Two full-scale beam-to-column connections in aprecast concrete frame were tested under uni-directional and bi-directional cyclic loadingthat simulated earthquake-type motions. Variablesincluded the detailing used at the joint to achievestructural continuity of the beam reinforcement,and the type of framing (whether two-dimensionalor three-dimensional). The most relevant feature ofthe connection is that conventional mild steelreinforcing bars or prestressing strands, rather thanwelding or special bolts, were used to achievebeam continuity. Specimen design followed thestrong-column–weak-beam concept. Beamreinforcement was purposely designed anddetailed to develop hinges at the joint faces and toimpose large inelastic shear force demands intothe joint. During specimen fabrication, the jointdetails enabled ease and speed of construction. Asexpected, the joint controlled the specimenfailure. In general, the performance of both beam-to-column connections was satisfactory. Jointstrength was 80 percent of that expected formonolithic reinforced concrete construction.Specimen behavior was ductile due to hoopyielding and bar pullout, while strength was nearlyconstant up to drifts of 3.5 percent.

Precast concrete has been widely accepted as a viablemeans of constructing safe, durable, reliable, high-quality, and cost-effective structural systems. Its fullimplementation in high seismic areas, however, has beensomewhat limited, mainly due to scarce design guidelinesas compared to those available for cast-in-place concretestructures. In particular, the lack of design provisions for

70 PCI JOURNAL

Sergio M. Alcocer, Ph.D.ProfessorInstitute of EngineeringUniversity of Mexico (UNAM)and Director of ResearchNational Center for Disaster Prevention(CENAPRED)Mexico City, Mexico

Rene Carranza, S.E.Director General

Servicios y Elementos Presforzados,S.A. de C.V. (SEPSA)Mexico City, Mexico

David Perez-Navarrete, M.Sc.Former Graduate StudentSchool of EngineeringUniversity of Mexico (UNAM)Mexico City, Mexico

Raul Martinez, S.E.Project Manager

Servicios y Elementos Presforzados,S.A. de C.V. (SEPSA)Mexico City, Mexico

Seismic Tests of Beam-to-Column Connections in a Precast Concrete Frame

seismic-resistant beam-to-column con-nections appropriate for precast con-crete frame construction is still appar-ent in North America (Mexico, UnitedStates, and Canada).1

Another obstacle to the use of pre-cast concrete construction in seismicareas has been the imposition of pre-scriptive provisions developed to pro-mote ductility in cast-in-place con-crete construction.2 This set of designand detailing requirements makes astandard methodology for establishingequivalence of energy dissipation andductility between precast concrete andcast-in-place systems desirable. In1999, such a standard methodologywas developed by an Innovation TaskGroup (ITG) of the American Con-crete Institute, which was published.3

Two design approaches are avail-able for the design of precast concretelateral force-resisting systems.4 Emu-lation of monolithic reinforced con-crete construction is the approachmost commonly adopted in codes.5-7

The alternative approach is the use ofthe unique properties of the precastconcrete elements interconnected byeither dry or wet connections.8

In the past, some precast concreteframed structures have performedpoorly in earthquakes because of inad-equate connection details.5 To gainconfidence in the use of precast con-crete in moment-resisting frames, sat-isfactory methods for connecting theprecast elements together were re-quired. Therefore, recommendationsdeveloped for the seismic design ofcast-in-place systems were straightfor-wardly adapted – that is, the objectiveof the design method was to emulatemonolithic construction.

The Mexico City Building Code(MCBC)6 permits the use of precastconcrete structural systems based onemulation of the behavior of mono-lithic cast-in-place concrete systems.Specific requirements on materialproperties, construction details, anddesign are given in the code.

The MCBC requires that the designcompressive strength of the concreteused in the connection be equal to orgreater than that of the adjoining ele-ments. Steel reinforcement in the con-nection, either longitudinal or trans-verse, must have a nominal yield

May-June 2002 71

strength limited to 414 MPa (60 ksi).Typical Mexican reinforcing steel fol-lows standards similar to ASTM A 615.

Connections have to be designedand detailed to resist 1.3 times theforces and moments obtained fromanalysis that act at their boundaries.The 1.3 factor is in addition to theusual load factors. Finally, all surfacesof precast concrete elements in contactwith the connections must be free oflaitance, intentionally roughened to afull amplitude of 5 mm (0.2 in.), andsaturated with water 24 hours prior to

casting the concrete in the connection.As is commonly done in other coun-

tries, the seismic-induced designforces recommended in the MCBC aresignificantly less than the inertiaforces induced if the structure were torespond in the elastic range to a majorearthquake. The design seismic-induced force is related to the achiev-able structural ductility through seis-mic response modification factors Q,which are equivalent to the force-reduction factors R in the UniformBuilding Code.7 Therefore, the MCBC

Fig. 1. Type of precast concrete frame construction studied.

Fig. 2. Specimen geometry and testing rig.

5120 mm5120 mm

4000

mm

4000

mm

72 PCI JOURNAL

design criterion is force-based.For ductile moment-resisting cast-

in-place frames, a Q value of 3 or 4 isused to determine the appropriate de-sign load from elastic design spectra.However, for both values of the seis-

mic response modification factors, thesame detailing requirements apply.The choice of using a Q of 3 or 4 isleft to the designer. The ultimate de-sign horizontal seismic forces typi-cally vary between 0.04g and 0.2g, de-

pending on the seismic zone, the soilcategory, the importance of the struc-ture, and the fundamental period of vi-bration of the structure. For precastconcrete frames, a Q of 2 or 3 is usedand the ultimate design horizontal

Fig. 3. Dimensions and detailing of Specimen J1.

Section A-A

Section B-B

May-June 2002 73

seismic-induced forces vary between0.04g and 0.3g.

As has been recognized,5 the chal-lenge in precast frame constructionlies in finding economical and practi-cal methods of connecting the precast

concrete elements together to ensureadequate stiffness, strength, ductilityand stability. A number of experimen-tal research programs conducted in re-cent years have significantly improvedour understanding of the behavior of

connections between precast concreteelements.9-18 For example, internal re-sisting mechanisms have been identi-fied and understood,10,14-16,18 and de-sign and detailing requirements havebeen developed.12,14,17,18

Fig. 4. Dimensions and detailing of Specimen J2.

74 PCI JOURNAL

In some research programs, precastelements have been connected at thebeam-to-column joint region;9-16,18 inothers, elements have been connectedat midspan (in the case of beams) and at midheight (in the case ofcolumns).15,16 Since one advantage ofusing one-way perimeter frames is thereduced complexity of the design andconstruction of the beam-to-columnjoints, many studies have focused on

the performance of perimeter precastconcrete frame joints.9-17

In emulation systems, precast beamlongitudinal reinforcement that is con-nected at the beam-to-column joint iscommonly spliced. Splicing can beachieved through proprietary steelsleeves11,15,16 or by lap splicing bars,which may be bent to form 90-degreestandard hooks.16,17 In other cases, con-tinuity at the joint has been achieved

Element

Precast columns and beamsCast-in-place joint

12.7 mm (0.5 in.) strand

No. 3 (9.53 mm)fyfu

Elongation at fractureNo. 4 (12.7 mm)

fyfu

Elongation at fractureNo. 5 (15.88 mm)

fyfu

Elongation at fractureNo. 8 (25.4 mm)

fyfu

Elongation at fractureNo. 10 (31.75 mm)

fyfu

Elongation at fractureNo. 12 (38.1 mm)

fyfu

Elongation at fracture

No. 4 (12.7 mm)fyfu

Elongation at fractureNo. 5 (15.88 mm)

fyfu

Elongation at fractureNo. 6 (19.1 mm)

fyfu

Elongation at fractureNo. 8 (25.4 mm)

fyfu

Elongation at fractureNo. 10 (31.75 mm)

fyfu

Elongation at fracture

Specimen J1Concrete, f ′c:

32 MPa (4650 psi)44 MPa (6370 psi)

Prestressing steel, nominal, fy:N/A

411 MPa (59,660 psi)716 MPa (103,830 psi)

10.3 percent

437 MPa (63,440 psi)700 MPa (101,550 psi)

9.6 percent

434 MPa (63,010 psi)691 MPa (100,270 psi)

11.6 percent

455 MPa (65,990 psi)719 MPa (104,250 psi)

13.7 percent

441 MPa (64,000 psi)726 MPa (105,250 psi)

15.5 percent

N/AN/AN/A

437 MPa (63,430 psi)700 MPa (101,550 psi)

9.6 percent

434 MPa (63,010 psi)691 MPa (100,270 psi)

11.6 percent

N/AN/AN/A

460 MPa (66,700 psi)729 MPa (105,680 psi)

13.0 percent

448 MPa (65,000 psi)731 MPa (105,960 psi)

12.7 percent

Specimen J2

42 MPa (6040 psi)40 MPa (5840 psi)

1872 MPa (271,520 psi)

456 MPa (66,140 psi)713 MPa (103,400 psi)

8.0 percent

424 MPa (61,510 psi)684 MPa (99,130 psi)

9.2 percent

413 MPa (59,880 psi)678 MPa (98,280 psi)

10.4 percent

454 MPa (65,850 psi)698 MPa (101,270 psi)

12.3 percent

N/AN/AN/A

431 MPa (62,440 psi)700 MPa (101,550 psi)

18.3 percent

424 MPa (61,510 psi)684 MPa (99,130 psi)

9.2 percent

N/AN/AN/A

420 MPa (60,870 psi)668 MPa (96,860 psi)

12.8 percent

458 MPa (66,420 psi)698 MPa (101,270 psi)

13.9 percent

476 MPa (68,980 psi)721 MPa (104,540 psi)

12.8 percent

Table 1. Material properties.

Reinforcing steel in precast elements:

Reinforcing steel in cast-in-place concrete:

Note: 1 in. = 25.4 mm; 1 psi = 0.006895 MPa.

by means of bonded post-tensioning,9

ungrouted post-tensioning,10,14 or mildreinforcing steel bars grouted insidethe joint but with an unbonded lengthadjacent to the column face.13

This paper reports on the behaviorof concrete frames under cyclic load-ing of a precast concrete beam-to-col-umn connection developed and used inMexico. In the frame under considera-tion, design and detailing were aimedat emulating monolithic construction,as is required in the MCBC.6 This in-vestigation was conducted to assessthe joint behavior and its stiffness, de-formation, and strength characteristicswhen subjected to large shear andbond force demands. It was consid-ered that such information would beuseful in assessing the validity of theemulation hypothesis.

The beam-to-column connectiondiscussed herein has already been usedin Mexico for low-rise commercialbuildings up to 25 m (82 ft) in height.The system consists of multistory pre-cast concrete columns that have opengaps at each floor level (see Fig. 1).Typically, columns are fabricated inone piece and extend over a number ofstories.

To form the gaps, concrete place-ment is interrupted at each floor level.These gaps allow placement of theprecast beams during erection. To im-prove horizontal shear transfer at thejoint, a 150 mm (5.9 in.) deep squarehole is left in the concrete in the col-umn below the joint. During place-ment of the joint concrete, concretefills the hole so that it works as a shearkey. Because the shear key is left un-reinforced, its strength is limited to theshear strength of the concrete.

To promote the escape of the en-trapped air in the joint concrete duringcasting and compacting, an invertedpyramid shape is formed in the precastcolumn above the joint. In addition, a12.7 mm (0.5 in.) diameter pipe thatruns from the pyramid vertex up toone side of the precast column is leftembedded.

Column longitudinal bars are madecontinuous through the gaps and areplaced near the corners to leave amplefree space for beam erection andplacement. The column height is lim-ited by the hauling truck’s load capac-

May-June 2002 75

ity and dimensions, as well as by high-way and urban traffic regulations. Ifneeded, the precast column units canbe spliced at a floor midheight.

Single-bay precast beams, eitherprestressed or conventionally rein-forced, are T- and inverted-T-beams.Beams are placed between columnsand seated on the cover concrete andpart of the core of the lower part of theprecast concrete column. The bottomsteel reinforcement protrudes from thebeam ends and extends into the joint.This reinforcement may be bent toform either standard 90-degree hooksor closed loops.

A precast concrete hollow-core one-way floor system is placed on top ofthe inverted-T precast beam elementsand spans between them, parallel tothe T-beams. The reinforcement isthen placed on top of the beam, in thetopping slab over the floor system, andin the beam-to-column joint core. Fi-nally, the concrete for the joint regionand the topping slab is placed.

TEST PROGRAMTwo full-scale beam-to-column con-

nections made with precast beams andcolumns were fabricated and tested.The specimens represented an interiorjoint of a lower story of a multistorybuilding. The test units modeled theregion from mid-column height belowthe joint to mid-column height abovethe joint and from midspan to midspanof beam on either side of the joint.

Specimen geometries are shown inFig. 2; dimensions correspond to thedistance between hinged supports. Di-mensions and details of the test unitsare presented in Figs. 3 and 4. Testunits consisted of beams 5.56 m (18.2ft) long framing into a column 4.00 m(13.1 ft) high at midheight.

Experimental variables studied werethe type of framing (two- and three-di-mensional) and the joint detailing forcontinuity of beam reinforcement.19,20

The intent of connecting the bottombars was to transmit tensile forces tothe far side of the joint so that a diago-nal compression strut could formalong the diagonal of the joint panelzone. Design concrete cover for all el-ements was 25 mm (1 in.) for bothspecimens.

Specimen J1 consisted of twobeams framing into the joint on oppo-site sides (2D construction); the bot-tom longitudinal steel reinforcementof the beams was terminated with 90-degree hooks at the joint. Continuityof this reinforcement through the jointwas achieved with hoops placedaround the extensions of the 90-degreehooks that protruded from the beamends. Hoops had a 90-degree bendwith an extension of six bar diameters.

The column of Specimen J1 wassquare with 500 mm (19.7 in.) sidesand was reinforced with eight No. 10(31.75 mm diameter) Grade 60 (fy =414 MPa or 60 ksi) continuous longitu-dinal bars and No. 4 (12.7 mm diame-ter) hoops at 100 mm (3.9 in.) spacing.The height of the gap in the column forbeam placement and casting of thejoint was 1000 mm (39.4 in.).

Specimen J1 had beams with a final500 mm (19.7 in.) square section. Thepartially precast beams had an in-verted-T shape and were reinforcedwith two No. 8 (25.4 mm diameter)bottom longitudinal bars and No. 3(9.53 mm diameter) stirrups spaced at100 mm (3.9 in.) on center. Bottombars protruded from the beam endsinto the joint with 90-degree hooks.

Additional No. 4 and No. 3 Grade60 longitudinal bars were placed in thecolumns and beams to ease the fabri-cation of the cages. Since the beam

width in the east-west direction wasthe same as the column width, a 250mm (9.8 in.) reduction was made atthe joint in order to fit the beam be-tween the column longitudinal bun-dled bars (see Fig. 3).

Once the column and the beamswere mounted on the test rig (see Fig.2), the continuity reinforcement of thebeam bottom longitudinal bars wasplaced. For Specimen J1, this rein-forcement consisted of four No. 5(15.88 mm diameter) Grade 60 hoopsplaced around the extensions of the90-degree hooks.

Hoops were proportioned so thatthey would remain elastic while trans-ferring the tensile axial force devel-oped along the beam bottom rein-forcement at ultimate bendingmoment. Since the nominal yieldstrengths of the hoop and longitudinalbars were the same, the total area ofhoop legs in the direction of loadingwas 1.5 times the total area of beambottom longitudinal bars.

To improve the confinement,strength, and deformability of the jointconcrete, No. 4 cross-ties were placedaround the joint through holes leftacross the beam width during precastfabrication. Several horizontal steelpipes were left embedded in thebeams, at the design joint hoop spac-ing, to allow placement of joint hoopsor cross-ties. Hoops were anchored

Specimen J1 Specimen J2Beam flexure

Beam flexural strength Mp+ 281 kN-m (2480 kip-in.) 276 kN m (2440 kip-in.)Beam flexural strength Mp- 540 kN-m (4780 kip-in.) 579 kN m (5120 kip-in.)

Beam shearShear corresponding to Mp 211 kN (47 kips) 226 kN (51 kips)

Vn (MCBC, beam) 431 kN (97 kips) 484 kN (109 kips)Column flexure

Axial force P 0 0Moment corresponding to Mp 411 kN-m (3640 kip-in.) 428 kN-m (3780 kip-in.)

Mn (moment-curvature analysis) 579 kN-m (5120 kip-in.) 785 kN-m (6940 kip-in.)Column shear

Shear corresponding to Mp 235 kN (53 kips) 244 kN (55 kips)Vn (MCBC, column) 633 kN (142 kips) 636 kN (143 kips)

Joint shearHorizontal shear force 1636 kN (368 kips) 1777 kN (399 kips)

Shear stress 6.5 MPa (949 psi) 7.1 MPa (1031 psi)Shear strength Vn,j, γ =15 2530 kN (569 kips) 2442 kN (549 kips)

(ACI-ASCE Committee 352)Shear strength Vn,j , γ =12 2070 kN (465 kips) 1998 kN (449 kips)

(ACI-ASCE Committee 352)Maximum interstory drift applied, percent 3.5 percent 3.5 percent

Table 2. Member forces and strengths.

Note: 1 psi = 0.006895 MPa; 1 kip = 4.448 kN; 1 kip-in. = 0.113 kN-m.

76 PCI JOURNAL

around column longitudinal bars withalternating 90-degree and 135-degreehooks.

Afterwards, continuous beam topreinforcement was placed through thejoint and the U-shaped beam stirrups,left anchored during precasting, werebent around. Beam top reinforcementconsisted of two No. 8 and two No. 10continuous bars (see Fig. 3). Finally,40 MPa (5800 psi) ready-mixed con-crete was placed in the joint and ontop of the precast beams.

In Mexico, all frames of a buildingare typically designed to resist theseismic-induced forces. Therefore,frames are commonly designed anddetailed as special moment-resisting

frames (SMRFs). To examine the be-havior of an interior joint of such astructural system layout, Specimen J2was tested.

In Specimen J2, the bottom longitu-dinal reinforcement of the beam wasinterrupted at the joint face and lap-spliced with a U-shaped prestressingstrand that extended from the beaminto the joint. Continuity wasachieved with a steel bolt insertedthrough the overlapping U-shapedprestressing strands at the joint mid-depth (see Fig. 4).

The Specimen J2 column had thesame dimensions and transverse rein-forcement as Specimen J1; however,continuous longitudinal reinforcement

was made of eight No. 12 (38.1 mmdiameter) Grade 60 bars. This bar sizeis readily available in Mexico.

In the east-west direction, beamswere similar to those of Specimen J1.In the north-south direction, T-beamswith a 200 x 500 mm (7.9 x 19.7 in.)cross section were used. These beamswere reinforced similarly to east-westbeams. In both the east-west andnorth-south beams, bottom longitudi-nal bars were interrupted at the joint(see Fig. 4).

To provide continuity to the beambottom reinforcement, looped (U-shaped) prestressing strands were lapspliced 125 mm (4.9 in.) with the de-formed longitudinal bars. A No. 12Grade 60 bar was inserted verticallythrough the intersection of the looped(U-shaped) low-relaxation Grade 270(fy = 1860 MPa) prestressing strandsthat protruded from the beams; the barwas anchored inside the square col-umn hole in the precast column belowthe joint.

To improve bar anchorage insidethe joint, a square steel plate was butt-welded in the upper part; further anal-ysis of strain gauge data indicated thatthe plate was not necessary.20 Pre-stressing strands were designed toyield when conventional beam bottombars yielded, i.e., when beam plastichinges formed. The vertical steel barwas designed to remain elastic underyield demands from the strands at ul-timate.

Similarly to Specimen J1, transversereinforcement was provided to confinethe joint concrete. No. 4 cross-tieswere anchored around column longitu-dinal bars with 90-degree and 135-de-gree hooks, alternated over the jointheight (see Fig. 4).

The beam top reinforcement in theeast-west direction of Specimen J2was the same as in Specimen J1. In thenorth-south direction of Specimen J2,two No. 8 and two No. 6 (19.1 mm di-ameter) bars were used.

To induce forces in the joint at lev-els near the shear strength, large beammoments had to be developed. For thisto happen, it was necessary to use alarge amount of beam steel areathrough the joint region. Because ofspace limitations, it was decided to uselarger but fewer bars (two No. 8 and

Fig. 5a. Displacement-controlled test sequence for Specimen J1.

Fig. 5b. Displacement-controlled test sequence for Specimen J2.

May-June 2002 77

two No. 10 bars), being cognizant that,especially for No. 10 bars, slippageand bond failure was more likely tooccur. Moreover, the No. 10 bars didnot comply with the minimum ratio ofcolumn depth to beam bar diameter.Analysis of strain gauge data indicatedthat bar slippage did not occur in ei-ther specimen.19,20

Specimens were built without thetopped hollow-core slabs generallyused in the system. It was assumedthat since the hollow-core slab actedas a one-way floor system in thenorth-south direction of the speci-mens, its contribution to joint capacitywas not significant.

Table 1 lists the unconfined com-pressive strength of the concrete foundfrom cylinder tests at the time of test-ing the beam-to-column joint units,the yield and ultimate tensile strength,and elongation at fracture of variousreinforcement sizes used. Concretecylinders were cured in the same envi-ronment as specimens. Calculatedmember capacities are provided inTable 2, based on the measured mate-rial strengths listed in Table 1 and onmeasured bar locations. The memberstrengths do not include strength re-duction factors.

The flexural strength Mp of thebeam plastic hinges was based on theassumption that concrete would followHognestad’s model. Column flexuralstrength, based on a moment-curvatureanalysis considering the effects ofconfinement and a maximum com-pressive strain of 0.0038, exceeded thecolumn moment corresponding to de-velopment of Mp in the beams. Col-umn flexural strengths exceeded theflexural demands by 40 and 83 percentin Specimens J1 and J2, respectively.Column shear forces corresponding tothe development of Mp in the beamwere 235 and 244 kN (52.8 and 54.9kips) for Specimens J1 and J2, respec-tively.

The calculated ratio of column flex-ural strength to beam flexural strength,based on measured material propertiesand dimensions and considering anequivalent rectangular stress block forthe concrete as well as assuming thatplane sections remain plane, was 1.53and 2.08 for Specimens J1 and J2, re-spectively.

The joint shear forces and strengthsare listed in Table 2. The horizontalshear forces were calculated assumingyielding of the beam top and bottomreinforcement and considering the col-

umn shear force. The joint shear stresswas obtained by dividing the jointshear force by the column area. Theshear stress at development of Mp inthe beam hinges was 6.5 MPa (949

Fig. 6a. Crack pattern of Specimen J1 at design drift of 2 percent.

Fig. 6b. Crack pattern of Specimen J1 at maximum drift of 3.5 percent.

78 PCI JOURNAL

psi), which corresponds to 0.99MPa (11.9 psi).

The joint shear strength was basedon the shear strength recommended inACI 352R,22 which is the basis of therequirements in the MCBC. Since thejoint was not effectively confined onall four vertical sides, a constant γequal to 1 in MPa units (or 12 in psiunits) was used to determine the stressat the shear strength, γ .

Such recommendations are, in ef-fect, applicable to monolithic construc-tion, but were applied to these casesfor comparison purposes. Therefore, itwas expected that a diagonal strut

′fc

′fc′fc

Fig. 7a. Joint damage of Specimen J1 at design drift of 2 percent.

Fig. 7b. Joint damage of Specimen J1 at maximum drift of 3.5 percent.

requirements prescribed in the MCBCwere intentionally not fulfilled in thespecimens.

1. Prestressing strands with a nomi-nal yield stress greater than 414 MPa(60 ksi) provided continuity at thejoint of Specimen J2.

2. Nominal joint shear strength ca-pacities and joint shear demands weremade approximately equal in both testunits. As is preferred in earthquake-resistant design, specimens were de-signed and detailed to achieve ductilepost-elastic deformations by flexuralyielding in beam plastic hinges at theface of the column. However, highshear demands in the joint were pur-posely introduced in order to assessthe beam-to-column joint shear capac-ity. Thus, while the longitudinal steelarrangement of columns was designedto reduce the likelihood of yielding,beam reinforcement and detailingwere designed to ensure plastic hing-ing and to impose large shear force de-mands into the joint.

3. The 1.3 factor, specified by theMCBC to augment the moments andforces obtained from analyses (in addi-tion to the usual load factors), was notincluded in the design of the test units.This was done because the specimensdid not represent any prototype struc-ture that would have been designed inaccordance with the MCBC. In con-trast, it was the one objective of theprogram to understand the joint load-carrying mechanisms and to assess thejoint capacity. It should be emphasizedthat in all structures designed in Mex-ico City with the connections describedherein, the 1.3 factor must be used.

TEST SETUP, LOADINGPROGRAM, AND

INSTRUMENTATIONThe experimental setup for Speci-

mens J1 and J2 is shown in Fig. 2.Column ends and beam ends werepinned. Horizontal displacementswere applied at column midheightabove the joint through a pin-endeddouble-acting actuator with the testunit rotating about a spherical bearingat the lower column midheight. Thebeams were connected at midspan po-sitions to the reaction floor throughpin-ended steel struts, allowing free

would resist the entire horizontal jointshear introduced. Based on the experi-mental results, mechanisms involvedin the joint shear transfer are discussedin more detail later in this paper.

Compliance with MCBC

Design and detailing complied withmost requirements for precast con-struction and monolithic ductileframes in the MCBC.6 The MCBC hasdesign and detailing requirements forSMRFs that are similar to those ofChapter 21 of ACI 318-9521 and of theUniform Building Code.7 Three design

May-June 2002 79

lateral translation of the beams but re-straining vertical movement and en-suring a moment pattern in the speci-men similar to that in a joint of abuilding frame under cyclic lateralloads. No axial load was applied to thecolumns since experimental evidencesuggests that axial load has no effecton the joint shear strength.23

Specimens were tested under a dis-placement-controlled cyclic load his-tory that was based on the interstorydrift which represented a severe loadcondition for a beam-to-column joint(see Fig. 5). The test pattern is basedon drift rather than ductility incrementsbecause ductility can be difficult to de-fine for systems incorporating compo-nents other than conventionally rein-forced concrete or mild reinforcingsteel components.14

The loading sequence originatedfrom recommendations made in a 1997draft of the ITG document.3 At eachdrift level, three cycles were applied.Bi-directional cycles up to 1.5, 2.5, and3.5 percent drift were applied to Speci-men J2, with the main loading direc-tion being east-west (beam with squarecross section). Specimens were sub-jected to much larger drifts than theirdesign drift capacities of 2 percent.

Specimens were instrumented withdisplacement, strain, and load trans-ducers.19,20 Displacement transducerswere attached to the beam and columnsurfaces at opposite faces to measurethe axial deformations and calculatemember rotations and curvatures. Aset of transducers was placed in thejoint region to obtain joint shear defor-mations. Resistive strain gauges werebonded to longitudinal and transversereinforcing bars of columns andbeams. A dense array of strain gaugeswas placed on the reinforcement of thejoint details tested, as well as on thejoint confining steel.

TEST RESULTSThe detailed test results, together

with a discussion of the importantpoints, are presented in this section.

Cracking Patterns

The crack patterns and photographsof joint damage at the design drift of

2.0 percent and at the end of the testare shown in Figs. 6 through 9 forSpecimens J1 and J2. In both struc-tures, most of the damage was con-centrated in the joint and in thebeams. Consistent with a strong-col-

umn–weak-beam system, the columndamage was minor.

For Specimen J1 (see Figs. 6 and 7),the beams exhibited few flexuralcracks near the column face. Spallingof cover concrete first developed at a

Fig. 8b. Crack pattern of Specimen J2 at maximum drift of 3.5 percent, east-west direction.

Fig. 8a. Crack pattern of Specimen J2 at design drift of 2 percent, east-west direction.

80 PCI JOURNAL

drift of 3.0 percent; spalling pro-gressed as drift increased. Under neg-ative bending moment (top fibers intension), a single vertical 10 mm (0.39in.) wide crack at the column face oc-curred. Under positive bending mo-ment, a 14 mm (0.55 in.) wide crackwas observed at the column face fol-lowing the contour of the precast con-crete beam.

During specimen demolition, in-clined cracking of the narrower por-tion of the beams inside the joint wasobserved. Such damage extendedfrom the interior of the joint, at thelower side of the column gap, up andout to the beams. This damage is at-tributed to pullout of the beam bottombars and beam rotation inside the jointthat was not concentrated at the col-umn face, as is commonly expected inmonolithic construction. Also, the ex-tent of beam damage at maximumdrift (3.5 percent) was less than wouldbe expected in monolithic reinforcedconcrete construction.

Column flexural cracking initiated ata drift of 0.5 percent and remainedminor for the remainder of the testing.First joint inclined cracking developedat a drift of 0.25 percent. Considerablejoint inclined cracking took placewhen cycling to drifts of 2.0 percent(see Figs. 6a and 7a). This extended upto a drift of 2.5 percent; the crack pat-tern stabilized and only a few cracksappeared at higher drifts. At maximumdrifts, joint cracks were well dis-tributed and oriented approximatelyparallel to the joint diagonal at a 45-degree angle (see Figs. 6b and 7b).

Specimen J2 showed cracking anddamage patterns similar to SpecimenJ1 (see Figs. 8 and 9). Beams showeda more uniform distribution of crack-ing, especially under positive bending(top fibers in compression). Thenorth-south beams exhibited crackingcomparable to that observed in mono-lithic construction (see Fig. 8c). Atsimilar drifts, beam crack widths inSpecimen J2 were smaller than inSpecimen J1.

At the design drift of 2.0 percent, nospalling was observed in the beamsand only light spalling was observed inthe joint region (see Figs. 8a and 9a).Joint inclined cracking first appearedat a drift of 0.25 percent. Cracking ex-

Fig. 8c. Crack pattern of Specimen J2 at maximum drift of 3.5 percent, north-south direction.

Fig. 9a. Joint damage of Specimen J2 at design drift of 2 percent, east-west direction.

Fig. 9b. Joint damage of Specimen J2 at maximum drift of 3.5 percent, east-west direction.

May-June 2002 81

Fig. 10a. Story shear versus interstory drift for Specimen J1.

Fig. 10b. Story shear versus interstory drift for Specimen J2, east-west direction.

82 PCI JOURNAL

tended until a drift of 0.5 percent, afterwhich it stabilized. At the same drifts,joint inclined cracks in Specimen J2were finer than in Specimen J1.Spalling of the joint cover concrete oc-curred at 3.0 percent drift. Similarly toSpecimen J1, under positive bending,the beam rotated inside the joint.

shears corresponding to joint failure,Vn,j, are also shown. The story shearscorresponding to joint failure werebased on the shear strength recom-mended by ACI 352R22 (see Table 2).

The story shears associated withthe formation of one or two plastichinges in the beams are indicated ashorizontal dashed lines. The occur-rences of the first joint diagonalcracking and yielding recorded atseveral locations during the tests arealso indicated in the figures. All ofthe recorded yielding took place atthe face of the column.

The experimental response indi-cated a hysteretic behavior at drifts to2.0 percent with comparatively lowstrength degradation among succes-sive cycles at equivalent drift levels.The hysteresis loops are nearly sym-metrical, even during bi-directionalloading of Specimen J2. Considerablepinching and severe stiffness degrada-tion is noticeable, especially at driftsof 3.5 percent. The hysteresis curvesof Specimens J1 and J2 are dominatedby the response of the most damagedelements, which were the joints andthe beams.19,20

Fig. 10c. Story shear versus interstory drift for Specimen J2, north-south direction.

Fig. 11. Equivalent viscous damping versus interstory drift, east-west direction.

Story Shear Versus Interstory Drift

The hysteresis curves for story shearversus interstory drift for SpecimensJ1 and J2 are presented in Fig 10. Thecomponents of the bi-directional cy-cles of Specimen J2 in the east-westand north-south directions are pre-sented in Figs. 10b and 10c. The story

May-June 2002 83

To assess the energy dissipation ca-pacity and the stability of the hys-teretic behavior, the equivalent vis-cous damping ratio, Heq, was plottedagainst the interstory drift during east-west loading for both specimens (seeFig. 11). In the plot, Heq was taken asthe ratio of the dissipated energywithin a cycle to 2π times the strainenergy measured at peaks of an equiv-alent linearly elastic system.

In general, damping ratios increasedwith drift. At equivalent drifts, damp-ing ratios decreased as much as 20percent. This reduction is attributed topinching in the curves; concrete dam-age and shear distress contributed tonarrowing the hysteresis loops. Damp-ing ratios for bi-directional cycleswere higher.

Since the story shears were limitedby the beam flexural strength and,more exactly, by pullout of the beambottom bars, the calculated storyshears corresponding to joint strength,Vn,j, in the east-west and north-southdirections for Specimens J1 and J2were never reached even at cycles to3.5 percent drift. Measured strengthsof Specimens J1 and J2 were about 80percent of those expected in a mono-lithic beam-to-column connection de-signed according to ACI 352R22 (seeFig. 10).

The calculated strength at firstyielding in the specimens is related toyielding of the beam bottom longitudi-nal bars. In Specimen J1, the systemreached the story shear associated withyielding not because the bottom barshad yielded, but because the connect-ing hoops had yielded. In SpecimenJ2, the bottom bars did not yield ei-ther, but because the top bar load pathwas much stiffer than the bottom one,the top bars reached strains within thestrain-hardening zone.19,20

Analysis of the strain gauge data in-dicated that beam top steel barsyielded prior to achieving the speci-men strength and before damage con-centrated in the joint.19,20 In SpecimenJ2, the maximum recorded steelstrains were greater than 0.014. Bot-tom longitudinal steel bars of thebeams remained elastic. Strainsrecorded on the joint continuity rein-forcement are discussed later in thepaper.

Fig. 12. Secant stiffness versus interstory drift angle.

Fig. 13a. Member contribution to interstory drift for Specimen J1, east-west direction.

Fig. 13b. Member contribution to interstory drift for Specimen J2, east-west direction.

84 PCI JOURNAL

Response Envelopes

Calculated story shear–drift re-sponse envelopes of Specimens J1 andJ2 are also shown in Fig. 10. Beamflexural behavior was assumed to con-trol the specimen response when theenvelopes were calculated. The effectof strain hardening of the beam longi-tudinal reinforcement on specimen

The measured initial lateral stiff-nesses of the specimens were one-halfof the calculated values. Calculatedstiffnesses were based on elastic andmonolithic behavior, as well as on un-cracked sections. For Specimen J1, thecalculated response showed higherstrength and stiffness; this is also no-ticeable at large drift levels sincestrain hardening and concrete confine-ment were considered in the predic-tion. Analysis of the strain gauge datadid not reveal any evidence of strainhardening of the beam longitudinal re-inforcement, nor any significant im-provement of concrete characteristicsin the beams next to the column due toconfinement.19

For Specimen J2, the measured en-velope in the east-west direction indi-cated less stiffness and strength thanthe calculated curve. However, the dif-ference between the actual and calcu-lated responses was less pronouncedthan those of Specimen J1. This wasbelieved to indicate that conventionalmild steel beam bottom reinforcement(Grade 60) was mobilized. Indeed,strain gauge analysis further confirmedthis reasoning.20 In the north-south di-rection of Specimen J2, agreement be-tween measured and calculated en-velopes was acceptable.

Stiffness Deterioration

In order to assess stiffness deteriora-tion, the secant stiffness was com-puted for each loading cycle. The se-cant stiffness was calculated using astraight line drawn between the maxi-mum load and corresponding driftpoints for the positive and negative di-rections in a loading cycle (peak-to-peak stiffness).

Both specimens exhibited a similarinitial lateral peak-to-peak stiffnessand a comparable rate of stiffness de-terioration (see Fig. 12). At a drift of2.0 percent, Specimens J1 and J2 hadretained 40 percent of their initialpeak-to-peak stiffness. This valuecompares favorably with ungroutedpost-tensioned beam-to-columnjoints14 and with monolithic connec-tions.23 The different joint detailingused to achieve continuity of beam re-inforcement in Specimens J1 and J2did not affect the trend observed.

Fig. 14. Strain on transverse leg of a continuity hoop of Specimen J1.

Fig. 15. Proposed plastic mechanism at failure of Specimen J1.

strength, stiffness, and inelastic de-formability were taken into account inSpecimen J1.

For Specimen J2, the calculationonly considered the contribution toflexural strength of strand yielding,whereas in the beam top reinforce-ment, the effect of strain hardening atthe column face was considered.

May-June 2002 85

Member Contribution to Interstory Drift

The relative components of inter-story drift (beam and column flexuraland shear deformations, and joint dis-tortion) were calculated.19,20 Membercontributions to drift during east-westpositive cycles are shown in Fig. 13.Note that the variation of measuredstory shear is also shown. The storyshear indicates the effect of thechanges in a given portion of thestructure on the strength of the speci-men, and is particularly important inunderstanding the mode of failure.

For both specimens, the columnscontributed least to total deformation.For Specimen J1, after cycles to 2.5percent, joint deformations con-tributed more to the total drift. This isconsistent with the damage observedin the joint. For Specimen J2, beamscontributed the most to total deforma-tion. Although the joint contributionwas smaller than in Specimen J1, itscontribution increased steadily in cy-cles between 1.5 and 2.5 percent.

Since the data became unreliableafter 2.5 percent due to spalling of thejoint concrete, it is not possible to de-termine the final member contribu-tions to total drift in Specimen J2.From the figures, it is clear that thestory shears remained nearly constantto very large drifts, despite the factthat the joint contributed significantlyto total drift. The ductile failure ob-served is explained by yielding of thecontinuity hoops (in Specimen J1) andbar pullout (in Specimens J1 and J2)inside the joint. Such mechanismscaused the joint softening.

Strains in the ContinuityReinforcement Within the Joint

As previously mentioned, the beambottom reinforcement in Specimen J1was made continuous through thejoint by a set of four No. 5 hoopsplaced around the extensions of the90-degree hooks that protruded fromthe beam ends. The hoops were de-signed to remain elastic when themaximum tensile force in the bottomreinforcement corresponding to thebeam flexural strength was developed.Strain gauges were bonded on hooplegs in directions parallel and orthog-

onal to the loading direction (east-west direction).19

Strains recorded in the continuityreinforcement demonstrated the bend-ing flexibility of the transverse legs(parallel to the north-south direction)of hoops when the 90-degree hook ex-tensions tried to pull away from thejoint. From the early stages of the test(see Fig. 14), the high bending flexi-bility of the transverse legs increasedthe beam rotation, concentrating it in-side the joint.

Subsequent plastification underbending further contributed to center-ing the rotation inside the joint andsoftening the joint core. The pulloutfailure mechanism of the beam hooked

bar in the joint caused the inclinedcracking observed in the narrow por-tion of the beam inside the joint whichwas mentioned earlier. The proposedplastic mechanism at failure of Speci-men J1 is depicted in Fig. 15.

The precast beam bottom reinforce-ment of Specimen J2 was made con-tinuous through the joint by means ofa No. 12 Grade 60 bar placed verti-cally in the intersection of the fourlooped 12.7 mm (0.5 in.) diameter,Grade 270 prestressing strands thatprotruded from the beam ends.

Strains recorded in the vertical No.12 bar at the beam soffit section and inthe east looped strand at the beam endsection (inside the joint) are shown in

Fig. 16. Strain on No. 12 vertical bar at beam soffit section of Specimen J2.

Fig. 17. Strains on east looped strand at beam end section of Specimen J2.

86 PCI JOURNAL

Figs. 16 and 17, respectively. Al-though the No. 12 bar, looped strands,and bottom beam bars remained elas-tic, damage patterns and hysteretic be-havior comparable to those of Speci-men J1 were observed. It is believedthat the continuity joint reinforcementof Specimen J2 was not stiff enoughto preclude the beam from rotating in-side the joint under positive bending,as was observed during the test ofSpecimen J2.

Clearly, the joint behavior of Speci-mens J1 and J2 was negatively af-fected by the beam rotation inside thejoint. The internal mechanisms of re-sistance (the main diagonal concretestrut and truss actions), joint stiffnessand toughness were harmed by the

development of a tensile strain fieldthat cracked the joint concrete early inthe test.

Joint Hoop Strains

Strains developed in the confiningsteel bars parallel to the loading direc-tion within the joint height are shownin Fig. 18. Results for Specimen J1 in-dicate that Positions 1 and 2 sustainedstrains exceeding yield at drifts largerthan 2.0 percent, with first yield beingrecorded at 2.0 percent drift. Strainsclose to 3ey were recorded at maximumdrift. The largest strains measuredwithin the lower half of the joint areconsistent with strains developed dueto the pullout of beam bottom bars.

In contrast, strains recorded in Spec-imen J2 were much lower than forSpecimen J1, with maximum values ofabout 1.5ey. Except for the bar in Posi-tion 1, all confining bars remainedelastic.

Joint Force Transfer Mechanism

In order to understand the mecha-nism involved in joint shear transfer,strain gauge data for Specimens J1and J2 were investigated in more de-tail. From “measured” forces in bars(calculated from strains and a cyclicstress-strain model curve), severalstrut-and-tie models were analyzed.Possible strut-and-tie models for thetwo specimens are shown in Figs. 19aand 19b. Column forces at each bun-dle of reinforcing bars were calculatedfrom moment-curvature analysesusing the column flexural momentscorresponding to the applied lateralshear force. From Figs. 19a and 19b,column and beam flexural compres-sion resultants combine to form a di-agonal strut. Hoop tensile forces, con-sidered to be lumped at jointmidheight, are equilibrated at the coreboundary by two inclined struts an-gling towards the centroid of the beamand column compression resultantsand by changes in the column rein-forcing bar forces.

The inclined struts in the lower halfof the joint are of particular interest. InSpecimen J1, the strut (with 235 kN)is balanced horizontally by tensileforces in the continuity hoops andbond along the horizontal short seg-ment of the bar protruding from thebeam end, and vertically through bondalong the hook extension of beam re-inforcement. This mechanism is simi-lar to that observed in pullout failuresof bars. In Specimen J2, the horizontalreaction of the inclined strut is givenby dowel action of the No. 12 barplaced to give continuity to the beambottom reinforcement.

For the struts within the lower halfof the specimen, premature crackingdue to bar pullout softens the joint re-gion, thus reducing its stiffness and re-sistance to shear forces. Such inclinedcracks at the lower half are supportedby the observed crack patterns of Figs.6 and 8.

Fig. 18. Joint hoop strains for Specimens J1 and J2.

May-June 2002 87

It is possible, based on strut-and-tiemodels, to determine how much of thenet force differential in the columnbundled bars was transferred by strut-and-tie action involving the hoop setsand how much was transferred by

shear-compression within the majordiagonal strut. From Fig. 19a, for ex-ample, the force differential in the col-umn bundled bars is 1108 + 1275 =2383 kN (536 kips). As shown, hoopforces assist in transferring only 991 +

1275 = 2266 kN (509 kips), or 95 per-cent of the total. It can be inferred thatthe remainder, 2383 – 2266 = 117 kN(26 kips), was transferred by shear-compression within the width of thediagonal strut.

Fig 19a. Possiblestrut-and-tie modelof joint region ofSpecimen J1 at strength.

Fig 19b. Possiblestrut-and-tiemodel of jointregion ofSpecimen J2 at strength.

88 PCI JOURNAL

CONCLUSIONSBased on the observations and re-

sults during fabrication, testing, anddata analysis of the two large-scaleprecast concrete beam-to-column con-nections, the following conclusionscan be made:

1. Precast concrete frames using thetwo tested beam-to-column connec-tions can be built easily and quickly.

2. The specimens exhibited ductilebehavior. The lateral load carrying ca-pacity was maintained nearly constantup to drifts of 3.5 percent, which arelarger than the maximum drift valuesallowed in most design codes aroundthe world.

3. The specimens were designed todevelop plastic hinges in the beamsnext to the columns and to imposejoint shear demands close to the nomi-nal joint shear strength for monolithicconstruction. As expected, the behav-ior was controlled by the joint at largedrifts. Joint degradation and stiffnessdecay were recorded after the beamtop steel reinforcement of the beamhad yielded in tension.

4. In both structures, beam rotationtook place inside and outside the joint.Joint mechanisms of resistance wereimpaired by the development of ten-sile strains due to beam rotation insidethe joint. Beam rotation inside thejoint does not usually occur in mono-lithic construction.

5. In Specimen J1, where hoopswere used to achieve continuity, pre-mature bending flexibility of hooplegs transverse to the loading direc-tion, as well as pullout of beam bot-tom bars, contributed to initial jointdamage.

6. Specimen J2 (in which continuityof bottom longitudinal reinforcementof the beams is provided by a steel barinserted through overlapping U-shaped prestressing strands) per-formed better than Specimen J1. Spec-imen J2 exhibited a more uniformdistribution of beam cracking andyielding under negative bending.

7. Continuity reinforcement in theform of hoops or U-shaped strandsshould be sufficiently strong and stiffto avoid plastic behavior under maxi-mum demands calculated from a ca-pacity design approach.

8. Joint shear strengths of Speci-mens J1 and J2 were 80 and 90 per-cent, respectively, of those expectedfor monolithic construction. More-over, the initial shear cracking oc-curred at lower levels of nominalshear stress than in monolithic con-struction. This phenomenon was at-tributed to premature beam rotation in-side the joint.

9. The test results showed that thestructural response of the precastframe was satisfactory. Although theconnections tested did not fully emu-

late monolithic construction, they canbe used in precast concrete frame sys-tems or in hybrid systems, providedthat their strength and stiffness aretaken into account.

10. To improve the cyclic behaviorof the connections, beam rotations in-side the joint should be minimized.One approach to accomplishing thisobjective is to force the concentrationof beam flexural rotations away fromthe column faces, i.e., relocate thebeam plastic hinges. An alternative toreducing beam rotations inside thejoint is to place unbonded post-ten-sioning tendons through the joint.

ACKNOWLEDGMENTThe support of the staff of the

Large-Scale Structures Testing Labo-ratory of the National Center for Dis-aster Prevention (CENAPRED) and ofServicios y Elementos Presforzados,S.A. de C.V. (SEPSA) is gratefully ac-knowledged. This investigation is partof a research program sponsored byCENAPRED and SEPSA. The con-nection system is protected under U.S.Patent No. 5 682 717 and MexicanPatent No. 94/09263.

The authors wish to express theirappreciation to the PCI JOURNAL re-viewers for their constructive com-ments.

May-June 2002 89

1. Vasconez, R. M., Naaman, A. E., and Wight, J. K., “Review ofResearch on the Design of Ductile Beam-to-Column Connec-tions for Precast Concrete Frames,” Report No. UMCEE 94-33, Department of Civil and Environmental Engineering, Uni-versity of Michigan, Ann Arbor, MI, October 1994, 173 pp.

2. Englekirk, R. E., “Seismic Design Considerations for PrecastConcrete Multistory Buildings,” PCI JOURNAL, V. 35, No. 3,May-June 1990, pp. 40-51.

3. ACI Innovation Task Group 1, “Acceptance Criteria for Mo-ment Frames Based on Structural Testing,” Provisional Stan-dard, American Concrete Institute, Farmington Hills, MI, 1999.

4. Ghosh, S. K., Nakaki, S. D., and Krishnan, K., “Precast Struc-tures in Regions of High Seismicity: 1997 UBC Design Provi-sions,” PCI JOURNAL, V. 42, No. 6, November-December1997, pp. 76-93.

5. Park, R., “A Perspective on the Seismic Design of PrecastConcrete Structures in New Zealand,” PCI JOURNAL, V. 40,No. 3, May-June 1995, pp. 40-60.

6. Reglamento de Construcciones Para el Distrito Federal (inSpanish), Departamento del Distrito Federal, Gaceta Oficialdel Distrito Federal, México, D.F., 1997.

7. UBC, Uniform Building Code, International Conference ofBuilding Officials, Whittier, CA, 1997.

8. Sucuoglu, H., “Effect of Connection Rigidity on Seismic Re-sponse of Precast Concrete Frames,” PCI JOURNAL, V. 40,No. 1, January-February 1995, pp. 94-103.

9. Cheok, G. S., and Lew, H. S., “Performance of Precast Con-crete Beam-to-Column Connections Subjected to Cyclic Load-ing,” PCI JOURNAL, V. 36, No. 3, May-June 1991, pp. 56-67.

10. Cheok, G. S., and Lew, H. S., “Model Precast Concrete Beam-to-Column Connections Subjected to Cyclic Loading,” PCIJOURNAL, V. 38, No. 4, July-August 1993, pp. 80-92.

11. Englekirk, R., “Development and Testing of a Ductile Connec-tor for Assembling Precast Concrete Beams and Columns,”PCI JOURNAL, V. 40, No. 2, March-April 1995, pp. 36-51.

12. Nakaki, S. D., Stanton, J. F., and Sritharan, S., “An Overviewof the PRESSS Five-Story Precast Test Building,” PCI JOUR-NAL, V. 44, No. 2, March-April 1999, pp. 26-39.

13. Palmieri, L., Saqan, E., French, C. W., and Kreger, M.E.,“Ductile Connections for Precast Concrete Frame Systems,”Paper No. SP162-13, Mete A. Sozen Symposium, ACI SP 162,American Concrete Institute, Farmington Hills, MI, 1996, pp.313-355.

14. Priestley, M. J. N., and MacRae, G. A., “Seismic Tests of Pre-cast Beam-to-Column Joint Subassemblages with UnbondedTendons,” PCI JOURNAL, V. 41, No. 1, January-February1996, pp. 64-81.

15. Priestley, M. J. N., “The PRESSS Program – Current Statusand Proposed Plans for Phase III,” PCI JOURNAL, V. 41, No.2, March-April 1996, pp. 22-41.

16. Restrepo, J. I., Park, R., and Buchanan, A. H., “Tests on Con-nections of Earthquake Resisting Precast Reinforced ConcretePerimeter Frames of Buildings,” PCI JOURNAL, V. 40, No. 4,July-August 1995, pp. 44-61.

17. Restrepo, J. I., Park, R., and Buchanan, A. H., “Design of Con-nections of Earthquake Resisting Precast Reinforced ConcretePerimeter Frames,” PCI JOURNAL, V. 40, No. 5, September-October 1995, pp. 68-80.

18. Kanoh, Y., “Review of Japanese Precast Concrete Frame Sys-tems Used as Building Structures,” Proceedings of a Seminaron Precast Concrete Construction in Seismic Zones, Japan So-ciety for the Promotion of Science and National Science Foun-dation, Tokyo, Japan, V. 2, October 1986, pp. 35-54.

19. Alcocer, S. M., Perez-Navarrete, D., and Gomez, A., “Estudiodel Comportamiento de Una Conexión Viga–Columna de Ele-mentos Prefabricados - Modelo J-E-” (in Spanish), ReportIEG/01/00, Centro Nacional de Prevención de Desastres (CE-NAPRED), Mexico, D.F., April 2000, 54 pp.

20. Alcocer, S. M., and Perez-Navarrete, D., “Estudio del Com-portamiento de Una Conexión Viga–Columna de ElementosPrefabricados - Modelo J-T-” (in Spanish), Report IEG/02/00,Centro Nacional de Prevención de Desastres (CENAPRED),Mexico, D.F., July 2000, 73 pp.

21. ACI Committee 318, “Building Code Requirements for Struc-tural Concrete (ACI 318-95),” American Concrete Institute,Farmington Hills, MI, 1995, 369 pp.

22. ACI-ASCE Committee 352, “Recommendations for Design ofBeam-to-Column Joints in Monolithic Reinforced ConcreteStructures (ACI 352R-91),” American Concrete Institute,Farmington Hills, MI, 1991, 18 pp.

23. Kurose, Y., Guimaraes, G. N., Zuhua, L., Kreger, M. E., andJirsa, J. O., “Study of Reinforced Concrete Beam-to-ColumnJoints Under Uniaxial and Biaxial Loading,” PMFSEL ReportNo. 88-2, Phil M. Ferguson Structural Engineering Laboratory,University of Texas at Austin, Austin, TX, December 1988,146 pp.

REFERENCES