Anda di halaman 1dari 10

INTERNATIONALJOURNALOFCIVILAND STRUCTURALENGINEERING Volume2,No 1,2011

Copyright2011AllrightsreservedIntegratedPublishingservices

Researcharticle

ISSN0976 4399

1 2 1 1 MohamadAhangar ,KhosroBargi ,HesamSharifian ,MortezaSafarnezhad 1Ph.D.student,DepartmentofCivilEngineering,CollegeofEngineering,University ofTehran 2Professor,DepartmentofCivil Engineering,CollegeofEngineering,Universityof Tehran hesam_sharifian@ut.ac.ir

Dynamicnonlinearbehavioroffixedoffshorejacketpiles

ABSTRACT The response of pile supported fixed offshore jackets is very sensitive under earthquake loadings. These structures shall be able to undergo the seismic loadings without any failure. To reach this purposes the dynamic analysis should be considered because of some factors like soil nonlinearity, energy dissipation, nonlinear behavior of foundations, discontinuity condition at pile soil interfaces and etc. the noticeable part of dynamic response of fixed offshore jackets is nonlinear behavior of piles. In this paper, innovative and newest approaches in nonlinear analysis together with ABAQUS modeling tobeusedtoextractthe usefulconclusionsrelatedtodynamicnonlinearresponseoffixedoffshorejackets. Keywords: Pile,nonlineardynamicresponse,interaction,seismicbehavior 1.Introduction Therecentdevelopmentofoffshoreindustries(inparticulartheexplorationandproductionof oilandgas)leadstogrowingdemandforrealisticpredictionsofoffshoreplatformsbehavior. Earthquake design ofoffshoreplatformsinseismicactiveareasisoneofthemostimportant parts in offshore platforms design. Evaluation of the pilesoilstructure interaction due to earthquake induced ground accelerations is an important step in seismic design of both the structureandpiles. Dynamic response will be defined as those characteristics of the structural system that can not be attributed to notime varying or static response. Inertial, damping and kinematics effects developed by loadings that have significant variations in time will be include in dynamic response. Dynamic response of offshore platforms due to earthquake excitations (and all cyclic loadings) would be inherently on nonlinear behavior because of the following reasons: 1 Nonlinearbehaviorofthesoilatawiderangeofshearstrain(Soilbehaviorwouldbe 4 linearonlyatverysmallshearstrainsofapproximatelylessthan 10 %). 2 PileSoil interaction, which is affected by different nonlinear behavior of interface elements(GappingandCaveineffect)fordifferenttypeofthesoils(clayandsand). 3 The seismic loading rates, which potentially could cause significant stresses and nonlinearmaterialbehaviorofpilesandsuperstructure. Dynamic response of piles in offshore platforms is a function of the characteristics of the loading, dynamic pilesoil interaction behavior and dynamic characteristics of the piles structural system. In recent years the interaction problem during earthquake loadings has receivedconsiderableattentionandstudiesindicatethenatureofgroundmotioninputandthe

ReceivedonJuly 2011publishedonSeptember2011

260

Dynamicnonlinearbehavioroffixedoffshorejacketpiles MohamadAhangar,KhosroBargi,HesamSharifian,MortezaSafarnezhad

mechanismofpilesoilinteractionplayanimportantroleindeterminationofplatformdesign loads. TheearliestsystematictheoreticalstudiesofdynamicsoilpileinteractionareduetoParmelee etal(1964),Tajimi(1966),Penzien(1970)andNovak(1974).Parmelee(1964)andPenzien (1970) employed a nonlinear discrete model and a static theory to describe e the dynamic elastic stress and displacements of fields. Tajimi (1966) used a linear viscoelastic stratum of the KelvinVoigt type to model the soil and in his analysis of the horizontal response he neglected the vertical component of the soil motion.Novak(1974)assumedlinearityandan elastic soil layer composed of independent infinitesimally thin horizontal layers extending to infinity. Recently investigators have begun to develop numerical methods in which all the soil, pile, superstructure and soilpilesuperstructure interfaces are modeled simultaneously together. Yegian and Wright (1973), Angelides and Rosset (1980), Randolph (1981), Faruque and Dessail (1982), Trochanis et al. (1988) and Wu(1997) used finite Element Method (FEM) whereas Sanches (1982), Kaynia and Kausel (1982) and Sen et al (1985) implemented Boundary Element Method (BEM) for dynamic response analysisofpiles.InbothFEMand BFM the soil is treated as a continuum media. Discritization of a three dimensional continuummediaandgenerationforamultitudeofdegreesoffreedominFEMandderiving of complex Green functions for complicated media in BEM, make both of these methods impracticalforseismicresponseanalysisofoffshoreplatforms. The finite and boundary element methods potentially provides the most powerful tools for conducting Seismic SoilPileStructure Interaction (SSPSI) analyses, but they have not yet been fully realized as a practical accepted method mainly due to their presumed excessive computational costs and their complexity for common pile dynamic response analysis. The mainadvantageofsuchapproachesisthecapabilityofperformingtheSSPSIanalysisofpile in a fully coupled manner, without resorting to independent calculations of site or superstructureresponse. The Beam on Nonlinear Winkler Foundation (ABAQUS) method is a dimplified approach thatcanaccountfornonlinearSoilPileStructureInteraction(SPSI)andhasprovenusefulin professional engineering and research practices. Trochanis (1991) showed that the response of laterally loadedpilespredictedusingaABAQUSformulationagreedwellwithstaticload test data and nonlinear three dimensional FEM. Boulanger et al (1999) showed that the results of seismic response of piles using ABAQUS modeling agreed well with centrifuge experimentaltests. ABAQUS models are the most versatile, economical and popular methods that can account for various complicated conditions in a simple manner. Basic components criteria in a ABAQUS models for dynamic response analysis of offshore piles are adopted in this paper. In ABAQUS method the pile is modeled as a series of a discrete beamcolumn members restingonaseriesofspringsanddashpotsindicatingthenonlineardynamicbehaviorofsoils. In seismic loadings, "free field" ground motion time histories are usually computed in a separate site response analysis and then applied to soilpile spring supports in ABAQUS models. A singular disadvantage of a ABAQUS mode is the two dimensional simplification of the soilpile contact, which ignores the radial ad threedimensional components of interaction.

InternationalJournalofCivilandStructuralEngineering Volume2Issue12011

261

Dynamicnonlinearbehavioroffixedoffshorejacketpiles MohamadAhangar,KhosroBargi,HesamSharifian,MortezaSafarnezhad

2.ModelDescription All above studies indicate that when a pile is subjected to highlevel lateral loading, the soil nonlinearity and relative movement at the pilesoil interface will strongly affect the pile behavior. Any model to be used for dynamic response analysis of piles should allow for variation of soil properties with depth. Nonlinear soil behavior, nonlinear behavior of pile soil interfaces, energy dissipation through radiation and hysteretic damping and soil strength degradation due to cyclic loads. During earthquake excitation all the components of a ABAQUS mode (representing the pile and surrounding soils) will be subjected to free field groundmotions. Figure1showsthegeneralviewofanABAQUSmodelanditsmaincomponentsindynamic nonlinear response analysis of offshore platforms. It is clear that for nonlinear dynamic response analysis of piles based on ABAQUS assumptions, each model should contain the followingitems: 1 Pilemodeling 2 SoilstiffnessandDampingmodeling 3 PileSoilinterfacemodeling 4 Freefieldexcitations 3.PileModeling The pile and surrounding soil layers are subdivided into a couple of discrete segments with pile nodes corresponding to soil nodes at the same elevation. Stiffness matrix of beam column elements are used to model to the structural stiffness matrix of each pile segment. These structural stiffness matrices of the pile segments will be assembled to build theglobal structuralstiffnessmatrixofthewholepile. 4.SoilModel InABAQUShypothesistherearetwodifferentmethodsofsolidmodelingasfollows: 1 hyperbolicstressstrainapproach 2 PYcurvesapproach 4.1PileSoilInterface One of the main sources in nonlinear dynamic response of piles is the relative movement of the soil and pile at interfaces. It is clear that each ABAQUS model shall include particular pilesoil interface elements to account for such relative movements. Behavior of these interface elements in compression and tension are quite different and therefore the pilesoil interface elements together with springs, dashpots and masses (if any) are usually modeled separatelyoneachsideofthepile.Accordingtodifferentbehaviorofcohesiveandcohesion lesssoils,thereshouldbedifferenttypeofinterfaceelements. Whenthetensilestressisdetectedinsoilsprings,theseinterfaceelementsshoulddetachpile nodes from the soil nodes and it means that a gap will be createdbetweenpileandthesoil. These gaps in cohesive soils (clay) will not be filled with the soils again and it means that therewouldbeagapdevelopment(permanentdisplacementofsoilnodes)duringearthquake excitationsinclaysoils(Matlock1978,Nogami1992).

InternationalJournalofCivilandStructuralEngineering Volume2Issue12011

262

Dynamicnonlinearbehavioroffixedoffshorejacketpiles MohamadAhangar,KhosroBargi,HesamSharifian,MortezaSafarnezhad

Figure1:GeneralViewofABAQUSmodelsfornonlineardynamicresponseanalysisof offshoreplatforms There would be cavein behavior incohesionlesssoils(sand)resultinginbackfillingofsand particles around the pile. It means that any developed gap in sand will be simultaneously filled with backfilledsoilagainandnopermanentgapswillbedeveloped.(ElNaggar2000). Considering different soil behavior in compression and tension (gapping in clay layers and cavein in sand layers), the soilreactions and the pile oil interface elements will be modeled separatelyonbothsideofthepiles.Generalviewsofsoilreactionversuspiledeflectionsfor cohesive andcohesionlesssoils(indicatinggappingandcaveinbehaviors)areshowninfigs. 2and3respectively

InternationalJournalofCivilandStructuralEngineering Volume2Issue12011

263

Dynamicnonlinearbehavioroffixedoffshorejacketpiles MohamadAhangar,KhosroBargi,HesamSharifian,MortezaSafarnezhad

Figure2:TypicalSoilreactionPiledeflection Behaviorforcohesivesoils(Gapping)soils(Cavein)

Figure3:TypicalsoilreactionPiledeflection Behaviorforcohesionless 4.2FreeFieldExcitations Earthquake induced loading on buried structures can be separated into two basic loading conditions of kinematic and inertial. These ABAQUS models (including pile, springs, dashpots and pilesoil interface elements) only deal with inertial loadings due to earthquake excitations. Kinematic loadings are an important part in dynamic response of piles due to seismic excitations in ABAQUS hypothesis. In these kinematic loadings, dynamic ground

InternationalJournalofCivilandStructuralEngineering Volume2Issue12011

264

Dynamicnonlinearbehavioroffixedoffshorejacketpiles MohamadAhangar,KhosroBargi,HesamSharifian,MortezaSafarnezhad

motions of the soil layers in free field due to earthquake excitations at bed rock should be determined. Results of such free field calculation (acceleration or displacement time histories at different soillayers)willbeusedastheinputexcitationatsupportnodesoftheABAQUSmodels.In such type of calculations, the free field motions are uncoupled from the pile ABAQUS models. In seismic loadings because of the different wave polarities in the near field and far filed, uncoupling the nonlinear pilesoil interaction in near field from the freefieldresponses wouldbeareasonableapproximation. Fen et al (1991) performed and extensive parametric study using an equivalent linear approach to develop dimensionless graphs for pile head deflections versus the free field response. Markis and Gazetas (1992) applied free field accelerations to a one dimensional ABAQUS model. Bentley (2000) performed a full three dimensional transient nonlinear dynamicanalysis(3Dwavepropagation)andcomparedtheresultswithequivalentlinear1D methods. In dynamic response analysis of piles, free field motions may be calculated by any desired wave propagation method such as equivalent linear (used in SHAKE91 software) or nonlinearprocedures.SHAKE(Schanbeletal1972)isstillcommonlyusedafter30yearsof itsreleaseanditisareferencecomputerprogramingeotechnicalearthquakeengineering.In SHAKEitisassumedthatthecyclicsoilbehaviorcanbesimulatedusinganequivalentlinear modelrepresentingthesoilstressstrainresponsebasedonKelvinVoigtmodel.Wang(1998) used SHAKE for free field motion analysis and showed an acceptable agreement between calculatedresultsandtherecordedresultsincentrifugetests. 5.SummaryandConclusion Dynamic soil reaction and pile head response to harmonic loads for both the PY and hyperbolicapproachesinABAQUSmodelswerecomparedbyElNaggarandBentley(2000). Apilewiththeoutsidediameterofd=0.5m,lengthofl=15mandanelasticmodulus ( Ep) equal to 35 GPa was used (as shown in Fig. 4). A parabolic soil profile with the ratio of E p /E =1000 at the pile base was assumed. The undrained shear strength of the clay was s assumedtobe25kPa.

Figure4:ThemodelusedbyElNaggarandBentley(2000)forcomparingPYcurveand hyperbolicapproaches Figure5showsthedisplacementtimehistoryofthepileheadforaharmonicloadwithsingle amplitude equal to 10 kN at a frequency of 2 Hz applied at the pile head. Fig 6 shows the

InternationalJournalofCivilandStructuralEngineering Volume2Issue12011

265

Dynamicnonlinearbehavioroffixedoffshorejacketpiles MohamadAhangar,KhosroBargi,HesamSharifian,MortezaSafarnezhad

calculated dynamic soil reaction (at 1.0 m depth) for a harmonic displacement of a single amplitudeequalto0.03d(0.015m)atafrequencyof2Hzappliedatthepilehead.

Figure5:Pileheadresponseduetoharmonicloadwithsingleamplitudeequalto10kNata frequencyof2Hz.

Figure6:CalculatedDynamicsoilreaction at1.0mdepthforharmonicdisplacementwith singleamplitudeequalto0.03datafrequencyof2Hz. InFigures5,6itisseenthathyperbolicandPYcurvemodelsshowverysimilarresponsesat the pile head displacement and soil reactions respectively and both stabilize after approximately five cycles. ElNaggar and Bentley (2000) also showed that dynamic soil reactions are in general larger than the static reactions because of the contribution from damping. Employing the same definition used for static PY curves, dynamic PY curves (which are frequency depended) can be established to relate the pile deflections to the correspondingdynamicsoilreactionsatanydepthbelowthegroundsurface.Thesedynamic PY curves can be implemented in equivalent static analyses, which are now used in earthquakeanalysisofoffshoreplatforms. Several implementations of Dynamic PY methods (different configuration of nonlinear springs and dashpots in Parallel and Series Radiation Damping) together with free field effects were compared with Centrifuge model tests (performed at University of California, Davis)byWangetal.(1998). Centrifuge tests were performed on samples of normally consolidated San Francisco Bay 3 3 Mud(densityofabout1700kg/m )withacrustofdensesand(densityofabout2100kg/m ) on the surface of clay. The water table was at the ground surface. General configuration,
InternationalJournalofCivilandStructuralEngineering Volume2Issue12011

266

Dynamicnonlinearbehavioroffixedoffshorejacketpiles MohamadAhangar,KhosroBargi,HesamSharifian,MortezaSafarnezhad

dimensionandinstrumentationsofthismodelareshowninFig.7.AscalefactorN=50was usedinthismodelandsoalldimensionsshowninFig.7shouldbefactoredbyascalefactor of50inprototypeunits.Moreexplanationsaboutcentrifugemodelingtestsandscalinglaws couldbereferredtoKutter(1992).Inprototypeterms,thismodelrepresentsasuperstructure massof1.44tonandpileheadmassof1.12tonsupportedbya317mmdiametersteelpipe havingawallthicknessof10mm.SantaCruzduringthe1989LomaPrietaEarthquakewas usedtoexcitethebaseofthecentrifugemode.

Figure7:ConfigurationofCentrifugemodeltest(Wangetal1998) Table1showstheconditionsandmethodsintwodifferentnumericalcasesusedbyWang etal(1998)forsimulationofthecentrifuge modeltestresults.All ABAQUS modelingand nonlinearanalysiscarriedoutusingDRAIN2D(DynamicResponseAnalysisofInelastic2D structures,version1.10Prakash &Powell1993).Dynamic motionsofthe free field(which are used as the input excitations of the supports in ABAQUS model) carried out using SHAKE(Schnabeletal,1972)whichusestheequivalentlinearprocedurefornonlinearsoil behavior Table1:Inputinformationfornumericalcases Damping Model CaseA Parallel Damping Coefficient
4B rv s

PYCurvesGeneration

PYStrength

Piecewise 1 P . x ult Linear (Matlock) Piecewise 4B rv s 1 P . x ult CaseB Series Linear (Matlock) Recorded and calculated spectral accelerations for above cases are shown in Figs. 8 a, b respectively. In Fig. 8a it is seen that only the peak acceleration for superstructure is predicted well by the results of CASE A meanwhile the peak acceleration for the pile head andfrequencycontentforbothlocations(superstructureandpilehead)arenotwellpredicted. In Fig. 8 b it is seen that the peak acceleration for the pile head and frequency content for both locations (superstructure and pile head) are well predicted and only the peak acceleration for superstructure is predicted dynamic response of the pile at pile head and superstructure responses) is seen in CASE B which series radiation damping is used. Comparing the response spectra for Cases A and B (in Figs. 8 a, b) it is concluded that
InternationalJournalofCivilandStructuralEngineering Volume2Issue12011

267

Dynamicnonlinearbehavioroffixedoffshorejacketpiles MohamadAhangar,KhosroBargi,HesamSharifian,MortezaSafarnezhad

parallel radiation damping acted to restrict the lateral movement of the pile head and therefore resulted in a stiffer system (higher frequency content) than for the series radiation damping.

Figure8a:Accelerationresponsespectra(CalculatedinCaseA)

Figure8b:Accelerationresponsespectra(CalculatedinCaseB) The highest peak moment in CASE a is about 29% more that CASE B meanwhile the location of the highest peak moment in CASE A (close to ground surface) is quitedifferent thatCASEB(about2mbelowgroundsurface).Itshouldbenotedthatpeakmomentatthe ground surface depends primarily on the superstructure acceleration and coincidence of the natural period of the system and predominant period of the shaking. Parallel radiation damping is likely to produce a stiffer system than series radiation damping and it allows forcestobypassthehystereticsystemthroughaparalleldashpot.Thereforeparallelradiation damping results in a more rapid reduction in bending moments withdepththaniscalculated usingseriesradiationdamping. 6.References 1. Matlok, H. (1970), Correlations for design of laterally loaded piles in soft clay. nd Proceedingofthe2 OffshoreTechnologyConference,Huston,Tex., 1,pp577588. 2. ElNaggar, M. H. and Bentley, K. J. (2000), Dynamic analysis for laterally loaded pilesanddynamicpycurves,CanadianGeotechnicalJournal,37,pp11661183.

InternationalJournalofCivilandStructuralEngineering Volume2Issue12011

268

Dynamicnonlinearbehavioroffixedoffshorejacketpiles MohamadAhangar,KhosroBargi,HesamSharifian,MortezaSafarnezhad

3. Kutter B. L., (1992), Dynamic centrifuge modeling of geotechnical structures, Transp.Res.Rec.1336,TransportationResearchBoard,Washington,D.C.,pp2430. 4. Markis, M., Gazetas, G.,(1992),Dynamicpilesoilpileinteraction.PartII,Lateral ans seismic response. Earthquake Engineering and Structural Dynamics, 21, pp 145 162. 5. Kaynia, A. and Kausel, E. (1982), Dynamic Stiffness and Seismic Response of Pile Groups,Rpt.R8203,MassachusettsInst.of Technology,Cambridge. 6. Gazetas, G. and Doprby, R. (1984) Horizontal response of piles in layered soils. JournalofGeotechnicalEngineering,ASEC,110(1),pp2040. 7. Fen, K., Gazetas, G., Kaynia, A., Kausel, E., and Ahmed, S. (1991), Kinematic response of singl piles and pile groups, Journal of Geotech. Eng., ASCE 117(12), 18601879. 8. ElNaggar,M.H.andNovak,M.(1996),NonlinearAnalysisfordynamiclateralpile rssponse.JournalofSoilDynamicsandEarthquakeEngineering,14(4),pp233244. 9. Bentley, K. J. and EiNaggar, M. H. (2000), Numerical analysis of kinematic responseofsinglepiles,CanadianGeotechnicalJournal,37,pp 13681382. 10. Angelides, D. C. and Roesset, J. M., (1980), Nonlinear dynamic stiffness of piles, Research report R 8013, Dept. of Civil Engineering, MIT, Cambridge, Massachusetts. 11. ElNaggar, M. H. and Novak, M. (1995), Nonlinear lateral interaction in pile dynamics,JournalofSoilDynamicsandEarthquakeEngineering,14(3),pp141157. 12. ElNaggar, M. H. and Novak, M. (1994), Nonlinear model for dynamic axial pile response.JournalofGeotechnicalEngineering,ASCE,120(2),pp308329. 13. American Petroleum institute. (2000), Recommended practice for planning, designingandconstructingfixedoffshoreplatforms.APIRecommendedPractice2A st (RP2A).21 ed.AmericanPetroleumInstitute,Washington,D.C.,pp4755. 14. Parmele, R. A., Penzien, J., Scheffey, C. F., Seed, H. B. and Thiers, G. R. (1964), Seismic effects on structures supported on piles extending through deep sensitive clays,Inst.Eng.Res.,UniversityofCalifornia,Berkeley,Rep.SEM642. 15. Kagawa, T., and Kraft, L. (1980), Seismic PY Responses of Flexible Piles, J. Geotech.Eng.,ASCE,106(8),pp899918.

InternationalJournalofCivilandStructuralEngineering Volume2Issue12011

269

Anda mungkin juga menyukai