Seismic Performance ofPrecast Segmental BridgeSuperstructures with InternallyBonded Prestressing Tendons

Sami Megally, Ph.D.Assistant Project ScientistDepartment of Structural EngineeringUniversity of California at San DiegoLa Jolla, California

Frieder Seible, Ph.D., P.E.Executive Associate Dean and Professor

Department of Structural EngineeringUniversity of California at San Diego

La Jolla, California

Manu GargGraduate Research AssistantDepartment of Structural EngineeringUniversity of California at San DiegoLa Jolla, California

Robert K. Dowell, Ph.D., P.E.Principal

Dowell-Holombo Engineering, Inc.San Diego, California

A research project is currently in progress toinvestigate the seismic performance of segment-tosegment joints of precast segmental concretebridges. This paper presents the experimental andanalytical results of two large-scale tests whichmodel the superstructure response of a prototypeprecast segmental bridge, post-tensioned withinternally bonded tendons, under fully reversedvertical cyclic displacements. The joints of the firsttest unit were epoxy bonded with no reinforcementcrossing the joints other than the prestressing steel.The second test unit had a reinforced cast-in-placedeck closure with the remaining portions of thejoints connected by epoxy. Both test units weresubjected to fully reversed cyclic loads simulatingearthquake vertical motions. It was found that bothtest units could undergo significant seismicdisplacements, but their failure modes wereconsiderably different. The paper also presentsresults from nonlinear finite element modeling ofthe simulated seismic response of the test units.The results showed that segment-to-segment jointscould undergo significant repeated opening andclosure before failure.

precast segmental concrete bridge construction ispopular because of its well-known advantages ofquality control and construction schedule over con

ventional cast-in-place construction. However, the popularity of precast segmental bridges is hampered in highseismic zones because of the lack of information on theirseismic performance.

40 PCI JOURNAL

The AASHTO Guide Specificationfor Design and Construction of Segmental Concrete Bridges1 permits theuse of precast segmental constructionin high seismic zones (Zones 3 and 4),provided that the precast segments areepoxy bonded. The same AASHTOGuide Specification’ also requires thatexternal tendons should achieve nomore than 50 percent of the superstructure post-tensioning.

In other words, fully bonded internal tendons should achieve at least 50percent of the post-tensioning. In addition to these requirements, precastsegmental bridge construction withoutmild steel reinforcement crossing thesegment-to-segment joints is not recommended in current practice in highseismic zones such as California.

The above-mentioned recommendations and restrictions of current practice are justified by the lack of information on seismic performance ofprecast segmental bridges. Thus, acomprehensive research project hasbeen developed by ASBI (AmericanSegmental Bridge Institute) andfunded by Caltrans (California Department of Transportation) to investigatethe seismic performance of precastsegmental bridge superstructures.

In order to investigate the seismicperformance of precast segmentalbridges, a large-scale experimental research project is currently in progressat the University of California, SanDiego (UCSD). This research projectconsists of the following three phases:

Phase I: To investigate the seismic performance of segment-to-segment joints in superstructures with different ratios of internal to externalpost-tensioning under simulated fullyreversed cyclic loading. In this firstphase, only superstructure joints closeto midspan in regions with high positive flexural moments and low shearsare considered. Phase I consists of thefollowing two parts:

(1) Phase I-A: Superstructures with100 percent internal post-tensioning(Test Units 1 OOINT and 1 OOINTCIP;see Table I).

(2) Phase I-B: Superstructureswith 100 percent external post-tensioning and superstructures with 50percent internal post-tensioning combined with 50 percent external post-

tensioning (Test Units 100EXT and5OINTI5OEXT, respectively). TestUnits 100EXT and 5OINTI5OEXT areidentical to Test Unit 100INT ofPhase I-A with the only differencebeing the percentage of external post-tensioning. Test Unit 100EXT, with100 percent external post-tensioning,does not satisfy the current requirement of the AASHTO Guide Specification’ that no more than 50 percentof post-tensioning should be achievedby external tendons.

• Phase II: To investigate the seismic performance of superstructurejoints close to the supports in regionswith high shears and negative flexuralmoments.

• Phase III: To investigate the system performance of segmental superstructure and columns under longitudinal seismic loading.

The experimental program and theanalytical model calibrations of thetwo tests from Phase I-A with 100percent internally bonded tendons arepresented in this paper. Each test unitconsisted of six precast segments,which were epoxy bonded. The entirejoint surfaces of the first test unit wereepoxy bonded with no mild steel reinforcement crossing the joints, whichdid not conform to the current recommended practice in high seismiczones. The second test unit had a reinforced cast-in-place deck closure atthe location of each joint, with theweb and bottom soffit of the segmentsepoxy bonded.

The major objectives of this firstphase of the research program are toinvestigate: (1) joint behavior in termsof opening and closure under repeatedcyclic loads simulating earthquake effects, (2) development of crack pattems, and (3) modes of failure.

Three-dimensional finite elementmodels of the test units were developed. The models took into accountconcrete cracking and crushing, opening and closure of segment-to-segment

joints and the inelastic characteristicsof prestressed and non-prestressedsteels. The finite element models werevalidated with the experimental resultspresented in this paper.

The calibrated analytical models arepowerful tools for parametric and design studies that provide useful information, which may be difficult to obtain experimentally. They also providea better understanding of the observedbehavior of segment-to-segmentjoints.

The main objective of the analyticalmodels described in this paper is todetermine the major characteristics ofjoint behavior. The results will beused to develop a comprehensiveglobal finite element model of bridgestructures; this global model can beused for analytical parametric studiesinvestigating different superstructuregeomethes, prestress levels, and seismic input variations.

PROTOTYPE STRUCTURETest units of Phase I of this experi

mental program are based on the prototype structure shown in Fig. 1. Thesuperstructure consists of three 100 ft(30.48 m) interior spans and 75 ft(22.86 m) exterior spans with a totallength of 450 ft (137.16 m) and is prestressed with a harped shape tendon(see Fig.la).

Because of its short spans, the prototype structure is constructed usingthe span-by-span method. Fig. lbshows the cross section of the prototype superstructure.

The prototype structure was designed according to the AASHTOGuide Specification for Design andConstruction of Segmental ConcreteBridges,1 the AASHTO-PCI-ASBISegmental Box Girder Standards forSpan-by-Span and Balanced Cantilever Construction2 and theAASHTO Standard Specifications forHighway Bridges.3

March-April 2002 41

Table 1. Test matrix.Phase P Test unit Test description

I AIOOINT j 100 percent internal post-tensioning

100TNTCIP 100 percent internal pest-tensioning and cast-in-place deck closure joints

I B100EXT 100 percent external post-tensioning

5OINT/5OEXT 50 percent internal + 50 percent external post-tensioning

450’

25T25125

Fig. 2. Joint and segment numbering of test units.

75’ .,. 100’ 100’ 100’.

75’

Th\

I I

I1 II1—1.30

—I

HamedCable Path

‘‘ 25j

(b) Cross section

Fig. 1. Prototype structure: (a) Elevation; and (b) Cross section.

EXPERIMENTAL PROGRAM Description of Test Unitsmiddle third of the prototype span in

This section gives a description of The critical location of the prototype which the tendon is horizontal (seethe test units and their method of con- structure for positive bending under Fig. la). For the laboratory tests, theystruction. The test setup and sequence dead load and seismic forces was were designed at two-thirds scale ofof loading of the test units are also de- found to be approximately at the prototype structure.scribed in this section. midspan.4 The test units model the Fig. 2 shows a typical test unit, sim

End Endsegment Test Zone segment

1 ii I

End supportframe (fixedbottom)

End supportframe (pinnedbottom)

42 PCI JOURNAL

9’ -0”

Fig. 3. Cross sectionof test units.

piy supported at its ends. The test zonehad a total length of 24 ft (7.32 m),which represented the center one-thirdportion of the prototype span. This testzone consisted of four 6 ft (1.83 m)long by 4 ft (1.22 m) deep precast segments (Segments 2 to 5 in Fig. 2).Each test unit was supported at itsends through precast end segments(Segments 1 and 6 in Fig. 2).

Fig. 3 shows the cross section andreinforcement of precast Segments 2through 5 (see Fig. 2). Half of the prototype box girder section was modeledand idealized in the shape of an equivalent I-section to simplify the testsetup.

The variable investigated in this experimental program (Phase I-A) wasthe presence of mild steel reinforcement crossing the segment-to-segmentjoints. Test Unit I O0INT used 100 per-

cent internal post-tensioning (bondedtendon) with no cast-in-place deckclosure joints, or in other words withno mild steel reinforcement crossingthe segment joints.

The segments of Unit 100INT wereconnected by Sikadur 31, SBA (Segmental Bridge Adhesive) slow-setepoxy, which was applied to the entirecross section of the segment-to-segment joints. Specifications of theepoxy are given in Table 2, in whichUcomp and bond are the compressiveand bond strengths, respectively.

Test Unit 1 OOINTCIP used 100 percent internal post-tensioning (bondedtendon) and reinforced cast-in-placedeck closures at locations of Joints J2,J3 and J4 (see Fig. 2). Details of thereinforced cast-in-place deck closurejoints of Test Unit 100INTCIP weresimilar to those used in the design of

the new East Span Skyway of the SanFrancisco-Oakland Bay Bridge.

Two different reinforcement detailswere incorporated in Unit IOOINTCIPat the cast-in-place deck joints. Benthairpin bars were used on one-half ofthe cross section and bars with mechanical anchors (welded heads) attheir ends were used in the second half(see Fig. 4). Both reinforcement details provide adequate anchorage ofthe reinforcing bars in the cast-in-place deck joints so their full yieldstrength can be mobilized. The objective was to study the effect of each ofthese details on the performance of thejoints.

The remaining portions of the jointsin Unit 100INTCIP, along the web andbottom slab, were connected by theslow-set segmental bridge adhesive(epoxy). An alternative to the use of

Table 2. Material properties.

f (cast-in-place Epoxyf (on day of testing) deck reinforcement) properties*

ksi (MPa) ksi (MPa) (Sikadur 31, SBA) ksi (MPa)Segment Segment Segment Cast-in-place Headed Bent u,,,,1,

Test unit No. 1 & 5 No.3 No.2,4 & 6 deck bars bars (3 days) (14 days)5.11 7.21 6.96 2.0 1.0T2

(i38) (6.9)

IOOINTCIP7.25 7.25 5.74 5.87 67.7 ] 75.6 2.0 1.0

(50.0) (50.0) (39.6) (40.5) (467) (521) (13.8) (6.9)Epoxy properties are those specified by the manufacturer (slow-set epoxy).

March-April 2002 43

1 .5”Dia. sleevesfor actuators rods

Duct for auxiliarytop prestressing

8 318” steel

16 Strandstotal prestressingsteel (Each A=0.2

#6 @ 12”

>#5 @6”

@8”

NDuct for auxiliary —

Bottom prestressing steel

-

---

-C.G. of Prestressing

L

reinforced cast-in-place deck closurejoints in Unit 100INTCIP was to post-tension the deck; however, post-tensioning of the deck was not in thescope of this experimental program.

Each test unit was post-tensionedwith 16 strands each of 0.6 in. (15.2mm) diameter with an ultimate tensilestrength of 270 ksi (1860 MPa). The

magnitude of the prestressing force,which is equal for all test units, wascalculated so that the concrete stressesresulting from post-tensioning are thesame as for the prototype structure. Except for the cast-in-place deck reinforcement in Test Unit 100INTCIP, thelayout of the reinforcing bars (Grade60) is the same for the two test units.

Table 2 gives the measured materialproperties of concrete and the reinforcement of the cast-in-place deckjoint (Test Unit 100INTCIP). InTable 2, f is the concrete compressivestrength on the day of testing of bothunits andf is the yield strength of themild steel reinforcing bars crossingthe segment-to-segment joints in Unit100INTCIP.

Construction of Test Units

As mentioned above, each test unitconsisted of six precast segments.Segments 1, 3 and 5 (see Fig. 2) werecast at the same time. This was followed by construction of Segments 2,4 and 6, which were match-castagainst Segments 1, 3 and 5. A bondbreaker was applied along the match-cast surfaces so that the segmentscould be separated after concretehardening. Fig. 5 shows a precast segment from Unit 100INT with shearkeys along the web and alignmentkeys in the deck and bottom slabs.

The segments of each test unit wereassembled on a wooden platform atthe UCSD Structures Laboratory. Theepoxy was applied to the joint surface

Deck reinforcement (East halt)

Deck reinforcement (West half)

End supportframe (fixedbottom)

) North

End supportframe (pinnedbottom)

Fig. 4. Reinforced cast-in-place deck joints of Test Unit 1 OOINTCIP.

/

Fig. 5. A precast segment from Test Unit 1 OOINT.

I J .

44 PCI JOURNAL

as shown in Fig. 6. After applicationof the epoxy and placement of eachsegment in its final position, the testunit was temporarily post-tensionedby four 1 in. (25.4 mm) diameter highstrength ASTM A 722 prestressingsteel bars (two bars in the top slab andtwo bars in the bottom slab). The temporary prestressing forces in the highstrength bars were determined suchthat the entire segment-to-segmentjoint surfaces would have a minimumcompressive stress of 40 psi.5

It should be remembered that afterjoining the segments of Unit 1 OOINTCIP, there was a gap in the deck at thelocation of Joints J2 to J4 (see Figs. 2and 4), which was closed later by acast-in-place concrete slab strip. Afterjoining of the precast segments andclosure of the cast-in-place deck joints(Unit I OOINTCIP), each test unit waspost-tensioned with a jacking force of720 kips (3203 kN). The effective prestressing force at the time of testingwas estimated to be about 600 kips(2669 kN).

The difference between the jackingand effective prestressing forces is dueto losses from anchor set, elastic shortening, creep and shrinkage of the concrete as well as relaxation of the prestressing steel. Fig. 7 shows Test UnitI OOINT during the post-tensioning operation, which was followed by grouting of the prestressing duct. Fig. 7 alsoshows the high strength bars used intemporary prestressing of the test units.

The temporary prestressing force inthe bottom slab was released after permanent post-tensioning of the testunit, whereas the temporary prestressing force in the top slab was releasedafter vertical loading of the test unit tosimulate the prototype dead loadstresses. The stressing and loading sequence was designed to avoid cracking of the units before the test. Afterpermanent post-tensioning, thewooden platform supporting the segments during assembly was removedand the test unit was mounted on thetwo end supports.

Test Setup

Fig. 8 shows the test unit and theload frame. Each test unit was simplysupported by a steel pin and steel links

March-April 2002

at its ends. At one end, the steel linkswere fixed at their bottom ends to restrain horizontal movement of the testunits. At the other end, the steel linkswere pinned at the bottom (rockerlinks) to allow rotation of the framelegs and horizontal movement of thetest units. The loads were transferredfrom the test units to the steel links bymeans of steel pins inside horizontalsteel pipes cast into the end segmentsat the neutral axis of the test units, allowing the ends of the test units to rotate freely.

Four vertical servo-controlled hydraulic actuators were used to apply

external loads to each test unit to simulate the effect of highway loadingand vertical seismic displacements onthe superstructure. As in the midspanjoint of the prototype span, themidspan joint of each test unit wassubjected to zero shearing force andthe highest bending moment.

At the beginning of the test, eachunit was loaded in the downward direction to a prescribed level so that thestresses in the midspan joint were similar to the stresses in the prototypestructure under combined dead load,superimposed dead load, as well asprestres sing primary and secondary ef

45

Fig. 6. Application of epoxy to the joint surface.

Fig. 7. Post-tensioning of test units.

Fig. 8. Test setup.

fects. This load level will be referredto as the reference load level throughout this paper. Each test was conducted as follows:

Stage I (Service Load Conditioning) — Only the two interior actuatorswere used in load control during thistest stage. Each test unit was loaded tothe reference load level at P = 74.5kips (331 kN), where P is the load pereach actuator. The temporary prestressing force in the top slab was released at this stage.

This was followed by cycling theload P between 112 and 65 kips (498and 289 kN) 100,000 times. The purpose of this test stage was to study the

maximum service load cycles.The upper and lower load limits

were calculated so that the stresses inthe test units at midspan would besimilar to the stresses at midspan forthe prototype structure under maximum and minimum service loads.Table 3 gives the estimated concretestresses at midspan for Units 100INTand 100INTCIP during Test Stage I aswell as the corresponding prototypeconcrete stresses.

Stage II (Seismic Test) — Thefour actuators were used in displacement control and the forces in the actuators were maintained equal

throughout this test stage. Each testunit was loaded to the reference loadlevel with P = 40.5 kips (180 kN). Thetest unit was then subjected to fully reversed cyclic displacement at midspanwith increasing amplitude to failure.

Three cycles were completed ateach displacement level up to 4 in.(102 mm) displacements. Beyond the4 in. (102 mm) displacement, only onecycle was performed at each displacement level. The applied displacementhistory during Test Stage II is shownin Fig. 9.

Electrical resistance gauges wereused to measure strains in the concreteand prestressing steel. Vertical dis

Table 3. Concrete stresses in the prototype structure and in test units during Test Stage I.- Concrete stresses,* psi (MPa) —

P 1 Test Unit Test Unit T Prototypekips 100INT 10O1NT structure

Load case (kN) Top Bottom Top Bottom Top Bottom,, 74.5 -473 -412 -462 -407 -475 -411DL + PT +

(331) (-3.26) (-2.84) (-3.19) (-2.81) (-3.28) (-2.83), 112 -710 -1.4 -699 +4 -1058 -0.2DL+PT±A +LL+T

__(498) (490) (001) (482) (003) (730) - (0)65 -413 -516 -402 -511 -413 -520DL + PT + — LL(289) (-2.85) (-3.56) (-2.77) (-3.52) (-2.85) (-3.59)

DL = Dead Load; PT = Prestressing primary moments; A = Prestressing secondary moments;LL = Live Load; T = Temperature gradient* Positive sign indicates tensile stresses.

performance of joints under repeated

46 PCI JOURNAL

placements along the span, joint openings at various locations, vertical sliding between the precast concrete segments at each joint and supportdisplacements were measured bymeans of linear potentiometers.

EXPERIMENTAL RESULTSThe major experimental results are

presented in this section. These experimental results include crack patterns, modes of failure, load-displacement response, performance ofjoints, strains in prestressing steel,cracking strength and flexural moment capacity.

Crack Patterns and Modesof FaIure

Table 3 indicates that both test unitswere subjected to very low tensilestresses at the midspan joint duringTest Stage I (Service Load Conditioning). Thus, no cracks or joint openingswere observed in both test units duringTest Stage I. Because of this linearelastic behavior of both test units during Test Stage I, only the results ofTest Stage II will be discussed here.

Test Unit 100INT — The firstcrack occurred during downward loading at the midspan joint (Joint J3, Fig.2) during the first displacement cycleto 0.25 in. (6.35 mm) amplitude (seeFig. 9). The opening of Joint J4 (seeFig. 2) also occurred during the samedisplacement cycle.

Shear cracks occurred in the webduring the same displacement cycle,between the supported ends of the testunit and the load application points(shearing force is approximately zeroat the midspan joint). Joint 12 (see Fig.2) opened during application of thedownward displacement to 0.5 in.(12.7 mm) amplitude. Thus, the threeinterior Joints J2, J3 and J4 (see Fig.2) opened during loading in the downward direction.

Few additional flexural cracks occurred inside Segments 3 and 4 (seeFig. 2) during subsequent downwardloading; however, the widths of thesecracks were very small and inelasticdeformations of the test unit were concentrated mainly at the midspan JointJ3. New shear cracks developed and

existing shear cracks propagated during subsequent displacement cycles.Shear cracks crossed the epoxy-bonded joints with no vertical slidingbetween adjacent precast segments.

The midspan Joint J3 was the onlyjoint that opened during upward loading of Unit 100INT. Joint 13 openedduring the 0.5 in. (12.7 mm) displacement cycle. Because Unit IOOINT didnot have any mild steel reinforcementcrossing the joints, the opening ofmidspan Joint J3 increased significantly under upward loading duringsubsequent displacement cycles. Fig.10 shows the midspan joint during thelast displacement cycle in the upwardloading direction before failure of TestUnit 100INT.

Under downward loading, themidspan Joint 13 opened significantlywith increased applied displacementuntil rupture of the prestressingstrands at the midspan joint at a displacement of about 4.8 in. (122 mm)and a total load of 490 kips (2180 kN).Fig. 11 shows the midspan joint afterfailure of Test Unit 100INT. Fig. 12shows a closeup view of the rupturedstrands.

Test Unit 100INTCIP — Development of the crack pattern duringdownward loading of Test Unit1 OOINTCIP was similar to the one described above for Unit 100INT. The

first flexural crack occurred atmidspan Joint J3 during the 0.25 in.(6.35 mm) displacement cycle. Diagonal shear cracks also occurred duringthe same displacement cycle. Joints J2and J4 (see Fig. 2) opened during the0.50 in. (12.7 mm) downward displacement cycle.

The first flexural crack, in the upward loading direction of Unit

CALTRANS

ECAST SEGMENTALBRIDGES

UNIT 1

SEISMIC TEST

AL LOAD -48 KIPS

ROER DISP -4.0 IN

CLE 1

OCTOBER 31 2000

-

March-April 2002 47

150

125

100

75

50

25

—‘ 4

AAAAAAAAAAAA.I lilA0

1

0-3

-4

-5

-2

0

I-25

-50

0.75

-100

-125

-150

Fig. 9. Loading protocol of Test Stage H.

Fig. 10. Midspan joint at maximumupward displacement before failure ofTest Unit 100INT.

CAL-TRANS

pRECAST SEGMENTALBRIDGES

UNIT1

SEISMIC TESt

*JTAL LCAD 490 (lPS!

GIRDER DISP 5,0 IN

cCLE

NOVEMBE°

Fig. 11. Midspan joint after failure ofTest Unit 100INT.

1 OOINTCIP, occurred at the midspanJoint J3 during the 0.75 in. (19.1 mm)displacement cycle. The first crack occurred at the interface between theprecast concrete and cast-in-placedeck closure joint, rather than at thecenter of Joint J3.

Additional flexural cracks occurredin the deck under upward loading during subsequent displacement cycles.Because of the continuous deck reinforcing bars in Unit 100INTCIP, several closely spaced and relatively nar

row cracks occurred in the deck underupward loading, rather than the singlewide crack at the midspan joint ofUnit IOOINT.

Fig. 13 shows the midspan joint atthe last displacement cycle in the upward loading direction before failureof Test Unit 100INTCIP. The figureshows the major crack which occurredat the interface between the precastconcrete and the deck closure joint;this crack propagated in the web at anangle towards the bottom soffit of the

test unit to join the crack which occurred at midspan under downwardloading. The midspan joint openingincreased significantly under increasing downward displacements.

Reversed cyclic loading of the deckreinforcement, across the midspanjoint, resulted in longitudinal cracksin the cast-in-place concrete, whichwas followed by buckling of the cast-in-place deck reinforcing bars at adownward displacement of about 3 in.(76.2 mm).

Fig. 13. Midspan joint at maximum upward displacement before failure ofTest Unit IOOINTCIP.

Fig. 14. Compression failure at midspan joint ofTest Unit 1 OOINTCIP.

3’

48 PCI JOURNAL

Table 4. Summary of experimental results.

Positive sign is for downward load and displacement.* Maximum total load or vertical displacement at 6 in. (152 mm) from midspan before failure.

With the concrete severely weakened by reversed cyclic loading, thebars experienced significant bucklingand pushed the cover concrete away,which resulted in reduction of the compression area and capacity of the deck.Buckling of the deck reinforcing barscontinued until compression failure ofthe deck at a downward displacementof about 5.85 in. (149 mm); the peaktotal load of Unit 100INTCIP was 480kips (2135 kN). Fig. 14 shows themidspan joint and buckling of the deckreinforcing bars after failure of TestUnit 100INTCIP. It is believed thatUnit I OOINTCIP could have sustainedfurther loading if buckling of the reinforcing bars was prevented; this couldhave been achieved by providingclosed stirrups around the deck top andbottom reinforcement.

Table 4 summarizes the experimental loads and displacements of bothtest units. In Table 4, cr and P, are,respectively, the total applied load atonset of joint opening and the maximum total load reached before failure.The vertical displacement at onset ofjoint opening and the maximum displacement measured at 6 in. (152 mm)from midspan are cr and A,4, respectively.

Load-Displacement Response

Figs. iSa and 15b show the historyof total applied load versus verticaldisplacement measured at 6 in. (152mm) from midspan of Test UnitsI OOINT and 1 OOINTCIP, respectively.The sign convention in Fig. 15 is positive for downward loading and dis

Pp j’p Onset of cracking Rupture of strands

++ 4’4’ of Joint J2

IOnset ofcrackingof

Joints J3 & 34Reference toe level

—-

Onset of cracking of Joint 33

Fig. 15. History of the total load versus vertical displacement measured at6 in. (152 mm) from midspan of the test units: (a) Test Unit 100INT; and(b) Test Unit 1 OOINTCIP.

Test unit

Per downkips (kN)

I0OTNT

JointJ3Joint J2 (midspan)

375(1668)

100INTCIP

Per upkips (kN)

Joint J4JointJ3

(midspan)

347(1544)

Test unit

Peak total load, Pkips (kN)

-93(-414)-103

IOOINT

I OOINTCIP

307 307(1366) I (1

324 361(1441) (1606)

Downin. (mm)Joint J3

(midspan)0.30

___________

(7.62)0.50

(6.35) (12.7)

Joint J20.53

r (13.5)0.40

(10.2)

_____

(458)

kr Up

________

in. (mm)Joint J3

Joint J4 (midspan)0.30 -0.04

(7.62) (-1.02)

Maximum* -_Minimum490 -93

(2180) j (-414)480 -327

(2135) (-1455)Peak displacement, 4,

_______

in. (mm)

1 -

Maximum* Minimum4.80 - -4.03(122)

_________

(-102)5.85 -4.61(149) (-117)

-0.05(-1.27)

2500

2000

Vertical displacennnt, t (nrn.150 .125 -100 -75 -50 .25 0 25 50 75 100 125 150 175 200

500

________________________

400

300

200

u 1000

-100

-200

-300

-6 -5 -4 -3

1500

1000

5000

0

-500

-1000

-2 -l 0 I 2 3 4Vertical displacennnt, i (in.)

(a) Test Unit 100INT

-1500

6 7 8

Vertical displacennnt, s (nest)

.50 -25 25 50 75 100 125 150 175 200

I-6 -5 -4 -3 -2 -l 0 1 2 3 4 5

Vertical displacenesnt, is (in.)

(b) Test Unit 100INTCIP

6 7 8

March-April 2002 49

placement. As Fig. 15 shows, the performance of Test Units 100INT and1 OOINTCIP was similar under downward loading. However, the performance of the two test units was substantially different under upwardloading.

Because Unit 100INT did not haveany mild steel reinforcement crossingthe joints, the total upward loaddropped after opening of the midspanjoint. The maximum upward load for

Unit 100INT was 93 kips (414 kN)which was the load at the onset ofmidspan joint opening. The strengthof the cast-in-place deck reinforcingbars of Unit 100INTCIP could be developed resulting in a maximum totalupward load of 327 kips (1455 kN).

Deck continuity had a strong influence on the hysteretic behavior of TestUnit 100INTCIP, compared to Unit100INT as Fig. 15 indicates. Test Unit100INT had almost no hysteretic be-

havior in the upward loading directionwhich indicates very low energy dissipation capability. Unit 100INTCIPhad better hysteretic behavior underupward loading because of yielding ofthe continuous deck reinforcement.

The energy dissipated during thecomplete loading cycle at 3 in. (76.2mm) displacement was calculated forboth test units from the hysteresisloops shown in Fig. 15. The energydissipation capability was quantifiedby the equivalent viscous dampingcoefficient as a ratio of critical damping.6 Damping ratios of Test Units100INT and 100INTCIP were determined to be 4.21 and 8.75 percent, respectively, which demonstrates the effect of deck continuity on energydissipation capability.

Another important issue is the permanent displacement after earthquakeoccurrence. This corresponds to thedisplacement at a total downward applied load of 162 kips (721 kN), whichis the reference load. Permanent vertical displacements were obtained fromFig. 15 at the reference load of 162kips (721 kN) for both test units afterthe first half of the 3 in. (76.2 mm)displacement cycle; the obtained values may be considered as the permanent displacements after this verticalcyclic displacement history.

The above-mentioned permanentresidual displacements were 1.17 and0.53 in. (29.7 and 13.5 mm) for TestUnits 100INT and IOOINTCIP, respectively. These displacement valuesrepresent two-thirds of the differencein permanent vertical displacementsexpected for the prototype structurebetween two sections at midspan andat one-third of the span; they are onlyreported here to show the effect ofcontinuous mild steel reinforcementover the joints on reduction of post-earthquake permanent displacements.

Performance of Joints

As mentioned earlier, Joints 12, 13and J4 opened during downward loading of Test Units 100INT andIOOINTCIP. Under upward loading ofUnit 100INT, only 13 opened, whereasTest Unit IOOINTCIP had severalclosely spaced cracks. Cracking of theconcrete cover adjacent to the joint

Epoxy Layer

Fig. 16. Cracking of concrete cover adjacent to a segment-to-segment joint.

Joint opening, 6bot ()20 30 40

z

0

0

-0.5 0.0 0.5 1.0 1.5

Joint opening, 6bot (in.)2.0 2.5 3.0

Fig. 1 7. Opening of midspan joint measured at bottom surface of the test units.

50 PCI JOURNAL

was observed rather than opening ofthe epoxy joints.

After termination of Test Stage II ofUnit 1 OOINT, one segment was cut asshown in Fig. 16 to investigate theopening of the epoxy joint versuscracking of the adjacent concretecover. Fig. 16 shows the epoxy layerand the failure surface in the adjacentconcrete cover. The dominant flexuralcrack adjacent to the joint occurredthrough the alignment and shear keys.

As mentioned earlier, the opening ofall joints was measured by means oflinear potentiometers. Examples ofjoint opening results are shown inFigs. 17 and 18, which provide thehistories of joint opening at midspanmeasured at the bottom and top surfaces of the test units, respectively.Fig. 17 shows that the joint opening inUnit 100INTCIP was wider than inUnit 100INT. Fig. 17 indicates thatpermanent joint openings (at referenceload level) of Unit 100INTCIP withmild steel reinforcement crossing thejoints were less than the permanentjoint openings of Unit 1 OOINT.

The effect of deck continuity onjoint opening is more pronounced inFig. 18, which shows the history ofjoint opening measured at the top surface of Units 1 OOINT and I OOINTCIP.Because Unit 100INT did not haveany continuous mild steel reinforcement in the deck, the joint openingwas significant. Despite its wide opening, the midspan joint of Unit 100INTwas closed completely upon unloadingduring each displacement cycle. Fig.18 shows the effect of continuousdeck reinforcement in controlling thewidths of cracks.

Fig. 19 shows the history of bendingmoment versus midspan joint rotationsmeasured for Test Units 100INT and100INTCIP. The joint rotation wasobtained from the joint openings measured at the top and bottom surfaces ofthe test units. The maximum rotationsof the midspan joint before failure ofTest Units 100INT and IOOINTCIPwere 0.035 and 0.05 1 radians, respectively.

Strains in Prestressing Steel

As mentioned earlier, electrical resistance gauges were used to monitor

the strains in the prestressing steel inTest Units 100INT and 100INTCIP.Unfortunately, all the strain gaugesmounted on the prestressing steel ofUnit 1 OOINT malfunctioned duringpost-tensioning. Some of the prestressing steel strain gauges malfunctioned during post-tensioning of TestUnit 100INTCIP.

Fig. 20 shows the strain history ofone of the surviving strain gauges in

the prestressing steel at midspan ofUnit 100INTCIP. The strain is plottedversus the number of loading cycles.In Fig. 20, Cycle 1 represents the timeduring which the strains wererecorded before starting Test Stage II.Fig. 20 also shows the strain in thetendon at midspan of Unit I OOINTCIPas obtained from the finite elementanalysis.

The yield strain level, represented

Joint opening, 6top (mm)20 30

‘a0,

0

. 00

I-’

z

0

01-’

-0.5 0.0 0.5 1.0

Joint opening, 8top

1.5 2.0 2.5

Fig. 18. Opening of midspan joint measured at top surface of the test units.

0,

I

5000

4000

3000

2000

I-1000

-2000

-0.06 -0.05 -0.04 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05 0.06

Joint rotation, 0 (iad.)

Fig. 19. History of midspan joint rotation versus bending moment.

March-April 2002 51

Fig. 20. History of strain

by a horizontal dashed line in Fig. 20,corresponds to the 0.2 percent offsetyield strain definition. The strain inthe tendon, r,, measured just beforestarting Test Stage II was about 5900micro strains. The correspondingstress, f (ksi), is calculated from thefollowing equation7 (assuming elasticmodulus, = 29000 ksi = 200 GPa):

0.975

10+ (118r) ]

(1)

The corresponding stress in the tendon before starting Test Stage II wasabout 171 ksi (1180 MPa) and the effective prestressing force was about588 kips (2616 kN). This measuredprestress force was close to the calculated 600 kip (2669 kN) force.

The strain in the tendon (see Fig.20) increased with applied loadinguntil the strain gauge malfunctionedduring the first cycle to 3 in. (76.2mm) displacement (downward loading). The strain recorded before malfunctioning of the strain gauge ex

ceeded 0.013 and the correspondingstress [Eq. (1)] was about 249 ksi(1717 MPa), or 92 percent of the ultimate tensile stress of the strands. It isbelieved that the stress in the tendonwas very close to the ultimate stress of270 ksi (1860 MPa) when the deckcompression failure occurred.

Fig. 20 indicates that plastic deformations of the tendon occurred beyond the 1.0 in. (25.4 mm) displacement cycles. Also, the strain inprestressing steel increased withdownward displacement and reducedat zero displacement.

In response to the upward displacements the strain also increased, but to asmaller degree than from downwardloading, and then reduced at zero displacement. The small increase in strainsduring upward loading was because theprestressing tendon was above the neutral axis, indicating that while the bottom soffit concrete strains were compressive the strains at the level of theprestressing steel were tensile.

Cracking Strength

The concrete cracking strength atjoint locations was determined usingthe known section properties, the experimental flexural moments at onsetof cracking, or joint opening, and theeffective prestressing force. Themidspan Joint J3 in Unit 100INT

opened under downward loading whenthe concrete reached a tensile stressof 3.25 (psi) [= 0.27.[j (MPa)].Joint J3 in Unit 100INTCIP opened at a concrete tensile stress of5.61 .,/j (psi) [= 0.47.m (MPa)]. itshould be mentioned that no tensilestresses are allowed to occur underservice loads according to Section9.2.2.2 of the AASHTO Guide Specification for Design and Constructionof Segmental Concrete Bridges.’

Cracking of the top surface in Unit100INTCIP occurred at a concrete tensilestiess of about 4.00J (psi) [= 0.33 j7,7(lViPa)1, which was relatively low considering the continuity of the deck. Theonset of cracking occurred in the deck atthe interface between the precast concrete and that of the cast-in-place deckclosure joint. The relatively weak interface between the precast and cast-in-place concretes resulted in this relativelylow cracking strength.

Flexural Moment Capacity

The calculated ultimate moment capacities of Test Units IOOINT andIOOINTCIP were 2993 and 2974 kip-ft(4058 and 4032 kN-m), respectively.The ultimate moment capacities ofboth units were determined accordingto the provisions of Article 9.17 of theAASHTO Standard Specifications .

The experimental peak flexural moments were 3126 and 3062 kip-ft(4238 and 4151 kN-m) for Units100INT and 100INTCIP, respectively.

The ratios of experimental to calculated moment capacity were 1.04 and1.03 for Units 100INT and 100INT-CIP, respectively. This indicates thatthe flexural strength of precast segmental bridge superstructures with internally bonded tendons can be accuratelyestimated using the equations of theAASHTO Standard Specifications.3

Deck Joint Reinforcement

As mentioned earlier, two differentdetails were used for the reinforcement of the cast-in-place deck joints(see Fig. 4) of Test Unit 100INTCIP.Bent hairpin bars were used on one-half of the cross section whereas barswith mechanical anchors (weldedheads) at their ends were used in thesecond half. Widths of cracks in the

30,000

o 25,000

20,000

024 6 8 10 12 14 16 18 20 22 24 26 28 30 32

Nuxther ofloading cycle

in prestressed stee’ at midspan of Test Unit 100INTCIP.

f,, = 29,000 r,1, 0.025 +

52 PCI JOURNAL

top surface of the deck were measuredat several locations in the two halvesof the cross section.

No difference was observed in performance of the two cross sectionhalves in terms of crack openings. Asmentioned earlier, failure of Unit100INTCIP was initiated by bucklingof deck mild steel reinforcing barswhich pushed the concrete cover awayand resulted in the deck compressionfailure.

Buckling of deck reinforcementcould have been prevented by use ofstirrups that enclose the top and bottom longitudinal bars of the deck. Although there is no difference betweenheaded and bent hairpin bars in termsof structural performance of the joints,use of headed bars would speed up theconstruction and make it easier especially if stirrups are used to enclosethe deck top and bottom reinforcingbars as recommended above.

FINITE ELEMENT ANALYSISThree-dimensional finite element

models of the test units were developed. The finite element models andthe major finite element results arepresented in this section.

Description of theFinite Element Models

Detailed finite element models weredeveloped for both test units (see Fig.21). Analyses were conducted usingthe general-purpose finite element program ABAQUS,8 interfaced with theANACAP9concrete material model.

The concrete was modeled as 3-D,eight-node, solid brick elements withstrain-hardening and strain-softeningcapabilities in compression, and tension cutoff with cracks that do notclose upon closure.9 Confinement effects were assumed to be negligibleand the unconfined concrete strengthwas taken as 7.5 ksi (51.7 MPa). Themodel was developed in a similar wayto the test units, with no solid elements crossing the joints between theprecast segments and no connectionbetween solid elements on either sideof the joints.

The joints were free to open by providing double nodes and compression-

only springs at all nodes in the crosssection at locations of joints. Prestressing steel was modeled by trusselements and connected to the concrete nodes at each 12-in. (305 mm)cross section, representing bondedstrands. All mild steel reinforcementwas modeled as 1-D sub-elements inthe solid concrete elements.

No mild steel reinforcement crossedthe joints of Unit IOOINT and thusthey served only to prevent failurewithin the precast segments and hadno impact on the overall load-displacement response of the structure.Test Unit 100INTCIP with the cast-in-place deck joint, required mild steelreinforcing bars to be placed acrossthe joints at the deck level; these mild

steel bars were activated and yieldedunder upward loading of the test unit.

At the joints, the prestressing steelwas not connected to the center nodes,but to nodes at sections 12 in. (305mm) on either side of the centerline.This represented an idealized unbonded length at the joints of 24 in.(610 mm), which was consistent withthe assumed plastic hinge length usedin designing the test units.4 Loadingwas applied to the models in displacement control as shown in Fig. 9.

Finite Element Results

Test Unit 100INT — Analysis results of Unit 100INT are given inFigs. 22 and 23. There was very close

only

Concrete solidelements

(a) Test Unit 100INT at 3 in. (76.2 mm) upward midspanvertical displacement

Compression onlyspring elements

(b) Test Unit 100INTCIP at 6 in. (152 mm) downward midspanvertical displacement

Fig. 21. Finite element models and deformed shapes of the test units: (a) Test Unit100INT at 3 in. (76.2 mm) upward midspan vertical displacement; and (b) Test Unit1 OOINTCIP at 6 in. (152 mm) downward midspan vertical displacement.

March-April 2002 53

agreement between the analysis andmeasured load-displacement responses(see Fig. 22). Analysis results showedthat the model behaved very similarlyto the test unit in terms of ultimateload, displacement at failure, andshape of the hysteretic response inboth the upward and downward loading directions. Rupture stress andstrain of the prestressing steel were assumed as 270 ksi (1860 MPa) and

0.04, respectively, which appeared tobe reasonable based on Unit 100INTload-displacement results in Fig. 22.

Strains of the prestressing steelcould be obtained from the finite element analysis but they are not shownhere because they had the same trendand observations as the strains obtained from the analysis of Unit100INTCIP (see Fig. 20). Fig. 23shows the prestressing steel stress his-

tory (stress versus loading cycle number) at midspan (Joint J3, see Fig. 2)from the analysis of Test Unit100INT. Plastic deformations of thetendon occurred beyond the 1 in. (25.4mm) displacement cycle. Of interestwith the permanent plastic deformations is that with increasing, unrecoverable, strains, the initial prestressforce was lost (see Fig. 23).

It is clear from Fig. 23 that beyond1 in. (25.4 mm) displacement, the initial prestress force started to reducesignificantly, and was completely lost(at zero displacement, or at the reference load level) following 4 in. (102mm) of downward displacement. Thismay have important consequences inthe design of precast concrete structures, where under a severe earthquakethe prestressing force is diminished orlost altogether before rupture of thestrand.

Test Unit 100INTCIP — Samplesof analysis results of Test Unit100INTCIP are given in Figs. 20 and24. Although the shape of the mono-tonic load-displacement results for theupward and downward loading directions matched the test results verywell, the cyclic analysis results did notfollow as closely. Failure from bothmonotonic and cyclic analyses wasfrom rupture of the prestressing steel,whereas the observed collapse duringthe experiment was from compressionfailure of the deck at the midspan cast-in-place joint.

As mentioned earlier, buckling ofthe cast-in-place deck reinforcementinitiated the deck compression failureof Unit 100INTCIP. The analysismodel was not able to catch this typeof failure because the cast-in-placedeck steel was modeled by 1 -D trusselements, which do not have the ability to buckle. It may be of interest tomodel the cast-in-place deck steel withbeam elements to allow buckling ofthe reinforcement.

Although the mode of failure wasdifferent than observed, the force anddisplacement at failure from the cyclicanalysis matched the measured resultsvery closely. Note that the monotonicanalysis showed rupture of the prestressing steel at a much earlier displacement, demonstrating that theanalysis model was able to spread the

‘ 200

0

.

01—

600 I I

I Pp PP Strand rupture \ Strand rupture

500 44’ 44 monotonic analysis)’ / (experiment)

4OOiH 1 SandruptureF

3001

-a’ 44,.- 1’

(cyclic analysis)

‘°:

Reference load leve

-100Downard displacement

-200— Finite element (Cyclic)——— Finite element (Monotonic)

-300’ Experiment

-400

Vertical displacennt, A (mn)-150 -125 -too -75 -50 -25 0 25 50 75 100 125 150 175 200

- 2500

___________________

- 2000

- 1500

1000z

5000

01-

- -500

- -1000

• -1500

-6 -5 -4 -3 -2 -l 0 1 2 3Vertical dispJacennt, A (in.)

4 5 6 7

Fig. 22. Load versus displacement analysis results of Test Unit 100INT.

300

275

250

225

200

175ft

150fO

C’D 125

100

75

50

25

0

0 2 4 6 8 10 12 14 16 18 20 22

Number ofloading cycles24 26

Fig. 23. Stress history analysis results of prestressing steel at midspan ofTest Unit 100INT.

54 PCI JOURNAL

strain penetration out further with increased number of cycles, due tocracking and damage of the elementsin the vicinity of the prestressing steel.

It is not clear at this point why theunloading stiffness was different during load reversal (may be due to theassumed prestressing steel model) andwhy, in the upward loading direction,the force at increasing peak displacements reduced and diverged from thetest results and those of the monotonicanalysis. The results from monotonicanalyses clearly demonstrate that thebasic model was correct.

Strain history results are provided inFig. 20. One strain gauge on the prestressing steel at the midspan joint remained intact to the first cycle to 3 in.(76.2 mm), allowing comparisons tothe cyclic strain analysis results atleast to 2 in. (50.8 mm) of downwardand upward displacement. The experimental and finite element variations ofthe strain with increased loading hadthe same trend as Fig. 20 indicates.

DESIGN IMPLICATIONS

This paper presents the experimental and analytical results of the testunits with internally bonded tendonsonly. Definitive recommendations forseismic design of precast segmentalbridge structures can only be madeupon completion of the three phases ofthe research project. However, based

on the results presented in this paper,

the following can be implied for seismic design of precast segmentalbridge superstructures post-tensionedwith internally bonded tendons:

Opening of the epoxy-bondedjoints, or cracking at the joint loca

tions, occurs when the concrete

reaches a tensile stress of about3J (psi) [= 0.25f (MPa)].Cracking of the deck with cast-in-place closure joints occurs at a relatively low concrete tensile stress ofabout 4j (psi) [= 0.33 J7 (MPa)].

• The flexural capacity of precast

segmental bridge superstructures canbe accurately predicted using the provisions of Article 9.17 of the

AASHTO Standard Specifications.3• Finite element analyses showed

that the effective prestressing force reduces after earthquake occurrence, es

March-April 2002

pecially if the superstructure segment-to-segment joints are subjected to significant joint openings or rotationsduring the earthquake (see Fig. 23).External post-tensioning may be agood alternative in which case less reduction in the effective prestressingforce is expected in external tendons.

To prevent the explosive compression failure of precast concrete superstructures with cast-in-place deck closure joints, the deck top and bottomlayers of longitudinal mild steel reinforcement should be enclosed bymeans of closed stirrups in the cast-in-place closure zone. The same shouldbe done if the ductility of the superstructures needs to be increased.

CONCLUSIONSA large-scale experimental program

and finite element studies are currently in progress to investigate theseismic performance of segment-to-segment joints of prestressed, precastsegmental bridge superstructures. Theexperimental program and the analytical model calibrations of two test unitswith internally bonded tendons arepresented in this paper. The joints ofthe first test unit were epoxy bondedwith no mild steel reinforcementcrossing the joints, whereas the secondtest unit had reinforced cast-in-place

deck closure joints with the remainingportions of the joints connected byepoxy.

The following conclusions can bedrawn:

1. Opening of an epoxy-bondedjoint occurs due to cracking of theconcrete cover adjacent to the jointrather than opening of the epoxy joint.The concrete cover adjacent to thejoint has relatively low crackingstrength compared to the concrete ofthe precast segments. The dominantflexural vertical crack adjacent to thejoint occurs through the alignment andshear keys.

2. Crack patterns for both units weresimilar under downward loading. Because of the mild steel reinforcementcrossing the joints, the deck of the second test unit experienced severalclosely spaced and relatively narrowcracks under upward loading in lieu ofone single wide crack at the midspanjoint of the first test unit with no mildsteel reinforcement crossing the joints.

3. The segment-to-segment jointscan experience significant repeatedopenings and closures under reversedcyclic loading without failure even ifthere is no mild steel reinforcementcrossing the joints. However, permanent deformations and joint openingsare reduced if there are mild steel reinforcing bars crossing the segment-to-

fa

w

Reirence load leve1

Vertical displacennt, z (aim)-150 .125 .11)0 .75 .51) -25 0 25 50 75 100 125 150 175 200

600I I

Strand rupture—s Deck compression 2500

500

____

Ii 2000

400 IIAk r*,

1500300

1000 —.

m200 Z

500.lo0

::: -sooH

.300

________________________________

(iJ

Downward displacementI,li

— Finite element (Cyclic)Finite element (Monotonic)Experiment

I I I I

-6 -5 -4 -3 -2 4 0 1 2 3Vertical displacement, t (in.)

.1000

-1500

4 5 6 7 8

Fig. 24. Load versus displacement analysis results of Test Unit 100INTCIP.

55

segment joints. The cast-in-place deckjoints, similar to those proposed forthe new East Span Skyway of the SanFrancisco-Oakland Bay Bridge, enhance the seismic performance of precast segmental bridges in terms of energy dissipation and reduction ofpermanent displacements and permanent joint openings.

4. The first test unit failed by rupture of the prestressing tendon,whereas compression failure occurredin the deck of the second test unit following buckling of mild steel reinforcement of the cast-in-place deckjoint.

5. The seismic response of precastsegmental bridge superstructures withcast-in-place closure joints will notdiffer if headed or bent hairpin bars

are used as longitudinal reinforcementin the closure joints. However, headedbars are recommended over bent hairpin bars for construction reasons.Buckling of the deck longitudinal reinforcement should be prevented bymeans of closed stirrups that confinethe top and bottom reinforcing bars.

6. Finite element analyses showedthat under severe earthquake loading,the prestressing force in the tendonscould diminish under repeated cyclingin the inelastic strain range.

ACKNOWLEDGMENTSThe authors want to express their

gratitude to Caltrans for funding ofthis research program under ContractNo. 59A0051.

Members of the AASHTO-PCIASBI technical research committeeare acknowledged for participating inthe development of this research project.

I. Muller International, San Diego,prepared the construction drawings ofthe test units.

Sika Corporation donated the epoxy.Dywidag-Systems International

(DSI), USA, donated the post-tensioning strands, high-strength bars andhardware.

Headed Reinforcement Corporation(HRC), California, donated the headedbars.

The authors wish to express theirappreciation to the PCI JOURNAL reviewers for their valuable and constructive comments.

REFERENCES1. AASHTO, Guide Speccation for Design and Construction of

Segmental Concrete Bridges, Second Edition, American Association of State Highway and Transportation Officials, Washington, DC, 1999.

2. AASHTO-PCI-ASBI, Segmental Box Girder Standards forSpan-by-Span and Balanced Cantilever Construction, American Association of State Highway and Transportation Officials, Washington, DC, Precast/Prestressed Concrete Institute,Chicago, IL, and American Segmental Bridge Institute,Phoenix, AZ, 1997.

3. AASHTO, Standard Speci.fications for Highway Bridges, 13thEdition, American Association of State Highway and Transportation Officials, Washington, DC, 1983.

4. Dowell, R. K., Silva, P. F., and Seible, F., “Precast Segmental

Bridge Performance in Seismic Zones: Phase I, Test Unit Design,” Progress Report submitted to Caltrans, April 2000, 30 pp.

5. ASBI, Recommended Contract Administration Guidelines forDesign and Construction of Segmental Concrete Bridges, American Segmental Bridge Institute, Phoenix, AZ, March 1995, pp.109-116.

6. Chopra, A. K., Dynamics of Structures, Prentice Hall, Inc.,Princeton, NJ, 1995, 729 pp.

7. Collins, M. P., and Mitchell, D., Prestressed Concrete Structures, Response Publication, Toronto, Canada, 1997.

8. Hibbitt, Karlson, and Sorenson, ABA QUS User’s and TheoryManual, Version 5.8, 1999.

9. ANATECH Corporation, ANA CAP-U User’s and Theory Manual, Version 2.5, 1997.

56 PCI JOURNAL