems plausible, at least in the long term, that the benefits of specimens in the experimental program. Conventional materials
increasing the allowable release stress will be used in the from Texas were used. Specific details of the concrete mixture
design phase. In making a decision to alter the release stress designs can be found elsewhere.18
in compression, it is essential to weigh these benefits and The prestressing strand used in all of the test specimens
their applicability with any negative implications associated was 1/2 in. (12.7 mm) diameter 270 ksi (1862 MPa) low-
with increasing the allowable stress. relaxation strand. All of the mild steel had a nominal yield
In 2006, Spencer and Chen17 compared the designs of strength of 60 ksi (414 MPa).
several standard Pennsylvania Department of Transportation
(PennDOT) bridge girders using an allowable release stress Test specimens
of 0.60f ci′ and 0.70f ci′ . Potential design benefits of increasing The 24 Series I beams consisted of scaled rectangular, T,
the release stress were focused on in their study. For adjacent and inverted-T sections designed to represent standard
and spread box beams and simply supported AASHTO I- TxDOT I, double-T, and U-girders, respectively.12 The
beams, few design advantages of using 0.70f ci′ were observed. maximum nominal compressive stress at the ends of the
The design of these members was typically governed by Series I beams ranged from 0.46f ci′ to 0.91f ci′ . Dimensions
PennDOT’s strict tensile stress limit at release of 3 f ci ′ along of the scaled section types are illustrated in Fig. 3 and summa-
the member. As such, increasing the allowable release stress in rized in Table 2. All of the beams were 15 ft (4.6 m) in length.
compression did little to improve their design. For Series I beams were fabricated by Castro et al.12 at the
continuous I-beams and simply supported bulb-tees, Ferguson Structural Engineering Laboratory at The University
however, design benefits were noted. Additional strands of Texas at Austin, Austin, TX. Prestressing strands were
were placed in the continuous I-beam, permitting a reduction stressed with a single-strand stressing jack in a straight
of its cross section. Similarly, additional strands were added profile. The stress in the strand was inferred from electrical
to the bulb-tee girders, allowing an increase in their maximum strain gauges attached to the strand and was confirmed with
span length by approximately 5%. The results of this study a pressure reading in the hydraulic ram. Concrete materials
agreed favorably with the previous design example. were batched at the laboratory and placed in the formwork.
Match-curing technology was used to match the temperature of
EXPERIMENTAL INVESTIGATION 4 x 8 in. (102 x 203 mm) concrete cylinders with the temperature
The experimental program consisted of two series of static of the hydrating beams. Thermocouples in beams were
tests. In Series I, 24 scaled rectangular, T, and inverted-T girders positioned near the geometric centroids of beam sections.
15 ft (4.6 m) in length were tested. In Series II, 12 full-scale At release, the strands were flame-cut in a symmetric pattern.
TxDOT Type A girders 40 ft (12.2 m) in length were tested. A more detailed account of the design and fabrication of the
The load due to first flexural cracking was experimentally Series I beams can be found elsewhere.12
evaluated in each test. The measured cracking loads were The 12 Series II beams consisted of full-scale TxDOT
compared to cracking loads estimated with the use of two Type A girders. The same nonstandard strand pattern was
used in all 12 specimens. Four of the 14 strands were harped Series II beams were fabricated at a Texas precast
to meet tension stress limits at the ends of the girders. The manufacturing plant in a small stressing bed. One beam
hold-down points were 5 ft (1.5 m) on either side of midspan. was constructed at a time. To achieve a targeted concrete
The only variable controlling the maximum compressive strength at release, approximately twenty-four 4 x 8 in.
stress at transfer was the compressive strength of the concrete at (102 x 203 mm) cylinders were cast with each beam. The
release, fci′ . The targeted concrete strength at transfer ranged cylinders were cured next to the specimen beneath a tarp
from 3900 to 5400 psi (26.9 to 37.2 MPa). The corresponding until they were tested according to ASTM C39-04. After
maximum compressive stress at transfer ranged from 0.75fci′ to initial set, two cylinders were tested every hour to map the
0.55fci′ . Dimensions and section properties of the Series II strength gain. When concrete strength was within 1000 psi
beams are illustrated in Fig. 3 and summarized in Table 3. All (6.9 MPa) of the targeted strength, two cylinders were tested
of the Series II beams were 40 ft (12.2 m) in length. more frequently until the average strength was as close to the
targeted value as possible. At this point, the strands were A load cell and a pressure transducer were used to measure
flame-cut in a symmetric pattern. the applied load. Both instruments provided consistent
results. Loads reported in this paper were measured at
Test procedure midspan by the load cell. The midpsan deflection was
The specimens were simply supported (Fig. 4 and 5) and recorded over the course of each test using either a string
were subjected to two concentrated loads using a hydraulic potentiometer or linear potentiometer. For the 24 Series I
specimens, four linear variable displacement transducers
ram at midspan and a stiff spreader beam. The constant
(LVDTs) and an electrical strain gauge attached to each
moment region was approximately 5 ft (1.5 m) in length for
prestressing strand measured applied midspan strains in the
all test specimens. Simple supports were achieved by section. For the 12 Series II specimens, only load and
sandwiching a round steel bar (ϕ = 2 in. [51 mm]) midspan deflection were monitored.
between two steel plates (6 x 8 x 1 in. [152 x 203 x 25 mm]). In each test, the measured cracking load was obtained
The bar at one support was allowed to roll freely. The bar at through visual inspection and examination of load-deflection
the other support was welded to the bottom plate. The load and load-strain plots. The load at which a change in stiffness
was increased incrementally to a level approximately 30% was apparent in these plots was termed the measured
higher than the cracking load, or until failure occurred. At cracking load. In general, cracking loads determined with the
each load step equal to or exceeding the cracking load, the load-deflection or load-strain plots were equal to or slightly
load was maintained, the beam was inspected, and the visible less than those determined through visual inspection. In
cracks were mapped on the beam. The Series I beams were some of the Series I beams, the roughened condition of the
tested at an age of approximately 3 years. The Series II concrete surface hindered the ability to detect first cracking.
beams were tested at an age of approximately 28 days. All The load at first cracking for these beams was obtained with
beams were stored outdoors until they were tested. load-deflection and various load-strain plots. A sample load-
ANALYTICAL INVESTIGATION
The loads required for first flexural cracking were
estimated using typical design methods. First, prestress
losses up to the time of the test were approximated with two
different prestress loss procedures. The two procedures
included the National Cooperative Highway Research
Program (NCHRP) Report 496 Detailed Prestress Loss
Method (2003) and the AASHTO LRFD Refined Loss of
Prestress Estimate (2005).19,20 AASHTO LRFD adopted the Fig. 6—Sample load-deflection plot for T2-91-5-T14.
recommendations of the NCHRP Report 496 in the 2005
specifications. As such, the two prestress loss procedures are
very similar. The main differences included the suggested
use of transformed section properties instead of gross section
properties and a different equation for the modulus of elasticity
of concrete in the NCHRP procedure. These procedures
were selected because they provide a loss estimate at any
time in the life of the girder, not just the total losses that
lump-sum methods provide. Because the Series II beams
were only 28 days old at the time of the test, this time-dependent
feature was required.
With the prestress loss estimates of each procedure, an
estimated cracking moment was calculated. To remain
consistent with each procedure, the same geometric properties
(gross or transformed) used in the loss calculations were also
used in the cracking moment calculations.
The estimated cracking moment, according to the
Fig. 7—Sample load-deflection plot for A73-T34.
AASHTO prestress loss procedure, is provided as Eq. (1).
The effective prestressing force is calculated as the initial
prestressing force minus losses due to elastic shortening,
Examples of the prestress loss calculations for each procedure
concrete creep, concrete shrinkage, and steel relaxation.
and the subsequent cracking moment estimate can be
Gross section properties are used throughout as is typically
found elsewhere.18
done by design engineers.
COMPARISON OF PREDICTIONS AND
I g ⎞ ⎛ P eff P eff e p y bg M g y bg
M cr - -------- + ----------------------- – --------------- + f r⎞
= ⎛ ------- (1) EXPERIMENTAL RESULTS
⎝ y bg ⎠ ⎝ A g Ig Ig ⎠ With the cracking moments from Eq. (1) and (2), the
corresponding midpsan load was calculated with statics.
The variables in Eq. (1) are defined in the Notation section This predicted cracking load was then compared to the
of this paper. measured cracking loads obtained from each beam test. To
The estimated cracking moment, according to the NCHRP facilitate the comparison, a percent difference formula was
prestress loss procedure, is provided as Eq. (2). Elastic shortening used to compare the two values (Eq. (3)). The estimated and
losses were automatically accounted for with the use of the measured cracking loads for all of the test specimens are
initial prestressing force and transformed section properties listed in Tables 4 and 5.
at release.19 Long-term losses due to creep, shrinkage, and
relaxation were accounted for with the net section properties, P measured – P predicted⎞
as per NCHRP Report 496. Accuracy CL = ⎛ -------------------------------------------------- × 100 (3)
⎝ P measured ⎠
I tt ⎛ P i ΔP LT P i e pti y bti ΔP LT e pnt y bnt where Pmeasured is the cracking load measured during static
M cr = ------- ------ – ------------ + ------------------------ – ------------------------------- –
y btt ⎝ A ti A nt I ti I nt test (kips or kN); and Ppredicted is the cracking load estimated
(2) analytically (kips or kN).
M g y bti
- + fr⎞
-------------- For each procedure, the value from Eq. (3) for all of the
I ti ⎠
test specimens was plotted versus the compressive stress at
release at the critical section (σb in Tables 4 and 5). The critical
The variables in Eq. (2) are defined in the Notation section section was defined as the location along the beam with the
of this paper. highest applied moment that was previously subjected to the
Measured cracking load, NCHRP Report 496 2003 AASHTO LRFD 2005
Specimen designation σb (x f ci′ )* kips (kN) kips (kN) % kips (kN) %
R1-52-1-T8 0.50 35 (156) 35.8 (159) –2.3 35.1 (156) –0.3
R1-52-1-T7 0.50 35 (156) 36.0 (160) –2.9 35.3 (157) –0.9
R1-50-1-T1 0.48 35 (156) 35.6 (158) –1.7 34.8 (155) 0.6
R1-49-1-T2 0.47 35 (156) 35.1 (156) –0.3 34.4 (153) 1.7
R1-46-1-T5 0.45 35 (156) 35.8 (159) –2.3 35.0 (156) 0.0
R1-48-1-T6 0.46 35 (156) 36.3 (161) –3.7 35.5 (158) –1.4
T1-68-2-T17 0.64 21 (93) 22.0 (98) –4.8 21.5 (96) –2.4
T1-62-2-T18 0.58 23 (102) 23.3 (104) –1.3 22.7 (101) 1.3
IT1-68-2-T20 0.66 55 (245) 53.8 (239) 2.2 53.0 (236) 3.6
IT1-73-2-T19 0.71 55 (245) 57.5 (256) –4.5 56.1 (250) –2.0
R3-75-3-T9 0.72 42 (187) 46.0 (205) –9.5 45.5 (202) –8.3
R3-78-3-T3 0.75 42 (187) 46.9 (209) –11.7 46.0 (205) –9.5
T2-79-3-T16 0.75 20 (89) 22.9 (102) –14.5 22.6 (101) –13.0
T2-86-3-T15 0.82 22 (98) 24.2 (108) –10.0 23.7 (105) –7.7
IT3-79-3-T21 0.77 63 (280) 67.4 (300) –7.0 65.5 (291) –4.0
IT2-76-3-T22 0.74 64 (285) 67.7 (301) –5.8 66.3 (295) –3.6
R3-78-4-T11 0.75 40 (178) 43.3 (193) –8.2 42.5 (189) –6.3
R3-83-4-T12 0.80 42 (187) 45.2 (201) –7.6 44.0 (196) –4.8
IT3-83-4-T24 0.81 58 (258) 63.9 (284) –10.2 61.3 (273) –5.7
R3-75-5-T10 0.72 40 (178) 44.9 (200) –12.3 43.7 (194) –9.3
R3-80-5-T4 0.77 42 (187) 46.4 (206) –10.5 44.9 (200) –6.9
T2-91-5-T14 0.87 20 (89) 22.0 (98) –10.0 21.4 (95) –7.0
IT3-79-5-T23 0.78 57 (254) 65.9 (293) –15.6 62.9 (280) –10.4
IT2-80-5-T13 0.77 58 (258) 65.3 (290) –12.6 62.9 (280) –8.4
*Linear-elastic stress calculation at prestress transfer at critical section.
Measured cracking load, NCHRP Report 496 2003 AASHTO LRFD 2005
Specimen designation σb (x f ci′ )* kips (kN) kips (kN) % kips (kN) %
A55-T25 0.55 62 (276) 64.3 (286) –3.7 60.6 (270) 2.3
A60-T26 0.60 63 (280) 63.5 (282) –0.8 59.7 (266) 5.2
A63-T27 0.63 63 (280) 63.6 (283) –1.0 60.0 (267) 4.8
A66-T28 0.66 63 (280) 63.8 (284) –1.3 60.5 (269) 4.0
A67-T29 0.67 60 (267) 62.5 (278) –4.2 58.6 (261) 2.3
A66-T30 0.66 59 (262) 63.1 (281) –6.9 59.4 (264) –0.7
A69-T31 0.69 60 (267) 62.6 (278) –4.3 58.9 (262) 1.8
A68-T32 0.68 58 (258) 62.8 (279) –8.3 59.0 (262) –1.7
A67-T33 0.67 63 (280) 63.0 (280) 0.0 59.5 (265) 5.6
A73-T34 0.73 57 (254) 62.8 (279) –10.2 59.5 (265) –4.4
A71-T35 0.71 63 (280) 63.1 (281) –0.2 59.8 (266) 5.1
A75-T36 0.75 60 (267) 62.6 (278) –4.3 59.2 (263) 1.3
*
Linear-elastic stress calculation at prestress transfer at critical section.
highest compressive stress at prestress transfer. For all of the the beams cracked at lower loads than predicted. The major
test specimens, the critical section was located directly difference between the two plots is the upward shift of the
beneath either of the two symmetric load points. Here, the data in the AASHTO plot with respect to the NCHRP plot.
compressive stress at release was slightly higher than at This shift is primarily due to the use of gross section properties
midspan of each specimen due to the effects of dead load. in the cracking moment equation of the former.
Comparison plots with predicted data, according to the
NCHRP prestress loss method and AASHTO prestress loss DISCUSSION
method, are given in Fig. 8 and 9, respectively. Data presented in Fig. 8 and 9 were analyzed, assuming
The same general trend exists in Fig. 8 and 9. As the that a cracking load prediction accuracy of ±5% was acceptable.
compressive stress at release increased above a certain limit, This level of accuracy seemed consistent with the techniques