Specially Funded R&D Program

PCISFRAD Project No. 5

Summary Paper

Design of Spandrel Beams

by

Gary J. Klein
Consultant
bliss, Janney, Elstner
Associates, Inc.
Northbrook, Illinois

Note: This summary paper is a slightly con- research and the preparation of the final reports
densed version of PCISFRAD Project No. 5, for each of the PCISFRAD projects were per-
"Design of Spandrel Beams." The full report is formed under the general guidance and direc-
available from PCI Headquarters at $8.00 to tion of selected industry Steering Committees.
firms supporting the sponsored research, However, it should be recognized that the re-
$12.00 to PCI Members (non-supporting firms) search conclusions and recommendations are
and $24,00 to non-PCI Members. those of the researchers. The results of the re-
The summary paper, and the full report, are search are made available to producers, en-
based on a research project supported by the gineers and others to use with appropriate en-
PCI Specially Funded Research and Develop- gineering judgment similar to that applied to any
ment (PCISFRAD) Program. The conduct of the new technical information.

76
 CONTENTS
1. Summary and Conclusions ........................... 78

2. Introduction ........................................ 79

3. Background Research ............................... 80

— General Design Considerations
— Flexure
— Shear and Torsion
-- Beam End Design
— Beam Ledges
— Beam Pockets
4. Finite Element Model Studies ......................... 87
— Spandrel Beam Behavior
— Transfer of Ledge Loads to Web

5. Load Tests ... ...................................... 91

—Test Specimens
—Test Procedure
— Behavior and Strength of Test Specimens

6. Analysis and Discussion .............................104

— General Design Considerations
— Flexure
— Shear and Torsion
— Beam End Design
— Beam Ledges
— Beam Pockets

7. Findings and Recommendations ......................112

Acknowledgments.....................................113
References...........................................114

Appendix A — Notation .................................115

Appendix B — Spandrel Design Checklist .................116

AppendixC — Design Examples .........................117

— Example 1. L-Beam for Parking Structure
— Example 2. Pocket Spandrel for Parking Structure

PCI JOURNAL/September-October 1986 77

 1. SUMMARY AND CONCLUSIONS
A study of the behavior and design 01 lateral displacement induced by bend-
precast spandrel beams was undertaken. ing about the weak principal axis.
This research project was primarily di- • Shear and torsion design proce-
rected toward spandrel beams com- dures for prestressed spandrels which
monly used in parking structures. Both consider a concrete contribution have
L-beams and pocket spandrels were in- been verified by two tests.
eluded in the study. • An approach for considering the
The research included background effect of the pocket on the shear strength
investigation of design practices, of pocket spandrels has been proposed.
analytical studies using finite element While the accuracy of this approach has
models, and full-scale load tests of two not been fully verified by tests, it is be-
L-beams and one pocket spandrel. All lieved to be conservative.
three test specimens were 72 in. high, 8 • With regard to detailing practices,
in. wide and 28 ft long. The target de- it was found that the torsional response
sign loads were based on 90 psf dead of deep spandrels is dominated by out-
load and 50 psf live load, which are typi- of-plane bending. The use of lapped-
cal for a double tee parking structure splice stirrups and longitudinal rein-
with 60 ft spans, forcing bars without hooks does not ap-
The background research revealed pear to have any detrimental effect.
that industry practices and published • Two independent design checks in
procedures vary with respect to several the end region of spandrels are recom-
fundamental aspects of spandrel beam mended. First, reinforcement should be
designn. Behavior near the end regions is provided to resist out-of-plane bending
not well understood, nor is the influence caused by the horizontal torsional
of connections to deck elements. In equilibrium reactions. This reinforce-
general, the design of beam ledges is not ment is not additive to the reinforce-
consistently handled; in particular, ment for internal torsion. Second, the
there is no consensus on the design of longitudinal reinforcement in the bear-
hanger reinforcement for ledge-to-web ing area should be sufficiently de-
attachment, Also, the ACI Building veloped to resist the external normal
Code (ACI 318-83) does not address force, in addition to the tension induced
combined shear and torsion in pre- by the vertical reaction,
stressed beams. Designers rely on sev- • The eccentricity of the ledge load
eral research reports that give design cannot be neglected in the design of
recommendations. hanger reinforcement for ledge-to-web
Ledge-to-web attachment and be- attachment, Nonetheless, not all of the
havior near the end region of spandrels load acting on the ledge is suspended
were identified as the key issues and from the web and the effective eccen-
were the primary focus of this research. tricity of the ledge load is significantly
The analytical studies and laboratory reduced due to torsion within the ledge.
testing program yielded several signifi- A design procedure which considers
cant findings: these effects has been recommended. In
• Contrary to several published de- addition, it was determined that hanger
sign examples, a critical section for reinforcement is not additive to shear
shear and torsion at the face of the sup- and torsion reinforcement.
port should be considered. • The PCI design equations for
• Connections to deck elements do punching shear strength of beam ledges
not substantially reduce torsion; how- may be unconservative. Further re-
ever, they are effective in restraining search in this area is recommended.

78
 In conclusion, this research has recommendations will be of immediate
clarified many of the questions relating benefit to the precast and prestressed
to spandrel beam design and the design concrete industry.

2. INTRODUCTION
Spandrel beams are one of the most parking structures to serve both load-
complex elements in precast construc- carrying and railing functions. Both
tion. Industry practices and published L-beams and spandrel beams with
procedures vary with respect to several pockets for tee stem hearings (pocket
fundamental aspects of their design. spandrels) were included in the pro-
PCI Specially Funded Research and gram. Fig. 1 shows typical cross sections
Development Project No. 5 investigated of these types of beams.
the behavior and design. of precast The findings of this research gener-
spandrel beams. ally apply to both prestressed and non-
The research program was primarily prestressed spandrels, but may not be
directed toward deep and slender span- applicable to spandrel beams of radi-
drels such as those commonly used in cally different geometric configuration

6" TO 40" B"TO 10"

m
x
0
a
0.
a

L- BEAM POCKET SPANDREL

Fig. 1. Typical spandrel sections studied in research program.

PCI JOURNALJSeptember-October 1986 79

 or load level. Furthermore, while this following objectives:
research is believed to be reasonably • Study of design requirements and
comprehensive, not all aspects of span- practices to determine the state-of-the-
drel beam design are covered. art of spandrel beam design.
In particular, the research does not • Analytical studies using finite ele-
address spandrel beam design as part of ment models of an L-beam and pocket
a lateral load resisting frame, nor the spandrel.
effects of volume change on design and • Full-scale tests of two L-beams and
detailing of spandrels. Also, handling one pocket spandrel designed using
and vehicular impact loads are not dis- state-of-the-art methods.
cussed. These considerations can be The following sections of this report
very important, but are considered be- describe the research, analyze the find-
yond the scope of this research. ings, and provide design recommen-
The research program included the dations.

3..BACKGROUND RESEARCH
The background research included a plate, must be considered as part of the
review of code requirements, published shear/torsion load. This consensus is
guides and research reports on spandrel contrary to the published design exam-
beam design. Questionnaires regarding ples in Section 4.4 of the PCI Design
design methods for L-heams and pocket Handbook' and Example 14.2 in the
spandrels were sent to industry design- PCA Notes on ACI 318-83_' ACI 318-83"
ers. The following discussion on span- does not address indirectly loaded
drel beam design is based on this re- beams; however, Section 11.1.2 of the
search. Commentary recommends special con-
sideration for concentrated loads near
General Design Considerations supports.
Equivalent Uniform Load — It is
Critical Section — In most precast common practice to simplify the
beams, the loads and reactions are analysis by replacing concentrated loads
applied at the top and bottom of the with equivalent uniform loads. Some
beam, respectively. Such beams are said designers increase the equivalent uni-
to be "directly loaded." Spandrel form floor load such that the shear and
beams, on the other hand, are indirectly torsion is correct at the critical section at
loaded, and the additional shear capac- the inside edge of the base plate, i.e.,
ity due to arch action near the support is the basic equivalent uniform load is
not available.' Therefore, design for multiplied by the ratio of grid span to
shear and torsion forces at a distance design span.
d(h/2 for prestressed spandrels) from the Eccentricity Contributing to Torsion
support may not be appropriate. Fig. 2 — Typically, the ledge loads are po-
shows potential critical inclined sec- sitioned at the centerline of bearing (al-
tions which carr y all the concentrated lowing for volume change and fabrica-
loads acting on the ledge rather than just tion and erection tolerances) or at the
loads farther than d from the support. outer one-quarter point of the ledge.
The consensus among designers is The former approach is generally pre-
that all loads acting on the ledge inside ferred because an increase in ledge
the critical section, based on inclined projection does not necessarily require
cracking from the edge of the beam base an increase in torsional load. The ec-

80
Fig. 2. Inclined failure planes in an "indirectly loaded" spandrel.

centricity contributing to torsion in a iWEB

spandrel is the distance from the cen-
terline of the web to the applied load, as
shown in Fig. 3.
Theoretically, the eccentricity should
be measured relative to the shear center,
which, for an uncracked L-beam section,
is slightly inside the centerline of the
web. However, this difference is negli-
gible in deep spandrels. Further, experi-
mental results are not consistent with
the theoretical prediction of shear
center location based on the uncracked ^— CONNECTION TO DECK
cross section.'
Influence of Deck Connections -
Prior to connection of the double tees or
topping to the spandrels, torsion can be SHEAR CEN'ER
computed as a product of the dead load
and the eccentricity between the — rt BEARING
e
applied load and centerline of the web. H
After connections to deck elements are
made, however, the applied live load
torsion may be partially counteracted by FRICTION AT BEARING
the horizontal force due to friction at the
bearing pads coupled with restraint at
the deck connections (Fig. 3). However,
most practitioners believe that it is in-
appropriate to rely on a soft bearing pad
for this purpose. In addition, recent re-
search6 indicates that the effective fric-
tion at the bearing pad may be 5 percent Fig. 3. Eccentricity contributing to torsion in
or less of the gravity load. spandrel beam.

PCI JOURNALJSeptember-October 1966 81

Flexure the influence of principal axes rotation
The flexural design of spandrels gen- on horizontal displacement of spandrels.
erally follows ACI and PCI procedures As shown in Fig. 4, a component of the
for bending about the horizontal and vertical load acts along the weak axis in-
vertical axes. However, L-shaped ducing an outward horizontal displace-
spandrel beams do not have symmetry ment. All loading prior to making dia-
about either axis. The principal axes are phragm connections can cause horizontal
rotated slightly from the vertical and displacement. Cleland 5 found that this
horizontal axes, as shown in Fig. 4. The was the most dominant behavior of long
influence of this rotation on bending slender spandrels and suggests a princi-
about the horizontal axis can be neg- pal axes analysis when the span length
lected for deep spandrel beams. For is 40 to 50 times the web width, de-
shallow spandrels, particularly those pending on the intermediate support
employing prestressing, this influence conditions.
should be considered. In general, detailing practice follows
Perhaps more important, however, is the ACI Code. One noteworthy excep-
tion pertains to Section 10.6.7 of ACI
318-83 which is applicable to nonpre-
stressed spandrels. This provision re-
quires that reinforcement be placed in
the side faces of webs more than 3 ft
deep. The reinforcement is to be dis-
tributed in the zone of flexural tension
with a spacing not more than the web
width, nor 12 in. Designers do not often
check this provision; instead, rein-
forcement in the side faces of the web is
designed to resist torsion or handling.

Shear and Torsion

Prestressed Spandrels — The ACI
Code does not address torsion in pre-
X p stressed concrete. A procedure for tor-
X -_ X sion design of prestressed concrete,
XP which is an extension of the AC1 provi-
sions of torsion for nonprestressed con-
crete, was developed by Zia and
McGee.' The second edition of the PCI
Design Handbook included a modified
VxP version of the Zia and McGee method"
The PCI procedure uses a simplified
method for computing torsional stress
which is conservative for most spandrel
beams.
A further refinement of these methods
was subsequently developed by Zia and
Hsu.9 While the general design ap-
Y VP proach follows that of Zia-McGee and
PCI, new expressions are proposed for
Fig. 4. Principal axes of an L-beam. torsion/shear interaction and minimum

82
torsion reinforcement. The Zia-Hsu quirement is primarily directed at hol-
equations are expressed in terms of low box sections and solid sections
forces and moments rather than nominal subjected primarily to torsion. In these
stresses, which is more consistent with members, the side cover spalls off, ren-
the current ACI Code. dering lapped-spliced stirrups ineffec-
Most designers follow one of these tive. This type of behavior is unlikely in
three similar procedures. Practices vary deep spandrel beams, and transverse
with respect to the design of longitudi- reinforcement is often provided by pairs
nal reinforcement for torsion. Some de- of lapped-spliced U-stirrups. Also, most
signers consider the prestressing strand designers feel that the stirrup spacing
to be part of the longitudinal reinforce- limit of 12 in. is not appropriate for deep
ment while others consider only the spandrels, and this limit is routinely ex-
mild reinforcing steel. In their original ceeded.
paper, Zia and McGee recommended Nonprestressed Spandrels — Torsion
that only the prestressing steel in excess design of nonprestressed concrete gen-
of that required for flexure, and located erally follows ACI Code requirements,
around the perimeter of closed stirrups, except for the detailing considerations
should be considered as part of the lon- discussed above.
gitudinal torsion steel. Pocket Spandrels — Typically, pocket
The third edition of the PCI Design spandrels need not be designed for tor-
Handbook2 describes a procedure de- sion. However, the pockets complicate
veloped by Collins and Mitchell, which the shear design. Design practices vary
is based on compression field theory. for considering the effect of the pocket;
This approach assumes that, after some designers neglect this effect. For-
cracking, the concrete can carry no ten- tunately, shear strength does not control
sion and that shear and torsion are car- the dimensions of deep pocket span-
ried by a field of diagonal compression. drels and often only minimum rein-
Because the concrete contribution is forcement is required. Welded wire fab-
neglected, this approach will generally ric is frequently used for web rein-
require somewhat more stirrup rein- forcement.
forcement depending on the selection of
the crack angle. The biggest difference,
however, is in the positive and negative Beam End Design
moment capacity requirements which Torsion Equilibrium — The eccentric
are based on the axial tension caused by load applied on the ledge produces tor-
shear and torsion. For the example sion in the spandrel which must be re-
shown in the PCI Design Handbook, sisted by reactions at the supports.
the required positive and negative Customarily, the web is connected to
bending strength at the face of the sup- the column to restrain rotation. Figs. 5a
port exceeds the midspan moment. and 5b show the torsion equilibrium
These requirements present consider- reactions for a normal and dapped con-
able detailing difficulties, and many de- nection, respectively.
signers do not feel they are valid for The torsional equilibrium reactions
deep spandrels. may require supplemental vertical and
Detailing practices for the torsional horizontal web reinforcement at the
reinforcement do not always follow ACI ends of the girder. Raths 1° and Osborn*
Code requirements. Section 11.6.7.3 re- prescribe similar methods for design of
quires that transverse reinforcement
consist of closed stirrups, closed ties or *Osborn, Andrew E. N., "Design of Ledger Gir-
spirals. However, the Commentary to ders," Draft Report for PCI Connection Details
the ACI Code indicates that this re- Committee, April 1984.

PCI JOURNALISeptember-October 1986 83

 hg h^

(a) (b)

Fig. 5. Torsion equilibrium reactions in spandrel beam.

this reinforcement. Vertical and lon- shear reinforcement. However, most de-
gitudinal steel,A,,, andAw, on the inside signers feel this reinforcement is not
face of the spandrel is calculated by: additive to reinforcement for internal
torsion. If the reinforcement for torsion
T,, is considered to function as A,^, and A,or
Ate„ = i = 1)
reinforcement, little or no supplemental
reinfprcement is required provided all
where
loads acting on the ledge are considered
T. = factored torsional moment at as part of the shear/torsion load.
end of girder (in.-lbs) Fig. 6 shows an alternative means to
d, = depth ofA,,,,, and A 1 steel from provide torsional equilibrium at the
outside face of spandrel (in.) support. In this case, the end reactions
f„ = yield strength ofreinforement are in close alignment with the ledge
(psi) (or effective prestress) loads. The projecting beam ledge is
= strength reduction factor = 0.85 treated as an upside-down corbel. Most
designers surveyed indicated that this
The use of 0 = 0.85 instead of 0.90
approach may Iead to excessive rolling
(flexure) compensates for the ratio of
of the spandrel beam at the support,
internal moment to total effective depth,
particularly where a soft bearing pad is
which is not in Eq. (1).
used.
Osborn recommends the bars he
evenly distributed over a height and Dapped-End Beams — Section 6.13 of
width equal to h. (see Fig. 5) from the the PCI Design Handbook presents de-
concentrated reaction point. sign criteria for dapped-end connec-
Because shear cracks may coincide tions. Research on dapped connections
with diagonal cracks due to out-of-plane under PCISFRAD Project No. 6, which
bending, A. should be added to the is being conducted concurrently with

84
 Raths1° uses all the hanger reinforce-
ment between ledge loads, but com-
putes the required reinforcement based
on the summation of moments about the
outside face of the spandrel, thus:
V„ (jd + a) (3)
4f id
where
A,h = area of transverse hanger rein-
forcement on inside face of
spandrel for each ledge Ioad (sq
in,)
Vaz = factored ledge load (kips)
a = distance from ledge load to
center of inside face reinforce-
V ment (in,)
jd = internal moment arm (in.)
Fig, 6. Beam end corbel behavior when (taken as d – ½ in.)
providing torsion equilibrium at support of ^i = strength reduction factor = 0.85
spandrel beam. Raths recommends an additional load
factor of 4/3 for design of hanger rein-

this project, is expected to recommend

modified procedures. Design of
dapped-end L-beams is often compli-
cated by reinforcement for torsion
equilibrium connections (Fig. 5b). Also,
the last blackout in a pocket spandrel
often interferes with the reinforcing for
the dapped end. The established design
procedures are modified as appropriate
to handle these special conditions.

Beam Ledges
Hanger Reinforcing — Fig. 7 illus-
trates a possible separation between the
ledge and web of an L-shaped spandrel.
Section 6.14 of the PCI Design Hand- -A sh {INSIDE LEG ONLY)
book and design examples by PCA 3 and
Collins and Mitchell" recommend
hanger reinforcement concentrated near
the ledge load given by:
POSSIBLE SEPARATION
BETWEEN LEDGE AND WEB
A,h = (2)
rV"
The notation is defined on the next Fig. 7. Ledge-to-web attachment showing
column above. hanger reinforcing.

PCI JOURNALSeptember-October 1986 85

 based on internal shear stress distribu-
tion, relative depth of the ledge, and
deflection compatibility.
There is no consensus among design-
ers on requirements for hanger rein-
forcement. Some designers do not check
ledge-to-web attachment, while others
use some combination of the above
methods. Furthermore, there is no
agreement as to whether or not hanger
reinforcement should be added to shear
and torsion reinforcement. The method
for designing hanger reinforcement
generally controls the quantity of trans-
verse reinforcement in the middle re-
gion of the spandrel, and can have a very
significant effect on material and fabri-
cation costs.
Ledge Punching Shear — The design
for punching shear in beam ledges gen-
UP erall y follows the procedures in Section
6.14 of the PCI Design Handbook. Some
designers follow a modified procedure
Fig. 8. Hanger reinforcement in pocket recommended by Raths; 1 ° based on un-
spandrels. published test results, this method con-
siders a lower ultimate stress on the
vertical shear plane along the inside
forcement. An alternate procedure for face of the web. Mirza, et al l ' - 13 and
using concrete tension as a means of KraukIis and Guedelliofer' have also
found that the PCI design equations
ledge-to-web attachment is also given in
his paper. may be unconservative.
Eq. (3) is based on sound principles of
statics, yet there are many existing Beam Pockets
spandrels that have performed well with
much less reinforcement than this It is customary to provide closed stir-
rups or U-bars in the plane of the web
equation would require. The only
for the entire tee stem load in pocket
known failures have occurred where
there was no hanger reinforcement. In spandrels. The hanger bars are typically
several instances, beams with ver y light located near the tee stem reaction, as
hanger reinforcement have survived shown in Fig. S. Therefore, Eq. (2) is
loading tests. used to determine hanger reinforcement
Further refinements of hanger rein- requirements. The concrete tensile
forcement design' I •* .t reduce the load stress at the reaction level is relatively
low so a horizontal crack at that location
that must be suspended from the web
is unlikely. Also, because hanger rein-
forcement is customarily used, punch-
• Osborn, Andrew E. N., "Design of Ledger Gir-
ing shear below the pocket is generally
ders," Draft Report for PC:I Connection Details
Committee, April 1984. not a concern.
t Sturm, Edward R., "Theory of Deflection Com- Prior to describing the experimental
patibility," Private correspondence with Andrew program, a summary of the finite ele-
Osborn, May 1984. ment model studies is given.

86
 4. FINITE ELEMENT MODEL STUDIES
Description considered so the analytical studies and
load tests modeled the same condition,
Finite element models of an L-beam although it should be noted that a direct
and pocket spandrel were analyzed. The connection between the column and
geometry of these models and the test deck is not necessarily required. Alter-
specimens was essentially the same. nately, the column can be indirectly
Refer to Figs. 13 and 14 for more de- connected to the deck through the span-
tailed information on the geometry of drel beam.
the beams.
The model studies had several objec-
tives: Spandrel Beam Behavior
• Investigate the deflections and ro- Fig. 9a shows the midspan deflection
tations caused by the eccentrically of the L-beam at service load without
applied load. any connections to deck elements. Note
• Determine the theoretical torsional the overall outward deflection due to
equilibrium reactions at the supports. the rotation of the principal axes. Con-
• Study the influence of connections nections to deck elements effectively
to deck elements on deformations and restrain this outward displacement, as
torsional equilibrium reactions. shown in Fig. 9b. Usually these con-
• Investigate the stresses across the nections are not made until all of the
ledge/web interface. dead load is in place. Similar plots for
Three-dimensional solid elements the pocket spandrel are shown in Fig.
were used with three degrees of free- 10. Due to the different cross-sectional
dom at each node. Cross sections shape and load eccentricity, the lateral
showing the finite element mesh are deflection is relatively small.
shown in Figs. 9 and 10. Fig. 11a shows the horizontal reac-
Service loads included beam (lead tions at the L-beam support without
load and a 16.8 kip tee stem reaction at 4 connections between the spandrel and
ft centers. The tee stem load was deck. These forces simply balance the
applied at 8 in. and 2 in. from the web external torsion due to the eccentrically
centerline for the L-beam and pocket applied loads. Fig. lib shows the hori-
spandrel, respectively. The restraints at zontal reactions with deck connections.
each end of the beam modeled a typical The deck connections in the midspan
spandrel beam support where the hear- region restrain the outward displace-
ing pad is placed at the centerline of the ment. The deck connections at the sup-
web, and lateral support is provided port work with the top corner connec-
near the hearing and at the top corners tions to restrain rotation. The net out-
of the bean. ward force between the deck and span-
For both the L-beam and pocket drel would be counteracted by the
spandrel, a second condition was ana- column-to-deck connection, If there
lyzed in which additional lateral re- were no column-to-deck connection, the
straint was provided near midheight of deck connection forces would tend to
the beam to simulate connections to balance, depending on the stiffness of
deck elements. There was no possibility the column.
of relative lateral movement between
the column restraints and deck ele- Transfer of Ledge Loads to Web
ments, simulating the case where there
is an independent connection between Stresses across a plane 3 in. above the
the deck and the column. This case was ledge/web interface were studied. (The

PCI JOURNALJSeptember-October 1986 87

w
00

F.E. LOA[ F.E. LOAD

MODEL TES" MODEL TESTS

VERT. (IN.) -0.060 -0.1: :RT. (IN.) -0.056 -0.131

HORIZ. (IN.) -0.089 -0.0C )RIZ. (IN.) -0.008 -0.057
ROT. (RAG) -0.00163 -0.01 )T. (RAO) -0.00169 -0.00282

(A) WITHOUT DECK CONNECTIONS (B) WITH DECK CONNECTIONS

Fig. 9. Midspan deflection of L-beam (superimposed dead load plus live load).
 F.E. LOAD F.E. LOAD
MODEL TESTS MODEL TESTS

VERT. (IN.) -0.053 -0.173 VERT. (IN.) -0.053 -0.146

HORIZ. (IN.) +0.024 +0.038 HORIZ. (IN.) 0.0 +0.013
ROT. (RAD) -0.00085 -0.00443 ROT. (RAD) -0.00083 -0.00346

(A) WITHOUT DECK CONNECTIONS (B) WITH DECK CONNECTIONS

QD Fig. 10. Midspan deflection of pocket spandrel (superimposed dead load plus live load).
 7.23 k - FINITE ELEMENT MO
6.51 k - LOAD TEST

7.96k
7.22k

796k

(a) WITHOUT DECK CONNECTIONS (b) WITH DECK CONNECTIONS

Fig. 11. Horizontal forces acting on L-bears.

(a) APPLIED LOAD

Pil
(b) STRESS AT
LEDGE/WEB INTERFACE
14.6k

(C) CALCULATED
RESULTANT FORCES

Fig. 12. Study of ledge region from finite element model.

90
geometry of the finite element mesh The resultant of these stresses can be
prevented investigation at the top of the computed by integrating stresses in the
ledge.) The results of that study are pre- individual elements near the ledge/web
sented in Fig. 12. As expected, the in- junction. As indicated in the figure, the
side face of the web is in tension. The resultant is slightly less than the applied
maximum tensile stress of 295 psi, ledge load and is shifted significantly
which occurs at the ledge load, is about towards the web centerline. These dif-
40 percent greater than the average ferences are equilibrated by shear and
stress. The compression in the outside torsion in the ledge itself. This mecha-
face of the web is significantly more nism is described in more detail in Sec-
uniform. tion 6.

5. LOAD TESTS
Two L-beams and one pocket span- ment for internal torsion.
drel were tested to study their behavior In view of the controversy regarding
and verify their strength. The tests were ledge-to-web attachment, alternate pro-
conducted in the structural laboratory of cedures were used for design of hanger
Wiss, Janney, Elstner Associates in reinforcement:
Northbrook, Illinois. • Hanger reinforcement for Speci-
men 1 was designed by Eq.(2), with a
Test Specimens one-sixth reduction in the load sus-
pended from the web based on relative
General — All three spandrels were ledge depth. All of the transverse rein-
72 in. high, 8 in. wide and 28 ft long. forcement between ledge loads was
The target design loads were based on considered to be effective, and hanger
90 psf dead load and 50 psf live load, reinforcement was not added to shear
which are typical for a double tee park- and torsion reinforcement.
ing structure with 60 ft spans. The reac- • Eq.(3) was used for design of the
tions at each stem of an 8 ft wide double hanger reintorceme nt in Specimen 2. A
tee were 16.8 kips. 7.4 percent reduction in the suspended
Design — The design of the test spec- load was taken based on an assumed
imens was based on the state-of-the-art parabolic shear stress distribution.
methods described in the background Again, all the hanger reinforcement
section, Shear and torsion design for the between ledge loads was considered
prestressed spandrels followed the pro- effective, and it was not added to
cedure recommended by Zia and Hsu. shear/torsion reinforcement.
ACI Eq.(11-10) (rather than Eq.(11-11) Hanger reinforcement for the pocket
or (11-13)1 was used to compute the spandrel (Specimen 3) was designed by
basic shear strength provided by the Eq.(2). In addition to a U-bar at the poc-
concrete section. Flexural design fol- ket, one wire on each side of the pocket
lowed ACI 318-83. Some reserve from the mesh reinforcing was consid-
flexural strength was required to meet ered to contribute.
the provisions of Section 18.8.3, which Design of the dapped-end connection
requires a bending capacity equal to at for the pocket spandrel basically fol-
least 1.2 times the cracking moment. lowed the PCI Design Handbook pro-
Reinforcement for torsional equilibrium cedure with two exceptions. First, due
was checked by Eq,(1). This reinforce- to relatively low stresses, there was no
ment was not added to the reinforce- special reinforcement provided for di-

PCI JOURNAUSeptember-October 1986 91

s
6' 1a «a J ^1r ^rru'sowi
8-^4 d' tee' i'-0' 9413 _ rY2" iY-O'
294 r^ ^ei2"

ELEVATION

#3

#4x2761

17.CLR
*3J SPECIMEN 2 ONLY
ap B e•
#4
'6 CHAMFER
#4 x 27' 6"

2 4 18 4'6 STRANDS
SQS3 RELIEVE.
26.9'
SECTION

Fig. 13. Dimensions and details of Specimens 1 and 2.

agonal tension in the extended end or are provided in Figs. 13 and 14. The
direct shear at the junction of the dap following features of the reinforcing
and the main body of the member. The details should be noted:
welded wire shear reinforcement, how- 0 Due to the different design
ever, was continued into the extended methods, Specimen 2 has twice as much
end. Second, the reinforcement for flex- hanger reinforcement across the
ure and axial tension in the extended ledge-web interface. This reinforcement
end was not continued past the potential was provided by partial height L-bars on
diagonal tension crack extending to the the inside face of the spandrel between
bottom corner of the beam. the stimips. These bars add about 4 per-
Details — The dimensions and rein- cent to the weight of the mild steel in
forcement details of the test specimens the beam.

92
 12.6-Wt4, W2.5
4, 4-W 2.9, W2.9 43'-6 .3-6') 4.4 -W29. W2.9
INSIDE FACE t INSIDE FACE

s:

L aX44' o

`2-46 ' *4 44J2-A16— q.

2-46 ELEVATION 2-#6

*4.2T-6°

72.6-W74, W2,5

4.4 -W2.9, W2.9

e4 HANGER BARS

4' CHAMFER
AT POCKET

SECTION

Fig. 14. Dimensions and details of Specimen 3.

• Closed ties formed in one piece by forcement of the pocket spandrel. The
overlapping 90 degree end hooks are mesh is not hooked around the main
used on the left half of the L-beams, reinforcement at the top and bottom of
Stirrups on the right side of the L-beams the beam, although the ACI Code re-
consist of lapped-spliced U-bars. quirements for development of web re-
• The longitudinal bars in the L- inforcement (Section 12.132.5) are
beams are not hooked at the ends. satisfied.
• At the right side of the L-beams, Materials — Design of the test speci-
two #5 bars are welded to a bearing mens was based on 5000 psi concrete, 60
plate. A #5 U-bar is used on the left side ksi reinforcing bars (ASTM A706), 270
of the L-beams. ksi stress-relieved strand, and ASTM
• Wire mesh is used for shear rein- A497 mesh. Concrete cylinders and re-

PCI JOUFiNAUSeptember-October 1986 93

 Table 1. Material strengths.

Concrete Reinforcing steel

Compressive
strength Bar Yield strength Tensile strength
Specimen fr (Psiyo, size f„ (ksi) fy (ksi)
1 5330 #3 78.9 98.7
2 5640 #4 70.4 103.7
3 6060 #6 64.2 98.1
(a) Average ofthree field-cured cylinders tested concurrently with load test (psi).

inforcing bar samples were tested to tested to failure in two phases. After a
determine actual strengths. The results failure near the end region in Phase 1,
are presented in Table 1. The yield the supports were moved in 4 ft from
strength of the #3 bars was much higher each end, and the specimen was re-
than expected. Ioaded to failure.

Test Procedure Behavior and Strength of Test

Specimens
Setup — The test setup is shown in
Fig. 15. The spandrels were supported Deflection and Rotation — Figs. 9 and
on rigid L-shaped frames which pro- 10 compare the measured deflections of
vided lateral restraint at the four corners the L-beam and pocket spandrel to those
of the beam. Load was applied at seven predicted by the finite element models.
points along the beam using specially Although the measured deflections are
designed double tees (and one single quite small, they are two to three times
tee). The test setup featured a remov- the predicted deflections. About half of
able connection between the spandrels the vertical deflection and some of the
and double tees. rotation may be attributed to deforma-
Instrumentation — Instrumentation tion of the bearing pads.
included load cells at two of the loading Fig. 16 shows a plot of stem reaction
points on the double tees, as well as all vs. midspan torsional rotation of Speci-
four horizontal reaction points. Three men 2. The stiffness of the beam is sig-
deflection transducers and one tiltmeter nificantly reduced after cracking was
were set up at midspan to monitor hori- observed.
zontal and vertical deflections and rota- Service Load Behavior — At service
tions. Finally, single element strain load, no cracks were observed in the
gauges were placed on selected rein- L-beams. However, minor cracks were
forcing bars as per Table 3. observed near the dapped-end connec-
Load sequence — Initially, each span- tion of the pocket spandrel. These
drel was incrementally loaded to service cracks, which are shown in Fig. 17a,
load (16.8 kips per tee stem) without the were all less than 10 mils (0.010 in.) in
connection between the double tees and width.
spandrels. After unloading, this se- Failure Patterns (Specimen 1) — The
quence was repeated with the deck cracking patterns that occurred during
connections in place. Finally, the beams loading to failure are shown in Fig. 18a.
were loaded to failure without the deck Diagonal cracks began to appear on
connections in increments of 2.5 kips Specimen 1 at a load of 25 kips per stem.
per tee stem. The third specimen was The crack at the ledge/web junction oc-

94
 (a) L-beams

(b) Pocket spandrel

Fig. 15. Test setup for L-beams and pocket spandrel.

PC[ JOURNAL/September-October 1986 95

rn

Table 2. Spandrel design and test results.

Calculated strength
Design force -
Specimen Design° Predicted' Test
Failure mechanism Units No." Service Ultimate x Nominal) 0=1 forced
Midspan flexure in.-kips 1 5,490 8,190 0.90 11,900T 13,730 10,520
2 5,490 8,190 0.90 11,900, 13,730 12,800
3-1 5,410 8,080 0.90 9,400' 10,440 8,150
Shear at support kips 1 68.0 101.4 0.85 111.1 145.2 130.3
Torsion at support in.-kips 470 709 793 1033 967
2 68.0 101.4 0.85 111.1 146.8 158.6
470 709 793 1033 1196
3-1 66.9 100.0 0.85 124.7 159.0 100.9
118 177 a 186
3-2 66.9 100.0 0.85 124.7 159.0 124.7
118 117 238
Lateral bending due to in,-kips 1 470 709 0.90 692 902 967
torsion equilibrium force 2 470 709 0.90 692 902 1196
3-1 118 177 0.90 246 273 186
Hanger reinforcement kips 3-1 66.9 100.0 0.9(1 95.0 113.0 100.9
at dapped end
Hanger reinforcement kips per 1 16.8 25.3 0.90 28.4k 41.5" 34.6'
for ledge load stem 2 16.8 25.3 0.90 26.8h 39.1'' 42.7
3-2 16.8 25.3 0.90 24.1 30.8) 47.6
Table 2 (cont.). Spandrel design and test results.

Calculated strength
Design force
Specimen Design° Predictede Test
Failure mechanism Units No. Service Ultimate chi (c x Nominal) 0=1 forced
Tee stem hearing' kips per 1 16.8 25.3 0.70 66.8 101.7 34.6
stem 2 16.8 25.3 0.70 66.8 107.6 42.7
3-2 16.8 25.3 0.70 66.8 115.6 47.6
Ledge punching shear kips per 1 16.8 25.3 0.85 61.7 74.9 34.6
at interiorbearingm stem 2 16.8 25.3 0.85 61.7 77.1 42.7
Ledge punching shear kips per 1 16.8 25.3 0.85 53.7 65.2 34.6
at exterior bearing" stem 2 16.8 25.3 0.85 53.7 67.1 42.7
3-1 and 3-2 indicate Phases I and 2 of the Specimen 3 load test, respectively.
Calculated nominal strength using state-of-the-art design equations and specified material properties (multiplied by
Calculated nominal strength using design equations and actual material properties (0 =1).
indicates failure at specified test force.
"Torsion design not required.
e Reserve flexural strength was required to meet the requirements of Section 18.8.3 of AC! 318-83 which requires a b
moment.
Hanger reinforcement designed by Eq. (2) with a one-sixth reduction in the load suspended from the web based on i
Hanger reinforcement designed by Eq. (3) with a 7.4 percent reduction in the load suspended from the web based of
Hanger reinforcement designed by Eq. (2); one wire on each side of pocket included.
Hanger reinforcement yield at 29.9 kips per stem.
Bearing design per PCI Eq. (6.8.1) withN„ = 0.
Using PCI Eq. (6.14.1).
Using PCI Eq. (6.14.2).
Table 3. Reinforcement strains.

Service load Factored load Max test load

Distance
Gage from load Load Strain Load Strain Load Strain
Location No.. (in.) ' percent b percent ° percent
Ledge hanger 1-1 0 16.9 0.004 27.3 0.239 34.6
reinforcement 1-2 12 16.9 0.001 27.4 0.120 35.6 3.211
(ne ar midspan) 1-3 24 16.9 0.0 27.4 0.223 34.6 2.235
1-4 12 16.9 0.0 27.4 0.245 34.6
1-5 0 16.9 0.003 27.4 34.6 `

Ledge flexure 1-6 24 16.9 -0.002 27.4 0.016 34.6 1) 015

reinforcement 1-7 0 16.9 -0.001 27.4 0.026 34.6 0.042

Ledge hanger 2-1 24 16.7 0.0 28.1 0.005 42.7 `

reinforcement 2-2 18 16.7 0.001 28.1 0.007 42.7 0.210
(near midspan) 2-3c 12
2-4 6 16.7 0.002 28.1 0.023 42.7 0.412
2-5 0 16.7 0.004 28.1 0.035 42.7

Ledge flexure 2-6 24 16.7 -0.002 28.1 -0.003 42.7 0.016

reinforcement 2-7 0 16.7 -0.001 28.1 0.007 42.7 0.034

Dapped end 3-1 8 16.7 0.056 24.9 0.130 - -

flexure reinforcement

flapped end 3-2 8 16.7 0.091 24.9 0.097 - -

hanger reinforcement 3-3 11 16.7 0.017 24.9 0.067 - -

Hanger reinforcement 3-4 6 16.7 0.006 24.9 0.101 46.8 0.414

at pocket (at midspan) 3-5 6 16.7 0.005 24.9 0.093 46.8 0.162
First number indicates specimen number.
Average ledge load (kips).
I Bad readings due to gauge failure or bending in bar apt crack.

curred at 27.5 kips. This crack immedi- served on the outside face. The crack at
ately opened to 20 mils and extended the ledge/web junction was restrained
end to end where it connected to in- by the additional hanger reinforcement,
clined cracks in the ledge. The ledge as shown in Fig. 20. At a load of 42.7
continued to separate from the web until kips per tee stem, punching shear fail-
the test was stopped at a ledge load of ures occurred at the first and sixth tee
34.6 kips per stem. At the end of the test, stem from the left. Fig. 21 shows the
the crack at the ledgetweb junction was punching shear failures, The failure
over Vs in, wide, as shown in Fig. 19. cone initiates behind the hearing pad.
Failure Patterns (Specimen 2) - As The failure surface is almost vertical
shown in Fig. 18b, a well developed near the top and inclined below the
pattern of inclined and "rainbow" ledge reinforcing. As a result, the ledge
cracking developed on the inside face of flexural reinforcement is not very well
Specimen 2. Typically, these cracks developed across the failure plane.
were less than 10 mils wide. Also, sev- Failure Patterns (Specimen 3) - The
eral 1 to 3 mil flexural cracks were ob- cracks which formed during Phase 1 of

98
 40

a_
Y
p 30
n

N
Q za
is

10

ROTATION (RADIANS)

Fig. 16. Stem reaction versus rotation (Specimen 2).

the Specimen 3 test are shown in Fig. 47.6 kips. The "rainbow" crack and
17b. Cracks near the dapped-end con- punching failure are shown in Fig. 17c.
nection which developed at service load Strength — Table 2 summarizes the
continued to lengthen and widen, and design force, calculated strength and
new inclined cracks formed. Cracks test force for several potential and actual
below the pockets began to form at tee failure mechanisms. The calculated
stem loads of 18 to 25 kips. As the load strengths are based on the equations
was increased, diagonal tension cracks used for design. Because the hanger re-
developed further from the support. inforcement for Specimens I and 2 was
These cracks typically initiated near designed using different equations, the
midheight of the beam. At a load of 26.5 calculated strength is roughly the same
kips per tee stem, a diagonal tension even though Specimen 2 had twice as
crack near the right support extended much hanger reinforcement.
down to the bottom corner of the beam The calculated strength is expressed
and failure occurred immediately, as as both a "design" strength and a "pre-
shown in Fig. 22. dicted" strength, The design strength is
In Phase 2 of the Specimen 3 test, a based on specified material properties,
wide "rainbow" crack formed at a load and includes the appropriate strength
of about 43 kips per tee stem. Appar- reduction factor. The predicted strength
ently this crack is due to a combination uses actual material properties and no
of diagonal tension due to shear and strength reduction factor.
vertical tension due to the tee stem As shown in Table 2, the spandrel
loads. The ultimate failure, however, beams were tested to a load near or be-
occurred when the concrete below the yond their predicted capacity for several
fifth pocket from the left punched out at of the primary failure mechanisms.

PCI JOURNALSeptember- October 1986

99
There were, however, several notable The ledge-to-web attachment strength
exceptions. of Specimen 1 was considerably less
The shear failure of Specimen 3 than predicted by Eq.(2). In contrast,
(Phase 1) occurred at the diagonal Specimen 2 showed no sign of a ledge-
cracking load, and the expected contri- to-web attachment failure, even though
bution from the shear reinforcing was the test force was slightly above the ca-
not realized. pacity predicted by Eq.(3). The strength

U U U U
(A) FRONT ELEVATION OF SPECIMEN 3
AT SERVICE LOAD

I EL 'N

Cs) FRONT ELEVATION OF SPECIMEN 3-(PHASE 1)

AT ULTIMATE LOAD

!Ntk U
(c) FRONT ELEVATION OF SPECIMEN 3-(PHASE 2)

AT ULTIMATE LOAD (END REGION CRACKS NOT SHOWN)

CRACK L='-Nil'

1-10 MU.
11-49 NIL
56 MU. OR MORE
---^- CRACK ON BACK (OUTBID:) FACE

Fig. 17. Crack patterns (Specimen 3).

100
 of the hanger reinforcement below the or near service load, factored load and
pocket of Specimen 3 (Phase 2) was well the maximum test load.
beyond the predicted capacity. Appar- At service load reinforcement, strains
ently, the shear strength of the concrete are insignificant except at the dapped-
below the pocket contributed. end connection of the pocket spandrel,
The most surprising result was the where the strain in the hanger rein-
punching shear failure at Specimen 2. forcement bar nearest the load is almost
Although the ledge loads were quite 0.1 percent. This strain level corre-
high, the punching shear strength was sponds to half the yield stress for a
only about 60 percent of the predicted Grade 60 bar. Even though the strain
capacity. levels in the ledge flexure and hanger
Horizontal Reactions — At service reinforcing are very low, they are
loads, the measured horizontal reactions noticeably higher at the ledge load.
at the supports were comparable to the At factored load, cracking of the
reactions predicted by the finite ele- ledge/web junction of Specimen 1 was
ment model, as shown in Fig. 1.1. accompanied by very high hanger rein-
Reinforcement Strain — Table 3 sum- forcement strain. In Specimen 2, this
marizes the reinforcement strain at cracking was limited to the vicinity of
gauged locations. Data are provided at the ledge load which is reflected in the

/ 1

I I

(A) FRONT ELEVATION OF SPECIMEN 1

AT ULTIMATE LOAD

i I 1

^^ ^1 j3/I+ Ill I ^ IS It r I t

`1 J 1, :SL.'. 4a1

(a) FRONT ELEVATION OF SPECIMEN 2

AT ULTIMATE LOAD

CRACK LEGENC:

-18 MIL
rf
:-49 N[L

- 5; .NIL OR MORE

- --- CRACK ON BACK (OUT$tOC) FACE

— Fig. 18: Crack patterns (Specimens 1 and 2).

PCI JOURNALJSeptember-October 1986 101

Fig. 19. Crack at ledge/web junction (Specimen 1).

Fig. 20. Crack at ledge/web junction (Specimen 2).

102
 (a) Tee stem at left support.

,4 ^^ ^^p^ ^
, l^k

1^ ^^5.'^r' A µd l ^{

(b) Sixth tee stem from left.

Fig. 21. Punching shear failures (Specimen 2).

PCI JOURNAUSeptember-Qctober 1986 103

 (a) Front. (b) Back.

Fig. 22. Shear failure (Specimen 3, Phase 1).

recorded strains. Strain in the ledge are approaching the yield strain. (Using
flexure reinforcement remains low at the 0.2 percent offset method, the yield
factored loads because there are no ver- strain of the #3 bars is about 0.5 per-
tical cracks at the ledge/web junction. cent.) The hanger reinforcing bars at the
Despite earls' cracking at the dapped- pocket on Specimen 3 are also near the
end connection, strain levels at factored yield strain. It should be noted that
loads are well below yield strain. these strains would exceed the nominal
At the maximum test load, the strain in yield strain of a Grade 60 bar. Strain in
the ledge hanger bars in Specimen 1 are the ledge flexure reinforcement remains
well into the strain hardening range. low at maximum test load, indicating the
The ledge hanger bars in Specimen 2 absence of ledge flexure cracks.

6. ANALYSIS AND DISCUSSION

General Design Consideration crack patterns which occurred in

Specimens I and 2 suggest a similar
Location of Critical Section — The possibility. Therefore, the shear and tor-
shear failure of Specimen 3, shown in sion design of spandrel beams should
Fig. 22, confirms the possibility of an in- consider a critical section at the face of
clined failure plane which carries all of the support. In addition, the transverse
the loads acting on the spandrel. The reinforcement spacing required for

104
shear and torsion at a particular section ment, however, is not very efficient in a
along the beam should be continued for deep dap.
a distanced inside that section. Recent research under PCIFSRAD
An alternate approach is to provide Project No. 6 emphasizes the impor-
separate hanger reinforcement to trans- tance of anchoring the primary flexural
fer the ledge loads to the top of the sec- reinforcement at dapped connections.
tion and design the spandrel as a di- This research concludes that the reac-
rectly loaded beam. However, the tion should be limited to the shear
former approach is more rational be- strength of the web (the lesser ofV,, and
cause it directly relates to the potential V), because the primary flexural rein-
failure planes. forcement is typically not anchored at
Influence of Deck Connections — As the bottom corner of the beam.
illustrated in Fig. 11, the connections to Predicting the strength of the concrete
deck elements do not substantially re- section is complicated by the pockets.
duce torsion. The only significant effect Hanson15 found that a conservative pre-
of the deck connections is the restraint diction of the strength of concrete joists
on lateral displacement induced by with square openings, but without stir-
bending about the weak principal axis. rup reinforcement, was obtained by cal-
culating the load at which cracking at
the corner of the opening develops, as-
Flexure suming the shear is distributed in pro-
With regard to flexure, both the portion to the area of the section above
strength and serviceability related be- and below the opening.
havior of the test specimens was satis- One approach to calculating this load
factory. It is worth mentioning, how- is to substitute b,,, (cl – h„) forb.d in AC!
ever, that flexural cracking of the L- Code equations for the shear strength of
beams only showed up on the back face. the concrete section [Eqs. (11-3) or
This observation is attributed to bend- (11-6) for nonprestressed spandrels, or
ing about the weak principal axis. Eqs. (11-10), (11-11) or (11-13) for pre-
stressed spandrels], where hp is the
height of the pocket, Similarly, the
Shear and Torsion strength provided by the shear rein-
Prestressed L-Beams — Specimens 1 forcement, V,, is given by:
and 2 were tested at load levels roughly
equal to the predicted capacity based on Ve = A,,ff(d – Ii) (4)
the Zia-Hsu equations, which was the s
basis for their design. There was no evi-
dence that the negative bending capac- which reflects an unfavorable crack
ity required by compression field theory pattern through the pocket region.
was needed. As discussed later, some The above approach is believed to be
level of positive bending capacity at the conservative for pocket spandrels, but is
face of the support is required. not universally applicable to beams with
Pocket Spandrels — The premature square openings. Using ACI Code Eq.
shear failure through the full section of (11-13) and substituting b. (d – hp) for
the pocket spandrel near the dapped hu,d, the predicted shear strength pro-
connection is attributed to poor anchor- vided by the concrete section of Speci-
age of the primary flexural reinforce- men 3 is 110 kips or 93 kips, depending
ment at the bottom corner of the beam. on whether or not the prestress is con-
It may have helped to extend the sidered to contribute to shear strength.
dapped-end flexural reinforcement be- These predictions are comparable to the
vond the inclined crack; this reinforce- failure load of 101 kips.

PCf JOURNALJSeptember-October 1986 105

 4- OA9fsd =NU h/d+Vu(O.5+a/d)
N0
Vu

Fig. 23. Forces acting an free body cut off by diagonal tension
cracks at support.

It is common practice not to use a Beam End Design

deep pocket for the tee stem nearest the
support. A welded bracket or Cazaly Torsion Equilibrium Reinforcement
hanger is used instead. In these cases, — The applied torsional load on Speci-
the he, term need not he included for de- mens 1 and 2 was beyond the predicted
sign ofthe end region. capacity of the torsion equilibrium
Detailing Practices — The torsional reinforcement required by Eq. (1). To
response of deep spandrels is domi- some extent eccentric bearing may have
nated by out-of-plane bending. There helped equilibrate the applied torsional
was no evidence of spalling of the side load. Nonetheless, the test results sup-
cover which can occur in compact sec- port the contention that reinforcement
tions subjected primarily to torsion. The for the torsion equilibrium reaction
use of lapped-splice stirrups in lieu of need not be added to the reinforcement
closed stirrups did not appear to have for internal torsion.
any detrimental effect, and the absence Longitudinal Reinforcement at End
of hooks on the longitudinal reinforce- — The premature failure near the dap-
ment did not lead to any apparent prob- ped connection points out a possible
lems. deficiency in the end region of spandrel
It is unlikely that there would have beams. Fig. 23 shows the forces acting on
been any improvement in shear strength a free body cut off by diagonal tension
of the pocket spandrel had the wire cracks at the support. Neglecting the
mesh been anchored by a bend at the distance from the top of the beam to the
longitudinal reinforcement, The failure compressive force, the developed force
is attributed to poor anchorage of the required at the face of the support is
primary flexural reinforcement, and given by:
there was no sign of an anchorage failure
of the wire fabric. OA, , = N„ hid + V,, (0.5 + aid) (5)

106
Fig. 24. Transverse forces acting on free body of ledge.

f
where d is the developed stress in the A design procedure for hanger rein-
longitudinal reinforcement at the face of forcement has been developed based on
the support, The remaining notation is the transverse forces acting on the free
defined in Fig, 23. body shown in Fig. 24. Summation of
For a dapped spandrel, a similar moments about the outside face of the
check of the free body forces across an spandrel gives:
inclined crack through the full section is
recommended. Typical cases are in- V„(d + a) –AV,br12 –OT,
Au, = (6)
cluded in the design examples in Ap- Of,,d
pendix C.
where
AV, = shear in ledge [Eq. (7)1
Beam Ledges ATr = torsion in ledge [Eq. (8)]
Hanger Reinforcement — The Ioad b i = width of ledge measured along
tests and analytical studies indicate that bottom of beam
the eccentricity of the ]edge load cannot = strength reduction factor =
be neglected in the design of hanger 0.85
reinforcement. Nontheless, not all of the Most of the notation used for hanger
load acting on the ledge is suspended reinforcement design is graphically de-
from the web, and the effective eccen- fined in Fig. 25. Similar to Eq. (1), the
tricity of the ledge load is significantly use of 0 = 0.85 instead of 0.9 compen-
reduced due to torsion within the ledge. sates for the ratio of internal moment
Design by Eq. (2) may be somewhat un- arm to total effective depth.
conservative, while use of Eq. (3) may The finite element model study ver-
be overly conservative. ified that the shear in the ledge, AVr,

PCI JOURNAL'September-October 1986 107

 AT r =V ,^ ey, (8)
Xxtr^

where
e = distance between applied load
and centerline of web
x = shorter overall dimension of rec-
tangular part of cross section
y = longer overall dimension of rec-
tangular part of cross section
(x 2 y) 1 d = b ( h j orb h,, whichever is
smaller
The use of y, in Eq. (8) is intended to
avoid assigning too much torsion to the
ledge. If closed stirrups are provided in
the ledgey, = 1.0; otherwise:

yi = —
^T,u = I (9)

where
TT = torsional moment strength pro-
vided by concrete
T. = factored torsional moment at

U^ critical section
Finally, if the end of the L-beam is
dapped, the end reaction will not
equilibrate Vr and T 1 . Therefore, for
Fig. 25. Notation for hanger reinforcement
design. dapped-end beams, the total hanger
reinforcement is given by:
IV.(d+a)
EA,,,= (10)
depends on the internal shear stress ct f,, d
distribution, which is calculated by in-
tegrating VQ/l from the top of the ledge For the L-heams included in this
to the bottom of the beam. In lieu of an study, Eq. (6) would require about 30 to
exact solution, the following expres- 60 percent more hanger reinforcement
sion, based on the parabolic shear than Eq. (2), depending on y,. As previ-
stress distribution in a rectangular ously noted, the use of Eq. (3) doubles
beam, gives a conservative approxima- the hanger reinforcement requirements
tion of AV,. compared to Eq. (2). Hanger reinforce-
ment is not additive to shear and torsion
reinforcement.
AVt =Vu (3 – 2h1 /h) h1/h)2 (7) The background research revealed
that at least four load tests of spandrel
where beams were conducted by precast pro-
h = overall height of beam ducers several years ago. During two of
hr – height of ledge these load tests, the Iedge of an L-beam
Observe that AT1 depends on the tor- separated from the web. A more detailed
sional strength of the ledge compared to discussion of these prior tests is pro-
the total torsional strength of the beam. vided in Project No. 5 report,' 1 which is
Accordingly: published separately. Similar to the test

108
 d PERIMETER OF CRITICAL SECTION {bo) .

CENTROID OF I A
CRITICAL SECTION F ' . . -

e^ FACE OF WEB
C

BRG.PAD

h2 /2 b h,/2

Fig. 26. Plan view of ledge showing eccentricity of ledge load

relative to critical section.

of Specimen 1, in these prior load tests a unexpected result of the load tests was
wide horizontal crack developed at the the early punching shear failures in the
ledge/web junction. In each case, the ledge of Specimen 2. As discussed in the
test was stopped before the ledge actu- background section, other researchers
ally fell off. All tests indicated the have found that the PCI equations for
ledge-to-web connection was very duc- ledge punching shear may he unconser-
tile despite very light hanger reinforce- vative. One reason may be that the PCI
ment. The behavior of these test speci- equations do not fully account for the
mens suggests that due to strain hard- eccentricity between the applied load
ening, forces in the hanger reinforce- and the centroid of the critical section.
ment approaching the ultimate tensile This eccentricity is shown in Fig. 26.
strength can be developed. The ultimate The analysis approach used to inves-
ledge loads, calculated using Eq. (6), are tigate transfer of unbalanced moment
comparable to the maximum test loads. between slabs and columns can be
The reinforcement ratio (AIsd, adapted to punching shear of beam
where s is the ledge load spacing) of ledges. The shear stress at the inside
these spandrels and Specimen 1 was edge of the ledge is given by:
roughly 100/f,,. This amount is similar to
the minimum requirement for structural
vC= V11 ^ + V. - 4 ^'7T (1i)
slabs. Because of the ductility demon- Jc
strated in these tests, a minimum rein-
forcement ratio of 100 /f,1 is recom- where
mended for hanger reinforcement. The bo = perimeter of critical section
effective distribution width for hanger er = distance between ledge load and
reinforcement is discussed later. centroid of critical section
Ledge Punching Shear — The most c = distance between centroid of

PCI JOURNAL/September-October 1986 109

 Fig. 27. Local failures related to punching shear strength of ledge.

critical section and inside face of shear resistance of the ledge. However,
ledge this study provides evidence that the
J = property of critical section anal- PCI design equations may be unconser-
ogous to polar moment of inertia vative in some situations, and further re-
(see Ref. 17) search is recommended.
This formula assumes that the full Distribution of Ledge Reinforcement
height of the ledge is effective and none — Prior to cracking, the L-bearn speci-
of the eccentricity is resisted by ledge mens showed evidence of higher stres-
flexure. The computed punching shear ses in the ledge hanger and flexure
capacity of Specimen 2 using Eq. (11) is reinforcement in the vicinity of the
40.5 kips, which is comparable to the applied load. The finite element model
failure load of 42.7 kips. Punching shear showed a similar concentration of stress.
capacity can be improved by increasing However, the hanger reinforcement
the ledge projection or depth. The use of strain was much more evenly dis-
developed ledge flexure reinforcement tributed after the horizontal crack at the
should also increase punching shear ledge/web junction had fully developed.
capacity. As the ledge separated from the web
Eq. (11) cannot be accurately applied along the entire length of Specimen 1, it
to conditions where flexural reinforce- was clear that all of the hanger rein-
ment developed across the critical sec- forcement between ledge loads was ef-
tion can help resist eccentricity. Also, fective. Ledge flexural cracks did not
shear and tensile stresses acting on the develop, so nothing was learned about
full section may reduce the punching the post-cracking distribution of strain

110
 Fig. 28. Local hanger reinforcement failure related to bending strength of ledge.

in ledge flexure reinforcement. flexural reinforcement and hanger

Of course, these results are only ap- reinforcement should also be concen-
plicable to L-beams with geometry and trated at the ledge load.
reinforcement similar to the test speci- Fig. 28 shows a local separation be-
mens. Local ledge failures are conceiv- tween the ledge and web related to the
able, particularly if the loads or load bending strength of the ledge. Assuming
spacing are not uniform. Fig. 27 shows the hanger reinforcement stress is
two local failures in which the ledge evenly distributed between ledge loads
flexure or hanger reinforcement as- (and neglecting 3V! ) the upward force
sumed to resist each ledge load is not between loads is equal to V„/s, where V„
fully effective. However, the shear and is the stem reaction and s is the ledge
torsional strength across the inclined load spacing. The corresponding sum of
failure planes abc and def contribute to the negative and positive bending mo-
the strength. ments in the ledge is equal to V„s18. The
Note that this contribution is related reinforcement required to resist this
to the punching shear strength of the bending moment is given by:
ledge. Even though the ledge rein-
forcement and shear strength may not be A,( = V°s (12)
fully additive, premature failures of the 8 djM
type shown in Fig. 27 are unlikely. On
the other hand, if the reinforcement at where
the ledge load is required to supplement A,,= ledge reinforcement in top or
the punching shear strength, the ledge bottom of ledge in addition to

PC] JOURNAL/September-October 1986 111

 reinforcement required for pri- Beam Pockets
mary moment
de = effective depth ofA,j During Phase 2 of the Specimen 3
0 = strength reduction factor = 0.85 test, the concrete below one of the beam
pockets punched out at a load of 47.6
Once again, use of a strength reduc- kips, The predicted failure load based
tion factor equal to 0.85 instead of 0.9 on yielding of the hanger reinforcement
compensates for the ratio of internal is 30.8 kips. The difference is apparently
moment arm to total effective depth. due to a punching shear strength con-
In summary, this research suggests tribution. Based on Eq. (11), the pre-
that all of the hanger reinforcement or dicted punching shear strength is 31.1
ledge flexure reinforcement between kips per stem. FuIIy developed inclined
ledge loads can be considered effective cracks below the pocket were observed
providing the punching shear and lon- at tee stem loads of 25 kips. These re-
gitudinal bending strength [Eq. (12)] of sults indicate that the strength con-
the ledge are adequate. Further testing tributions from hanger reinforcement
should be carried out to verify this as- and punching shear are not fully addi-
sertion, tive.

7. FINDINGS AND RECOMMENDATIONS

The following paragraphs describe — Au approach for considering the ef-
the findings based on the background fect of the pocket on the shear strength
research, analytical studies, and load of pocket spandrels has been proposed.
tests described herein. While the accuracy of this approach has
• Critical Section — Because span- not been fully verified by tests, it is be-
drel beams are loaded near the bottom, a lieved to be conservative.
critical section for shear and torsion at • Detailing Practices — The torsional
the face of the support should be con- response of deep spandrels is domi-
sidered. nated by out-of-plane bending. The use
• Influence of Deck Connections - of lapped-splice stirrups and longitudi-
Connections to deck elements do not nal reinforcing bars without hooks does
substantially reduce torsion, however, not appear to have any detrimental ef-
they are effective in restraining lateral fect.
displacement induced by bending about • Beam End Design — Two inde-
the weak principal axis. pendent design checks in the end region
• Shear and Torsion of Prestressed of spandrels are recommended. First,
L-Bearns — Methods which consider a reinforcement should be provided to re-
concrete contribution for shear and tor- sist out-of-plane bending caused by the
sion design of prestressed spandrels, horizontal torsional equilibrium reac-
such as the Zia-McGee or the Zia-Hsu tions. This reinforcement is not additive
methods, have been verified by two to the reinforcement for internal torsion,
tests. Design methods based on com- and very little supplemental steel will
pression field theory are somewhat be required provided a critical section
more conservative, particularly with re- for shear and torsion at the face of the
gard to the requirement for negative support is considered. Second, the de-
bending strength at the face of the sup- veloped force in the primary longitudi-
port. nal reinforcement at the face of the sup-
• Shear Strength of Pocket Spandrels port, or bottom corner of a dapped-end

112
connection, should equilibrate the • Ledge Punching Shear — PCI de-
applied normal force, as well as the sign equations for the punching shear
axial force induced by the vertical re- strength of beam ledges may be uncon-
action. servative. Further research in this area
• Ledge Hanger Reinforcing — The is recommended.
eccentricity of the ledge load cannot be In closing, it should be re-emphasized
neglected in design of hanger rein- that this study has focused on spandrel
forcement for ledge-to-web attachment. beams as load-carrying components. In
Nonetheless, not all of the load acting on this regard, the research has gone a long
the ledge is suspended from the web way toward the understanding and res-
and the effective eccentricity of the olution of several fundamental aspects
ledge load is significantly reduced due of spandrel beam design. The findings
to torsion within the ledge. A design generally apply to both prestressed and
procedure which considers these effects nonprestressed reinforced spandrels
has been recommended. Load tests commonly used in buildings and park-
conducted under this program and by ing structures. However, forces from
others have verified this procedure. In frame action, volume change, handling
addition, it was determined that hanger and vehicular impact were not dis-
reinforcement is not additive to shear cussed, and the report does not fully ad-
and torsion reinforcement. Minimum dress tolerances, corrosion protection or
hanger reinforcement amounts are rec- connection details. These factors must
ommended and distribution of ledge also he carefully considered during the
reinforcement is discussed. design process.

ACKNOWLEDGMENTS

Throughout the study, the Steering by J. W. Peters & Sons, Inc. of Bur-
Committee for PCISFRAD Project No. 5 lington, Wisconsin. Their performance
provided helpful guidance and per- in this difficult and precise task is a
spective. In particular, Ned Cleland, credit to their talent as a precast pro-
Alex Aswad, and Kamal Chaudhari con- ducer.
tributed significantly through their con- Also, the author wishes to express his
structive comments. appreciation to Susan Klein of Susan
The support of Wiss, Janney, Elstner Klein Graphic Design for her help in
Associates, Inc. in conducting this re- preparing the graphic figures.
search is gratefully acknowledged. The Finally, this research was funded by
author would like to specifically thank the PCI Specially Funded Research and
John Hanson, John Fraczek, Lilia Gli- Development Program. The author
kin, Dirk Heidhrink, and Doris Nelson wishes to thank the administrators and
for their assistance. contributors to that program who made
The test specimens were fabricated this research possible.

NOTE: Discussion of this paper is invited. Please submit your

comments to PCI Headquarters by May 1, 1987.

PCI JOURNAUSeptember-October 1986 113

 REFERENCES
1. MacGregor, James G. (Chairman), "The crete," Preprint 3424, ASCE Chicago
Shear Strength of Reinforced Concrete Exposition, October 1978.
Members," by the Task Committee on 10. Raths, Charles H., "Spandrel Beam Be-
Masonry and Reinforced Concrete of the havior and Design," PCI JOURNAL,
Structural Division," Journal of the V. 29, No. 2, March-April 1984, pp.
Structural Division, ASCE, V. 99, No. 62-131.
ST6, Proceedings Paper 9791, June 1973, 11. Collins, Michael P., and Mitchell, Denis,
pp. 1091-1187. "Shear and Torsion Design of Pre-
2. PCI Design Handbook — Precast and stressed and Non-Prestressed Concrete
Prestressed Concrete, Third Edition, Beams," PCI JOURNAL, V. 25, No. 5,
Prestressed Concrete Institute, Chicago, September-October 1980, pp. 85-86.
Illinois, 1985- 12. Mirza, Sher Ali, and Furlong, Richard
3. Notes on ACI 318-83, Fourth Edition, W., "Serviceability Behavior and Failure
Portland Cement Association, Skokie, Mechanisms of Concrete Inverted T-
Illinois, 1984, pp. 14-28. Beam Bridge Bentcaps," ACI Journal,
4. ACI Committee 318, "Building Code Proceedings V. 80, No. 4, July-August
Requirements `or Reinforced Concrete 1983, pp. 294-304.
(ACI 318-83)," American Concrete In- 13. Mirza, S. A., and Furlong, R. W., "Design
stitute, Detroit, Michigan, 1983. of Reinforced and Prestressed Concrete
5. Cleland, Ned M., "Identification of Sec- Inverted T Beams for Bridge Structures,"
ondary Behavior in Combined Bending, PCI JOURNAL, V. 30, No.4, July-August
Shear, and Torsion of Reinforced Con- 1985, pp. 112-137. See also Discussion of
crete Ledger Beams," PhD Dissertation, Ref. 13 by Basile G. Rabbat and S. A.
University of Virginia School of En- Mirza and R. W. Furlong in PCI JOUR-
gineering and Applied Science, August NAL, V. 31, No. 3, May-June 1986, pp.
1984. See also "Behavior of Precast 157-163.
Reinforced Concrete Ledger Beams," by 14. Krauklis, A. T., and Guedelhofer, O. C.,
Cleland, Ned M., and Baber, Thomas T., Discussion of "Spandrel Beam Behavior
PCI JOURNAL, V. 31, No. 2, March- and Design" (Ref. 10), PCI JOURNAL,
April 1986, pp. 96-117. V. 30, No. 5, September-October 1985,
6. Iverson, James K., and Pfeifer, Donald pp. 171-174.
W., "Bearing Pads for Precast Concrete 15. Hanson, John M., "Square Openings in
Buildings," PCI JOURNAL, V. 30, No. 5, Webs of Continuous Joists," PCA Re-
September-October 1985, pp. 128-154. search and Development Bulletin, Port-
7. Zia, Paul and McGee, W. Denis, "Tor- land Cement Association, Skokie, II-
sion Design of Prestressed Concrete," linois, 1969, 14 pp.
PCI JOURNAL, V. 19, No. 2, March- 16. Klein, Cary J., "Design of Spandrel
April 1974, pp. 46-65. Beams," PCISFR.AD Project No, 5, Pre-
8. PCI Design Handbook — Precast and stressed Concrete Institute, Chicago, 11-
Prestressed Concrete, Second Edition, linois, 1986, 100 pp.
Prestressed Concrete Institute, Chicago, 17. Rice, Paul F., et al, Structural Design
Illinois, 1978. Guide to the ACI Building Code, Third
9. Zia, Paul and Hsu, Thomas, "Design for Edition, Van Nostrand Reinhold Co.,
Torsion and Shear in Prestressed Con- Inc., New York, N.Y., 1985, 477 pp.

114
 APPENDIX A-NOTATION
a = shear span, distance between f, = yield strength of nonprestressed
concentrated load or reaction and reinforcement
hanger reinforcement f, = ultimate tensile strength of
A, = area of flexural tension rein- reinforcement
forcement h = overall height of section
A,,, = area of hanger reinforcement h, = height of ledge
A,, = area of reinforcement in top or hp = height of pocket in pocket span-
bottom of ledge in addition to drel
reinforcement required for pri- h, = height of beam effective in re-
mary moment sisting bending due to torsional
A, = area of shear reinforcement equilibrium reactions
A,,t = area of longitudinal web rein- j = ratio of internal moment arm to
forcement for bending due to tor- total effective depth
sionalequilibrium reactions J, = property of critical section
A,,, = area of vertical web reinforce- analogous to polar moment of in-
ment for bending due to torsional ertia
equilibrium reactions axial force at bearing
b = bearing width of concentrated s = spacing of shear or torsion rein-
ledge load forcing
b! = width of ledge measured along s = spacing of ledge loads
bottom of beam T. = torsional moment strength pro-
= perimeter ofcritical section vided by concrete
bu = web width Tj = torsional moment in ledge
c = distance from extreme fiber to T', = factored torsional moment at
neutral axis critical section
d = distance from extreme compres- VC = shear strength provided by con-
sion fiber to centroid of flexural crete
tension reinforcement Vr = shear in ledge
dr = effective depth of ledge rein- V, = factored shear force
forcement V„ = factored reaction
e = distance from centerline web to x = shorter oversll dimension of
ledge load rectangular cross section
er = distance from centroid of critical y – longer overall dimension of rec-
section for shear to ledge load tangular cross section
f = compressive strength of concrete A = symbol for difference
= square root of compressive yi = reduction factor for torsion in
strength of concrete ledge
f,r = developed stress in primary = capacity reduction factor
flexural reinforcement E = summation symbol

PCI JOURNAUSeptember-October 1986 115

 APPENDIX B - SPANDREL DESIGN CHECKLIST
The following checklist items are —Due to vehicular impact
presented in an order of their usual con- • Sweep due to strand eccentricity
sideration in the design process of span- • Principal axis analysis for slender
drel beams. Note that these items are L-beams
not necessarily given in their order of
importance. Some of these design con-
Shear and Torsion
siderations are illustrated in Appendix
C; however, due to the limited scope of • Eccentricity contributing to torsion
research under PCISFRAD Project No. • Minimum and maximum torsion
5, many of the items listed below are not • Transverse reinforcement
addressed. The reader is directed to the • Longitudinal reinforcement
appropriate section of the PCI Design
Handbook and Ref. 10 for discussion of Beam End Design
design considerations outside the scope
of this research. • Torsion equilibrium reinforcement
• Longitudinal reinforcement at end
• Beam hearing design
Dimensions • Dapped end design
• Span
• Web height and width Ledge Design
• Ledge depth and projection • Tee stem bearing
• Daps and blockouts • Punching shear:
—At interior reaction
Loads —At outside reaction
• Ledge flexure
• Dead and live • Hanger reinforcement
• Frame action • Ledge distribution reinforcement
• Volume change
• Vehicular impact
Details
• Column and deck connnections
Flexure • Reinforcement details:
• Service load stresses: — Anchorage/development
—At release — Spacing
— In service — Tolerance and clearance
• Flexural strength • Corrosion protection:
• Minimum reinforcement —Concrete cover
• Out-of-plane bending: — Protection of exposed plates
— During handling — Protection of end of strand
— During erection • Inserts for handling

116
 APPENDIX C - DESIGN EXAMPLES
To illustrate the proposed design pro- y„R =5.0in.
cedure of spandrel beams, two design At release (7 percent loss)
examples are presented. Example 1 _f, Jt
covers an L-beam while Example 2 Computed (psi) 483 -215
treats a pocket spandrel. Both examples Allowable (psi) 2100 -355
pertain to a parking structure but the
basic principles apply to other types of In service (17 percent loss)
structures. J;, fr .fi fe
Computed (psi) 166 525 148 430
EXAMPLE 1 - L-BEAM FOR Allowable (psi) -424 2250
PARKING STRUCTURE Ultimate strength:
An„=0.612in.'
Design Loads A, = Four #4 bars = 0.80 in .2
Stem reactions OM1 = 9243 (prestress) plus 2654 (mild
Dead load (90 psi) _ steel reinforcement) = 11,897 in.-kips
0.09(60/2)4 = 10.8 kips M, = 8245 in.-kips < 11,897 in.-kips
Live load (50 psi) = 1.2MMcr = 1.2 (7.5 y f7 + ff )Zb
0.05 (60/2)4 = 6.0 kips = 1.2 (7.5 +430)
= 16.8 kips 9406/1000
Total service load
= 10,840 in.-kips < 11,897
Factored load = 1.4 x 10.8 + 1.7 x 6.0
in.-kips
25.3 kips
Equivalent uniform load Refer to Fig. C1 for the L-beam geom-
etry and design data.
Service: w = 16.8/4 + 0.675
= 4.88 kips/ft
Factored: w,, =25.3/4 + 1.4 x 0.675 Shear and Torsion
= 7.27 kips/ft
The basic uniform loads are increased The shear and torsion design follows
by the ratio of grid span to design span. the Zia-Hsu method A See Fig. C2 for
Grid span = 28.0 ft the bending, shear and torsion diagrams.
Shear span = 27.0 ft Shear and torsion properties of section:
Service (adjusted):
Element x xZy
w = 4.88 x28/27 = 5.06 kips/ft
Factored (adjusted): Web (above ledge) 8 60 3840
w,, = 7.27 x 28/27 = 7.54 kips/ft Ledge 12 14 2016
J x2y = 5856
Flexure b,,d= 8x66.6=533in.2
The following is a summary of the C t = b,^dllx'i^
flexure design. Refer to Section 4.2 of = 5353/5856
the PCI Design Handbook for details of = 0.091 in.-'
the design procedure. Minim um torsion
Yt = Vi + 10 f,,.l f
Service load moment = 5533 in.-kips
Note: The moment computed using the _ 110x 48/500O
adjusted equivalent uniform load is = 1.14
about 2 percent greater than the value = 4, (0.5,1 YtY.x2J)
computed using concentrated loads. = 0.85 (0.5 000,/5 x 1.14 x 5856)
Prestress: Four ' -in, diameter strand, = 201 in.-kips < 708 in.-kips.

PCI JOURNAL/September-October 1986 117

 60 SPAN
$' 6'
4•
BAG

3" TOPPING (NORMAL WT)

8DT 24 (NORMAL WT)

DESIGN DATA SECTION PROPERTIES

= 5000 psi A = 648 in.'
= 3500 psi I = 307,296 in'
f, = 60 ksi y = 32.67 in.
f og = 270 ksi Ze = 9406 in'
( 1/2 in. diameter stress-reNeved strand) Z, = 7813 in?
clearance to stirrups = 1 1/4 in. weight — 0.675 kips/ft

Fig. C1. L-beam geometry and design data.

fm
Therefore, torsion design is re- Note: Strand is not developed at sup-
quired. port, therefore, = 0 and y, = 1.0.
Maximum torsion
C = 12-10 (f,If,) TG = 2V T :x2t!(yt-0.6)
= 12-10(148/5000) = 2 5000x5856(1.0-0.6)
= 11.7 = 331, 000 in.-Ibs = 331 in.-k ips
(113) C yr ,,7' 1 x'y
V^, = V'1^ 1 + [(V, T,,)1(T^V.)]'
Tmaz = = 131.9r ^1+[ 131.9x708133 xx101.8)]
^I1+ Cy,V^30C,T„
= 44.8 kips
— (113)11.7x1.14500 0 x5856
TT=T/d1+[T,V^/
— Y 1+ 14x101.8)/(30x0.091x708)
L091x708 = 3311 0 + [(331x101.8)/(131.9/708 ]'
= 1540 in.-kips > 708 in.-kips (ok) = 311 in.-kips
Shear and torsion strength of concrete At quarter point:
r—
f
At Support: Mtr — 28( 6 VTr +fur)
Vt = V = (3.5 J + 0.3f )b.d +Vp = 9406(6 v' + 430)11000
= (3.5 V, 5(X)0+0.3x0)8x66.8 + 0 = 8035 in.-kips
= 131,900lbs=13I.9kipscVii V^ = V,, = 0.6 TJr b^d +V, Mcr/M,

118
 (j SUPPORT
13.5 r Cr SPAN

M. 0
(in-k)

vu
(kips}

Tu
(In-k)

Fig. C2. Moment, shear and torsion diagrams.

= 0.6 v 5000x8x66.6/1000 + = 5221(1.5x5x69x60)

50.9x803516184 = 0.017 in?/in. = 0.20 in.2 /ft
8 8.7 in.-kips (A z + 2A 1 )Is = 0.23 + 2x0.20
T,' = 2 00x5856(1.14 - 0.6)11000 = 0.63 in.21ft
= 447 in ,-kips Min(A„+2At)Is
Vc = 88.7/,1T+ + [(88.7x354)/(447x50.9)]2 = 50(bW If,,) (1 + 12f^.If,' 200b,,1f,
= 52.0 in.-kips = 50(8/60,000) (1 + 12x148/5000)
TT = 4471 ç I + [ 447x50.9 /(88.7x3540) ]r = 0.009 in 2/in. = 0.11 in?/ft
= 362 in.-kips See "Beam End Design" for selection
Transverse reinforcement of reinforcement.
At support: At quarter point:
A„Is = (Vk / - VV)ldf,, A a, /s = (50.910.85 - 52.0)1(66.6x60)
= (101,8'0.85-44.8)/(66.6x60) = 0.002 in.2 = 0.02 in.2 /ft
= 0.019 in 2/in. = 0.23 in.2/ft T, = 354/0.85 - 362 = 54 in.-kips
Ts = T,I4 - T^ A,ls = 541(1.5x5x69x60)
= 708/0.85-311 = 0.002 in.r/in. = 0.02 in.5/ft
= 522 in.-kips (A„ + 2A,)!s = 0.02 + 2x0.02 = 0.06 in?/ft
= 0.66 + 0.33 y1/x 1 1.5 Min (A. + 2A1 )Is = 0.11 in.2lft (controls)
= 0.66 + 0.33x6915 = 5.2 Use #3 bars at 12 in.; 0.11 in./ft,
Therefore, a r = 1.5. Longitudinal reinforcement
A,ls = Tla, x, y, f, At = (2A,/s) (x l +y,) Ref. 9, Eq. (7)

PCI JOURNAL/September-October 1986 119

 Summary of the required transverse and longitudinal reinforcement.

A, A,
Location 2A,/s [Eq. (7) ] T. VM I Eq. (8)1 A,
(in=/in.) (in. 2 ) (in.-kips) (kips) (in.2)

At support 0.034 2.52 708 101.8 0.05 2.52

At quarter point 0.002 0.15 354 50.9 1.90 1.90

or Longitudinal reinforcement at end

At ' 440x T. - 2A, + N„=0.2V„=0.2x101.8=20.4 kips
1 yI)
f,, T^ +Vu 13C, s , a =5 +(h-d)=5 +(72 -66.6)= 10.4 in.
Ref. 9, Eq. (8) 4A,f,,, =N„ hld + V„ (0.5 + aid)
whichever is greater, where: = 20.4x72166.6 + 101.8(0.5 +
2A,/s tin Eq. (8)] _- 50b.,, (1 + 12f„,Ife)/f„ 10.4/66.6)
= 0.009 in.2/in. 88.9 kips
A summary of the required transverse Summarized at the top of the next
and longitudinal reinforcement of the page are the developed stresses and
L-beam at the support and quarter point forces of the provided reinforcement.
is given above.
Use seven #4 bars each side of the Ledge Design
web;Ar = 2.80 in.2 Bearing, punching shear and ledge
flexure
Beam End Design The following is a summary of the
ledge design following the PCI Design
Torsion equilibrium reinforcement Handbook procedures (refer to Part 6 of
d,=8- 1.25 - 0.5 = 6.25 in. the Handbook).
h,,=72- 12-6=54 in. Bearing: Bearing reinforcement is not
A max A wl _ required.
T. 708
= 1.11 in.x Punching shear: Punching shear
2 ct fv d, 2x0.85x60x6.25 strength is about twice the 25.3-kip stein
A,^1h, = 1.11/54 = 0.021 in.2/in. = 0.25 reaction. Note that the apparent inaccu-
in.2/ft racy of the PCI equations here is not a
Ai ls = 0.20 in 2i t (see transverse rein- concern. Also, the 42.7 kip test result is
forcement calculated previously). much greater than the stem reaction.
Therefore, Ledge flexure: A. = 0.50 in." distributed
AIs controls. evenly between stem reactions. Use #4
(A,, + 2 A,, )ls = 0.23 + 2x0.25 = 0.76 bars at 12 in.;A, = 0.80 in.2
in.2/ft. Use #4 stirrups at 6 in. = 0.80 Hanger reinforcement
in.2/ft. Six #4 bars in web above ledge;
=1.20in2>1.11in2 V„= 25.3 kips
Therefore, the specified longitudinal AV1 =V„(3-2h,h)(htl1)2
reinforcement is adequate. =25.3(3-2x12172)(12/72)'
= 1.9 kips
Reinforced concrete bearing = TC IT,, - 311/708 = 0.44
Based on Section 6.9 of the PCI De- AT, = V.e ya WY)^d^/1x2ff
sign Handbook,A f +A, = 1.02 in. R Use = 25.3x8x0.44x2016/5856
two #7 bars welded to the bearing plate. = 30.7 in.-kips
Again, refer to the Handbook for details d = 8 - 1.25 - 0.25 = 6.5 in.
of the design procedure. a =4+ 1.25+0.25= 5.5in.

120
 Summary of the developed stresses and forces of provided reinforcement.

Reinforcement Developed stress Developed force

Four #4 bars 60x8/12=40ksi 0.9x40x0.8 = 28.8 kips
Four' in.
diameter strand 150x10125=60ksi 0.9x60x0.61 = 32.9kips
Two #7 bars 60 ksi 0,9 x 60 x 1.20 = 64.8 kips
(welded to bearing plate) 126.5 kips
(ok)

Summary of the required transverse reinforcement.

Parameter Formula Near support Midspan

Shear/torsion (0.5A, +A,)Is 0.32 0.11 (min)
Torsion equilibrium (0.5A. +A.)/s 0.38 —
Hanger reinforcement A,,, (per ft) 0.26 0.20
Provided #4 bars at 6 in. #3 bars at 6 in.
(0.40) (0.22)

A = [V,,(d+a) — AVrbt 12- The two #4 bars at the end of the

AT1 ]I(0fvd) ledge are not required for the basic
= [25.3(6,5+5.5) — 1.9x14/2 — 30.7]/ flexural moment. However, the bars are
0.85x60x6.5 needed to help resist 1.2Mer . Therefore,
= 0.78 in.2 they may he considered as A„ 1 rein-
Near support use #4 stirrups at 6 in., 3 ft forcement.
tributary length at end reaction.
Ae5 = 0.4 in. 2/ft (3 ft) = 1.20 in.!
Midspan: Add #3 L-bars at 12 in.; alter-
nate with #3 stirrups at 12 in. EXAMPLE 2- POCKET
A,5 =2x0.11x4 =0.88in2 SPANDREL FOR PARKING
Minimum:Afn = 100.sdlf„ STRUCTURE
= 100x48x6.5/60,000
This example illustrates the design of
= 0.52 in.2
shear, end region, and hanger rein-
A summary of the transverse rein-
forcement for a dapped pocket spandrel.
forcement (inside face, in. 2/ft) is given Note that the pocket is provided near
above. the dapped end. Often this pocket is
Ledge distribution reinforcement omitted due to detailing difficulties. A
Punching shear strength is adequate. welded bracket or Cazaly hanger is used
Therefore, all hanger reinforcement and instead.
ledge flexure reinforcement between Shear and bending forces are identical
ledge loads are considered effective, to those used in Example 1 (Fig. C2).
provided the flexural strength of the Refer to Fig. C3 for framing details and
ledge is adequate. design data.
dr =12--3=9in. In addition, the following is given:
AF, =Vysl80djf5
ftt= 167psi
= 25.3x48/(8x0,85x9x60) 904 psi (at pocket)
= 0.33 in.' d = 67.0 in.

PCI JOURNALi September-October 1986 121

 60 SPAN

m
R, 21 611

n :.. • 3" TOPPING (NORMALWT)

I 8DT 24 (NORMAL WT)

ry BRG

6 SPACES AT 4'-0'=24-011

DESIGN DATA FULL SECTION AT POCKET

= 5000 psi A = 576 in?
fF, = 3500 psi 432 in'
= ! = 248,832 in.° 204,288 in.
60 ksi (bars) y, = 36.0 in.
fy = 70 ksi (WW F) 40.7 in.
Z, = 6912 in .3
f 270 ksi 5023 in .3
Z, = 6912 in, s5za in. 3

Fig. C3. Pocket spandrel geometry and design data.

Shear and Torsion

reactions at the supports.
Torsion at support Shear strength of concrete
Stem reaction = 25.3 kips; e = 2.0 in. At support:
7'. = 7x25.3x2.012 = 177 in.-kips V = Vice = (3.5 v i.' + 0.3f,) ba. (d –ha)
V

Inside outer reaction: _ (3.50 4 0+ 0)8(67.0 –

Tu = 5x25.3x2.0/2 = 127 in.-kips 240/1000
Minimum torsion = 85.1 kips
=,j1+1Of,^•If, At quarter point (see Section 11.4.2 of
=V1+10x167/5000=1.15 ACI 318-83 Commentary):
x'y = 82 x72 = 4608 Mrr =Z1,(6,J, +.f,,)
Tmfx = 43(0.5 VTl y Ix 2 [/ ) = 5023 (6^ 5(?((1 + 904)/1000
= 0.850„5 V^' ^00x1.15x4608/1000
50 = 6672 in.-kips (at pocket)
= 159 in.-kips
Ve = V1 = 0.6 .e b.(d –hg) +V. Mcr./M,
Therefore, torsion design is not re- = 0.6 \ 3OOp x8(67.0 – 24.0)/1000 +
quired inside the outer reaction. Design 50.9x667216184
the end region for torsion equilibrium = 69.5 kips <V„

122
 2-18 AT CENTERLINE
WELDED TO END ANGLE

I!*Wi END DETAIL

N" 2-#7
2-#7
FORCE MODEL be

65"
Ik
/ (Vc+V.)
101.8k 25•3k

T,
d

FORCE MODEL de FORCE MODEL fg

Fig. C4. Dapped end detail and force models.

Shear reinforcement h8 =38-6= 32 in.

A, h = u l0 — V, )I (d — hp )ff A,c„/s = A Is = 0.23/32
= 01.810.85 — 85.1)/(67.0 - = 0.0072 in?/in.
24.0)60 = 0,086 1n2/ft
= 0.16 in?/ft or 0.013 in 21in. Use an additional layer of 6x6 —
MinAnIs = 50b,^1f, W4.OxW4.0 inside face, each end.
= 50x8/60,000 = 0.0067 in.2iin. A,,,,;/s =A,,,Is = 0.08 in.s/fl:
= 0.080 in.=/ft Dapped end design
Use one layer of 12 x 6 — W2.0 x W4.0 Dapped end design is based on the
each face, full length. end detail and equilibrium force models
Ads = 2x0.08 = 0.16 in.2lft shown in Fig, C4. It should be noted,
however, that the reinforcement scheme
and design procedure have not been va-
Beam End Design lidated by load tests.
Torsion equilibrium reinforcement For direct shear, see Section 6.13.2 of
d, = 8.0 — 1.5 = 6.5 in. the PCI Design Handbook.
A,,1, = A = T /2Of. d, µe = 1000 X bh ,LIV E 3.4
= 177/(2x0.85x70x6.5) = 1000xlx8x38x1.4/(101.8x1000)
= 0.23 in.' =4.18µe=3.4

PCI JOURNAUSepternber-October 1986 123

 A. = Nv l(cb}) = 20.4/(0.85x60)
= 2.06 in.2 (four #7 bars)(ok)
= 0.40 in.' EM,, = 0 -* T„ _ (20,4x36 + 101.8x17)133
A. = 2VWl(3^f .,,) +A„ = 74.7 kips
= 2xI01. 8/ (3x0.85x60x3.4) + 0.40 A„ = 74.7!(0.85x60)
=0.79in2 = 1.46 in.2 (two #8 bars)(ok)
Two #8 bars p rovided; A s = 1.44 in.2 Full section (force model fg)
Ah= 0.5(4,-A„)
Hanger reinforcement is not effective
= 0.5(0.72 - 0.40) = 0.16 in.2 due to bend. NeglectA„ reinforcement.
6x6 - W4.OxW4.0 provided I Mo = 0 -* T, = (20.4x36 + 101.8x67.5 -
Ah = 3.0x0.08 = 0.24 in.2
25.3x49.5)/65
Crack at re-entrant corner (force model = 97.7 kips
be): Neglect inclined hanger reinforce- From PCI Design Handbook Fig.
ment. 4.10.4j,, = 170 ksi.
IF0.->T,,=V,,= 101.8 kips 0A,B fpr = 0.9x0.61x170 = 93.3 kips
A,,, = T„,,/ca f„ = 101.81(0.85x60) (say ok)
= 2.00 in.2 Check depth of compression block
Four #7 bars provided, A,, = 2.40 in.2 = I Fh10.85hfc
= (97.7 - 20.4)/0.85x8x5 - 2.3 in.
0 -* T„=(20.4x36+
a12=2.312-1.2 in. <2 in. (ok)
101.8x'9.5)/33
= 51.6 kips
A„ = T„/0f,=5I.61(0.85x60)= 1.01 in.2 Hanger Reinforcement
Two #8 bars provided;A B = 1.44 in.2 At pocket
Crack at bottom corner (force model de) A,,, = V,,4 f„ = 25.31(0.90x60) = 0.47 in?
Neglect vertical hanger reinforcement Use one #4 U-bar (slope of 1 to 4) at
(not effective at bend).
each pocket (plus two W4.0 wires).
IFr = 0--*T =V„/cos14 = 101.8fcos14
Am = 2x0.20(cosl4) + 2x0.04 = 0.47 in.2
= 104.9 kips Idh = 1200c4/ `I f^ = 1200x0.51 5000
A■n = 104.91(0.85x60)
= 8.5 in. (ok)

Metric (Si) Conversion

Factors
1 It = 0.3048 m
1 in. = 25.4 mm
1 kip = 4448 N
fib =4.448N
1 psi = 0.006895 MPa
1 ksi = 6.895 MPa
1 psf = 4.882 kgf /m2

124