Behavior of Precast, PrestressedConcrete Pile to Cast-in-PlacePile Cap Connections

Kent A. Harries, Ph.D.Assistant ProfessorDepartment of Civiland Environmental EngineeringUniversity of South CarolinaColumbia, South Carolina

Michael F. Petrou, Ph.D.Associate ProfessorDepartment of Civil andEnvironmental EngineeringUniversity of South CarolinaColumbia, South Carolina

This investigation studied the capacity of squarepile-to-pile cap connections where the precast,prestressed pile is simply embedded in the cast-in-place pile cap. Both experimental and analyticalresults are presented. It is shown that the plainembedment can develop the flexural capacity ofthe pile without distress to the pile cap orconnection region provided that a sufficientembedment length is furnished. Equations fordetermining the required embedment length areprovided. For design purposes, it is recommendedthat a plain embedment length equal to the widthof the embedded pile be used.

piles, particularly those embedded in soft soils, may besubjected to large lateral deflections in the event of anearthquake. The lateral deflections can result in high

local curvature and moment demands at various locationsalong the pile length as shown in Fig. 1. Of particular concern is the behavior at the pile-to-pile cap interface.

At this location, very high moment demands result fromthe assumed fixity of the pile-to-pile cap connection. Inorder for this behavior to occur as assumed, the connectionmust be able to transmit lateral forces to the pile and remainessentially rigid. For this discussion, it is assumed that thepile cap may translate but not rotate. If rotation is permitted,the demands on the connection are reduced.

82 PCI JOURNAL

Severe pile damage has been observed in past earthquakes.”2Pile design for seismic loading assumes thatthe pile can develop and maintain itsmoment capacity through large deformation demands. Indeed, significantresearch1’2has shown that well-detailed precast, prestressed piles can develop and maintain large moments.

Although a pile may be detailed toresist large forces, it is also necessarythat the pile-to-pile cap connection beable to transfer these forces. There areonly a few published investigationswhich report the behavior of the pile-to-pile cap connection. These studiesare summarized further on in thispaper.

The objective in designing the pile-to-pile cap connection is to provide aconnection capable of developing themoment demands on the pile while remaining essentially rigid. Conservatively, this requires the connection tobe able to develop the theoretical capacity of the pile while remaining elastic. In this paper, the specific case ofprecast, prestressed piles embedded incast-in-place pile caps is considered.

PILE EMBEDMENT DETAILSThere are a number of options for

detailing pile-to-cap connections. Fig.2 shows the connection detail currentlyused by the South Carolina Department of Transportation (SCDOT). It isreported that this detail costs close to$800 per pile to fabricate. The objective of the study presented here was toaddress this issue and determine if lessexpensive details could provide adequate lateral load resisting capacity.

There are a variety of details proposed and used in the embedment region of piles in cast-in-place pile caps.Fig. 3 shows a number of these detailswhich are described as follows:

A. No treatment; the pile is simplyembedded in the pile cap.

B. Roughening the exterior of thepile (using a rotary or chipping hammer, for instance) to provide additional mechanical bond between thepile and pile cap.

C. Grooving (cut or cast in place)the pile surface to provide additionalmechanical bond.

D. Embedding vertical dowels in thedriving head of the pile (after driving).

Fig. 2. Pile anchorage detail required by SCDOT.

E. Drilling horizontal dowelsthrough the pile.

F. Confining the immediate embedment region with hoop or square spiralreinforcement.

G. Confining the immediate embedded region with round spiralreinforcement.

H. Exposing the strands and embedding them in the cast-in-placeconcrete. Often, the wires will be

“broomed” (separated) or twistedopen to form an annular space (aso-called “olive” anchorage) to improve their development.

Typically, embedments will includea combination of these details. For example, the SCDOT detail shown inFig. 2 incorporates Details B, D, andG. Each additional detail has an associated cost in terms of both money andtime.

(a) Pile partially exposed (b) Pile embedded in soil

Fig. 1. Bending of long piles due to horizontal ground motion (adapted from Joenand Park,’ 1990).

16”

36”

#3 square spiral

12” to 20”

Surface of pilerougbened to 0.25”amplitude overentire embedment

July-August 2001 83

Studies of Pile Embedment Details

Joen and Park’ reported tests of sixpile-to-pile cap connection types. Thepiles were tested under combined axialload and reversed cyclic lateral loads.The axial load was kept constant at0.2Agfc’ for all tests. All piles testedwere 16 in. (406 mm) octagonal piles.

Two specimens were provided witha 32 in. (813 mm) embedment havingDetails B and G (see Fig. 3). Anothertwo specimens had 24 in. (610 mm)embedments with Details G and H(exposed strand left straight). A fifthspecimen had a 36 in. (914 mm) embedment with Details G and H (exposed strand provided with an “olive”anchorage). The final specimen wasprovided with only a 2 in. (51 mm)embedment and Details D and G.

The theoretical capacity of the pilewas obtained in each test and only thesixth detail showed significant distressto the pile cap and the embedment region, which led to a significant decay ofthe load-deflection response of the pile.’

Sheppard2summarized the results ofboth experimental and post-earthquake

field investigations. Two embedmentdetails were presented as being adequate for the pile-to-pile cap connection to behave in a desirable manner.The first suitable detail is H; the second is D. It is implied that confiningDetail G is also provided.

Curiously, the details presented bySheppard show minimal embedmentof the pile, similar to the sixth specimen described by Joen and Park.’ Assuch, it would appear that the detailsrecommended by Sheppard may beinadequate for severe seismic loading. No experimental results concerning this aspect are reported bySheppard.

quate to develop the theoretical moment capacity of the pile. Note that thescope of this study is restricted todriven precast, prestressed piles embedded in cast-in-place pile caps.

There are two proposed models,namely, Mattock and Gaafar3 andMarcakis and Mitchell,4 for determining the capacity of the pile-to-pile capconnection. Both models assume thata rigid body (pile) is embedded in acast-in-place concrete monolith (pilecap). Both models are based on themobilization of an internal momentarm between bearing forces C and C,,as shown in Fig. 4.

Mattock and Gaafar3

A parabolic distribution of bearingstresses is assumed for Cb, and Cf iscomputed by a uniform stress equal to0.85f. The bearing stresses are disthbuted over the width of the embedded pile, b. Following these assumptions and calibrating the calculatedstresses against experimental data, therequired embedment length, Le, maybe determined from:

PLAIN PILE-TO-PILE CAPEMBEDMENT

The objective of this study is to investigate the behavior of a plain embedment (Detail A in Fig. 3). In thisdetail, the capacity of the pile is developed along the length of the embedment. The pile-to-pile cap connectionshould be designed such that it is ade

Fig. 3. Proposed pile embedment details.

84 PCI JOURNAL

0.66

I3ibLegiven by Mattock and Gaafar as thewidth of the element into which (inthis case) the pile is embedded.

This value is intended to account for

thespreading of the compressive

urn sstresses away from the embedment asindicated in Fig. 4(b). For a single pile

(la)in a pile cap, this value is taken as thewidth of the pile cap. For a pile group,this value may be conservatively takenas the pile spacing.

Marcakis and Mitchell4

Using slightly different assumedstress distributions shown in Fig. 4(b),Marcakis and Mitchell4 proposed thefollowing expression for determiningthe required embedment length, Le. ThiS

expression has also been calibratedagainst experimental data:

v = 0.85f’ b’(Le — c)3 .6e

1+Le - C

(consistent units)

where e is the eccentricity from thepoint of zero moment to the center of

the effective embedment as shown inFig. 4(b).

Marcakis and Mitchell define b’based on a “strut-and-tie” approach asbeing the effective width to the assumed “tie” steel, limited by a valueof 2.5b [see Fig. 4(b)].

Eq. (2) has been adopted in Chapter 6 of the PCI Design Handbook6for the design of embedded structuralsteel haunches or brackets in precastconcrete. The same method has alsosuccessfully been applied to the design of moment-resisting connectionsfor steel beams embedded in concrete core walls7 and may be reasonable extended to the embedment ofany essentially rigid body in a concrete embedment.

EXPERIMENTAL PROGRAMThe objective of this experimental

program is to demonstrate that no special details are needed when embed-

(2) ding prestressed piles into cast-in-place pile caps provided that theembedment length is sufficiently long.The pile-to-pile cap connection must

Fig. 4. Analytical methods for determining capacity of embedment.

066

/3ibLe

0.58 — 0.22/3a

‘ (MPa units)0.88 +

L-c

(1 b)

where a is the shear span of the pile(distance from pile cap to assumedpoint of zero moment) and f3 is theconcrete stress block factor defined inACT 3 19-99, Section l0.2.7.3.

It is suggested that the shear span beincreased by an amount equal to theconcrete cover, c, to account for possible spalling of the soffit of the pile capas shown in Fig. 4. The value of b’ is

L - c aJ

Spalled coverp concrete

,frU

L

load spreading

L,

LIII IIC

hPoint of

zero moment

Assumedstrain

distribution

6b

zero moment b’ <2.5b

g1= 0.003

1/3(LeC)

XI

Xi E=0.003

0.85f’

(a) Mattock and Gaafar (1982)

Assumed ctfstress f3x6

distribution

f3, Xf

— 0.85f

(b) Marcakis and Mitchell (1980)

July-August 2001 85

- .

Fig. 5. Prestressed concrete pile [18 in. (450 mm) square] used in present study.

be adequate to develop the probablemoment capacity of the embedded pilewithout significant deterioration.

The connection must remain stiffenough so that rotation of the pilewithin the embedment does not contribute significantly to the overall driftof the pile-to-pile cap assembly. Additionally, the deterioration of the connection region should not allow theexpected hinging of the pile to migrateinto the embedded region.

In this program, piles are embeddedinto cast-in-place pile caps. The embedded portions of the piles are notprepared in any particular way; theyare simply placed within the pile capforms and the pile cap concrete isplaced around them. No dowels or

confinement reinforcement beyondthat provided by the cast-in-place pilecap are provided.

Pile Details

Two 18 in. (450 mm) square by 18ft (5.49 m) long piles were fabricatedsimultaneously in the 40 ft (12.2 m)prestressing bed located in the University of South Carolina Structures Laboratory. These piles were prepared foran ongoing study of strand slippage.8The piles used in this study had thelowest observed strand end slip of all22 piles available to this experimentalprogram.

The 28-day concrete compressivestrength of the Type I ready-mixedconcrete was 6700 psi (46.2 MPa). The

piles were prestressed with eight ‘2 in.(12.7 mm) diameter low-relaxationstrands. Each strand had an initial prestressing force of 31 kips (138 kN),equivalent to 0.75f. The strands werereleased 43 hours after casting whenthe concrete compressive strength was4500 psi (31 MPa). The strands wereconfined with W6 (0.276 in.) plainwire spiral. The strand layout and spiral details are shown in Fig. 5.

The predicted nominal moment capacity of the piles corresponding to anaxial load of 200 kips (890 kN) is 285ft-kips (386 kN-m). All predictionspresented in this paper were madeusing the plane section analysis program RESPONSE-2000.9

Pile Cap Details

Fig. 6.Detail of

24 in. (610 mm)embedment.

Each pile was embedded in a 7 x 3 x7 ft (2.14 x 0.92 x 2.14 m) cast-in-place pile cap. Each pile cap was reinforced with No. 7 bars on the top andbottom and No. 3 ties at 6 in. (152mm) spacing in the transverse direction and through the depth of the pilecap (see Fig. 6).

The concrete compressive strengthand primary reinforcing details of eachpile cap was different for each test.The details of the first pile test are representative of what may be expectedin the field in South Carolina. The details for the second test were purposely selected to represent very poorin situ conditions. Pile cap data areprovided in Table 1.

L4 ...L 6” 6” ,_,3”J

rT T Ti3’,

5 tUI11S

11 r(th1.

6”

ims @ 1.5” (5” pitch beyond)

6”

respiml

3,,

1/2’ dia. low relaxation strandtensioned to 31 kips each

Driving Head

i’k86 PCI JOURNAL

Embedment Details Table 1. Pile cap and embedment details and requirements.

The embedment length of Pile No.I was selected to be 24 in. (610 mm),the value desired by the SCDOT. Theembedment length of Pile No. 2 was18 in. (457 mm). This value was feltby the SCDOT to be the minimumacceptable embedment length. Required embedment lengths calculatedfrom Eqs. (1) and (2) are shown inTable 1.

Based on current practice, and assuming typical pile cap material properties and details, an embedmentlength of approximately 12 in. (305mm) is required to develop the 285 ft

kip (386 kN-m) capacity of the pilesused. These calculations are based ona shear span (a in Fig. 4) equal to 12 ft(3.66 m). Based on these calculations,it is expected that the embedment

lengths provided are sufficient to develop the piles used. A photograph ofthe embedment region of Pile No. 1prior to casting the pile cap is shownin Fig. 6.

Fig. 7. Test setup to simulate seismic loading of pile-to-pile cap assembly.

I Concrete Embedment Required embedment length

I compressive length —

Specimen Primary steel jstrength provided Eq. (1) Eq. (2)

Pile No. 1 6 - No. 7 bs 5000 psi 24 in. 12.4 in. 11.9 in.

top and bottom (34.5 MPa) (610 (316 mm) 5303 mm)

Pile No. 2 4-No. 7 bars 3000 psi 18 in. 14.0 in. 14.8 in.top and bottom (20.7 MPa) (457 mm) (356 mm) (376 mm)

C.LP.pile cap

Lateral load reaction frame

6Okip lateralload rams

J9restressea nile

146”

4-

(a) Elevation view

Axial loadstrongback

200 kip axialload rams

(b) Plan view

F axial loadP0.lfc’AgP=200kips

(c) Test set-up.

July-August 2001 87

Test Setup

The pile-to-pile cap assembly wastested as a cantilever beam in the horizontal position. The test setup isshown in Fig. 7. Each pile-to-pile capassembly was tested under constantaxial load and reversed cyclic lateralload. The 200 kip (890 kN) axial load,equal to 0.O92Agfc’, was applied usinga load following mechanism that doesnot impose secondary moment effectson the column.

The axial load was kept constantthroughout the test using a regulatedhydraulic power supply. The reversedcyclic lateral loads were applied at adistance of 146 in. (3.7 m) from thesoffit of the pile cap. This loading isrepresentative of a pile having a shearspan of 146 in. (3.7 m) or point of fixity approximately 24 ft (7.3 m) belowthe pile cap soffit.

This shear span was thought to be typical of partially exposed 18 in. (457 mm)piles [see Fig. 1(a)]. Shorter piles willplace less demand on the embedment region and longer 18 in. (457 mm) pilesare not common. Pile deflections wererecorded at the point of application ofthe lateral load.

Load History andExperimental Observations

The lateral load is cycled threetimes at each load or displacementlevel. The load-deflection response ofeach pile is shown in Fig. 8. The predicted pile moment capacity of 285 ftkips (386 kN-m) is shown as a horizontal dotted line. The piles werecycled at two elastic load levels; ±7kips (±85 ft-kips) and ±14 kips (±170ft-kips) [(±3 1 kN (±115 kN-m) and±62 kN (±230 kN-m)].

In each test, the piles were first observed to crack at the pile-to-pile capinterface at an applied load of 13.9kips (62 kN), corresponding to an applied moment of 169 ft-hips (230 kNm) at the crack location. The predictedload to cause cracking of the piles is136 ft-kips (185 kN-m).

Loading continued to the “yield”load level. Yielding was defined as asignificant change in the load-deflection response of the pile. The yielddisplacement, 6, was found to be approximately I in. (25.4 mm), corresponding to a pile drift of 0.7 percent.The applied load to cause a deflectionof 6 1 in. was 20.2 kips (90 kN) or246 ft-hips (332 kN-m) and 17.8 hips(79 kN) or 217 ft-hips (293 kN-m) forPile No. 1 and No. 2, respectively.

The predicted moment to cause significant nonlinearity of the pile sectionresponse is 240 ft-hips (326 kN-m). Atthis load, cracks were observed in thepiles at approximately 12 and 26 in.(305 and 660 nun) from the pile capsoffit.

Beyond yield, the piles were cycledthree times each to displacementsequal to 1.56, 26 and 36. Pile No. 2was also cycled to 2.56 in sequence.No lateral or axial load capacitydecay or stiffness degradation wasnoted through these cycles as can beseen in Fig. 8. The peak load valuesrecorded during testing were 24.9kips (111 kN) or 303 ft-kips (410 kNm) and 21.5 kips (96 kN) or 262 ftkips (354 kN-m) for Pile No. I andNo. 2, respectively.

‘While attempting the initial cycle to46, Pile No. 1 experienced a flexuralfailure due to crushing of the extremecompression concrete (see Fig. 9). Atthis point, the axial capacity of the pilewas affected. The axial load regulatorwas unable to sustain the 200 hip (890kN) axial load without further drivingthe compressive failure. The finalmonotonic load response to a peak deflection of approximately 5 in. (127mm), corresponding to a 3.4 percentdrift, is shown in Fig. 8(a).

Pile No. 2 was cycled once at 3.56and twice at 46 before failing whilebeing pushed to 56 for the first time.In this case, the failure appeared to berupture of at least one of the strands.After testing, the pile was broken off

30

20

10

ci

0

-o.2 -10

-20

-30

30

20

.10

0

0.2 -10

-20

24

tip displacement, A (in.)

0 2

tip displacement, A (in.)

Fig. 8. Applied load versus deflection responses for piles tested.

88 PCI JOURNAL

from the pile cap and no strand ruptures were evident. It is believed, inthis case, that the loud noise and accompanying decrease in lateral loadthat occurred on the cycle to 5(5 mayhave resulted from the catastrophicslipping of a strand. Two views of PileNo. 2 near the end of testing areshown in Fig. 10.

Pile No. 2 behaved somewhat differently from Pile No. 1. Most of thedeflection of Pile No. 2 was accountedfor by the significant opening of thecrack at the pile-to-pile cap interfaceas shown in Fig. 10(a). The deflectionof Pile No. 1, on the other hand, derived from the more uniform openingof cracks at the interface and at 12 and16 in. (305 and 406 mm) from thepile-to-pile cap interface. It is believedthat this differing behavior is entirelydue to the piles and is not related tothe pile-to-pile cap connections.

Pullout Test

It is possible that piles may experience tensile loads during a seismicevent. It is clearly undesirable for thepile to separate from the pile cap inthese instances. The question arises asto whether the cyclic loads imposedon the embedment region causes a“ratcheting” induced degradation ofthe embedment region resulting in the

possibility of the pile pulling out ofthe pile cap.

To test this hypothesis, an ad hocpullout test was devised. After thecompletion of the reversed cyclic loadtests, Pile No. 1 was fitted with a collar and an attempt was made to remove the pile from the pile cap. A direct axial tension of 75 kips (334 kN)was applied to the pile. There was noevidence of distress or movement ofthe pile away from the pile cap at thisload. This test was not repeated forPile No. 2 since it was assumed that astrand had been ruptured.

Fig. 9.Pile No. 1[24 in. (610 mm)embedment] atdisplacement of÷5 in. (127 mm)(3.4 percentdrift).

SUMMARY OFEXPERIMENTAL RESULTS

Each pile behaved very much as wasexpected. Observed cracking, yielding,and ultimate capacities were very closeto those predicted for the 18 in. (457mm) piles. There was no observabledamage to the embedment region in either test. Pile curvature data measuredrelative to the pile cap indicated thatneither rotation of the pile cap nor rotation of the embedment region had asignificantly measurable effect on therecorded deflections of the pile.

Fig. 10. Pile No. 2 near end of test.

(a) Pile No. 2 [18 in. (457 mm) embedment] atdisplacement of -4.5 in. (114 mm) (3.1 percent drift).

(a) Pile No. 2 [18 in. (457 mm) embedment] atdisplacement of +5.5 in. (140 mm) (3.8 percent drift).

July-August 2001 89

In essence, the pile cap provided thedesired rigid end condition for thecantilevered pile. It is clear that the 24and 18 in. (610 and 457 mm) embedments provided were sufficient to develop the capacity of the piles with nospecial embedment details. This conclusion was expected based on past research of embedded members in cast-in-place concrete.3’4’7

For practical reasons, it is not believed that embedment lengths shorterthan 18 in. should be specified. TheSCDOT expects the tolerance on embedment lengths to be ±6 in. (±152mm). Similarly, embedment lengthslonger than 24 to 30 in. (610 to 762mm) are also impractical without significantly affecting the design andconstruction of pile caps. For instance,pile caps in South Carolina have beenstandardized to be 36 in. (914 mm)deep, making the longest practical embedment length approximately 30 in.(762 mm).

Application to Other Pile Sizes

The experimental program has indicated that practical construction issuesare more likely to control the specified embedment length than are capacity requirements. Fig. 11 shows

curves generated from Eqs. (1) and(2) for the moment capacity of theembedment for varying embedmentlengths and square pile sizes. Figs.11(a) and (b) show the capacity toembedment length relationships forpiles having a shear span of 12 ft(3.66 m) [similar to Fig. 1(a)], whileFigs. 11(c) and (d) show the similarrelationship for piles having a shortshear span of 4 ft (1.22 m) [Fig. 1(b)].

Note that for all calculationsshown in Fig. 11, the concrete compressive strength, f, of the pile caphas been assumed to be 5000 psi(34.5 MPa), the load spreading factor[see Fig. 4(b)] is 2 (thus b’ = 2b) and3 in. (76 mm) of concrete cover hasbeen assumed.

Both Eqs. (1) and (2) yield similarresults. Eq. (2) tends to result inslightly more conservative embedmentcapacity values.

The calculated embedment capacities are lower for shorter shear spans,where the shear-to-moment ratio at thepile-to-pile cap interface is high [seeFigs. 11(c) and (d)]. As the shear spanincreases, the embedment capacity increases at a decreasing rate. For thegeometry shown in Fig. 11, increasingthe shear span beyond 12 ft (3.66 m)

has little effect on the capacity of theembedment.

Fig. 12 shows the same resultsfrom Eqs. (1) and (2) for 18 and 36in. (457 and 914 mm) square piles.Superimposed on these relationshipsis the range of probable pile momentcapacities.

Based on the data shown in Figs.11 and 12, it is proposed that providing a minimum embedment lengthequal to the width of the pile willconservatively result in an embedment having sufficient capacity to develop the pile. Furthermore, such anembedment may reasonably be assumed to provide a rigid end condition for the top of the pile. Certainly,the pile embedment requirement maybe significantly reduced for piles having a long shear span.

Additional Embedment Details

The inclusion of additional embedment details, such as those discussedpreviously and shown in Fig. 3, willincrease the capacity of the embedment to some extent. For instance, themethod for designing embedded steelhaunches contained in the PCI DesignHandbook includes guidance for determining the additional embedment

(a) Equation (1)f’5000psi

— ;7 ‘b”2b;c3in, — - -

shear span, a=l2fe 24’

Fig. 11. Embedment capacity predictions of varying pile sizes having varying embedment lengths.

90

6000

__

5000

4()

3000

2000

E “ 1000

500

18

EzL2’

4l7,... ‘-‘ 5000‘

333, ‘3,4000

250a)

167q

“12 18 24

83

embedment length, Le (inches)I

embedment length, Le (inches)

embedment length, Le (inches) embedment length, Le (inches)

PCI JOURNAL

capacity resulting from the inclusionof horizontal dowels [see Fig. 3(e)]. Itis felt, however, that these additionaldetails will not significantly enhancethe capacity of the embedment eventhough they may be beneficial in providing post-failure continuity in theevent of extremely large lateral deflections of the pile.

Application to Other Pile Shapes

The previous discussion applies tosquare piles. It is felt, however, thatthe discussion may be extended toother typical pile shapes using the projected pile width in place of the squarepile dimension b. It is cautioned, however, that unlike square piles, round oroctagonal piles will develop bearingforces directed radially from the embedment. This may result in greaterdeterioration of the pile cap and embedment region.

It is not believed that this discussioncan be extended beyond prestressed

concrete piles. For instance, it is notadvocated that the analyses presentedhere are applicable to large caisson-to-pier cap connections despite the geometric similarities. No analysis or experiments of other pile shapes or sizeshas been carried out. Data from embedment tests on smaller embeddedsections6 having a width less than 12in. (305 mm) do correlate well withthe presented equations.

Strand Development LengthIn this study, it has been assumed

that the design capacity of the pile-to-pile cap embedment is based on developing the capacity of the pile. Implicitin this assumption is that the capacityof the pile is actually attainable at thepile-to-pile cap interface. This requiresfull development of the prestressingstrand at this location.

With relatively short embedmentlengths this may not be possible.8 Ithas been suggested that a pile should

have an embedment length equal tothe strand development length to ensure that the capacity of the pile isavailable at the face of the pile cap.’°

It is felt that the provision of anembedment length equal to thestrand development length is impractical in most cases because it wouldresult in very deep pile caps. Notethat the maximum moment demandon a pile may not occur at the pile-to-pile cap interface’ (see Fig. 1), inwhich case full development of thestrand at the face of the pile cap maybe unnecessary.

The embedment calculations proposed here will result in a conservative pile cap design. This is desirablebecause it leads to a “weak pile, strongpile cap” behavior that permits easierinspection and repair in the event ofdamage from an earthquake.

CONCLUSIONS ANDRECOMMENDATIONS

Based on the results of this investigation, the following conclusions and recommendations can be made. Theseconclusions apply to square prestressedpiles smaller than 36 in. (914 mm) embedded in cast-in-place pile caps.

1. The simple plain embedment of aprecast, prestressed pile into a cast-in-place pile cap can be designed to develop the required capacity of the pile.

2. The required plain embedmentlength may be conservatively determined from either Eqs. (1) or (2). Eq.(2) is proposed in Chapter 6 of the PCIDesign Handbook6for the design ofembedded steel haunches or brackets inprecast concrete. This is analogous tothe embedment of precast, prestressedpiles in cast-in-place pile caps.

3. Conservatively, the required embedment length to develop the flexuralcapacity of a pile may be taken as thewidth of the pile. A minimum absoluteembedment length of 12 in. (305 mm)is recommended.

4. Due to the prestressing strandsnot being developed at the pile-to-pilecap interface, the flexural capacity ofthe pile may not be available at this location. This condition must be investigated by the designer.

5. The pile-to-pile cap embedmentlength proposed here should be inter-

333

250- ‘;;;‘0

167

Cl)8)

18 24embedment length, Le (inches)

Fig. 12. Comparison of embedment capacity and pile capacity.

July-August 2001 91

preted as the minimum embedment required to attain the theoretical capacityof the pile. The flexural capacity at thepile-to-pile cap interface is determinedfrom the flexural capacity of the pile,which is affected by the strand development length provided at this location.

ACKNOWLEDGMENTSThis investigation was funded by

the South Carolina Department ofTransportation (SCDOT).

The authors would like to thank theentire staff of the USC Structures Laboratory for their assistance in preparing the piles, pile caps, and assistingwith the tests; and SMI-Owen Steelfor assisting with the fabrication of thereaction frame.

The authors would also like to acknowledge Terry Koon, the SeismicSpecial Projects Engineer at SCDOT,Jeff Mulliken, a Project Engineer atthe LPA Group in Columbia, South

Carolina, and Lewis Ryan of UnitedContractors in Chester, South Carolina. Their assistance is greatly appreciated.

Lastly, the authors wish to thank thePCI JOURNAL reviewers for theirhelpful and constructive comments.

The opinions, findings, and conclusions expressed in this paper are thoseof the authors and do not necessarilyreflect those of the South CarolinaDepartment of Transportation.

REFERENCES

1. Joen, P. H., and Park, R., “Simulated Seismic Load Tests on Prestressed Concrete Piles and Pile-Pile Cap Connections,” PCIJOURNAL, V. 35, No. 6, November-December 1990, pp. 42-61.

2. Sheppard, D. A., “Seismic Design of Prestressed Concrete Piling,” PCI JOURNAL, V. 28, No. 2, March-April 1983, pp. 2 1-49

3. Marcakis, K., and Mitchell, D., “Precast Concrete Connectionswith Embedded Steel Members,” PCI JOURNAL, V. 25, No.4, July-August 1980, pp. 88-116.

4. Mattock, A. H., and Gaafar, G. H., “Strength of EmbeddedSteel Sections as Brackets,” ACI Journal, V. 79, No. 2, March-April 1982, pp 83-93.

5. ACI Committee 318, “Building Code Requirements for Structural Concrete (ACI 318-99),” American Concrete Institute,Farmington Hills, MI, 1999.

6. PCI Design Handbook: Precast and Prestressed Concrete, FifthEdition, PrecastlPrestressed Concrete Institute, Chicago, IL, 1999.

7. Harries, K. A., Mitchell, D., Cook, W. D., and Redwood,R. G., “Seismic Response of Steel Beams Coupling Reinforced Concrete Walls,” ASCE Journal of the Structural Division, V. 119, No. 12, December 1992, pp.3611-3629.

8. Wan, B., Petrou, P., Harries, K. A., and Hussein, A. A., “TopBar’ Effect in Prestressed Concrete Piles,” submitted for publication to ACI Structural Journal.

9. Bentz, E. C., and Collins, M. P., “RESPONSE-2000 Reinforced Concrete Sectional Analysis Using the Modified Compression Field Theory Computer Program, Release 1.0.0.1,”University of Toronto, Toronto, Ontario, Canada.

10. Shahawy, M.A., and Issa, M., “Effect of Pile Embedmenton the Development Length of Prestressing Strands,” PCIJOURNAL, V. 37, No. 6, November-December 1992, pp.44-59.

APPENDIX - NOTATIONa = shear span of pile taken as distance from pile cap soffit

to point of zero momentb’ = effective width of concrete compression blockb width of pile, also bearing width of the embedmentc = depth of concrete coverCf = resultant bearing (compressive) force in embedmente = eccentricity from midspan of beam to center of

embedment

f’ = specified concrete compressive strength

Le = embedment length of pile inside pile capV, = shear force on pile

Xb = length of compression block at back of embedmentXf = length of compression block at front of embedment

= ratio of average concrete compressive strength tomaximum stress

ô = deflection at yield stress of pile

92 PCI JOURNAL