ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 103-S75

Effect of Duct Type on Shear Strength of Thin Webs

by Aurelio Muttoni, Olivier L. Burdet, and Eckart Hars

A series of 16 specimens of concrete web panels 23.6 x 23.6 x 4.9 in. struts acting in compression and of stirrups acting in tension.
(600 x 600 x 125 mm) was tested in compression to investigate the The design shear strength is limited by the strength of either
effect of the presence of various types of post-tensioning ducts on or both components, and it is desired that the compressive
the strength of the shear-induced compression struts. Most panels strength of concrete be sufficient to avoid a brittle failure
were cast in the laboratory, but some were extracted from an actual
bridge girder, which had been previously loaded, allowing
mode. As Fig. 2 shows, the inclined concrete compression struts
investigation of the effect of web cracking on the ultimate strength. are crossed by the post-tensioning tendons, which decreases
The presence of a duct in a web, whether injected or not, decreases the their load-carrying capacity. When considering an inclined
compressive strength of the panel. This effect is most pronounced for cross section of a web (Fig. 3(a)), the presence of an empty duct
non-injected ducts, but is also much larger for injected plastic has the effect of deviating the compression field around the
ducts than for injected steel ducts. The effect of web cracking further void, which induces transverse tensile stresses in the immediate
decreases the strength of web panels containing post-tensioning vicinity of the tendon duct. At the other extreme, the presence
ducts, and this effect can be estimated using classical strength of a very stiff injected tendon duct (Fig. 3(b)) attracts a large
reduction formulas. part of the effort, which also induces transverse tensile
stresses, at a larger distance from the duct. In addition, if the
Keywords: post-tensioning duct; shear strength; testing; web.
surface of the duct is smooth, sliding can occur, increasing
the splitting effect. In reality, the stiffness of the combined
INTRODUCTION
system duct/injection grout/tendon is typically not very large,
Post-tensioning is a system for introducing a prestressing
force in a structure after the concrete has been cast and hardened.
This disposition has the advantage over flat-bed prestressing
in that there is no need for massive anchoring blocks, as the
tendons are directly stressed against the structure itself. It is
also the most practicable solution for the prestressing of cast-
in-place structures, and for ensuring continuity between
precast girders. To be able to stress the tendons, however,
they need to be separated from the concrete so that large
strains can be applied to them. This is usually done by
inserting the tendons in ducts that are later filled with an
injection grout or by a corrosion-inhibiting grease or wax. Fig. 1—Disposition of tendon ducts in web of bridge girders.
The presence of the duct in the web of a girder has an influence
on the strength of the section (Fig. 1). Whereas this effect is
generally negligible for the bending and axial compressive
strength of post-tensioned girders, it can be significant for
webs, especially in area of high shear (Fig. 2). This is true
whether the duct is left empty, filled with a soft material such as
grease, or filled by a stiff injection grout.

RESEARCH SIGNIFICANCE
The presence of ducts decreases the strength of webs in
concrete girders, and current design codes account for this
effect. Most test results available at the present time are for
corrugated steel ducts, which are increasingly being substituted
by high-density polyethylene (HDPE) ducts. This paper presents
information on the behavior of web girders with post-
tensioning ducts using steel or HDPE ducts. The replacement
of steel ducts by HDPE ducts decreases the strength of thin
webs. This paper also investigates the combined effect of the Fig. 2—Tendon duct crossing diagonal shear compression
presence of post-tensioning ducts and of cracking, based on struts.
tests performed on specimens extracted from real bridge girders.

INFLUENCE OF PRESENCE OF ACI Structural Journal, V. 103, No. 5, September-October 2006.

MS No. 05-247 received September 9, 2005, and reviewed under Institute publication
DUCTS IN CROSS SECTION policies. Copyright © 2006, American Concrete Institute. All rights reserved, including
In reinforced and prestressed concrete, shear is carried the making of copies unless permission is obtained from the copyright proprietors. Pertinent
discussion including author’s closure, if any, will be published in the July-August
after cracking by the combined action of inclined concrete 2007 ACI Structural Journal if the discussion is received by March 1, 2007.

ACI Structural Journal/September-October 2006 729

 losses during the tensioning of the cables as well as limiting
ACI member Aurelio Muttoni is Professor and Head of the Structural Concrete
Laboratory, Ecole Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland. fretting fatigue effects between the cable and the duct. In
He received his diploma and PhD in civil engineering from the Swiss Federal Institute addition, this type of duct is required to obtain electrically
of Technology in Zurich, Switzerland, in 1982 and 1989, respectively. His research isolated post-tensioning tendons that are sometimes neces-
interests include the theoretical bases of the design of reinforced concrete structures,
shear and punching shear, fiber-reinforced high-strength concrete, soil-structure sary to completely protect the post-tensioning system
interaction, and the conceptual design of bridges. from corrosion. The soft material (EHDPE ≈ 145 – 290 ksi
Olivier L. Burdet is a Senior Researcher and Lecturer at EPFL. He received his
[1000 – 2000 MPa]) that these plastic ducts are made of
diploma in civil engineering from EPFL and his PhD from the University of Texas at significantly changes the state of stresses around the injected
Austin, Austin, Tex., in 1983 and 1990, respectively. His research interests include the ducts and further lowers the strength of the web that contains
serviceability behavior of structures and the modeling of the behavior of concrete
structures using stress fields.
them, with compression struts tangent to the outer surface of
the duct. Its low friction coefficient adds to this effect.
Eckart Hars is a PhD candidate at EPFL. He received his diploma in civil engineering
from the Technische Universität, Hamburg-Harburg, Germany, in 2000. His research
interests include the shear behavior of post-tensioned bridge girders. STATE OF THE ART AND CURRENT
CODE PROVISIONS
Several researchers have investigated the phenomena
related to the presence of empty or injected tendon ducts in
the web of girders.1-9 Most tests were performed on panel
elements subjected to compression (Fig. 4) representing a
section of a web subjected to a diagonal compression field
induced by shear (Fig. 2). The main parameter for these
studies was usually the ratio δ.
Gaynor1 tested 66 concrete cylinders, with δ values
between 0.1 and 0.2, containing solid steel sections instead
of a duct and reported a strength reduction compared with
plain concrete cylinders.
Leonhardt2 investigated 52 panels, some with a lateral
eccentricity of the duct and a varying inclination of the duct
Fig. 3—Effect of presence of ducts in webs on compression with respect to the direction of the loading. It was found that
field (after Leonhardt2). both these parameters have little influence on the strength.
The presence of two ducts side-by-side in a panel leads in
average to a smaller strength reduction than the presence of
a single duct of a double diameter, provided that the spacing
between ducts is at least one duct diameter. The measurement
of the compressive strain distribution over the thickness of
the panel at the level of the tendon showed that strains
parallel to the load are larger near the duct than on the side
surface, regardless of whether the ducts are injected or not.
The proposed effective width formula (Eq. (1) and (2)) is a
linear function of the duct diameter, distinguishing between
injected and empty ducts.
For injected ducts
Fig. 4—Web panel with post-tensioning duct.
η D = 1 – 2--- δ (1)
so that the behavior is an intermediate case, closer to that of an 3
empty duct.
In addition, the presence of post-tensioning ducts in thin For non-injected ducts
webs also has some adverse effects on the quality of the
concreting in the immediate vicinity of the duct, with less ηD = 1 – δ (2)
dense concrete immediately underneath the duct and cracks
along the tendon path due to compaction. In the case of injection
of the ducts with cement grout, voids are also possible in the where the strength reduction factor ηD is the ratio of the
upper part of the duct, although this problem does not necessarily strength of the specimen with a duct to the strength of an
appear in zones of high shear in beams where the tendons equivalent specimen without a duct. In the case of Leonhardt and
usually have a larger inclination. several authors, this parameter is often expressed as an effective
Results from the literature have established that the effect of width factor for the web, which leads to the same result.
the presence of post-tensioning ducts in the webs can be Leonhardt also tested specimens with solid steel bars
significant, especially in cases where the ratio δ = Σ∅D /bw of instead of injected ducts. The results were not significantly
the total duct section to the overall width of the web is large1-9 different, showing that the stiffness of the injected tendon is
(δ > 0.2). This effect was neglected in early versions of design not a governing parameter.
codes dealing with post-tensioning, but was later included.10-15 Clarke and Taylor3 conducted a series of tests aimed at
In recent years, classical corrugated steel post-tensioning verifying Leonhardt’s proposal. They tested 64 panels and varied
ducts have been increasingly replaced by plastic ducts made the diameter and inclination of the duct as well as its stiffness.
of HDPE. These ducts have the advantage of lower friction The results generally confirmed Leonhardt’s provisions.

730 ACI Structural Journal/September-October 2006

 Chitnuyanondh,4 Campbell et al.,5 and Campbell and
Batchelor6 tested 52 panels. Chitnuyanondh4 reports that the
strength of panels with injected cavities does not significantly
differ from panels with injected steel ducts. The tested panels
contained one or two ducts placed on top of one another, with
spiral reinforcement around the ducts for 16 panels and without
spiral reinforcement for the remaining panels. No other
passive reinforcement was provided. Increasing the spacing
between the two ducts increases the strength of the panel.
When the spacing between ducts reaches one times the duct
diameter, the strength reduction compared with a specimen
with only one duct is negligible for injected ducts, but it
remains substantial for empty ducts. Placing a spiral around
the duct increases the strength, with a more pronounced
effect for empty ducts. Unreinforced panels with an empty Fig. 5—Concrete strength reduction factor η D for panels
duct and δ > 0.5 had a strength 30% lower than extrapolated with post-tensioning ducts (δ = Σ∅D / bw) and corresponding
from Leonhardt’s formula (Eq. (2)). During loading, they code provisions.
cracked over their whole length; the two remaining parts then
failed in buckling. The authors attributed these findings to the Table 1—Diameter correction factor k (Eq. (3)):
high duct diameter to panel thickness ratio, which was outside current and code provisions
of the parameter range of Leonhardt’s investigations. k
Rezai-Jorabi and Regan7 tested 15 panels and varied the
Code Year Empty Steel Plastic
duct diameter while keeping the panel dimensions constant.
AASHTO 2004 0.5 0.25
Their measurements of the strain distribution over the thick-
ness of the panel confirmed observations by Leonhardt.2 BS 5400-4 1990 1 0.67
Ganz et al.8 tested 14 panels with steel or HDPE ducts. CEB MC90 1993 1.2 0.5
The authors observed that the use of plastic ducts reduces the CEB MC78 1978 0.5 0.5
ultimate strength more than the use of steel ducts, although EC 2 2004 1.2 0.5 1.2
only by a small amount and in the same range as the test EC 2 1992 — 0.5
scatter. It must be noted that the duct diameter to panel thickness
ratio was only δ = 0.2 in this series.
As can be readily seen in Fig. 5, a significant decrease of
A few authors performed comparative tests on beams with
the strength of panel elements is observed for large values of
tendons (stressed or not) compared with beams without
δ. This decrease is stronger for empty ducts than for injected
tendons.4,7,9 The presence of inclined post-tensioning
ones. In most cases, the decrease indicated by current codes
tendons, causing a favorable compression of the section and
is in good accordance with experimental results, with the
a decrease of the shear force in the web by the amount taken
notable exception of the AASHTO code, which underestimates
by the inclination of the tendon, profoundly changes the
the effect of the presence of the ducts.
observed behavior and makes comparing structures with
post-tensioning and structures without post-tensioning difficult. Previous versions of the same codes did not take this effect
In spite of these difficulties, it must be noted that the into account, having thus a constant ηD = 1, which was clearly
observed strength reduction was much less for beams than unconservative. Others accounted only for a reduction of
for panels; in some cases9, beams with injected ducts even half the duct diameter, while empty ducts were not
reached a higher strength than their reference beams. mentioned15 or not distinguished from injected ducts.14
Regarding the loss of strength in the presence of plastic
Summary of results from literature ducts, only the latest version of Eurocode 2 takes that effect
All comparable results from the aforementioned test series into account, with a rather conservative value. This parameter
from the literature were compiled and formatted to allow for should be considered in further revisions of design codes.
their direct comparison. The test results from the current
study (described as follows) are also included. The results EXPERIMENTAL PROGAM
are presented as the ratio ηD of the strength of the panel Within the framework of an ongoing research project on
element with a tendon duct to the strength of an identical the shear capacity of thin webs containing post-tensioning
panel without a duct. Physically, the ratio ηD is a strength tendons, two series of tests were performed on panel
reduction factor to be applied to the concrete strength, in specimens. The first series consisted of 12 specimens cast in
combination with other factors (to account for cracking, size the laboratory, and the second consisted of four panel
effect, and slenderness). The results are summarized in Fig. 5 specimens extracted from an existing bridge built in 1967
(tests with spiral reinforcement omitted). and replaced in 2003.
Current code provisions (from AASHTO,10 BS 5400,11 The first series (laboratory panels) contained various types
CEB-fip Model Code 90,12 and EC213) are also shown in the of ducts: empty duct (W7 and W8), injected steel duct (W5
figures. The code provisions are in the form and W6), injected HDPE duct (W1, W2, W9, and W10), duct
and tendon extracted from an existing bridge (1967 steel
ηD = 1 – k · δ (3) duct; W11 and W12) (Fig. 6). Two reference panels without
duct (W3 and W4) were also tested. Each panel was 23.6 x
The diameter correction factor k is given in Table 1 for 23.6 x 4.9 in. (600 x 600 x 125 mm) in dimensions (Fig. 7(a)).
these codes. The laboratory specimens were provided with passive

ACI Structural Journal/September-October 2006 731

 Fig. 8—Profiled I-girder with two parabolic post-tensioning
cables in web (panel extraction and panel loading direction).

Fig. 6—Types of ducts used in experimental program.

Fig. 9—Specimen in testing machine with surface-mounted

displacement transducers.

150 mm2]) were inserted into the ducts. Three days after
casting, they were injected with non-shrink grout with a w/c
of 0.33, except for Specimens W7 and W8 which were kept
ungrouted. The measured grout strength fg′ is given in Table 3.
Specimens W11 and W12 included injected grouted tendons
Fig. 7—Geometry and reinforcement of tested specimens. ducts that were extracted from beams of the existing bridge.
The purpose of the inclusion of these specimens was to
assess the influence of older duct types on the load-carrying
Table 2—Measured properties of passive
capacity of thin webs and to compare the results with those
reinforcing steel
from panels directly extracted from the existing bridge girders.
f ′y, ksi (MPa) 73.8 (509) The second test series (bridge panels) consisted of two
f ′t , ksi (MPa) 86.6 (597) cracked specimens containing two tendons (1967 steel duct)
f ′t /f ′y 1.17 placed on top of one another at a spacing of more than one
Es, ksi (MPa) 29,300 (202,000)
duct diameter (W21 and W22, Fig. 7(b)) and two reference
specimens without tendons and with no visible cracking
(W23 and W24), also extracted from the existing bridge
Table 3—Measured cylinder strength of concrete girders. The panels with tendons were cut from the bridge
and injection grout at 14, 21, and 28 days* girder after it had been tested in the laboratory. They were
thus extensively cracked, although the girder had failed in
t, days 14 21 28
shear at the opposite side of the beam. The orientation of the
Concrete f ′c , psi (MPa) 5030 (34.7) 5310 (36.6) 5410 (37.3) panels W21 and W22 was chosen so that the shear cracks are
Grout f ′g , psi (MPa) 3920 (27.0) 1060 (28.7) 4670 (32.2) running parallel to the direction of the loading (Fig. 8). The
*Note: Cylinder dimensions = ∅/h = 6.3/12.6 in. (160/320 mm). uncracked reference panels W23 and W24 were extracted in
a vertical orientation. The amount of passive reinforcement
reinforcement similar to that of the existing bridge to make in the bridge panels was comparable to that of the laboratory
a direct comparison possible. Table 2 shows the properties of panels (Fig. 7(b)).
the reinforcing steel used for the laboratory specimens. All Table 4 shows the main parameters for all test specimens,
specimens were cast horizontally in a single batch with a including the estimated concrete strength at the time of testing.
normal strength concrete (fc′ ≅ 5200 psi [36 MPa], Ec ≅ 4600 ksi Each specimen was tested in a high-capacity universal
[31,500 MPa]). The maximum concrete aggregate size was testing machine (Fig. 9), at a constant speed of 3.7 × 10–5 in./s
0.63 in. (16 mm) and the water-cement ratio (w/c) was 0.55. (0.9 × 10–3 mm/s), corresponding to approximately 0.3 kips/s
All specimens were tested at ages between 14 and 30 days. (1.3 kN/s) in the linear part of the loading curve. The load
Table 3 gives the measured concrete strength at 14, 21, and was introduced through a thin 0.4 to 0.6 in. (10 to 15 mm)
28 days. Seven 7-wire-strands 0.6 in. (A = 7 x 0.23 in.2 [7 x layer of cement-based high-strength mortar placed at the

732 ACI Structural Journal/September-October 2006

 Fig. 11—Stress reduction factor ηD for tested specimens.

the slenderness of the panels and shows that the strength was
not increased by local confinement effects due to friction at
Fig. 10—Average compression stress versus average strain. the loading plates. Specimens W7 and W8 with empty ducts
reached the lowest value (38% of the average strength of
reference specimens W3 and W4). The largest values for
base and on top of the specimen. Surface measurements were specimens with injected ducts were reached by specimens with
made on the concrete using 34 surface-mounted displacement steel ducts (W5 and W6, approximately 87%), while the
transducers on all sides of the specimen (Fig. 9). The
specimens with HDPE ducts reached the lowest value (W1,
measurement base was 4.9 in. (125 mm) vertically, 9.8 in.
W2 and W9, and W10; 63%). The angle of inclination β of
(250 mm) horizontally, and 4.5 in. (115 mm) transversally,
the duct does not seem to have influenced the strength (64%
with a range of ±0.08 in. [2 mm] and a non-linearity of 1%. The
compared with 63%), but seems to have caused a more
ambient temperature during testing was approximately 20 °C.
ductile behavior in the load-deformation response (W9 and
W10). Finally, one of the specimens with tendons retrieved
EXPERIMENTAL RESULTS
from an existing bridge reached values significantly lower
Figure 10 shows the stress-strain curve for all tested
specimens, based on average vertical strain measurements than specimens with comparable steel ducts (W11; 71%),
(Fig. 9) and the applied load. The initial branch of the load- probably because the injection grout was precracked by the
deflection curve is linear for all specimens, followed by a previous loading of the tendon and by the extraction process.
loss of stiffness until the maximum load is reached. In most The other specimen with an extracted tendon reached a
cases, the failure was sudden, with little or no post-peak higher value (W12; 82%).
strength. Table 4 and Fig. 11 give the ratio ηD of the ultimate The measurement of the transverse expansion at the level
load carried by each specimen, after deduction of the of the tendon gives valuable information about the way in-
contribution of the steel reinforcement, compared with the plane cracking occurs, although most cracks remain invisible
average strength of the two reference specimens without ducts, to the naked eye up to high levels of loading. As Fig. 12
also after deduction of the contribution of the longitudinal shows, specimens without a duct exhibited limited transverse
reinforcement. No reduction was applied to the specimens expansion until very high levels of loading. The expansion
W21 and W22 that were extracted from the existing bridge, as was mainly caused by Poisson’s ratio. Specimens with ducts
the passive reinforcement runs diagonally and is insufficiently started to diverge from the behavior of reference specimens
anchored, thus contributing little to the load-carrying capacity. at fairly low load levels, with a lateral expansion of 5.9 × 10–3
to 11.8 × 10–3 in. (0.15 to 0.30 mm), mainly due to thin splitting
Results from laboratory panels cracks, reached at 80% of ultimate. The behavior of panels
Specimens W3 and W4, the two solid specimens, reached containing HDPE ducts is clearly different from the behavior
the highest load capacity, corresponding to 92% of the of panels with steel ducts, with a more rapid development of
concrete compressive strength. This reduction results from inner cracking. It also differs from that of empty ducts.

ACI Structural Journal/September-October 2006 733

Table 4—Tested specimens: main parameters and results
First series, laboratory panels Second series, bridge panels
Specimen W1 W2 W3 W4 W5 W6 W7 W8 W9 W10 W11 W12 W21 W22 W23 W24
HDPE None Steel Steel HDPE Steel 1967 Steel 1967 None
Duct
Injected — Injected Empty Injected Injected Injected —
∅D, in. (mm) 2.48 2.48 0 0 2.44 2.44 2.44 2.44 2.48 2.48 2.36 2.36 2.36 2.36 0 0
(63) (63) (0) (0) (62) (62) (62) (62) (63) (63) (60) (60) (60) (60) (0) (0)
δ 0.50 0.50 0 0 0.50 0.50 0.50 0.50 0.50 0.50 0.48 0.48 0.45 0.45 0 0
Injected Yes Yes — — Yes Yes No No Yes Yes Yes Yes Yes Yes — —
f ′c , psi (MPa) 5278 5452 5250 5415 5076 5154 5352 5116 5434 5304 4984 5329 7818 6846 6962 6846
(36.4) (37.6) (36.2) (37.3) (35) (35.5) (36.9) (35.3) (37.5) (36.6) (34.4) (36.7) (53.9) (47.2) (48) (47.2)
t, days 22 30 21 28 16 18 25 17 29 23 14 24 years 36 years 36 years 36 years 36 years
β, degrees 0 0 — — 0 0 0 0 34 34 0 0 52 50 — —
NR, kips (kN) 386 396 607 627 501 538 248 228 389 388 413 508 464 433 719 736
(1718) (1763) (2700) (2790) (2228) (2393) (1103) (1013) (1733) (1725) (1838) (2258) (2066) (1924) (3200) (3276)
NRs, kips (kN) 29 31 46 45 42 46 27 21 22 23 36 41 0 0 23 21
(129) (139) (203) (202) (187) (205) (120) (91) (100) (100) (161) (181) (0) (0) (102) (94)
NRc, kips (kN) 357 365 561 582 459 492 221 207 367 365 377 467 464 433 696 715
(1589) (1624) (2497) (2588) (2040) (2188) (982) (921) (1633) (1625) (1677) (2076) (2066) (1924) (3098) (3182)
N Rc
------------ , psi (MPa) 3072 3140 4829 5005 3946 4231 1900 1781 3158 3142 3242 4015 3720 3479 6117 6153
bw ⋅ c (21.2) (21.6) (33.3) (34.5) (27.2) (29.2) (13.1) (12.3) (21.8) (21.7) (22.4) (27.7) (25.6) (24) (42.2) (42.4)

N Rc
---------------------- 0.58 0.58 0.92 0.92 0.78 0.82 0.35 0.35 0.58 0.59 0.65 0.75 0.48 0.51 0.88 0.90
b w ⋅ c ⋅ f c′
ηD 0.63 0.62 1.00 1.00 0.84 0.89 0.38 0.38 0.63 0.64 0.71 0.82 0.54 0.57 0.99 1.01

Fig. 13—Side surfaces of laboratory panels after testing.

Results from bridge panels

The stress-strain curve, the strength reduction factor ηD
and the transverse elongation measurements of the bridge
panels are shown in Fig. 10(b) to 12(b), respectively. Their
behavior was similar to that of the laboratory panels. The two
cracked specimens with tendons reached 56% of the strength
of their reference panels, much lower than the laboratory
specimens with injected steel ducts.
This reduction is mainly caused by the previous loading in
shear and the resulting cracks parallel to the compression
field (Fig. 8). Using the strength reduction factor by Vecchio16

f ce 1
----- = ------------------------------------------------------ (4)
f c′ ε1
1 + 0.27 ⋅ ⎛ ------- - – 0.37⎞
⎝ –εp ⎠
Fig. 12—Average compression stress versus transverse
expansion at duct location. one can estimate the reduction in strength caused by the
transverse strains from the testing of the bridge, which were
Shortly before the ultimate load was reached, splitting measured at approximately ε1 = 3.5‰, and taking εp = –2‰,
cracks became visible on the side faces. At the ultimate load, to approximately 0.7fc′ .
the specimen was almost split in two parts (Fig. 13). On the Assuming that the same deformation ε1 would have been
front faces, no cracking was observed before the ultimate imposed to Specimens W11 and W12, their strength would
load, except for specimens with empty ducts, for which have been reduced by a factor 0.73, yielding ηD values of
cracks appeared at approximately 70% of ultimate. 0.52 and 0.60, respectively, which compare very well with

734 ACI Structural Journal/September-October 2006

the average measured ratio of 0.56 measured on the cracked c = side dimension of panel element
panels W21 and W22. Ec = modulus of elasticity of concrete
EHDPE = modulus of elasticity of high-density polyethylene
The measured transverse expansion of the reference bridge Es = modulus of elasticity of passive reinforcement
panels W23 and W24 was similar to that of the reference fc′ = concrete cylinder strength
laboratory panels W3 and W4. The bridge panels exhibited fce = effective concrete strength in presence of transverse strains
a much softer behavior, likely caused by the precracking fg′ = grout cylinder strength
imposed by the previous loading of the panels (Fig. 12). The ft′ = tensile strength of passive reinforcement
fy′ = yield stress of passive reinforcement at 0.2% offset
transverse crack pattern after testing of Specimens W21 and k = diameter correction factor
W22 is similar to that of panels W11 and W12. The fact that N = compression force on panel element
two tendons were present in the panels does not appear to NR = ultimate load of panel element
have had a significant effect. NRc = portion of ultimate load carried by concrete
NRs = portion of ultimate load carried by passive reinforcing steel;
estimated on basis of concrete surface strain measurements
Proposal for code improvement t = age
Based on the observed behavior from both laboratory and tD = duct wall thickness
bridge panels, the authors propose that a strength reduction w = transverse expansion
in the form of ηD = 1 – k ⋅ δ be applied with k being equal to β = tendon angle of inclination relative to normal to compression strut
0.4 for steel ducts, 0.8 for plastic ducts, and 1.2 for empty ducts. δ = ratio of total width of all ducts to panel thickness in critical section
ε = strain
These values are valid for normal-strength concrete elements. ε1 = maximum concrete strain
The effect of the more stringent requirements implied by εp = reference concrete strain at peak stress (–2‰)
the findings of the present study may be somewhat mitigated ηD = strength reduction factor
in practical cases by the fact that the shear strength of webs Σ∅D = sum of duct diameters in critical section
of post-tensioned girders is not necessarily limited by the ∅D = duct diameter
strength of the compressive struts, but rather by the tensile
strength of the stirrups. In critical cases, the disposition of a REFERENCES
1. Gaynor, R. D., “Effect of Horizontal Reinforcing Steel on the Strength
suitable reinforcement transverse to the web in the vicinity of Molded Cylinders,” ACI JOURNAL , Proceedings V. 62, No. 7, July 1965,
of the tendon can prevent this mode of failure without pp. 837-840.
requiring other design changes. 2. Leonhardt, F., “Abminderung der Tragfähigkeit des Betons infolge
stabförmiger, rechtwinklig zur Druckrichtung angeordnete Einlagen (Reduction
of Concrete Compressive Strength due to Rods Inserted Perpendicularly to the
SUMMARY AND CONCLUSIONS Loading),” Festschrift Rüsch, Berlin, Germany, 1969, pp. 71-78. (in German)
Two series of 12 and four specimens of web panels have 3. Clarke, J. L., and Taylor, H. P. J., “Web Crushing—A Review of
been tested in the laboratory to investigate the effect of duct Research,” Technical Report, Cement and Concrete Association, London,
type and of web cracking on the ultimate compressive 42-509, 1975, 16 pp.
strength of web girders crossed by post-tensioning tendons. 4. Chitnuyanondh, L., “Shear Failure of Concrete I-Beams with Prestressing
Ducts in the Web,” PhD thesis, Queen’s University, Kingston, Canada,
The ratio of the diameter of the duct to the thickness of the 1976, 245 pp.
web was approximately 0.5. For that value, the observed loss 5. Campbell, T. I.; Batchelor, B.,; and Chitnuyanondh, L., “Web Crushing
of strength was significant, comparable for panels with steel in Concrete Girders with Prestressing Ducts in the Web,” PCI Journal,
ducts to values obtained in previous studies—approximately V. 24, No. 5, 1979, pp. 71-87.
13%. Panels with HDPE plastic ducts had a much larger loss 6. Campbell, T. I., and Batchelor, B., “Effective Width of Girder Web
Containing Prestressing Duct,” Journal of the Structural Division, ASCE,
of approximately 37%. V. 107, 1981, pp. 733-744.
Precracked panels extracted from an actual bridge with 7. Rezai-Jorabi, H., and Regan, P. E., “Shear Resistance of Prestressed
steel tendons exhibited a much lower strength than initially Concrete Beams with Inclined Tendons,” The Structural Engineer, V. 64B,
uncracked laboratory panels without tendons (loss of No. 3, 1986, pp. 63-75.
8. Ganz, H. R.; Ahmad, A.; and Hitz, H., “Load Transfer through Concrete
approximately 44%), the difference being attributed to the Sections with Grouted Ducts,” Report 242e, VSL, Bern, Switzerland, 1992,
effect of cracking on the concrete strength. The influence of 20 pp.
the type of steel duct (1967 versus current) is not significant. 9. Fairbairn, E. M. R., and Trinh, J. K. L., “Influence des Câbles Relevés
Current code provisions mostly give a correct estimate of sur la Résistance de l’Ame des Poutres Soumises à des Actions Tangentes—
the loss of compressive strength of the compression struts in Etudes Expérimentales (Influence of Inclined Post-Tensioning Cables on
the Web Strength of Girders Subjected to Tangential Loads—Experimental
the presence of steel ducts, with the exception of AASHTO, Studies),” I.T.B.T.P Annals, No. 400, Series B 203, Paris, France, 1981,
which underestimates this effect. pp. 14-20. (in French)
Only the Eurocode 2 explicitly distinguishes between steel 10. AASHTO, “AASHTO LRFD Bridge Design Specifications,” 3rd
and plastic ducts. Based on the observed behavior from the Edition, Washington, D.C., 2004, 1450 pp.
tested panels, the authors propose that a strength reduction in 11. BS 5400-4:1990, “Steel, Concrete and Composite Bridges—Part 4:
Code of Practice for Design of Concrete Bridges,” London, 1990, 79 pp.
the form of ηD = 1 – k ⋅ δ be applied for normal-strength 12. Comité Euro-International du Béton (CEB), “CEB-FIP Model Code
concrete, with k being equal to 0.4 for steel ducts, 0.8 for 1990,” Thomas Telford Publishing, London, 1993, 460 pp.
plastic ducts, and 1.2 for empty ducts. 13. EUROCODE 2, “Design of Concrete Structures—Part 1-1: General
Rules and Rules for Buildings,” CEN, EN 1992-1-1, Brussels, Belgium,
2004, 225 pp.
ACKNOWLEDGMENTS 14. Comité Euro-International du Béton (CEB), “CEB-FIP Model Code
The authors want to acknowledge the gracious support and funding of the for Concrete Structures,” Information Bulletin, 124/125-E, Paris, France,
Swiss Federal Roads Authority which made the experimental and theoretical 1978, 348 pp.
studies possible and VSL International for providing post-tensioning material
15. EUROCODE 2, “Design of Concrete Structures—Part 1-1: General
for the laboratory specimens.
Rules and Rules for Buildings,” CEN, prENV 1992-1-1, Brussels, Belgium,
1992, 252 pp.
NOTATION 16. Vecchio, F. J., “Disturbed Stress Field Model for Reinforced Concrete:
A = area Formulation,” Journal of Structural Engineering, ASCE, V. 126, No. 9, 2000,
bw = web thickness; panel element thickness pp. 1070-1077.

ACI Structural Journal/September-October 2006 735