Numerical Simulations of Geotechnical Works in Bangkok ... - Plaxis

Numerical Simulations of Geotechnical Works in Bangkok Subsoil UsingAdvanced Soil Models Available in Plaxis and Through User-Defined ModelPornkasem Jongpradist, Trin Detkhong & Sompote Youwai, King Mongkut’s University of Technology Thonburi, ThailandIn this article, three constitutive models with enhancing levels of complexity are used to simulate three types of geotechnicalworks (embankment construction, deep excavation and tunneling) on/in Bangkok subsoil conditions. In dense urbanenvironments where land is scarce and buildings are closely spaced, one of the main design constraints in these projects is toprevent or minimize damage to adjacent structures due to excessive movements induced from the construction. To eliminateor reduce the possibility of such damage, an effective method is needed to accurately ‘‘predict” the construction-inducedmovement for such complex condition.eliminate or reduce the possibility»Toof such damage, an effective method isneeded to accurately ‘‘predict” the constructioninducedmovement for such complex condition.The magnitude of the settlement and lateralmovement and their distributions depend ona number of factors, such as soil profile and itsgeotechnical engineering properties, stiffness ofstructure and support system. The finite elementmethod (FEM) is often used to predict groundmovements induced by such soil-structureinteraction problems. The interaction betweenexisting structures and underground activities is acomplex phenomenon in which the behavior of thesurrounding ground is one of the main aspects tobe taken into account. Consequently, a reasonablesoil model is crucial in order to estimate themagnitudes and distribution of the strains. Theconstitutive model frequently used in numericalsimulation of an underground work in currentpractice is linear elastic perfectly plastic with aMohr–Coulomb (MC) yield criterion. The greatestadvantage of MC is that only five parameters,which includes two elastic parameters (i.e.,Young’s modulus E and Poisson’s ratio n) and threeplastic parameters (i.e., friction angle φ, cohesion cand dilatancy angle y),are sufficient in describingthe behavior. Moreover, the parameters can beeasily determined. However, the model doesnot take into account the fundamental aspectsof soil behavior, such as variation of modulusaccording to different stress state and modulus inloading and unloading conditions or stress pathdependency. Therefore, in general, the numericalresults by MC are only in good agreement withparticular field observation at a certain strainrange for each case. To achieve good agreementwith the measured values, the parametersare usually obtained from back calculation bycurve fitting with history records. Althoughthis technique could improve the accuracy ofprediction in practice, it is only applicable for aspecific range of each work. To overcome suchshortages, it is necessary to consider at least anelasto-plastic model with isotropic hardening.In addition to previously mentioned aspects,the non-linear pre-failure initial stiffness of soilat small strain range is necessary for analysis ofsome geotechnical earthworks (Burland, 1989).Recently, with the advancement in constitutivemodel development and implementation in Plaxis,the Hardening Soil model (HS) (Schanz et al., 1999)with small strain stiffness (HS-small) has beenproven to be effective in analysis of geotechnicalworks. Moreover, with the advanced feature inlater versions of Plaxis, the implementation ofany desired constitutive model via user definemodelsubroutine is possible. This facilitates theengineers to utilize more advanced soil models.In this article, three constitutive models withenhancing levels of complexity are used tosimulate three types of geotechnical worksFigure 1: Typical soil profile of Bangkok subsoil and typical section and position of geotechnical works6 Plaxis Bulletin l Autumn Issue 2012 l www.plaxis.nl

γsatν ‘ φ ‘ c E’ ψR interSoil parameter[kN/m3] [-] [°] [ kPa] [kPa] [°] [-]MC model 16 0.33 22 5 10000 0 1Table 1: Soil parameters of Bangkok soft clay for MC modelE ref oedE ref 50 Eref urG ref 0γ 0.7m p refSoil parameter[kPa] [kPa] [kPa] [kPa] [-] [-] [kPa]HSsmall model 10000 10000 30000 22560 1x10-4 1 100Table 2: Soil parameters of Bangkok soft clay for HS-small modelBasic parameters [unit in parenthesis]N* [-] λ* [-] κ* [-] φ c[deg] r [-]1.85 0.17 0.043 24 0.2Intergranular strain extension parametersm R[-] m T[-] R [-] β r[-] χ [-]5.25 5.25 0.0001 0.2 6Table 3: Soil parameters of Bangkok soft clay for Masin’s hypoplastic model for claySoil layer Weather Crust Soft Clay Medium Clay Stiff Clay Sandγ [kN/m3] 17 16 18 18 20satn [-] 0.32 0.33 0.33 0.33 0.3φ [°] 22 22 22 22 36c [kPa] 8 5 10 18 0E’ [kPa] 6000 5000 20000 60000 80000y [°] 0 0 0 0 0R inter1 1 1 1 0.7Table 4: General MC soil model parameters for Bangkok subsoil (Wonglert et al., 2008)(embankment construction, deep excavationand tunneling) on/in Bangkok subsoil conditions.These include MC, HS-small and a hypoplastic(HP) model (Masin 2005). The first two models areavailable in the Plaxis model library, whereas, thelast one was a user-defined model subroutine.First, the calibration of selected models on thebasis of the extensive in situ test results for stiffclay and laboratory test results for soft clay iscarried out. The analysis results are compared withtriaxial test results. Then analyses of three kindsof geotechnical works are carried out with thosethree soil models with a single set of parametersfor each model. The impact of selecting a soilmodel for geotechnical work analysis on accuracyof the predictions of soil displacements ishighlighted.Subsoil Condition and Soil ParametersBangkok Soft Clay was deposited in marineconditions at the delta of the rivers in the ChaoPhraya Plain. The typical Bangkok subsoil isshown in Figure 1. It consists of made groundwith a thickness of 2.0 m over a 13.0 m thick softclay layer. The 12.0 m thick first stiff clay layer isencountered below the soft clay layer at the depthof 15.0 m to 27.0 m Beneath the first stiff claylayer is the 8.0 m thick dense sand layer and the6.0 m thick second stiff clay layer at the depthsof 35.0 m to 41.0 m respectively. The 19.0 m thicksecond dense sand layer at the depth of 60.0 m isunderneath these laters. The soil properties usedin the analyses are mainly determined from localinvestigated data correlation from comprehensivein-situ tests of previous mass transit projects (Prustet al., 2005) and previous laboratory tests (Shibuyaet al., 1997; Theramast, 1998 ; Uchaipichat, 1998).Note that the various soil models are appliedto the only soft clay layer, whereas, other layersare assumed to behave as MC model in order tohighlight the influence of soft clay model. Tables1-3 tabulate the soil parameters of soft clay for MC,HS-small and HP model, respectively. The valuesof other soil layers are listed in Table 4. Figure1 also shows the typical section and position ofgeotechnical works on/in Bangkok subsoil.www.plaxis.nl l Autumn Issue 2012 l Plaxis Bulletin 7

Numerical Simulations of Geotechnical Works in Bangkok Subsoil Using Advanced Soil Models Available in Plaxis and Through User-Defined ModelNumerical Analysis and ResultsAll problems which are from well-documentedcase histories having reliable monitored data areanalyzed by PLAXIS 2D assuming plane straincondition with the appropriate analysis condition.These include the embankment construction(Bergado et al., 1994), deep excavation (Teparaksaet al., 1999) and tunneling (Suwansawat, 2002).The consolidation analysis is carried out forembankment construction since the monitoringdata was obtained at 300 days after construction.For analyses of deep excavation and tunneling,undrained conditions are assumed since themeasured data were obtained during and at afew days after construction. The extension of thefinite mesh is wide enough and suitable boundaryconditions are assumed for all model boundariesto ensure the accuracy of the analysis.Embankment ConstructionFigure 2 depicts the finite element mesh,geometry and dimension of the problem for theembankment construction case. The case is theconstruction of 4.0 m high test embankment atAIT (Asian Institute of Technology) in 1992. Thedata to be compared is the settlement of theembankment measured at the level of the originalground using surface settlement plates for 300days after the construction.The comparison between measured data (reddot) and analysis results by three different soilmodels (color lines) as embankment loadingincreases is illustrated in Figure 3. It is seen that,for this problem in which the soil behavior is mainlygoverned by compression loading, the resultsfrom HP are best comparable with measured data.While those of HS-small and MC have acceptableagreement with measured data in low to moderateloading range.Deep ExcavationThe finite element mesh, geometry and dimensionof the problem for underground construction ofBangkok Metropolitan Hospital (BMH) which is a14 m deep excavation using 20 m high diaphragmwall as retaining structure, is depicted in figure4. Figure 5 shows the comparison between theFEM analysis results and observed values of theexcavation works at the excavation depth of 2.0,5.7, and 11.0 m. For this problem which the soilbehaviors are mainly governed by compressionunloading and extension unloading paths, it showsthat the tendencies of analytical results with HPand HS-small models are satisfactory in terms ofmagnitude and the shape of the wall movementwhile those of MC model are over-estimated.Figure 6 shows the maximum lateral wallmovement from analyses compared with theobservation results of all cases (Oriflame Building,MRT Bang Sue, China tower and TPI Building). Itwas shown that the predicted results by HP andHS-small models give a satisfactory accuracy withthe observed data. Especially, when the maximumlateral wall movements are lower than 20.0 mm. Itcan be seen that, for the constitutive soil modelswhich take the small-strain stiffness into theaccount, the lateral wall movement is accuratelypredicted. However, the simple constitutive soilmodel such as elastic-perfectly plastic (MC) is stillable to accurately predict the lateral deformationonly for specific strain range of each problem.TunnelingFor tunneling work, the ground surfacesettlements were measured by settlement marker.The case study is construction of the Mass RapidTransit (MRT) tunnel having an inner diameter of5.8 m at three sections including CS-8E, CS-8Cand CS-9A with the depth from ground surface at21.0, 19.0 and 17.0 m, respectively. As an example,the boundary condition and mesh of FEM analysisfor MRT tunnel at point CS-8E is shown in Figure7. The results of ground surface settlement areshown in Figure 8 for comparisons of the FEanalysis results and the observed values. It showsthat the analytical results from all models givesatisfactory tendencies in terms of the magnitudeand the shape of the settlement profile. However,the results from HP are closed to measured valuesat near-centerline (0.0 - 18.0 m), whereas, those ofMC are in good agreement with measured ones inthe distance of 20.0 - 40.0 m from centerline.The results of surface settlement at the tunnelcenter (maximum), 5.5 and 11.0 m from the tunnelcenter are compared with measurement data atpoint CS-8E, CS-8C and CS-9A as shown in Figure9. From FE analyses, HSsmall and HS model givea highly accurate prediction for a wide range ofobserved settlement. The results from MC modelseem to be more scattered.Figure 2: Finite element mesh for embankment construction caseFigure 3: Comparison of settlement between measured data (red dot)and analysis results by three different soil models (color lines)Figure 4: Finite element mesh for deep excavation caseFigure 5: Comparison of wall movement of BMH construction betweenmeasured data and analysis results by three different soil models8 Plaxis Bulletin l Autumn Issue 2012 l www.plaxis.nl

Numerical Simulations of Geotechnical Works in Bangkok Subsoil Using Advanced Soil Models Available in Plaxis and Through User-Defined ModelDiscussionThe analyses of three kinds of geotechnicalworks carried out in this article shows theimpact of soil model on the simulations. Usingmore sophisticated soil models considerablyimproves the prediction of movements. The HPand HSsmall models which include non-linearityat prefailure and high stiffness under very smallstrain, particularly the HP which has stress-pathdependent stiffness, give satisfactory accuracy ofmovement prediction for all three types of workcovering the wide range of observed data. MCmodel over-predicts the deformation for analysisof embankment and excavation work, particularly,at range of small movement. However, theadvanced models are applied to only the soft claylayer in this study. By applying to the other claylayers, the analysis results can be further improved(Rukdeechuai et al., 2009).References• Bergado, D.T., Long, P.V., Loke, K.H. and Werner,G. (1994), Performance of reinforced embankmenton soft Bangkok clay with high strengthgeotextile reinforcement, Geotextiles andGeomembranes, Vol. 13(6-7), pp. 403-420.• Burland, J.B. (1989), Ninth Laurits Bjerrum memoriallecture: Small is beautiful---The stiffnessof soils at small strain, Canadian GeotechnicalJournal, Vol. 26(4), pp. 499-516.• Masin D. (2005), A hypoplastic constitutivemodel for clay, International Journal for Numericaland Analytical Method in Geomechanics,Vol. 29, pp. 311-336.• Prust, R.E., Davies, J., and Hu, S. (2005), Pressuremeterinvestigate for mass rapid transit inBangkok-Thailand, Journal of the transportationresearch board, Transportation research ofthe national academies, Washington D.C., No.1928, pp. 207 -217.• Rukdeechaui, T., Jongpradist, P., Wonglert, A.and Kaewsri, T.(2009), Influence of Soil Modelson Numerical Simulation of Geotechnical Worksin Bangkok Subsoil., EIT Research and DevelopmentJournal, Vol. 20(3), pp. 17-28.• Shibuya, S., Hanh, L.T., Wilailak K., Lohani T.N.,and Tanaka H. (1997), Characterizing stiffnessand strength of soft Bangkok clay from in-situand laboratory tests, First Int. Conf. on SiteCharacteristics.• Teparaksa, W., Thasnanipan, N., Tanseng, P.(1999), Analysis of lateral wall movement fordeep excavation in Bangkok subsoils, Proceedingof Civil and Environmental engineeringconference “New Frontiers and Challenges”,Bangkok, pp. 67-76• Theramast N. (1998), Characteristic of pseudoelasticshear modulus and shear strength ofBangkok clay, M. Eng. Thesis, AIT, Thailand.• Uchaipichat, A. (1998), Triaxial Tests on SoftBangkok Clay with Different Applied StressPaths. M. Eng. Thesis, AIT, Thailand.• Schanz, T., Vermeer, P.A. and Bonnier, P.G. (1999)Formulation and verification of the Harening soilmodel, in Beyond 2000 in computational Geotechnics,A.A. Balkema, Roterdam, Netherlands,pp. 281-290• Suwanawat, S. (2002), Earth pressure balance(EPB) shield tunneling in Bangkok groundresponse and prediction of surface settlementsusing artificial neural networks., PhD. Thesis,Massachusetts Institute of technology, Cambridge,USA.• Wonglert, A., Jongpradist, P., Kalasin, T.(2008),Wall Movement Analysis of Deep Excavationin Bangkok Subsoil considering Small StrainStiffness, Journal of Research in Engineeringand Technology, Vol.5(4), pp. 393-405.Figure 6: Comparison between computed maximum wall displacementby different soil models and measured data of all case studiesFigure 7: FE mesh for tunneling caseFigure 8: Comparison of surface settlement at point CS-8E betweenmeasured data and analysis results by three different soil modelsFigure 9: Comparison between computed settlements by differentsoil models and measured data of all case studieswww.plaxis.nl l Autumn Issue 2012 l Plaxis Bulletin 9