Modeling of L-Shaped, Precast,
Prestressed Concrete Spandrels
Tarek Hassan, Ph.D.
Assistant Professor
Faculty of Engineering
Ain Shams University
Cairo, Egypt
Gregory Lucier
Research Engineer
Constructed Facilities Laboratory
North Carolina State University
Raleigh, N.C.
Sami Rizkalla, Ph.D., P.Eng.
Distinguished Professor of Civil,
Construction, and Environmental
Engineering and Director
Constructed Facilities Laboratory
North Carolina State University
Raleigh, N.C.
Paul Zia, Ph.D., P.E., FPCI
Distinguished University Professor
Emeritus
North Carolina State University
Raleigh, N.C.
78 PCI JOURNAL
This paper presents the results of nonlinear finite
element analyses conducted to model the behavior
of L-shaped, precast, prestressed concrete spandrels
constructed with open web reinforcement. The
finite element model was calibrated using experi-
mental results from recent tests of slender, L-shaped,
precast, prestressed concrete spandrels. Detailed
correlative studies between analytical and experi-
mental results are presented, demonstrating the
capability of the finite element program to describe
the observed experimental behavior.
The feasibility of using open web reinforcement
in compact, L-shaped, precast, prestressed concrete
spandrels to achieve a more construction-friendly
reinforcement scheme is also examined. Five differ-
ent web reinforcement configurations for the com-
pact spandrels were studied in order to evaluate
the contribution of closed stirrups to the spandrels’
shear-torsion behavior.
The behavior, ultimate load-carrying capacity,
and mode of failure of both the slender and compact
L-shaped precast, prestressed concrete spandrels
are presented. For loading values near the ultimate,
the out-of-plane bending behavior of compact,
L-shaped, precast, prestressed concrete spandrels
is strongly influenced by the web-reinforcement
configuration. Results from the analysis show that
for long-span, compact spandrels, open web rein-
forcement can be used effectively to resist torsional
forces throughout the member.
D
espite past research, there still
exists a need to study the be-
havior of L-shaped, precast,
prestressed concrete spandrels when
subjected to different combinations
of torsional, flexural, and shear loads.
Industry methods and published pro-
cedures vary significantly with respect
to several fundamental aspects of the
design and detailing of such members.
Current U.S.1 and Canadian2 provisions
for the design of members for compat-
ibility torsion are simple to use and
conservative for design, but they often
result in areas of heavily congested re-
inforcement within a beam.
Significant potential exists for reduc-
ing the complexity of L-shaped, precast,
prestressed concrete spandrel designs
by removing closed ties from slender
members. Limited tests on full-scale
L-shaped spandrels3 revealed the pos-
sibility of reducing the transverse rein-
forcement at their end regions. Elastic
theory (assuming an uncracked sec-
tion) is a necessary tool for proportion-
ing the member. However, an analysis
of the post-elastic behavior—including
stiffness, deformation, and cracking
patterns—is essential for evaluating
the complete response of the member
to different loading conditions.
Knowledge of the complete response
of an L-shaped spandrel to different
loading conditions is critical for as-
sessing the amount of the transverse
reinforcement needed at the member
ends. Test results have shown that the
torsional stiffness of a member is great-
ly affected by cracking and by the in-
teraction among torsional, flexural, and
shear loads.3 Figure 1 shows a typical
L-shaped spandrel that is used in park-
ing structures.
A unified procedure for the design
of prestressed concrete members for
shear and torsion was originally devel-
oped by Zia and McGee in 1974.4 Their
design procedures were derived from
a comprehensive set of test data and
were coordinated with existing design
practice. Further refinement of these
procedures was subsequently proposed
by Zia and Hsu.5,6
Although these procedures are com-
monly used, research data have never
validated them for slender spandrels,
which are typically used in practice.
March–April 2007 79
Lateral
restraint
at ends (typ.)
Vertical reaction
Acting loads
Double-tee beams
Fig. 1. Typical spandrel used in parking structures
Recent efforts to classify spandrel be-
havior include a study by Rahal and
Collins,7 which describes a procedure
to calculate compatibility torsion in
spandrels. Their procedure relies on
modified compression field theory to
calculate the cracked torsional and
flexural stiffnesses for sections subject-
ed to various combinations of stress re-
sultants. Rahal and Collins’ procedure
was capable of predicting the response
of concrete members where the effect
of compatibility torsion is dominant.
The American Concrete Institute’s
ACI 318-051 requires closed stirrups to
be placed throughout a concrete mem-
ber subjected to combined shear and
torsion. According to this document,
closed stirrups are mandatory to avoid
spalling of the concrete cover. Test re-
sults by several researchers3,8 showed
that this type of behavior is unlikely to
occur in deep spandrels.
Recently, the Precast/Prestressed
Concrete Institute (PCI), and many PCI
Producer Members, have questioned
the need for closed stirrups along the
entire length of a slender spandrel.
It should be noted that in the precast
concrete industry, common detailing
practices for torsional reinforcement
in deep spandrels do not usually fol-
low the ACI requirements. Transverse
reinforcement is often provided in
L-shaped spandrels with pairs of lap-
spliced, mild-steel, U-shaped stirrups.8
Unfortunately, widespread, full-scale
experimental testing to examine the
Vertical reaction
Spandrel beam
influence of various web reinforcement
configurations in L-shaped spandrels is
prohibitively expensive.
Therefore, the use of nonlinear finite
element analysis coupled with limited
experimental studies is a powerful tool
for predicting the behavior and failure
modes of L-shaped, precast, prestressed
concrete spandrels. The complex com-
bination of stress resultants that de-
velop in the member due to bending,
shear, and torsion, as well as the size
effect of the L-shaped spandrel’s slen-
der web, dictate the intricacy of such
analyses.
This paper presents the results of
nonlinear finite element analyses con-
ducted to simulate the behavior of L-
shaped, precast, prestressed concrete
spandrels. The main objective of the
current study was to develop reliable
and computationally efficient finite
element models (FEMs) to analyze L-
shaped, precast, prestressed concrete
spandrels subjected to combined bend-
ing, shear, and torsion. Results from
previous testing were used to calibrate
the FEM. Once a model was validated,
it was used to investigate the response
of compact, L-shaped, precast, pre-
stressed concrete spandrels designed
with open web reinforcement.
The behavior, ultimate load-carry-
ing capacity, and failure mode of both
slender and compact, L-shaped, pre-
cast, prestressed concrete spandrels
are presented. The influence of the
lateral deck ties and several different
 22' 9" TO CENTERLINE W4.0 by W4.0
2 @ 8" 28" 4 @ 8" 28" 4 @ 8" 28" 5 @ 8" 9 @ 6" 15" 4" x 4" STARTS 7' FROM EA. END
W4.0 by W4.0 6"x6" CONT.
#4 L-BAR #4 U-BARS 8' LONG (TYP. OF 1 ABOVE)
(TYP.)

#4 U-BARS 6' LONG (TYP. OF 1 ABOVE)

#4 U-BARS 2' LONG (TYP. OF 2 EA. END)

#4 U-BARS 3' LONG (TYP. OF 2 ABOVE)
20" 5 SPA. @ 8" 20" #4 U-BAR 5' LONG
C.L
#4 C-BAR (TYP.) WELDED LEDGE DETAIL (TYP.) #6 BAR CONT.
(2) #4 C-BARS
NOTE: = STRAND PULLED TO 22,500 LB
SP3 (2 LOCATIONS EA. END)
= STRAND PULLED TO 15,800 LB
17 TOTAL STRANDS ON 2" GRID

22' 9" TO CENTERLINE W4.0 by W4.0

43 @ 6" 15" 6"x6" CONT.
#3 L-BAR (TYP.) W4.0 by W4.0
6"x6" CONT.

#4 BARS 6' LONG

(TYP. OF 4 EA. END)

#4 U-BARS 2' LONG (TYP. OF 2 EA. END)

#4 U-BARS 6' LONG (TYP. OF 3 EA. END)

20" 5 SPA. @ 8" 20"
C.L #6 BAR CONT.
#4 C-BAR (TYP.) (2) #4 C-BARS
WELDED LEDGE DETAIL (TYP.) (2 LOCATIONS EA. END)
NOTE: = STRAND PULLED TO 22,500 LB
SP4 = STRAND PULLED TO 15,800 LB
17 TOTAL STRANDS ON 2" GRID

Fig. 2. Reinforcement details of spandrels SP3 and SP4. Note: ' = ft; " = in.; 1 ft = 304.8 mm; 1 in. = 25.4 mm; 1 lb = 0.00448 kN;
#4 = 12M; #5 = 16M; #6 = 19M.

Plane of symmetry

Load (TYP.)

Support

Fig. 3. Mesh dimensions used in the finite element model. Note: ' = ft; " = in.; 1 ft = 304.8 mm; 1 in. = 25.4 mm.
Fig. 3 Mesh dimensions used in the finite element model
Note: 1 in.=25.4 mm
80 PCI JOURNAL
web reinforcement configurations on
the out-of-plane behavior of compact,
L-shaped, precast, prestressed concrete
spandrels is also discussed. Plane of symmetry

Validation of the FEM

The first reinforced concrete FEM
that included the effects of cracking
was developed in 1967.9 Cracks were Transfer length
modeled by separating the nodal points
of the finite-element mesh, thus creat-
ing a discrete crack model. With the Plan view at the end
change of topology and the redefinition
of nodal points, the narrow bandwidth
of the stiffness matrix was destroyed,
resulting in increased computational
effort. Moreover, the lack of generality Isometric view Support
in crack orientation has made the dis-
crete crack model unpopular. The need
for a crack model offering automatic
generation of cracks and complete gen- Fig. 4. Layout of the prestressing strands for spandrels SP3 and SP4.
erality in crack orientation, without the
need for redefining the finite element
topology, has led the majority of inves- an engineering approximation to the Several attempts8,12 have been made
tigators to adopt other crack models. concrete’s actual behavior and permits in the past few years to model the be-
In the current study, the ANATECH the analysis of concrete structures up to havior of L-shaped, precast, prestressed
Concrete Analysis Program (ANA- and during failure. In the smeared-crack concrete spandrels using finite element
CAP)10 was used to model the behavior approach, the modulus and strength of analysis. Nevertheless, the complex
of the L-shaped, precast, prestressed the concrete in the direction normal to behavior of these spandrels under com-
concrete spandrels. The concrete materi- an open-crack surface is zero, but the bined bending, shear, and torsion lim-
al model in ANACAP has evolved over shear modulus and shear strength re- ited the previous analyses to modeling
the past 30 years and is based on smeared main intact. The shear modulus is grad- only linear-elastic behavior.
cracking methodology for the treatment ually reduced, however, as crack widths Two L-shaped, precast, prestressed
of concrete tensile cracking.11 Model- increase. This gradually reducing shear concrete spandrels, denoted span-
ing of the compressive behavior of the resistance is critical to the continued drels SP3 and SP4, were selected
concrete follows the generally accepted load resistance of the structure. from the literature to validate the
principles of computational plasticity,
though these principles are modified for
Table 1. Materials Properties Used in Finite Element Analysis
the unique and computationally demand-
ing aspects of concrete response.
Property SP3 SP4
Cracks are assumed to form perpen-
dicular to the directions of the largest
tensile strains. Multiple cracks are al- Concrete compressive strength, psi 5790 7190
lowed to form at each material point,
but they are constrained to be mutually Modulus of rupture of concrete, psi 456 509
orthogonal. At the onset of cracking, the
normal stress across the crack is reduced,
Yield strength of welded wire reinforcement, psi 98,000 98,000
and the distribution of stresses around
the crack is recalculated through iteration
of equilibrium equations. This recalcula- Yield strength of conventional mild-steel reinforcing bars, psi 64,500 64,500
tion allows stress redistribution and load
transfer to the reinforcement. Once a Yield strength of prestressing strands, psi 243,000 243,000
crack forms in the model, the direction of
the crack remains fixed and it can never
Prestressing losses, % 15 15
heal. However, a crack may close to re-
sist compression and then reopen. Note: Modulus of elasticity of all conventional and prestressing steel is 29,000 ksi. 1 psi = 0.006895 MPa.;
The smeared-crack model represents 1 ksi = 6.895 MPa.

March–April 2007 81
FEM. The spandrels measured 45 ft Modeling the Concrete Spandrels mesh was chosen so that elements
6 in. (13.87 m) long from end to end. would maintain acceptable aspect ra-
Figure 2 shows cross-sectional dimen- Because geometry and loading of the tios while accurately representing ge-
sions and reinforcement details of both members were symmetrical about their ometry, loading conditions, and sup-
spandrels. A detailed description of the midspans, half of each spandrel was port conditions. Figure 3 shows the
testing of these two specimens is re- modeled using 20-node brick elements, finite-element mesh dimensions used
ported in this issue of the PCI Journal each node having three translational in the FEM.
and elsewhere.3 degrees of freedom. The finite-element
Modeling the Prestressing and Mild-
Steel Reinforcement
Mid-span deflection (mm)
0 6 13 19 25 32 38 44 51 57 64 The prestressing force in each mem-
Mid-span deflection (mm) ber was applied gradually to the span-
200 890
0 6 13 19 25 32 38 44 51 57 64 drel ends in the model to replicate the
180 Experimental 801
200 890 transfer length of the strands. This was
ANACAP
160 712 accomplished by splitting each strand
(kips) (kips)

Experimental

(kN) (kN)
180 801
into 10 small strands. Each small
140 ANACAP 623
160 712 strand has one-tenth the area of the

of spandrel
of spandrel

120 534 original strand, but all occupy virtually

140 623
the same location in the spandrel.

of spandrel
of spandrel

100 445 The first of the 10 strands started at

120 534

End reaction
the spandrel end, and the 10th started at
End reaction

80 356
100 445 a distance equal to the transfer length.
60 267

End reaction
The remaining eight strands started
End reaction

80 356
40 178 at equal, incremental distances be-
60 267 tween the spandrel end and the transfer
20 SP3 89
40 178 length, as shown in Fig. 4. The rein-
0 0 forcement was modeled as individual
20 SP3 89
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5 subelements within the concrete ele-
0 0 ments. The stress and stiffness of the
Mid-span deflection (in)
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5 mild-steel reinforcing bar subelements
Fig. 5. Load-deflection Fig. 5 Load-deflection
behavior of spandrel
Mid-span
behavior of SP3
SP3.
deflection (in)
were superimposed on the concrete el-
ement in which the reinforcing bar re-
Fig. 5 Load-deflection behavior of SP3 sided. The analytical model accounted
Mid-span deflection (mm)
0 6 13 19 25 32 38 44 51 57 64
for every mild-steel reinforcing bar
Mid-span deflection (mm) used in each of the spandrels.
200 890
0 6 13 19 25 32 38 44 51 57 64 Simulation of the Applied Load
180 Experimental 801
200 890
ANACAP Load was applied to the spandrel
160 Experimental 712
(kips) (kips)

180 801 ledge at each double-tee stem as a uni-

(kN) (kN)

140 ANACAP 623 form pressure acting over the stem bear-
160 712
of spandrel
of spandrel

ing area. The analysis was conducted

120 534
140 623 using an incremental-iterative solution
of spandrel
of spandrel

100 445 procedure, in which the applied load

120 534
was incrementally increased. The load-
End reaction

80 356
End reaction

100 445 ing increment was set to 1 kip (4.448

60 267 kN) per step. Within each step, equi-
End reaction

80 356
End reaction

librium was achieved and iteration was

40 178 repeated until internal equilibrium con-
60 267
20 SP4 89 ditions were sufficiently fulfilled and
40 178
convergence was obtained. At the end
0 0 of each step, the program adjusted the
20 SP4 89
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5 stiffness matrix to reflect any nonlinear
0 0 changes in the spandrel’s stiffness.
Mid-span deflection (in)
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5 The self-weight of the spandrel,
Fig. 6 Load-deflection behavior
Mid-span deflection (in) of SP4 loading jacks, and spreader beams,
along with the weight of the double
Fig. 6. Load-deflection Fig. 6 Load-deflection
behavior behavior of SP4
of spandrel SP4. tees, were introduced at the first load-

82 PCI JOURNAL
ing step. Applied loads were then in-
Crack potential values ranged from 82% to 92%
creased to failure. Plane
prior to development of shear crackof symmetry
(Concrete is about to crack along the diagonal)
Materials and Boundary Conditions Support

Table 1 summarizes the mate-

rial properties used in the FEM for
spandrels SP3 and SP4. The spandrel
model employed the same bound-
ary conditions as those implemented
Crack potential values ranged from 82% to 92%
in the laboratory tests. In the model, prior to development of shear crack
the spandrel was restrained vertically (Concrete is about to crack along the diagonal)
Support
throughout its width for the first 12 in.
(305 mm) along both ends to simulate Plane of symmetry
the bearing pads used at the laboratory
spandrels’ ends. Lateral restraint was
provided throughout the width of the
spandrel, 6 in. (152 mm) from each
end and 12 in. (305 mm) from the top
and bottom of the spandrel. This lat- Crack potential values ranged from
zero to 20% after development of
eral restraint simulates the tiebacks diagonal shear crack
provided by the threaded rods during
laboratory testing of the actual span- Support
Plane of symmetry
drels. A symmetry boundary condi- Fig. 7. Cracking potential of spandrel SP3 with an end reaction of 60 kip (267 kN)
tion was applied at midspan for each (above) and 100 kip (445 kN)(below).
analysis because only half of each
spandrel was modeled.
with the exception of the effect of creep sibly be attributed to the instruments
as discussed previously. used to obtain the vertical deflection
Results and Discussion Crack potential values ranged from
Fromzero
thetofigures,
20% afteritdevelopment
is observed of that measurements. As the spandrel rotates
the predicted diagonal shear crack stiffness
post-cracking and deflects vertically, a component
Deflections
is slightly lower than the measured of the lateral deflections is included
Support
Figures 5 and 6 plot the predicted values, especially for spandrel SP4. A in the vertical measurements. This
and measured vertical end reactions significant portion of this error can pos- error, inherent to obtaining vertical
versus midspan deflections for span-
drels SP3 and SP4, respectively. It Plane of symmetry
should be noted that the load was held
during testing for several relatively
long periods of time, including a 24-
hour period, causing a small amount
of creep, which is reflected by the pro-
gressive increase in residual deflections
upon each unloading cycle. This short-
term creep behavior was not simulated
in the ANACAP program and, thus, the
increases in deflection at various load
levels are not seen in the FEM-predict- R=70 kips (311 kN)

ed behavior. It should also be noted that

the end reactions plotted for both span-
drels represent the externally applied
loads and do not include the dead load
of the system. Linear behavior was pre- R=80 kips (356 kN)
dicted for both specimens up to the ini-
tiation of the first crack at a load level of
95 kip (423 kN). Predictably, this ini-
tial behavior was followed by a nonlin-
ear behavior up to failure. In general,
R=95 kips (423 kN)
the FEM-predicted behavior is in good
agreement with the measured values, Fig. 8. Predicted crack
Fig. pattern at different
8 Predicted loading
crack pattern stages.
at different loading stages

March–April 2007 83
measurements from a rotating cross-
section that is moving both vertically
and laterally, is discussed elsewhere.13
Contributions of the double tees at
greater load levels could also result
in the higher spandrel stiffness values
than the predicted values.
Crack Pattern
Cracking potential is defined as the
ratio of the principal concrete tensile
200
200
180
180
160
(kips)
stress to the tensile strength of the con- ing potential for spandrel SP3 with
crete at any given point in the analy- an end reaction of 60 kip (267 kN).
sis (expressed in terms of percentage). The figure clearly shows the tendency
Concrete cracking will occur when the of the concrete to crack along a diago-
cracking potential reaches a value of nal near the end of the spandrel. Figure
100%. At this stage, the principal ten- 7 also shows the cracking potential of
sile stress at a given location is equal to spandrel SP3 with an end reaction of
the tensile strength of the concrete. 100 kip (445 kN). At an end reaction
After cracking, the cracking poten- of 100 kip, the shear crack has already
tial will drop to zero in the vicinity of developed because the cracking poten-
the crack. Figure 7 depicts the crack- tial in the marked area has been reduced
to zero.
Although these figures are shown for
spandrel SP3 only, spandrel SP4 had a
890
890 nearly identical cracking pattern. Figure 8
Experimental shows the predicted cracking patterns
Experimental 801
ANACAP 801 for the spandrel at various loading
ANACAP 712 stages. The FEM effectively captures

160 712
spandrel(kips)

the observed deflection behavior. In the

(kN)
140 623

(kN)
140 623 model, the top of the spandrel rotates

spandrel
spandrel

120 534

spandrel
120 534 forward at midspan, the ledge rotates
100 445 back, and the entire cross section de-
100 445 flects downward.
ofof

of of
reaction
80 356
reaction

reaction
80 356
reaction

Rotation
60 267
60 267 Figures 9 and 10 show the predicted
End
End

40 178 End rotations of spandrels SP3 and SP4 at

End

40 178
20 SP3 89
their quarter spans, respectively. FEM-
20 SP3 89 predicted rotations compare well with
0 0 the measured values up to failure. The
0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 figures clearly illustrate the capability of
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Rotation at quarter span (degrees) the FEM to reasonably predict the out-
Rotation at quarter span (degrees) of-plane deflections of the spandrels.
Fig.
Fig. 9. PredictedFig. 9 Predicted
rotations rotations
at quarter at quarter
span for span
spandrel SP3.for SP3
9 Predicted rotations at quarter span for SP3 Shear Stresses
Figure 11 illustrates the predicted
shear stresses for spandrels SP3 and
200 890 SP4 along the front face of the span-
200 890
Experimental drels. High shear stresses were ob-
180 Experimental 801
180 ANACAP 801 served at the junction of the ledge and
160 ANACAP 712 the spandrel web. Spandrel SP4 expe-
(kips)

(kN)

160 712
(kips)

rienced slightly higher shear stresses

(kN)

140 623
spandrel

140 623 than spandrel SP3 did at different load-

spandrel

spandrel
spandrel

120 534 ing stages. This increase could be at-

120 534
100 445
tributed to the distribution of the web
of of

100 445 reinforcement at the ends of the span-

of of

reaction

80 356
reaction

drel. Spandrel SP4 had relatively uni-

reaction

80 356
reaction

60 267
form web reinforcement, whereas in
60 267 spandrel SP3, the web reinforcement
End
End

40 178 was more concentrated at the ends.

End
End

40 178
20 SP4 89 Failure Mode
20 SP4 89
0 0 In the laboratory, both spandrels SP3
1.5 2 002.5 3 0.5 3.5 1 4 4.5 5 0 and SP4 failed along a skewed-diago-
1.5 2 02.5 3 0.5 3.5 1 4 4.5 5 nal crack and experienced a horizontal
Rotation at quarter span (degrees)
Rotation at quarter span (degrees) separation across the diagonal crack
Fig. 10 Predicted rotations at quarter span for SP4 extending across the top of the web.
Fig.rotations
Fig. 10. Predicted 10 Predicted rotations
at quarter spanatforquarter span
spandrel for SP4
SP4. Compression shear failure at the end

84 PCI JOURNAL
regions of the spandrels was the gov-
erning mode of failure for both speci-
mens.3
Failure in the FEM ultimately oc-
curred in both spandrels due to crush-
ing of the concrete along the primary 40 kip (178 kN) 40 kip (178 kN)
compressive strut, as shown in Fig.
12 for spandrel SP3 (spandrel SP4
was virtually identical). Analysis was
terminated when the principal com-
pressive strains along the compres-
40 kip (178 kN) 40 kip (178 kN)
sive strut reached a value of 0.002, as 100 kip (445 kN) 100 kip (445 kN)
recommended by modified compres-
sion field theory.7 The predicted fail-
ure loads for spandrels SP3 and SP4
are within 3% of the measured values.
Table 2 summarizes the predicted ul- 100 kip (445 kN) 100 kip (445 kN)
160 kip (712 kN) 160 kip (712 kN)
timate loads and deflections for both
specimens. 13 ft (4.0 m) 13 ft (4.0 m)

SP3 SP4
Influence of Deck Ties
Deck ties consisting of steel Fig.
Fig. 11.11Shear
Shearstress
stress
160 distribution
kipdistribution
(712 kN) for spandrels
for specimensSP3
SP3and
andSP4
SP4
160 atatkip
different
(712 kN)
different loading stages
loading stages.
plates of dimensions 3 in. × 6 in. Note:
Note: 1 1psi
psi=0.006895
= 0.006895 MPa
MPa.
13 ft (4.0 m) 13 ft (4.0 m)
× 3⁄8 in. (76 mm × 152 mm × 0.5 mm)
SP3 SP4
were used to connect the double tees to Plane of symmetry
the spandrel webs in the actual speci-
Fig. 11 Shear stress distribution for specimens SP3 and SP4 at different loading stages
mens. To investigate the influence of Note: 1 psi=0.006895 MPa
the lateral restraint provided by deck
ties on the predicted behavior of the Plane of symmetry
spandrel, the FEM incorporated lateral
springs at the spandrel front face at the
center of these plates. The stiffness of
the springs was set to 21,750 kip/in.
(3809 kN/m), which is equivalent to
EA/L of a given steel plate, in which
E is the elastic modulus of the steel, A
is the cross-sectional area of the plate, Compressive strains Support
and L is the length of the plate. Specimen SP3 after exceed 0.002
It should be noted that using spring failure3
supports simulates an upper bound-
Fig. 12 Typical principal compressive strain at ultimate
Compressive for SP3 and SP4
strains
ary condition for the lateral stiff- exceed 0.002
Support
Specimen SP3 after
ness provided in the actual test. failure3
Figure 13 shows the predicted load-
deflection behaviors with and without Fig. 12 Typical principal compressive strain at ultimate for SP3 and SP4
deck ties for spandrel SP3. The finite-
element analysis demonstrated the lat-
eral restraint provided by the deck ties Fig. 12. Typical principal compressive strain at ultimate for spandrels SP3 and SP4.
had a minor effect on the stiffness of
the spandrel.
This discrepancy could be attrib- Table 2. Results of the Finite Element Analysis for Specimens SP3 and SP4
uted to the fact that the location of
SP3 SP4
the deck ties within the spandrel web
nearly coincides with the center of ro- Experimental ANACAP Experimental ANACAP
tation of the web. Figure 14 shows the
Ru, kip 174 174 177 171
FEM-predicted lateral displacements
at midspan at the bottom of spandrel ∆ver, in. 1.98 2.22 1.66 1.83
SP3. The lateral restraint provided by Note: Ru = the end reaction of the spandrel at ultimate; Δver = the vertical deflection at midspan at ultimate; 1 kip =
the deck ties reduces the post-cracking 4.448 kN; 1 in. = 25.4 mm.

March–April 2007 85
 Alternative Cross-
Mid-span deflection (mm)
Sectional Dimensions
0 6 13 19 25 32 38 44 51 57 64
200 890 Compact Sections
SP3 Mid-span deflection (mm) While the previous analysis focused
180 801
0 6 13 19 25 32 38 44 51 57 64 on slender, L-shaped spandrel cross
160
200 712
890 sections (d/b of 7.5), the following
spandrel (kips)

End reaction of spandrel (kN)

140 SP3 623 analysis is related to compact, L-shaped
180 801
cross sections (d/b of 1.75), in which d
120
160 534
712 and b are the depth and the width of the
spandrelof(kips)

End reaction of spandrel (kN)

100 445 spandrel web, respectively. This study
140 623
relies on the validated analytical model
Endofreaction

80
120 356
534 discussed previously to investigate the
60
100 267
445
influence of various shear and torsion
Experimental reinforcement schemes on the behavior
End reaction

40
80 178
356 of compact spandrels.
ANACAP: No deck ties
60
20 267
89 Five different reinforcement schemes
ANACAP: With deck ties
Experimental were considered. Because the research-
40
0 178
0 ers desired to compare the transverse
ANACAP: No deck ties
20 0 0.25 0.5 0.75 1 1.25 1.5 1.75 2
ANACAP: With deck ties
2.25 2.5
89 reinforcing schemes in the slender and
Mid-span deflection (in) compact L-shaped spandrels to one
0 0 another, the cross-sectional dimen-
Fig.13. 0 0.25load-deflection
13 Predicted
Predicted 0.5 0.75 behavior
1 1.25 with
1.5 1.75without2 the 2.25 ties
2.5 for SP3 sions and prestressing levels were kept
Fig. load-deflection behavior with and
and without the deck
deck ties for
spandrel SP3. Mid-span deflection (in) constant for all five cases. All analyses
Mid-span lateral displacement (mm) were conducted using a 45 ft (13.7 m)
Fig. 13 Predicted load-deflection behavior with and without the deck ties for SP3 span.
-127 -114 -102 -89 -76 -64 -51 -38 -25 -13 0
200 890
The compact section geometry and
Mid-span lateral displacement (mm) reinforcement layouts were proposed,
SP3
180 -127 -114 -102 -89 -76 -64 -51 -38 -25 -13 0 801 designed, and detailed by the PCI Pro-
200 890 ducer Members sponsoring the study.
160 712
End reaction of spandrel (kN)
(kips) (kips)

180
SP3
801 Longitudinal reinforcement complied
140 623 with ACI 318-05 requirements. Shear
160 712
of spandrel

End reaction of spandrel (kN)

and torsion design of the first reinforce-

120 534
140 623 ment case (utilizing closed stirrups) fol-
100 445 lowed the procedure recommended by
of spandrel

120 534 Zia and Hsu.5

80 356
End reaction

100 445 The remaining four reinforcement

60 267 configurations are variations of the first.
80 356
End reaction

Experimental Figure 15 shows the reinforcement de-

40 178
60 ANACAP: Without Lateral Ties 267 tails of the proposed compact section.
20 Experimental
ANACAP: With Lateral Ties 89 All details shown in the figure, with the
40 178
ANACAP: Without Lateral Ties exception of the web reinforcement, are
0 0
20 ANACAP: 89 common to all other spandrels evalu-
-5 -4.5 -4 -3.5With-3Lateral Ties -2
-2.5 -1.5 -1 -0.5 0
ated in this study. Figure 16 shows the
0 Mid-span lateral displacement (in) 0
details of the transverse reinforcement
-5 -4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0
used in all five cases.
Fig. 14 Predicted lateral displacements at midspan
Mid-span lateral displacement (in) for SP3
Cases 1 and 2 are included to dem-
Fig. 14 Predicted lateral displacements at midspan for SP3
onstrate the efficiency of open vertical
Fig. 14. Predicted lateral displacements at midspan for spandrel SP3. stirrups with 90-degree hooks at the
top and bottom. Case 1 also serves as
a basis for comparison with the other
lateral displacements 45% to 65%, de- As expected, the actual behavior of the four cases because it is the only case
pending on the load level. spandrel falls between the two extreme currently accepted in common prac-
The finite-element analysis indi- cases considered in the analysis. Such tice. The influence of hooking the ver-
cates that the only significant effect a phenomenon indicates that the as- tical web reinforcement at the front
of the deck ties is the restraint of lat- sumed spring stiffness was much high- face of the spandrel is investigated by
eral displacements induced by bending er than the actual stiffness provided by comparison of cases 3 and 4.
about the weak axis of the spandrel. the deck ties. In these cases, welded-wire rein-

86 PCI JOURNAL
forcement (WWR) was utilized as tor- Vertical Deflections for the respective precast, prestressed
sional-shear reinforcement at the back concrete spandrel. Linear behavior
face of the spandrel. WWR was se- Figure 18 shows the vertical-load- was predicted up to the initiation of
lected that had the same steel area per deflection behaviors of the five com- the first flexural crack at an end reac-
linear foot in the transverse direction as pact, L-shaped spandrels for the dif- tion of 45 kip (200 kN), followed by a
was provided in the second case. The ferent reinforcement configurations. nonlinear behavior to failure. All five
final case, case 5, was reinforced iden- Identical precracking and postcracking cases demonstrate substantial ductil-
tically to case 4. However, additional stiffenesses were predicted, regardless ity prior to failure. While the deflec-
top horizontal reinforcement connect- of the web reinforcement configura- tion behavior of the spandrel certainly
ing the transverse reinforcement at the tion. does not provide great insight into
front and back face of the spandrel was All five load-deflection curves dem- the effectiveness of a particular shear
provided. onstrate a typical flexural response and torsion reinforcement configura-
Half of the compact L-shaped, pre-
cast, prestressed concrete spandrel
was modeled using 1472 twenty-node 16"
2#11
brick elements, as shown in Fig. 17.
2" 6'
The spandrel web was divided into 1#6
12"
four equal layers within its thickness
3#5 12" 6'
to accurately model the shear-torsion- #3 2@4" c/c each end & 16"
2#11
Bal. @16" c/c #4 2@4" c/c each end &
al stress distribution within the width 28"
Bal. @8" c/c 6'
2"
of the spandrel. For all cases, the de- 2" 1#6
12"
#4 8@4" c/c each end &
sign concrete compressive strength Bal. @12" c/c 3#5 12" 6'
2"end & 1#4 8"
and modulus of elasticity were taken #3 2@4" c/c each
2" c/c #4 2@4" 4' each end &
28" Bal. @16" 4" c/c
as 6000 psi (41 MPa) and 4200 ksi 2" (TYP.) Bal. @8" c/c

(29 GPa), respectively. Grade 60 2" 24"

#4 8@4" c/c each end &
mild-steel reinforcement with a yield Bal. @12"0.5"
c/c (12.7 mm) diameter prestressing strands, Pull=31.56
8" kips (802 kN)
2" 1#4
2" 4'
strength and modulus of elasticity of 4"
0.5" (12.7 mm) diameter prestressing strands, Pull=31.56 kips (802 kN)
2" (TYP.)
Debonded at 4 ft from beam ends
60 ksi (414 MPa) and 29,000 ksi (200 24"
GPa), respectively, was utilized as the Fig. 15 Reinforcement
0.5"details of diameter
(12.7 mm) the compact section
prestressing strands, Pull=31.56 kips (802 kN)
Fig. 15. Reinforcement details
Note: 1 in=25.4 mm; 1 of the compact
ft=304.8 mm; #4=12M;section. Note:
#5=16M; ' = ft; " = in.; 1 ft =
#6=19M
non-prestressed reinforcement. 304.8 mm; 1 in. = 25.4 mm;0.5" 1 (12.7
lb = mm) diameterkN;
0.00448 prestressing
#4 = strands, #5
12M; Pull=31.56
= 16M;kips (802
#6 = kN)19M.
Debonded at 4 ft from beam ends
Seventeen 0.5-in.-diameter (13 mm) 16"
low-relaxation strands with a nominal Fig. 15 Reinforcement details of the compact section
Note: 1 in=25.4 mm; 1 ft=304.8 mm; #4=12M; #5=16M; #6=19M
cross-sectional area of 0.167 in.2 (107
mm2) were used within the spandrel. 28" 16"
#4 2@4" c/c each end & #4 2@4" c/c each end &
Prestressing strands were modeled Bal. @8" c/c Bal. @8" c/c
#3 2@4" c/c each end &
Bal. @16" c/c
using the same approach as described 8"
for spandrels SP3 and SP4. Prestress- 28" #4 2@4" c/c each end & #4 2@4" c/c each end &
24" Bal. @8" c/c Bal. @8" c/c
#3 2@4" c/c each end &
ing losses of 15% were assumed in the Bal. @16" c/c
8"
analysis.
Two prestressing strands were 24"

debonded for the first 4 ft (1219 mm)

at each end of the spandrel to avoid 6x6 / W4xW4 #4 2@4" c/c each end & 6x6 / W4xW4 #4 2@4" c/c each end &
Bal. @8" c/c Bal. @8" c/c
crushing of the concrete in the end re- Mesh Mesh
gion. The spandrel was restrained ver- 6x6 / W4xW4 #4 2@4" c/c each end & 6x6 / W4xW4 #4 2@4" c/c each end &
Bal. @8" c/c Bal. @8" c/c
tically throughout the width of the web Mesh Mesh

for the first 12 in. (305 mm) along the #4 2@4" c/c each end &
ends. Lateral restraints were provided 6 Bal. @8" c/c

in. (152 mm) from each end at the top #4 2@4" c/c each end &
Bal. @8" c/c
and bottom of the spandrel. 6x6 / W4xW4 #4 2@4" c/c each end &
Nine spring supports were provided Mesh Bal. @8" c/c
#4 2@4" c/c each end &
along the length of the spandrel to sim- 6x6 / W4xW4
Bal. @8" c/c
Mesh
ulate deck ties. The springs were posi-
tioned at the top front face of the span-
drel with an axial stiffness of 21,750 Fig. 16 Different web reinforcement configurations for the compact section
kip/in. (3809 kN/m), as discussed. Note: 1 in=25.4 mm; 1 ft=304.8 mm; #4=12M; #5=16M; #6=19M
Fig. 16 Different web reinforcement configurations for the compact section
Load was applied gradually using a Fig. 16. Different
Note: web reinforcement
1 in=25.4 mm; 1 ft=304.8configurations for the
mm; #4=12M; #5=16M; compact section. Note:
#6=19M
step-by-step analysis, as described for ' = ft; " = in.; 1 ft = 304.8 mm; 1 in. = 25.4 mm; 1 lb = 0.00448 kN; #4 = 12M;
spandrels SP3 and SP4. #5 = 16M; #6 = 19M.

March–April 2007 87
 A

A
Z
7x4"
Z
x4"
X
12" 16.125" 12x4.6875" 8x7.03125" 8x7.03125" 8x7.03125"
X 3.75" 3.75" 3.75" 3.75" 1.875"
12" 16.125" 12x4.6875" 8x7.03125" 8x7.03125"
270" 8x7.03125"
3.75" 3.75" 3.75" 3.75" 1.875"
270" A
A
Elevation of the spandrel beam
Elevation of the spandrel beam
4x4"
4x4"

7x4"
7x4" Z
Z 8"
8"
Y
Y

6x4" 6x4"

SECT. A-A SECT. A-A

Fig. 17. Finite-element-model mesh dimensions used in modeling the compact spandrels. Note: ' = ft; " = in.; 1 ft = 304.8 mm;
1 in. = 25.4 mm
Fig. 17 Mesh
Fig. 17 Mesh dimensions useddimensions
in modelingused in modeling
the compact the beams
spandrel compact spandrel beams
Note: 1 in.=25.4 mm 1 in.=25.4 mm
Note:
Lateral Displacements
Mid-span deflection (mm) deflection (mm)
Mid-span When lateral displacements at mid-
0 25 51 76 102 127 152 178 203 229 254
0 25 51 76 102 127 152 178 203 229 span
254 at ultimate load are considered,
140 623
140 the influence
623 of the five reinforcement
configurations becomes much more
120 pronounced, as shown in Fig. 19. 534
120 534
In the FEM, lateral displacements are
End reaction of spandrel (kN)
End reaction of spandrel (kip)

100 445 predicted at the bottom edge of the web End reaction of spandrel (kN)
End reaction of spandrel (kip)

100 Case: 1
on the445back face of the spandrel. Dis-
Case: 1
80 356 placements toward the ledge side are
Case: 2 considered
80 356 positive, while those away
60 Service load level Case: 2
267 from the ledge side are negative. While
Service load level
Case: 3 the ultimate end reactions sustained by
60 267
40 178
Case: 3 the five cases are all similar, the lateral
Case: 4 displacements predicted for each case
40 178
vary substantially.
20 89
Case: 4 Case 1 (using closed stirrups) dem-
Case: 5
20 onstrates
89 the least lateral displacement
0 0
Case: 5 of all cases. The maximum predicted
0 1 2 3 4 5 6 7 8 9 10 lateral displacement at midspan was
0 0
Mid-span deflection (in) about 0.8 in. (20 mm). Absence of the
0 1 2 3 4 5 6 7 8 9 10
Fig. 18 Predicted load-deflection behavior using different web reinforcement configurations
hooks on the front vertical web rein-
Fig. 18. Predicted load-deflection behavior using differentdeflection
Mid-span web reinforcement
(in) forcement (case 4) resulted in larger
configurations. lateral deformations of the spandrel
Fig. 18 Predicted load-deflection behavior using different web reinforcement configurations
than in other cases. The maximum
lateral displacement in this case was
tion, the analysis indicates that all five load, while cases 1, 2, and 5 sustained a nearly three times that predicted using
reinforcement cases were sufficient slightly higher end reaction of 105 kip closed stirrups.
for preventing premature end-region (467 kN). Ultimate vertical deflections This behavior demonstrates that the
failures. for the five cases ranged from 5.5 in. lateral and torsional stiffness of the
Cases 3 and 4 sustained an ultimate to 6.8 in. (140 mm to 173 mm), with member is significantly influenced by
applied end reaction of approximately cases 1, 2, and 5 outperforming cases the amount of reinforcement crossing
100 kip (445 kN), not including dead 3 and 4. the top and bottom faces of the web.

88 PCI JOURNAL
Interestingly, the lateral displacement
results from case 5 are nearly identi- Lateral displacement at mid-span(mm)
cal to those from case 1. Therefore, -64 -57 -51 -44 -38 -32 -25 -19 -13 -6 0
Lateral displacement at mid-span(mm)
the reinforcement crossing the top web 140 623
face is more significant than that cross- -64 -57 -51 -44 -38 -32 -25 -19 -13 -6 0
ing the bottom web face. On the other 140 623
120 534
hand, under service load, the lateral

End reaction of spandrel (kN)

End reaction of spandrel (kip)
displacement of case 4 is about 0.4 in. 120 534
100 445
(10 mm), almost twice that of the other

End reaction of spandrel (kN)

End reaction of spandrel (kip)
four cases. 100 Case: 1 445
80 356
Crack Pattern Case: 1 2 Service load level
Case:
80 356
A similar crack pattern was predicted 60 267
Case: 2 Service load level
for all five cases, regardless of the web Case: 3
60 267
reinforcement configuration. Flexural 40 178
cracks were initiated at an end reaction Case: 3
Case: 4
of 45 kip (200 kN), as shown in Fig. 40 178
20 89
20. These cracks were first initiated at Case: 4
Case: 5
the back face of the spandrel as a result 20 89
0 0
of the out-of-plane bending behavior Case: 5
-2.5 -2.25 -2 -1.75 -1.5 -1.25 -1 -0.75 -0.5 -0.25 0
of the spandrel. The cracks started to 0 0
Lateral displacement at mid-span(in)
propagate toward the ledge of the span- -2.5 -2.25 -2 -1.75 -1.5 -1.25 -1 -0.75 -0.5 -0.25 0
drel as the applied load was increased.Fig. 19 Predicted lateral displacements
Lateral at midspan
displacement using different web reinforcement configurations
Fig. 19. Predicted lateral displacements atat mid-span(in)
midspan using different web reinforcement
Localized cracks around the spring configurations.
supports were also observed Fig. 19 Predicted lateral displacements
as the at midspan using different web reinforcement configurations
Plane of
result of stress concentrations at these symmetry
locations. Diagonal cracks at the span- Plane of
drels’ ends started to appear shortly symmetry
after the initiation of the flexural cracks
at an end reaction of 55 kip (245 kN).
As the load was increased, the cracks
were further extended and diagonal
tension cracks developed farther from
the support.
In general, extensive diagonal and
rainbow cracking was predicted by
46 kips (205 kN)
the FEMs along the front faces of the
spandrels due to the combined torsion- 46 kips (205 kN)
al and shear stresses. The back faces
of the spandrels showed rather evenly
spaced vertical cracking, mostly due to 56 kips (249 kN)
the flexural effect (because the stresses
due to torsion and shear counteracted 56 kips (249 kN)
each other). The vertical cracks were
tallest toward the center and gradually
decreased in height toward the end of 70 kips (311 kN)
the spandrel. Minor diagonal cracks
were also predicted by the FEM at the Fig. 20 Typical crack pattern at different loading stages
70 kips (311 kN)
back faces of the spandrels toward their
ends. Fig. 20. Typical
Fig.crack patterncrack
20 Typical at different
patternloading stages.
at different loading stages

Shear Stresses
Figure 21 shows the ultimate shear
stress distributions at the ends of each
spandrel for the different reinforce-
ment configurations. The use of open
vertical stirrups with 90-degree hooks
at the top and bottom did not have any
detrimental effect on the induced shear 4 and 5 indicates that absence of the
stresses in the spandrels (compared horizontal top web reinforcement in-
with the case with closed stirrups). The creases the concrete shear stress 20%.
FEM predicted the same level of stress It was also observed, by comparing the
for both cases 1 and 2. induced shear stresses in cases 3 and 4,
A direct comparison between cases that the presence of hooks enhances the

March–April 2007 89
 behavior and reduces the shear stresses
20%. Obviously, this is because the
hooks provided more anchorage for the
web reinforcement.
Case 1 Case 2
Failure Mode
Flexural failure due to crushing of
the concrete at the midspan section
of the spandrel was predicted by the
1500 psi
FEM for all five cases. Failure loads
1500 psi
were nearly identical for all speci-
mens. Cases 1 and 4 exhibited the
highest and lowest ultimate load-car-
rying capacity, respectively. Never-
theless, the variation of the ultimate
load between these two extreme cases
Case 3 Case 4 was less than 12 kip (54 kN), which
corresponds to approximately 6% of
the capacity of the spandrel. Finite-
element analysis was terminated when
the principal compressive strains ex-
ceeded 0.003 according to ACI 318-
1900 psi 2300 psi 05.
It was observed that the principal
compressive strains were much higher
at the front face of the spandrel than
at the back face due to out-of-plane
bending behavior of the spandrel. Such
Case 5 behavior was highly pronounced for
the spandrels analyzed without deck
ties. At the onset of flexural failure,
the maximum principal compressive
strains along the diagonal compression
strut were less than 0.002, which is
1900 psi recommended by other researchers for
shear compression failure.7
Forced Shear
Fig. 21 Shear stress distribution at the end of the spandrel using different reinforcement configurations at an end Failure Mode
reaction of 99 kips (440 kN)
Fig.1 21.
Note: psi =Shear stress
0.006895 MPadistribution at the end of the spandrel using different To further
examine the influence
reinforcement configurations at an end reaction of 99 kip (440 kN). Note: 1 psi = of the different web reinforcement
0.006895 MPa.
configurations on the shear-torsional

Table 3. Results of the Finite Element Analysis for Cases 1, 2, and 4 for Compact Sections
Case
Flexural Reinforcement Ru, kip Failure Mode
No.

1 Normal 104 Flexural failure

1 Nine #11 bars were added at midspan 133 Shear-compression failure

2 Normal 104 Flexural

2 Nine #11 bars were added at midspan 123 Shear-compression failure

4 Normal 99 Flexural

4 Nine #11 bars were added at midspan 110 Shear-compression failure

Note: Ru = the end reaction of the spandrel at ultimate; 1 kip = 4.448 kN.

90 PCI JOURNAL
strength of compact, L-shaped span- Midspan section
drels, additional top and bottom flex-
ural reinforcement was provided at Midspan section
midspan. Placement of this additional
reinforcement was limited to between
the quarter points L/4 and the midpoint
L/2 to eliminate the possibility of af-
fecting the shear-torsion strength of the
spandrels at their end regions (L is the
span of the spandrel).
It was intended that this additional
reinforcement would prevent the flex- Zone of additional
ural failure mode observed previously, flexural
reinforcement
Zone of additional
allowing a mode governed by shear
flexural
and torsion to develop. Cases 1, 2, and reinforcement
4 were all reanalyzed with the addition-
al flexural reinforcement, and Table 3 Principal compressive strains
summarizes the results of the analysis. along the diagonal exceeds 0.002
In all three of these cases, fail- Support reaction
Principal compressive strains
ures occurred in the end regions and along the diagonal exceeds 0.002
Support reaction
were due to crushing of the con- Fig. 22. Typical principal compressive strain distribution at ultimate after adding
Fig.
crete along the primary compressive22 Typical principal compressive strain distribution at ultimate after adding flexural reinforcement at mids
flexural reinforcement at midspan.
strut, as shown in Fig. 22. Finite-
Fig. 22 Typical principal compressive strain distribution at ultimate after adding flexural reinforcement at m
element analysis was terminated when Lateral displacement at mid-span(mm)
the principal compressive strains along -64 -57 -51 -44 -38 -32 -25 -19 -13 -6 0
the compressive strut reached a value Lateral displacement at mid-span(mm)
140 623
of 0.002.7 Figure 23 shows the predict- -64 kips
133 -57(592-51
kN) -44 -38 -32 -25 -19 -13 -6 0
ed lateral displacements at midspan. 140 123 kips (547 kN) 623
120 534
The maximum predicted end reaction 133 kips (592 kN)
End reaction of spandrel (kip)

End reaction of spandrel (kN)

for the case with closed stirrups (case 120 110 kips
123 (489
kips kN)kN)
(547
1) was 133 kip (592 kN), which did not 100 445534
End reaction of spandrel (kip)

End reaction of spandrel (kN)

include dead load. 110 kips (489 kN)
100
80 356445
Finite-element analysis indicated
that using open vertical stirrups with Case: 1
80 356
90-degree hooks instead of closed stir- 60 267
rups did not have a dramatic effect on Case: 1
60 Case: 2
the strength of L-shaped spandrels. For 40 178267
case 2, the FEM predicted a reduction
Case: 2 Additional flexural reinforcement
of 8% in the ultimate load-carrying 20
40
89
178
Case: 4 are provided
capacity of the spandrel. Using open,
Additional flexural reinforcement
unhooked web reinforcement (case 4)
020 Case: 4 are provided 0 89
reduced the shear capacity of the span-
-2.5 -2.25 -2 -1.75 -1.5 -1.25 -1 -0.75 -0.5 -0.25 0
drel 17% compared with case 1. Based 0 0
on these results, the analysis indicates Lateral displacement at mid-span(in)
-2.5 -2.25 -2 -1.75 -1.5 -1.25 -1 -0.75 -0.5 -0.25 0
that it is possible to use open web re- Fig. 23 Predicted lateral displacements and failure loads for Cases 1, 2 and 4 after adding
inforcement effectively in compact Lateral displacement at mid-span(in)
flexural reinforcement
L-shaped spandrels, provided that the Fig. Fig.23
23.Predicted
Predictedlateral
lateraldisplacements
displacementsand andfailure
failureloads
loads for
for Cases
cases 1,1,2,
2 and 4 after
after adding
designer accounts for reductions in the adding flexural
flexuralreinforcement
reinforcement.
shear-torsion strength of the spandrel.

prestressed concrete spandrels In this case, web reinforcement

Conclusions
subjected to combined shear, configurations have a trivial ef-
Based on the results of this investi- bending, and torsion. fect on serviceability as well as
gation, the following conclusions are • For the compact, L-shaped span- on the spandrel’s ultimate load-
drawn: drels spanning 45 ft (13.7 m), carrying capacity.
• FEM is capable of accurately typically used by the precast/ • The out-of-plane bending be-
predicting the response, up to prestressed concrete industry, havior of compact, L-shaped
failure, of L-shaped, precast, flexural failure controls design. spandrels is highly dependent

March–April 2007 91
 on the configuration of the web
reinforcement. The absence of
hooks in the front vertical web
reinforcement (as in case 4) may
result in larger lateral deforma-
tions of the spandrel compared
with spandrels using closed stir-
rups, without reductions in load-
carrying capacity.
• Deck ties reduce the lateral dis-
placements induced in L-shaped
spandrels typically caused by
bending about the weak axis of
the spandrel. The presence of
ties does not have any significant
effect on a spandrel’s ultimate
load-carrying capacity or its fail-
ure mode.
• The use of open vertical stirrups
with 90-degree hooks at the top
and bottom did not have any det-
rimental effect on the induced
shear stresses at spandrel ends
(compared with closed stirrups).
• The absence of horizontal top
web reinforcement increases the
shear stress in the spandrel 20%.
Conversely, the presence of
hooks in the web reinforcement
at the front face enhanced the
spandrel’s behavior and reduced
its shear stresses 20%.
• Using additional reinforcement to
prevent flexural failure led to com-
pression shear failure at the end
regions of the compact L-shaped
spandrels. Finite-element analy-
sis indicated that the use of open,
unhooked web reinforcement re-
duces the spandrel’s shear strength
17% compared with a closed-stir-
92 PCI JOURNAL
rups configuration. The spandrel’s
shear strength reduction is about
half as much when open vertical
stirrups with 90-degree hooks re-
place closed stirrups.
Acknowledgments
This project was conducted while
Tarek Hassan was a visiting scholar at
North Carolina State University. The
project was jointly sponsored by Harry
Gleich of Metromont Corp. in Green-
ville, S.C., and Don Logan of Stress-
con Corp. in Colorado Springs, Colo.
The authors are grateful for the sup-
port and guidance provided by all of
the PCI Producer Members throughout
the duration of the project. In addition,
the authors acknowledge the efforts of
Gary Klein of Wiss, Janney, Elstner
Associates Inc. for his valuable com-
ments during the research program.
References
1. ACI Committee 318. 2005. Building Code
Requirements for Structural Concrete
(ACI 318-05) and Commentary (ACI
318R-05). Farmington Hills, MI: ACI.
2. CSA Standards Committee 23. 2004.
Design of Concrete Structures (A23.3-
04). 5th ed. Rexdale, ON, Canada: Ca-
nadian Standards Association (CSA).
3. Lucier, G., S. Rizkalla, and P. Zia. 2006.
Behavior of Full-Size Spandrel Beams.
Technical report IS-06-01, Constructed
Facilities Laboratory, North Carolina
State University, Raleigh, NC.
4. Zia, P., and W. D. McGee. 1974. Tor-
sion Design of Prestressed Concrete.
PCI Journal, V. 19, No. 2 (March–
April): pp. 46–65.
5. Zia, P., and T. T. C. Hsu. 1978. Design
for Torsion and Shear in Prestressed
Concrete. Paper presented at the
American Society of Civil Engineers
(ASCE) convention, October 16–20, in
Chicago, IL, reprint #3424.
6. Zia, P., and T. T. C. Hsu. 2004. Design
for Torsion and Shear in Prestressed Con-
crete Flexural Members. PCI Journal, V.
49, No. 3 (May–June): pp. 34–42.
7. Rahal, K. N., and M. P. Collins. 2006.
Compatibility Torsion in Spandrel Beams
Using Modified Compression Field The-
ory. ACI Structural Journal, V. 103, No.
3 (May–June): pp. 328–338.
8. Klein, G. J. 1986. Design of Spandrel
Beams. Report on PCI Research Proj-
ect No. 5. Chicago, IL: PCI.
9. Ngo, D., and A. C. Scordelis. 1967. Fi-
nite Element Analysis of Reinforced
Concrete Beams. ACI Structural Journal,
V. 64, No. 3 (May–June): pp. 152–163.
10. ANATECH Corp. 2003. ANATECH
Concrete Analysis Program (ANACAP)
Version 2.2.3 Reference Manuals.
11. Rashid, Y. R. 1968. Analysis of Pre-
stressed Concrete Pressure Vessels.
Nuclear Engineering and Design, V. 7,
No. 4: pp. 334–344.
12. Yazdani, N., and J. Ach. 2004. Behav-
ior of Thin Spandrel Beams under Tor-
sion. PCI Daniel P. Jenny Fellowship
report, Florida State University, Talla-
hassee, FL.
13. Cleland, N. M. 1984. Identification
of Secondary Behavior in Combined
Bending, Shear, and Torsion of Rein-
forced Concrete Ledger Beams. Ph.D.
diss., University of Virginia.