Design of Spandrel Beams, G.J. Klein

Specially Funded R&D Program

PCISFRAD Project No. 5

Summary Paper

Design of Spandrel Beamsby

.

Gary J. KleinConsultantbliss, Janney, Elstner

Associates, Inc.Northbrook, Illinois

Note: This summary paper is a slightly con-densed version of PCISFRAD Project No. 5,"Design of Spandrel Beams." The full report isavailable from PCI Headquarters at $8.00 tofirms supporting the sponsored research,$12.00 to PCI Members (non-supporting firms)and $24,00 to non-PCI Members.The summary paper, and the full report, arebased on a research project supported by thePCI Specially Funded Research and Develop-ment (PCISFRAD) Program. The conduct of the

research and the preparation of the final reportsfor each of the PCISFRAD projects were per-formed under the general guidance and direc-tion of selected industry Steering Committees.However, it should be recognized that the re-search conclusions and recommendations arethose of the researchers. The results of the re-search are made available to producers, en-gineers and others to use with appropriate en-gineering judgment similar to that applied to anynew 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 CONCLUSIONSA study of the behavior and design 01

precast spandrel beams was undertaken.This research project was primarily di-rected toward spandrel beams com-monly used in parking structures. BothL-beams and pocket spandrels were in-eluded in the study.

The research included backgroundinvestigation of design practices,analytical studies using finite elementmodels, and full-scale load tests of twoL-beams and one pocket spandrel. Allthree test specimens were 72 in. high, 8in. wide and 28 ft long. The target de-sign loads were based on 90 psf deadload and 50 psf live load, which are typi-cal for a double tee parking structurewith 60 ft spans,

The background research revealedthat industry practices and publishedprocedures vary with respect to severalfundamental aspects of spandrel beamdesignn. Behavior near the end regions isnot well understood, nor is the influenceof connections to deck elements. Ingeneral, the design of beam ledges is notconsistently handled; in particular,there is no consensus on the design ofhanger reinforcement for ledge-to-webattachment, Also, the ACI BuildingCode (ACI 318-83) does not addresscombined shear and torsion in pre-stressed beams. Designers rely on sev-eral research reports that give designrecommendations.

Ledge-to-web attachment and be-havior near the end region of spandrelswere identified as the key issues andwere the primary focus of this research.The analytical studies and laboratorytesting program yielded several signifi-cant findings:

• Contrary to several published de-sign examples, a critical section forshear and torsion at the face of the sup-port should be considered.

• Connections to deck elements donot substantially reduce torsion; how-ever, they are effective in restraining

lateral displacement induced by bend-ing about the weak principal axis.

• Shear and torsion design proce-dures for prestressed spandrels whichconsider a concrete contribution havebeen verified by two tests.

• An approach for considering theeffect of the pocket on the shear strengthof pocket spandrels has been proposed.While the accuracy of this approach hasnot been fully verified by tests, it is be-lieved to be conservative.

• With regard to detailing practices,it was found that the torsional responseof deep spandrels is dominated by out-of-plane bending. The use of lapped-splice stirrups and longitudinal rein-forcing bars without hooks does not ap-pear to have any detrimental effect.

• Two independent design checks inthe end region of spandrels are recom-mended. First, reinforcement should beprovided to resist out-of-plane bendingcaused by the horizontal torsionalequilibrium reactions. This reinforce-ment is not additive to the reinforce-ment for internal torsion. Second, thelongitudinal reinforcement in the bear-ing area should be sufficiently de-veloped to resist the external normalforce, in addition to the tension inducedby the vertical reaction,

• The eccentricity of the ledge loadcannot be neglected in the design ofhanger reinforcement for ledge-to-webattachment, Nonetheless, not all of theload acting on the ledge is suspendedfrom the web and the effective eccen-tricity of the ledge load is significantlyreduced due to torsion within the ledge.A design procedure which considersthese effects has been recommended. Inaddition, it was determined that hangerreinforcement is not additive to shearand torsion reinforcement.

• The PCI design equations forpunching shear strength of beam ledgesmay be unconservative. Further re-search in this area is recommended.

78

L- BEAM

mx00.aa

POCKET SPANDREL

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

2. INTRODUCTIONSpandrel beams are one of the most

complex elements in precast construc-tion. Industry practices and publishedprocedures vary with respect to severalfundamental aspects of their design.PCI Specially Funded Research andDevelopment Project No. 5 investigatedthe behavior and design. of precastspandrel beams.

The research program was primarilydirected toward deep and slender span-drels such as those commonly used in

parking structures to serve both load-carrying and railing functions. BothL-beams and spandrel beams withpockets for tee stem hearings (pocketspandrels) were included in the pro-gram. Fig. 1 shows typical cross sectionsof these types of beams.

The findings of this research gener-ally apply to both prestressed and non-prestressed spandrels, but may not beapplicable to spandrel beams of radi-cally different geometric configuration

6" TO 40" B"TO 10"

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

PCI JOURNALJSeptember-October 1986 79

or load level. Furthermore, while thisresearch is believed to be reasonablycomprehensive, not all aspects of span-drel beam design are covered.

In particular, the research does notaddress spandrel beam design as part ofa lateral load resisting frame, nor theeffects of volume change on design anddetailing of spandrels. Also, handlingand vehicular impact loads are not dis-cussed. These considerations can bevery important, but are considered be-yond the scope of this research.

The research program included the

following objectives:• Study of design requirements and

practices to determine the state-of-the-art of spandrel beam design.

• Analytical studies using finite ele-ment models of an L-beam and pocketspandrel.

• Full-scale tests of two L-beams andone pocket spandrel designed usingstate-of-the-art methods.

The following sections of this reportdescribe the research, analyze the find-ings, and provide design recommen-dations.

3..BACKGROUND RESEARCH

The background research included areview of code requirements, publishedguides and research reports on spandrelbeam design. Questionnaires regardingdesign methods for L-heams and pocketspandrels were sent to industry design-ers. The following discussion on span-drel beam design is based on this re-search.

General Design ConsiderationsCritical Section — In most precast

beams, the loads and reactions areapplied at the top and bottom of thebeam, respectively. Such beams are saidto be "directly loaded." Spandrelbeams, on the other hand, are indirectlyloaded, and the additional shear capac-ity due to arch action near the support isnot available.' Therefore, design forshear and torsion forces at a distanced(h/2 for prestressed spandrels) from thesupport may not be appropriate. Fig. 2shows potential critical inclined sec-tions which carry all the concentratedloads acting on the ledge rather than justloads farther than d from the support.

The consensus among designers isthat all loads acting on the ledge insidethe critical section, based on inclinedcracking from the edge of the beam base

plate, must be considered as part of theshear/torsion load. This consensus iscontrary to the published design exam-ples in Section 4.4 of the PCI DesignHandbook' and Example 14.2 in thePCA Notes on ACI 318-83_' ACI 318-83"does not address indirectly loadedbeams; however, Section 11.1.2 of theCommentary recommends special con-sideration for concentrated loads nearsupports.

Equivalent Uniform Load — It iscommon practice to simplify theanalysis by replacing concentrated loadswith equivalent uniform loads. Somedesigners increase the equivalent uni-form floor load such that the shear andtorsion is correct at the critical section atthe inside edge of the base plate, i.e.,the basic equivalent uniform load ismultiplied by the ratio of grid span todesign span.

Eccentricity Contributing to Torsion— Typically, the ledge loads are po-sitioned at the centerline of bearing (al-lowing for volume change and fabrica-tion and erection tolerances) or at theouter one-quarter point of the ledge.The former approach is generally pre-ferred because an increase in ledgeprojection does not necessarily requirean increase in torsional load. The ec-

80

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

centricity contributing to torsion in aspandrel is the distance from the cen-terline of the web to the applied load, asshown in Fig. 3.

Theoretically, the eccentricity shouldbe measured relative to the shear center,which, for an uncracked L-beam section,is slightly inside the centerline of theweb. However, this difference is negli-gible in deep spandrels. Further, experi-mental results are not consistent withthe theoretical prediction of shearcenter location based on the uncrackedcross section.'

Influence of Deck Connections -Prior to connection of the double tees ortopping to the spandrels, torsion can becomputed as a product of the dead loadand the eccentricity between theapplied load and centerline of the web.After connections to deck elements aremade, however, the applied live loadtorsion may be partially counteracted bythe horizontal force due to friction at thebearing pads coupled with restraint atthe deck connections (Fig. 3). However,most practitioners believe that it is in-appropriate to rely on a soft bearing padfor this purpose. In addition, recent re-search6 indicates that the effective fric-tion at the bearing pad may be 5 percentor less of the gravity load.

iWEB

^— CONNECTION TO DECK

SHEAR CEN'ER

e — rt BEARING

H

FRICTION AT BEARING

Fig. 3. Eccentricity contributing to torsion inspandrel beam.

PCI JOURNALJSeptember-October 1966 81

FlexureThe flexural design of spandrels gen-

erally follows ACI and PCI proceduresfor bending about the horizontal andvertical axes. However, L-shapedspandrel beams do not have symmetryabout either axis. The principal axes arerotated slightly from the vertical andhorizontal axes, as shown in Fig. 4. Theinfluence of this rotation on bendingabout the horizontal axis can be neg-lected for deep spandrel beams. Forshallow spandrels, particularly thoseemploying prestressing, this influenceshould be considered.

Perhaps more important, however, is

X-_XP

VxP

Y VP

Fig. 4. Principal axes of an L-beam.

the influence of principal axes rotationon horizontal displacement of spandrels.As shown in Fig. 4, a component of thevertical load acts along the weak axis in-ducing an outward horizontal displace-ment. All loading prior to making dia-phragm connections can cause horizontaldisplacement. Cleland 5 found that thiswas the most dominant behavior of longslender spandrels and suggests a princi-pal axes analysis when the span lengthis 40 to 50 times the web width, de-pending on the intermediate supportconditions.

In general, detailing practice followsthe ACI Code. One noteworthy excep-tion pertains to Section 10.6.7 of ACI318-83 which is applicable to nonpre-stressed spandrels. This provision re-quires that reinforcement be placed inthe side faces of webs more than 3 ftdeep. The reinforcement is to be dis-tributed in the zone of flexural tensionwith a spacing not more than the webwidth, nor 12 in. Designers do not oftencheck this provision; instead, rein-forcement in the side faces of the web isdesigned to resist torsion or handling.

Shear and TorsionPrestressed Spandrels — The ACI

Code does not address torsion in pre-stressed concrete. A procedure for tor-sion design of prestressed concrete,which is an extension of the AC1 provi-sions of torsion for nonprestressed con-crete, was developed by Zia andMcGee.' The second edition of the PCIDesign Handbook included a modifiedversion of the Zia and McGee method"The PCI procedure uses a simplifiedmethod for computing torsional stresswhich is conservative for most spandrelbeams.

A further refinement of these methodswas subsequently developed by Zia andHsu.9 While the general design ap-proach follows that of Zia-McGee andPCI, new expressions are proposed fortorsion/shear interaction and minimum

X pX

82

torsion reinforcement. The Zia-Hsuequations are expressed in terms offorces and moments rather than nominalstresses, which is more consistent withthe current ACI Code.

Most designers follow one of thesethree similar procedures. Practices varywith respect to the design of longitudi-nal reinforcement for torsion. Some de-signers consider the prestressing strandto be part of the longitudinal reinforce-ment while others consider only themild reinforcing steel. In their originalpaper, Zia and McGee recommendedthat only the prestressing steel in excessof that required for flexure, and locatedaround the perimeter of closed stirrups,should be considered as part of the lon-gitudinal torsion steel.

The third edition of the PCI DesignHandbook2 describes a procedure de-veloped by Collins and Mitchell, whichis based on compression field theory.This approach assumes that, aftercracking, the concrete can carry no ten-sion and that shear and torsion are car-ried by a field of diagonal compression.Because the concrete contribution isneglected, this approach will generallyrequire somewhat more stirrup rein-forcement depending on the selection ofthe crack angle. The biggest difference,however, is in the positive and negativemoment capacity requirements whichare based on the axial tension caused byshear and torsion. For the exampleshown in the PCI Design Handbook,the required positive and negativebending strength at the face of the sup-port exceeds the midspan moment.These requirements present consider-able detailing difficulties, and many de-signers do not feel they are valid fordeep spandrels.

Detailing practices for the torsionalreinforcement do not always follow ACICode requirements. Section 11.6.7.3 re-quires that transverse reinforcementconsist of closed stirrups, closed ties orspirals. However, the Commentary tothe ACI Code indicates that this re-

quirement is primarily directed at hol-low box sections and solid sectionssubjected primarily to torsion. In thesemembers, the side cover spalls off, ren-dering lapped-spliced stirrups ineffec-tive. This type of behavior is unlikely indeep spandrel beams, and transversereinforcement is often provided by pairsof lapped-spliced U-stirrups. Also, mostdesigners feel that the stirrup spacinglimit of 12 in. is not appropriate for deepspandrels, and this limit is routinely ex-ceeded.

Nonprestressed Spandrels — Torsiondesign of nonprestressed concrete gen-erally follows ACI Code requirements,except for the detailing considerationsdiscussed above.

Pocket Spandrels — Typically, pocketspandrels need not be designed for tor-sion. However, the pockets complicatethe shear design. Design practices varyfor considering the effect of the pocket;some designers neglect this effect. For-tunately, shear strength does not controlthe dimensions of deep pocket span-drels and often only minimum rein-forcement is required. Welded wire fab-ric is frequently used for web rein-forcement.

Beam End Design

Torsion Equilibrium — The eccentricload applied on the ledge produces tor-sion in the spandrel which must be re-sisted by reactions at the supports.Customarily, the web is connected tothe column to restrain rotation. Figs. 5aand 5b show the torsion equilibriumreactions for a normal and dapped con-nection, respectively.

The torsional equilibrium reactionsmay require supplemental vertical andhorizontal web reinforcement at theends of the girder. Raths 1 ° and Osborn*prescribe similar methods for design of

*Osborn, Andrew E. N., "Design of Ledger Gir-ders," Draft Report for PCI Connection DetailsCommittee, April 1984.

PCI JOURNALISeptember-October 1986 83

hg h^

(a) (b)

Fig. 5. Torsion equilibrium reactions in spandrel beam.

this reinforcement. Vertical and lon-gitudinal steel,A,,, andAw, on the insideface of the spandrel is calculated by:

Ate„ = i =T,, 1)

where

T. = factored torsional moment atend of girder (in.-lbs)

d, = depth ofA,,,,, and A 1 steel fromoutside face of spandrel (in.)

f„ = yield strength ofreinforement(psi) (or effective prestress)

= strength reduction factor = 0.85

The use of 0 = 0.85 instead of 0.90(flexure) compensates for the ratio ofinternal moment to total effective depth,which is not in Eq. (1).

Osborn recommends the bars heevenly distributed over a height andwidth equal to h. (see Fig. 5) from theconcentrated reaction point.

Because shear cracks may coincidewith diagonal cracks due to out-of-planebending, A. should be added to the

shear reinforcement. However, most de-signers feel this reinforcement is notadditive to reinforcement for internaltorsion. If the reinforcement for torsionis considered to function as A,^, and A,orreinforcement, little or no supplementalreinfprcement is required provided allloads acting on the ledge are consideredas part of the shear/torsion load.

Fig. 6 shows an alternative means toprovide torsional equilibrium at thesupport. In this case, the end reactionsare in close alignment with the ledgeloads. The projecting beam ledge istreated as an upside-down corbel. Mostdesigners surveyed indicated that thisapproach may Iead to excessive rollingof the spandrel beam at the support,particularly where a soft bearing pad isused.

Dapped-End Beams — Section 6.13 ofthe PCI Design Handbook presents de-sign criteria for dapped-end connec-tions. Research on dapped connectionsunder PCISFRAD Project No. 6, whichis being conducted concurrently with

84

V

Fig, 6. Beam end corbel behavior whenproviding torsion equilibrium at support ofspandrel beam.

Raths 1 ° uses all the hanger reinforce-ment between ledge loads, but com-putes the required reinforcement basedon the summation of moments about theoutside face of the spandrel, thus:

V„ (jd + a) (3)

4f id

whereA,h = area of transverse hanger rein-

forcement on inside face ofspandrel for each ledge Ioad (sqin,)

Vaz = factored ledge load (kips)a = distance from ledge load to

center of inside face reinforce-ment (in,)

jd = internal moment arm (in.)(taken as d – ½ in.)

^i = strength reduction factor = 0.85Raths recommends an additional load

factor of 4/3 for design of hanger rein-

this project, is expected to recommendmodified procedures. Design ofdapped-end L-beams is often compli-cated by reinforcement for torsionequilibrium connections (Fig. 5b). Also,the last blackout in a pocket spandreloften interferes with the reinforcing forthe dapped end. The established designprocedures are modified as appropriateto handle these special conditions.

Beam Ledges

Hanger Reinforcing — Fig. 7 illus-trates a possible separation between theledge and web of an L-shaped spandrel.Section 6.14 of the PCI Design Hand-book and design examples by PCA3 andCollins and Mitchell" recommendhanger reinforcement concentrated nearthe ledge load given by:

A,h = rV"

(2)

-A sh {INSIDE LEG ONLY)

POSSIBLE SEPARATIONBETWEEN LEDGE AND WEB

The notation is defined on the nextcolumn above.

Fig. 7. Ledge-to-web attachment showinghanger reinforcing.

PCI JOURNALSeptember-October 1986 85

Fig. 8. Hanger reinforcement in pocketspandrels.

forcement. An alternate procedure forusing concrete tension as a means ofledge-to-web attachment is also given inhis paper.

Eq. (3) is based on sound principles ofstatics, yet there are many existingspandrels that have performed well withmuch less reinforcement than thisequation would require. The onlyknown failures have occurred wherethere was no hanger reinforcement. Inseveral instances, beams with very lighthanger reinforcement have survivedloading tests.

Further refinements of hanger rein-forcement design' I •* .t reduce the loadthat must be suspended from the web

•Osborn, Andrew E. N., "Design of Ledger Gir-ders," Draft Report for PC:I Connection DetailsCommittee, April 1984.

tSturm, Edward R., "Theory of Deflection Com-patibility," Private correspondence with AndrewOsborn, May 1984.

based on internal shear stress distribu-tion, relative depth of the ledge, anddeflection compatibility.

There is no consensus among design-ers on requirements for hanger rein-forcement. Some designers do not checkledge-to-web attachment, while othersuse some combination of the abovemethods. Furthermore, there is noagreement as to whether or not hangerreinforcement should be added to shearand torsion reinforcement. The methodfor designing hanger reinforcementgenerally controls the quantity of trans-verse reinforcement in the middle re-gion of the spandrel, and can have a verysignificant effect on material and fabri-cation costs.

Ledge Punching Shear — The designfor punching shear in beam ledges gen-erally follows the procedures in Section6.14 of the PCI Design Handbook. Somedesigners follow a modified procedurerecommended by Raths; 1° based on un-published test results, this method con-siders a lower ultimate stress on thevertical shear plane along the insideface of the web. Mirza, et al l ' - 13 andKraukIis and Guedelliofer' have alsofound that the PCI design equationsmay be unconservative.

Beam PocketsIt is customary to provide closed stir-

rups or U-bars in the plane of the webfor the entire tee stem load in pocketspandrels. The hanger bars are typicallylocated near the tee stem reaction, asshown in Fig. S. Therefore, Eq. (2) isused to determine hanger reinforcementrequirements. The concrete tensilestress at the reaction level is relativelylow so a horizontal crack at that locationis unlikely. Also, because hanger rein-forcement is customarily used, punch-ing shear below the pocket is generallynot a concern.

Prior to describing the experimentalprogram, a summary of the finite ele-ment model studies is given.

UP

86

4. FINITE ELEMENT MODEL STUDIES

DescriptionFinite element models of an L-beam

and pocket spandrel were analyzed. Thegeometry of these models and the testspecimens was essentially the same.Refer to Figs. 13 and 14 for more de-tailed information on the geometry ofthe beams.

The model studies had several objec-tives:

• Investigate the deflections and ro-tations caused by the eccentricallyapplied load.

• Determine the theoretical torsionalequilibrium reactions at the supports.

• Study the influence of connectionsto deck elements on deformations andtorsional equilibrium reactions.

• Investigate the stresses across theledge/web interface.

Three-dimensional solid elementswere used with three degrees of free-dom at each node. Cross sectionsshowing the finite element mesh areshown in Figs. 9 and 10.

Service loads included beam (leadload and a 16.8 kip tee stem reaction at 4ft centers. The tee stem load wasapplied at 8 in. and 2 in. from the webcenterline for the L-beam and pocketspandrel, respectively. The restraints ateach end of the beam modeled a typicalspandrel beam support where the hear-ing pad is placed at the centerline of theweb, and lateral support is providednear the hearing and at the top cornersof the bean.

For both the L-beam and pocketspandrel, a second condition was ana-lyzed in which additional lateral re-straint was provided near midheight ofthe beam to simulate connections todeck elements. There was no possibilityof relative lateral movement betweenthe column restraints and deck ele-ments, simulating the case where thereis an independent connection betweenthe deck and the column. This case was

considered so the analytical studies andload tests modeled the same condition,although it should be noted that a directconnection between the column anddeck is not necessarily required. Alter-nately, the column can be indirectlyconnected to the deck through the span-drel beam.

Spandrel Beam BehaviorFig. 9a shows the midspan deflection

of the L-beam at service load withoutany connections to deck elements. Notethe overall outward deflection due tothe rotation of the principal axes. Con-nections to deck elements effectivelyrestrain this outward displacement, asshown in Fig. 9b. Usually these con-nections are not made until all of thedead load is in place. Similar plots forthe pocket spandrel are shown in Fig.10. Due to the different cross-sectionalshape and load eccentricity, the lateraldeflection is relatively small.

Fig. 11a shows the horizontal reac-tions at the L-beam support withoutconnections between the spandrel anddeck. These forces simply balance theexternal torsion due to the eccentricallyapplied loads. Fig. lib shows the hori-zontal reactions with deck connections.The deck connections in the midspanregion restrain the outward displace-ment. The deck connections at the sup-port work with the top corner connec-tions to restrain rotation. The net out-ward force between the deck and span-drel would be counteracted by thecolumn-to-deck connection, If therewere no column-to-deck connection, thedeck connection forces would tend tobalance, depending on the stiffness ofthe column.

Transfer of Ledge Loads to WebStresses across a plane 3 in. above the

ledge/web interface were studied. (The

PCI JOURNALJSeptember-October 1986 87

w00

F.E. LOA[MODEL TES"

VERT. (IN.) -0.060 -0.1:HORIZ. (IN.) -0.089 -0.0CROT. (RAG) -0.00163 -0.01

F.E. LOADMODEL TESTS

:RT. (IN.) -0.056 -0.131)RIZ. (IN.) -0.008 -0.057)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. LOADMODEL TESTS

VERT. (IN.) -0.053 -0.173HORIZ. (IN.) +0.024 +0.038ROT. (RAD) -0.00085 -0.00443

F.E. LOADMODEL TESTS

VERT. (IN.) -0.053 -0.146HORIZ. (IN.) 0.0 +0.013ROT. (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).

Pil(b) STRESS AT (C) CALCULATED

LEDGE/WEB INTERFACE RESULTANT FORCES

14.6k

(a) APPLIED LOAD

796k

7.23 k - FINITE ELEMENT MO

6.51k - LOAD TEST

7.96k7.22k

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

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

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

90

geometry of the finite element meshprevented investigation at the top of theledge.) The results of that study are pre-sented in Fig. 12. As expected, the in-side face of the web is in tension. Themaximum tensile stress of 295 psi,which occurs at the ledge load, is about40 percent greater than the averagestress. The compression in the outsideface of the web is significantly moreuniform.

The resultant of these stresses can becomputed by integrating stresses in theindividual elements near the ledge/webjunction. As indicated in the figure, theresultant is slightly less than the appliedledge load and is shifted significantlytowards the web centerline. These dif-ferences are equilibrated by shear andtorsion in the ledge itself. This mecha-nism is described in more detail in Sec-tion 6.

5. LOAD TESTSTwo L-beams and one pocket span-

drel were tested to study their behaviorand verify their strength. The tests wereconducted in the structural laboratory ofWiss, Janney, Elstner Associates inNorthbrook, Illinois.

Test SpecimensGeneral — All three spandrels were

72 in. high, 8 in. wide and 28 ft long.The target design loads were based on90 psf dead load and 50 psf live load,which are typical for a double tee park-ing structure with 60 ft spans. The reac-tions at each stem of an 8 ft wide doubletee were 16.8 kips.

Design — The design of the test spec-imens was based on the state-of-the-artmethods described in the backgroundsection, Shear and torsion design for theprestressed spandrels followed the pro-cedure recommended by Zia and Hsu.ACI Eq.(11-10) (rather than Eq.(11-11)or (11-13)1 was used to compute thebasic shear strength provided by theconcrete section. Flexural design fol-lowed ACI 318-83. Some reserveflexural strength was required to meetthe provisions of Section 18.8.3, whichrequires a bending capacity equal to atleast 1.2 times the cracking moment.Reinforcement for torsional equilibriumwas checked by Eq,(1). This reinforce-ment was not added to the reinforce-

ment for internal torsion.In view of the controversy regarding

ledge-to-web attachment, alternate pro-cedures were used for design of hangerreinforcement:

• Hanger reinforcement for Speci-men 1 was designed by Eq.(2), with aone-sixth reduction in the load sus-pended from the web based on relativeledge depth. All of the transverse rein-forcement between ledge loads wasconsidered to be effective, and hangerreinforcement was not added to shearand torsion reinforcement.

• Eq.(3) was used for design of thehanger reintorceme nt in Specimen 2. A7.4 percent reduction in the suspendedload was taken based on an assumedparabolic shear stress distribution.Again, all the hanger reinforcementbetween ledge loads was consideredeffective, and it was not added toshear/torsion reinforcement.

Hanger reinforcement for the pocketspandrel (Specimen 3) was designed byEq.(2). In addition to a U-bar at the poc-ket, one wire on each side of the pocketfrom the mesh reinforcing was consid-ered to contribute.

Design of the dapped-end connectionfor the pocket spandrel basically fol-lowed the PCI Design Handbook pro-cedure with two exceptions. First, dueto relatively low stresses, there was nospecial reinforcement provided for di-

PCI JOURNAUSeptember-October 1986 91

s

8-^4 d' tee'6' 1a «a J ^1r ^rru'sowii'-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 STRANDSSQS3 RELIEVE.26.9'

SECTION

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

agonal tension in the extended end ordirect shear at the junction of the dapand the main body of the member. Thewelded wire shear reinforcement, how-ever, was continued into the extendedend. Second, the reinforcement for flex-ure and axial tension in the extendedend was not continued past the potentialdiagonal tension crack extending to thebottom corner of the beam.

Details — The dimensions and rein-forcement details of the test specimens

are provided in Figs. 13 and 14. Thefollowing features of the reinforcingdetails should be noted:

0 Due to the different designmethods, Specimen 2 has twice as muchhanger reinforcement across theledge-web interface. This reinforcementwas provided by partial height L-bars onthe inside face of the spandrel betweenthe stimips. These bars add about 4 per-cent to the weight of the mild steel inthe beam.

92

12.6-Wt4, W2.54, 4-W 2.9, W2.9 43'-6 .3-6')INSIDE FACE t

4.4 -W29. W2.9INSIDE FACE

L aX44' o

s:

`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' CHAMFERAT POCKET

SECTION

Fig. 14. Dimensions and details of Specimen 3.

• Closed ties formed in one piece byoverlapping 90 degree end hooks areused on the left half of the L-beams,Stirrups on the right side of the L-beamsconsist of lapped-spliced U-bars.

• The longitudinal bars in the L-beams are not hooked at the ends.

• At the right side of the L-beams,two #5 bars are welded to a bearingplate. A #5 U-bar is used on the left sideof the L-beams.

• Wire mesh is used for shear rein-

forcement of the pocket spandrel. Themesh is not hooked around the mainreinforcement at the top and bottom ofthe beam, although the ACI Code re-quirements for development of web re-inforcement (Section 12.132.5) aresatisfied.

Materials — Design of the test speci-mens was based on 5000 psi concrete, 60ksi reinforcing bars (ASTM A706), 270ksi stress-relieved strand, and ASTMA497 mesh. Concrete cylinders and re-

PCI JOUFiNAUSeptember-October 1986 93

Table 1. Material strengths.

Concrete Reinforcing steel

Compressivestrength Bar Yield strength Tensile strength

Specimen fr (Psiyo, size f„ (ksi) fy (ksi)

1 5330 #3 78.9 98.72 5640 #4 70.4 103.73 6060 #6 64.2 98.1

(a) Average ofthree field-cured cylinders tested concurrently with load test (psi).

inforcing bar samples were tested todetermine actual strengths. The resultsare presented in Table 1. The yieldstrength of the #3 bars was much higherthan expected.

Test ProcedureSetup — The test setup is shown in

Fig. 15. The spandrels were supportedon rigid L-shaped frames which pro-vided lateral restraint at the four cornersof the beam. Load was applied at sevenpoints along the beam using speciallydesigned double tees (and one singletee). The test setup featured a remov-able connection between the spandrelsand double tees.

Instrumentation — Instrumentationincluded load cells at two of the loadingpoints on the double tees, as well as allfour horizontal reaction points. Threedeflection transducers and one tiltmeterwere set up at midspan to monitor hori-zontal and vertical deflections and rota-tions. Finally, single element straingauges were placed on selected rein-forcing bars as per Table 3.

Load sequence — Initially, each span-drel was incrementally loaded to serviceload (16.8 kips per tee stem) without theconnection between the double tees andspandrels. After unloading, this se-quence was repeated with the deckconnections in place. Finally, the beamswere loaded to failure without the deckconnections in increments of 2.5 kipsper tee stem. The third specimen was

tested to failure in two phases. After afailure near the end region in Phase 1,the supports were moved in 4 ft fromeach end, and the specimen was re-Ioaded to failure.

Behavior and Strength of TestSpecimens

Deflection and Rotation — Figs. 9 and10 compare the measured deflections ofthe L-beam and pocket spandrel to thosepredicted by the finite element models.Although the measured deflections arequite small, they are two to three timesthe predicted deflections. About half ofthe vertical deflection and some of therotation may be attributed to deforma-tion of the bearing pads.

Fig. 16 shows a plot of stem reactionvs. midspan torsional rotation of Speci-men 2. The stiffness of the beam is sig-nificantly reduced after cracking wasobserved.

Service Load Behavior — At serviceload, no cracks were observed in theL-beams. However, minor cracks wereobserved near the dapped-end connec-tion of the pocket spandrel. Thesecracks, which are shown in Fig. 17a,were all less than 10 mils (0.010 in.) inwidth.

Failure Patterns (Specimen 1) — Thecracking patterns that occurred duringloading to failure are shown in Fig. 18a.Diagonal cracks began to appear onSpecimen 1 at a load of 25 kips per stem.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

Table 2. Spandrel design and test results.

Calculated strengthDesign force -

Design° Predicted'Specimen TestService UltimateFailure mechanism Units No." x Nominal) 0 = 1 forced

Midspan flexure in.-kips 1 5,490 8,190 0.90 11,900T 13,730 10,5202 5,490 8,190 0.90 11,900, 13,730 12,8003-1 5,410 8,080 0.90 9,400' 10,440 8,150

Shear at support kips 1 68.0 101.4 0.85 111.1 145.2 130.3Torsion at support in.-kips 470 709 793 1033 967

2 68.0 101.4 0.85 111.1 146.8 158.6470 709 793 1033 1196

3-1 66.9 100.0 0.85 124.7 159.0 100.9118 177 a 186

3-2 66.9 100.0 0.85 124.7 159.0 124.7118 117 238

Lateral bending due to in,-kips 1 470 709 0.90 692 902 967torsion equilibrium force 2 470 709 0.90 692 902 1196

3-1 118 177 0.90 246 273 186Hanger reinforcement kips 3-1 66.9 100.0 0.9(1 95.0 113.0 100.9at 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

rn

Table 2 (cont.). Spandrel design and test results.

Calculated strengthDesign force

Design° PredictedeSpecimen TestFailure 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 34.6stem 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 47.6Ledge punching shear kips per 1 16.8 25.3 0.85 61.7 74.9 34.6at interiorbearingm stem 2 16.8 25.3 0.85 61.7 77.1 42.7Ledge punching shear kips per 1 16.8 25.3 0.85 53.7 65.2 34.6at 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 byCalculated 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 bmoment.Hanger reinforcement designed by Eq. (2) with a one-sixth reduction in the load suspended from the web based on iHanger reinforcement designed by Eq. (3) with a 7.4 percent reduction in the load suspended from the web based ofHanger 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.

GageDistancefrom load

Service load Factored load Max test load

Load Strain Load Strain Load StrainLocation No.. (in.) ' percent b percent ° percent

Ledge hanger 1-1 0 16.9 0.004 27.3 0.239 34.6reinforcement 1-2 12 16.9 0.001 27.4 0.120 35.6 3.211(near 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 34.61-5 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) 015reinforcement 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 0.4122-5 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 0.016reinforcement 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 0.414at 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-ately opened to 20 mils and extendedend to end where it connected to in-clined cracks in the ledge. The ledgecontinued to separate from the web untilthe test was stopped at a ledge load of34.6 kips per stem. At the end of the test,the crack at the ledgetweb junction wasover Vs in, wide, as shown in Fig. 19.

Failure Patterns (Specimen 2) - Asshown in Fig. 18b, a well developedpattern of inclined and "rainbow"cracking developed on the inside face ofSpecimen 2. Typically, these crackswere less than 10 mils wide. Also, sev-eral 1 to 3 mil flexural cracks were ob-

served on the outside face. The crack atthe ledge/web junction was restrainedby the additional hanger reinforcement,as shown in Fig. 20. At a load of 42.7kips per tee stem, punching shear fail-ures occurred at the first and sixth teestem from the left. Fig. 21 shows thepunching shear failures, The failurecone initiates behind the hearing pad.The failure surface is almost verticalnear the top and inclined below theledge reinforcing. As a result, the ledgeflexural reinforcement is not very welldeveloped across the failure plane.

Failure Patterns (Specimen 3) - Thecracks which formed during Phase 1 of

98

40

a_Yp 30

n

NQ zais

10

ROTATION (RADIANS)

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

the Specimen 3 test are shown in Fig.17b. Cracks near the dapped-end con-nection which developed at service loadcontinued to lengthen and widen, andnew inclined cracks formed. Cracksbelow the pockets began to form at teestem loads of 18 to 25 kips. As the loadwas increased, diagonal tension cracksdeveloped further from the support.These cracks typically initiated nearmidheight of the beam. At a load of 26.5kips per tee stem, a diagonal tensioncrack near the right support extendeddown to the bottom corner of the beamand failure occurred immediately, asshown in Fig. 22.

In Phase 2 of the Specimen 3 test, awide "rainbow" crack formed at a loadof about 43 kips per tee stem. Appar-ently this crack is due to a combinationof diagonal tension due to shear andvertical tension due to the tee stemloads. The ultimate failure, however,occurred when the concrete below thefifth pocket from the left punched out at

47.6 kips. The "rainbow" crack andpunching failure are shown in Fig. 17c.

Strength — Table 2 summarizes thedesign force, calculated strength andtest force for several potential and actualfailure mechanisms. The calculatedstrengths are based on the equationsused for design. Because the hanger re-inforcement for Specimens I and 2 wasdesigned using different equations, thecalculated strength is roughly the sameeven though Specimen 2 had twice asmuch hanger reinforcement.

The calculated strength is expressedas both a "design" strength and a "pre-dicted" strength, The design strength isbased on specified material properties,and includes the appropriate strengthreduction factor. The predicted strengthuses actual material properties and nostrength reduction factor.

As shown in Table 2, the spandrelbeams were tested to a load near or be-yond their predicted capacity for severalof the primary failure mechanisms.

PCI JOURNALSeptember- October 1986 99