Numerical Modeling of Soil-Pile Axial Load Transfer Mechanisms in Granular Soils

Transcription

1 EDF Séminaire LaMSID Numerical Modeling of Soil-Pile Axial Load Transfer Mechanisms in Granular Soils Sofia Costa D Aguiar Under the supervision of : Prof. Arézou Modaressi 1 Prof. Jaime Santos Ecole Centrale Paris, France Laboratoire MSSMAT 2 - Instituto Superior Técnico, Portugal 15/01/2009 1

2 I Introduction Statement of the Problem Q T = Q s +Q b Q T Pile resistance mobilization Shaft resistance: Q s Base resistance: Q b Berezantzev et al. (1961) 2

3 I Introduction Statement of the Problem Q T = Q s +Q b Q T Pile resistance mobilization Berezantzev et al. (1961) Q s Q b Depends on: - Nature of soil - Initial soil conditions: - Initial state (undisturbed soil) - Installation effects - Residual loads - Pile-soil Interface (Shaft) Relative movements for resistance mobilization - Type of loading (Monotonic, cyclic, static or dynamic) -Time effects 3

4 I Introduction Background and Motivations CONVENTIONAL DESIGN METHODS - Predictions: - Theoretical solutions, Empirical methods - Broad range of predictions; Low reliability IC 50 prediction event Dunkirk sand (Jardine et al. [2005]) ISC2 prediction event Porto residual soil (Santos et al. [2005]) 4

5 I Introduction Background and Motivations CONVENTIONAL DESIGN METHODS - Predictions: - Theoretical solutions, Empirical methods - Broad range of predictions; Low reliability NEEDS - Improve predictive methods: economies - Understanding of all interaction mechanisms (installation and loading) - Reliable and rational design method 5

6 I Introduction Objectives (challenges) Q T Pile Interface Q s Soil - Improve the knowledge of the load transfer mechanism - 3D numerical modeling of Pile-Soil Interface and soil behavior - Determine load settlement response (base and shaft) to axial loads of non-displacement piles Berezantzev et al. (1961) Q b - Contribute to the understanding of some of the installation effects: - Soil stress history - Residual loads - Friction fatigue - Apply the model to an in-situ case study 6

7 Short Outline I - Introduction II - Theory formulation and Numerical modeling III - Soil-Structure Interface behavior IV - Non-displacement Piles in Toyoura sand V - ISC 2 Experimental site VI - Final remarks 7

8 Short Outline I - Introduction II - Theory and Numerical modeling III - Soil-Structure Interface behavior IV - Non-displacement Piles in Toyoura sand V - ISC 2 Experimental site VI - Final remarks 8

9 II Theory and Numerical modeling Employed and developed Tools Numerical Tools GEFDYN FEM code [Aubry et al 1986] SDT / Matlab post-processing ECP Elastoplastic Constitutive Models Multimecanisms model ECP or Hujeux model, (2D and 3D) [Aubry et al. 1982, Hujeux 1985] Interface model, [Aubry et al. 1990] (2D) 3D Axisymmetric formulation of the Multimecanisms model 3D Interface model Cyberquake model, [Mellal 1997, Modaressi et al. 1997] (1D) Newly implemented 9

10 II Theory and Numerical modeling Employed and developed Tools ECP elastoplastic Constitutive models: Theoretical principles: Effective stress principle; Incremental plasticity; Critical state concept; Mohr Coulomb type failure criterion; q k p co Deviatoric primary yield surface of the k plane (3 planes): b=0 b=1 p k Isotropic yield surface: Progressive mobilization of shear: Roscoe s dilatancy law 10

11 II Theory and Numerical modeling Employed and developed Tools 3D Axisymmetric formulation of the ECP multimechanisms model: Induced anisotropy depends on the choice of the deviatoric mechanisms planes direction 3D CV shear test (axisymmetric conditions) y Active mechanism plane y τ s * z 3 x z 2 z x y y y x z 1 * τ s shear stress x x

12 II Theory and Numerical modeling Employed and developed Tools 3D Axisymmetric formulation of the multimechanisms model: Induced anisotropy depends on the choice of the deviatoric mechanisms planes direction 3D CV shear test (axisymmetric conditions) y r 3 τ s * z 2 z x y y θ x 1 * τ s shear stress x Deviatoric mechanisms: Transformation of three ortogonal planes: New -q formulation xy, p xy -q rθ, p rθ -q xz, p xz -q rz, p rz -q yz, p yz -q θz, p θz x θ z r y

13 II Theory and Numerical modeling Employed and developed Tools 3D ECP Interface Hypotheses: - Isotropy in the sliding plane: σ n τ t t=0 n t τ s - The tangential flow rule of the sliding plane: s Zero-thickness interface elements - The normal plastic displacement rate variation: 13

14 Short Outline I - Introduction II - Theory and Numerical modeling III - Soil-Structure Interface behavior Background Strategy for parameters identification Simulation of direct shear tests Calibration of ECP interface model IV - Non-displacement Piles in Toyoura sand V - ISC 2 Experimental site VI - Final remarks 14

15 III Interface behavior Background Soil-Structure Interface behavior depends on: Pile nature Soil nature Soil initial state Boundary conditions R t pile surface roughness D 50 soil grain size σ n0 initial normal stress D r initial relative density Shear tests: CNL - Constant normal load CV - Constant volume CNS - Constant normal stiffness R n - Normalized surface roughness R t D 50 t - interface layer thickness [Uesugi & Kishida 1986] 15

16 III Interface behavior Strategy for parameters identification Boundary conditions, CV, CNS, CNL Initial normal stress, σ no Initial relative density Type of loading Interface layer thickness, t Similitude with soil behavior Uesugi et al. [1988], Boulon and Nova [1990] Normalized roughness, R n ECP soil model parameter s identification strategy (extensively studied at ECP) (Hicher & Rahma [1994]... Kordjani [1995]... Lopez [2003]...) 16

17 III Interface behavior Strategy for parameters identification Boundary conditions, CV, CNS, CNL Initial normal stress, σ no Initial relative density Type of loading Interface layer thickness, t Similitude with soil behavior Uesugi et al. [1988], Boulon and Nova [1990] Normalized roughness, R n Taken into account implicitly in the parameters 17

18 III Interface behavior Simulation of Direct Shear Test Boundary conditions, CV, CNS, CNL Initial normal stress, σ no Initial relative density Type of loading Interface layer thickness, t Normalized roughness, R n Observation t Controlling parameters Quasi-steady state Dilatancy at the critical state Elasticity Friction angle at the critical state Shifting for lower u s Reduction Domain reduction Unaffected τ s (kpa) - Shear stress t FEM model (solid element) CNL shear test: (ECP multimechanisms) σ n0 =100 kpa u s u s (mm) - Tangential displacement 18

19 III Interface behavior Simulation of Direct Shear Test Boundary conditions, CV, CNS, CNL Initial normal stress, σ no Initial relative density Type of loading Interface layer thickness, t R n = R t D 50 [Uesugi & Kishida 1986] Normalized roughness, R n Pile R t A B C Surrounding soil s A B C Contraction / Dilation at the interface zone due to shearing Surrounding soil Rough interface t 0 D 50 t 0 + u n 19

20 III Interface behavior Calibration of ECP interface model Normalized roughness, R n, (calibration) rough φ = 30 smooth φ = 14.5 u n u s Rough / smooth CNS shear test k=1000kpa/mm Experimental Data from Fioravante [2002] 20

21 III Interface behavior Calibration of ECP interface model Normalized roughness, R n, (calibration) R n - Affects the failure mode: R n <R ncrit smooth (φ int << φ soil ) R n >R n crit rough (φ int = φ soil ) rough smooth φ = 30 φ = 14.5 FEM model Experimental Data from Fioravante [2002] Simulation CNS shear test k=1000kpa/mm 21

22 Short Outline I - Introduction II - Theory and Numerical modeling III - Soil-Structure Interface behavior IV - Non-displacement Piles in Toyoura sand Validation with physical models Soil-Pile interaction: parametric studies Cyclic loading at constant force V - ISC 2 Experimental site VI - Final remarks 22

23 IV Non-Displacement Piles Centrifuge tests Static load test HYDRAULIC ACTUATOR - Tested sand: Toyoura - Installation method Non-displacement (in store before sand pluviation) - Applied Load Compression - Soil density DR 93% - Roughness R n =0.01 and Prototype L=7.4m ; D=0.3m - Acceleration 50g,30g [Fioravante 2002] ISMES-GEO Centrifuge LVDT LVDT d m [Colombi 05] LVDT LOAD CELL 20 d m 20 d m 23

24 IV Non-Displacement Piles Toyoura sand Soil parameters calibration: (ECP multimechanism model) Sand γ min [kn/m 3 ] γ min [kn/m 3 ] e max [-] e min [-] G s [-] D 50 [mm] Toyoura TS40(Dr=40%) TS90(Dr=93%) Drained triaxial test (p o =100kPa) Fukushima and Tatsuoka [1984] 24

25 IV Non-Displacement Piles Numerical model Model and boundary conditions : Soil 7.4 m 13.5 m Interface Pile 8.5 m SAME SIZE AS PROTOTYPE 25

26 IV Non-Displacement Piles Validation with experimental results (1/2) Load settlement response Q s Q T =Q s +Q b Q b Rough, R n =0.45 Smooth, R n =0.01 Q s - Shaft resistance s pile head displacement d pile diameter Q b - Base resistance Experimental Data from Fioravante [2002] Simulation Reduction of R n : reduction of Q s Q T 26

27 IV Non-Displacement Piles Validation: Load Transfer mechanisms Rough, smooth interface stress paths : Simulation rough smooth 1 2 1,7m 3.4m 6.0m Control points

28 IV Non-Displacement Piles Validation with experimental results (2/2) Load transfer coefficient (β=τ/σ v0 ) R n =0.45 (Rough interface) τ = β. σ v0 β (τ/σ vo ) 0.9m 2.9m 4.6m 6.4m Control points Experimental Data from Fioravante [2002] Simulation 28

29 IV Non-Displacement Piles Validation with experimental results (2/2) β Normalized shear (τ/σ vo ) Load transfer coefficient (β=τ/σ v0 ) R n =0.45 (Rough interface) β decreasing with depth: - progressive inhibition of dilatancy - Increase in ambient stress Final local β (τ/σ vo ) 0.9m 2.9m 4.6m 6.4m Control points [Fioravante 2002] 29

30 IV Non-Displacement Piles Critical state approach Importance of the soil initial state parameter? - Regarding the interface roughness e CSL - Regarding the pile length e 1 e 2 p o /p co - Regarding the type of loading - Pile resistance for a soil with previous loading history p 0(1) p co p 0(2) ln p State parameter: p o /p co 30

31 IV Non-Displacement Piles Influence of soil initial state Rough soil-pile interface : Toyoura sand: different initial states, TS24, 40, 93 Q s - Shaft resistance Q b - Base resistance 31

32 IV Non-Displacement Piles Influence of soil initial state Rough soil-pile interface : 1) Different initial states, TS24, 40, 93 2) Initial stress: z=1.61 e TS24 TS40 TS93 p co(ts24) p co(ts40) p co(ts93) CSL p 1 ln p 1 (1.6m) Control points 32

33 IV Non-Displacement Piles Influence of soil initial state Rough soil-pile interface : 1) Different initial states, TS24, 40, 93 2) Different initial stress e Final local load transfer coefficient: β = τ/σ vo TS24 TS40 TS93 p co(ts24) p co(ts40) p co(ts93) CSL p (L) ln p p=0 L p (L) 33

34 IV Non-Displacement Piles Parametric studies Influence of : Soil initial state Pile-soil surface roughness Pile length In : Shaft resistance Base resistance Average load transfer coefficient: β avg β avg = Q s /(As.σ vo avg ) rough TS24, 40, 93 34

35 IV Non-Displacement Piles Parametric studies Influence of : Soil initial state Pile-soil surface roughness Pile length In : Shaft resistance Base resistance Average load transfer coefficient: β avg β avg = Q s /(As.σ vo avg ) Smooth interface : No soil-pile interaction rough smooth TS24, 40, 93 35

36 IV Non-Displacement Piles Parametric studies Influence of : Soil initial state Pile-soil surface roughness Pile length In : Shaft resistance Base resistance End bearing capacity factor: N*q N q =q b /σ vo TS24, 40, 93 36

37 IV Non-Displacement Piles Cyclic Load transfer Mechanisms Motivation: Reduction of radial and shear stress during pile penetration Linked to: cyclic stress history Main issue: Identify the main mechanisms controlling friction fatigue Key issues: - Effect of the load level (5 amplitudes) - Effect of the number of cycles - Effect of cycling in subsequent reload to failure Q T (t) Q T Q s (t) t Q b (t) 37

38 IV Non-Displacement Piles Cyclic Load transfer Mechanisms Effect of the load level: - Reduction of shaft resistance with N; - Progressive increase of the permanent settlement (s) - Progressive mobilization of base resistance - Different levels of residual loads Q s - Shaft resistance Q b - Base resistance 38

39 IV Non-Displacement Piles Cyclic Load transfer Mechanisms Analysis of volume changes in the soil: 4.4m Control point (N=5, Q T =1300 kn) Mechanism 1 39

40 IV Non-Displacement Piles Cyclic Load transfer Mechanisms Analysis of volume changes in the soil: 4.4m Control point zoom (N=5, Q T =400 kn) Mechanism 2 40

41 IV Non-Displacement Piles Cyclic Load transfer Mechanisms Analysis of volume changes in the soil: 4.4m Control point zoom (N=5, Q T =200 kn) Mechanism 3 41

42 IV Non-Displacement Piles Cyclic Load transfer Mechanisms In reload: soil-pile system improved with the previous application of cycling? M3 M2 - shaft resistance M1 M1 Q s - Shaft resistance M1 M2 M3 initial state State after cycling (z=-4.4 m) Q b - Base resistance 42

43 IV Non-Displacement Piles Cyclic Load transfer Mechanisms Is the soil-pile system improved with the previous application of cycling? - Shaft may not recover - Base can be improved; - Total resistance beneficiates: from previous cycling at large load amplitude; - Friction fatigue: compensated by the base mobilization during cycling M1 Q s - Shaft resistance M3 M2 M1 Q T - Total resistance Q b - Base resistance 43

44 Short Outline I - Introduction II - Theory and Numerical modeling III - Soil-Structure Interface behavior IV - Non-displacement Piles in Toyoura sand V - ISC 2 Experimental site ISC 2 Site description Residual soil modeling Static load test Dynamic load test VI - Final remarks 44

45 V ISC 2 Used data 2 nd International Conference on Site Characterization - University of Porto, Portugal International Prediction Event on the Behavior of Bored, CFA and Driven Piles Viana da Fonseca et al. (2004) Static and Dynamic Load tests - Bored Piles - CFA Piles - Driven Piles in situ tests - SPT, CPT, DMT, PMT, CH + Laboratory tests - Triaxial compression, extension; Resonant column and oedometer + Used to calibrate soil model s parameters (D Aguiar (2008)) 45

46 V ISC 2 Residual soil modeling Laboratory tests carried out in undisturbed samples: Triaxial tests Specimen S2/1 S2/5 S2/6 Depth (m) S r (%) σ cv σ ch φ pp = 32º K 0 = 0.5 experimental simulation 46

47 V ISC 2 Residual soil modeling Laboratory tests Specimen S5/2 Depth (m) 6.3 γ (kn/m 3 ) 17.8 e S r (%) 73 Specimen S5/1 S5/3 Depth (m) S r (%) σ cv σ ch simulated measured Oedometer test column test Resonant column test simulated with drained cyclic shear test 47

48 V ISC 2 Static Load Tests Static load: Bored and CFA Pile (E9 and T1) Assumed hypothesis: no modeling of installation effects Rough interface Bored pile (E9) CFA pile (T1) Simulated Experimental Q T Q T Q s Q s Q b Q b Bored Pile: E=20GPa CFA Pile: E=40GPa 48

49 V ISC 2 Static Load Tests Static load: Bored and CFA Pile (E9 and T1): Load distribution along depth Bored pile E9 CFA pile T1 Simulated Experimental Pile instrumentation Pile Axial Load (N) distribution along depth Pile Axial Load (N) distribution along depth 49

50 V ISC 2 Dynamic load test modelling Different simulations Drop h Drop hammer h h height: h 1 pile Wave propagation soil Blow 1 Blow 2 Blow 3 Blow 4 Simulation 1 Simulation 2 Simulation 3 Simulation 4 Simulation 1 Independent blow simulation + Simulation 2 + Simulation 3 + Simulation 4 Sequentially applied blow simulation 50

51 V ISC 2 Dynamic load test records Forcing functions Drop hammer pile soil Drop hammer: 80kN Forces at the pile head for the DLT, bored pile (E9) Blow m 0.8 m 1.1 m 1.4 m Fall heights for bored pile (E9) 51

52 V ISC 2 Independent blow simulation Velocities at the pile head measured computed 52

53 V ISC 2 Sequentially applied blows simulation Complete load history ( pre-blow + blow 1,2,3 and 4) Consistency of the dynamic load tests results measured computed Velocity at the pile head 53

54 V ISC 2 Sequentially applied blows simulation Influence of the loading history in each blow response ( pre-blow + blow 1 / blow 1) Velocity at the pile head Q s - shaft resistance Seq. ( pre-blow + blow 1) Indp. (blow1) Q b - base resistance 54

55 Short Outline I - Introduction II - Theory and Numerical modeling III - Soil-Structure Interface behavior IV - Non-displacement Piles in Toyoura sand V - ISC 2 Experimental site VI - Conclusions and further research 55

56 VI Conclusions and further research Conclusions The main issue: 3D non-linear numerical modelling of soil-pile load transfer mechanisms of non-displacement piles axially loaded The base of the work : Field observations In-situ Computer simulations In-silico In-vitro Laboratory experimental tests adapted from Barends [2005] Aiming to include the main physical constrains of the soil-pile interaction problem in a FEM model 56

57 VI Conclusions and further research Conclusions Spoted key aspects: Soil-pile surface normalized roughness Similitude between soil and soil-structure interface Soil initial state State parameter Soil loading history Allowing: The validation of a complex model but sufficiently flexible and adapted to real case studies Gained important insight, even if qualitative, concerning installation effects However: Parameters identification 57

58 VI Conclusions and further research Further research Soil-Structure interface model: More validations for cyclic loading and different soils; Water presence (coupled hydromechanical simulations): Soil model is adapted; Adapt the interface model; Validations; Installation effects of displacement piles 58

59 Thank you 59

60 Appendix 60

61 II Theory and Numerical modeling Concluding Remarks The GEFDYN FEM code Enhanced: 3D Axisymmetric formulation of the 3D ECP soil model (Hujeux model) 3D interface model based on the critical state concept Validated: With other published numerical results 61

62 III Interface behavior Concluding Remarks Soil-Structure interface behavior: 3D ECP Interface model - Validated with laboratory CNS shear test - Correctly captures: - Contractancy/dilatancy - Influence of normal stress - Coupling of shear and normal displacements - Critical state Advantage of the 3D ECP interface: - Same physical principles of the ECP soil model Inconveniences of the 3D ECP interface: - Large number of parameters Coherent strategy for parameters identification Flexibility in application 62

63 IV Non-Displacement Piles Concluding Remarks Newly implemented model: - Adapted for pile applications - Validated with centrifuge test results (non-displacement piles) - Flexible application to different soil-pile interface conditions Parametric studies in Toyoura sand: - Spot the key features of soil-pile interaction - normalized roughness - soil initial state - Proposal of β and N q parameters Friction fatigue: - Three different cyclic mechanisms control the rate of friction degradation - Related to the state parameter - Amplitude of the first load cycle - Number of cycles - Soil memory in the reload resistance (parallel with installation effects) 63

64 V ISC 2 Concluding Remarks ISC 2 Experimental site: non-displacement piles Consistent results compared with in-situ static load tests results (bored and CFA piles) Consistency of the modeling based on: - Non linear behavior of: - Soil - Interface - Important work in parameters identification - Correct assumption of soil-pile interface (rough interface) Compared performance of bored and CFA piles Estimation of CFA pile residual loads 64

65 Employed and developed Tools Comparison between Multimechanism and ECP interface models: 65

66 II Theory and Numerical modeling Employed and developed Tools ECP elastoplastic Constitutive models : Parameters 66

67 II Theory and Numerical modeling Employed and developed Tools Multiméchanism Interface Elasticity Yielding function Volumetric hardening Deviatoric hardening Plastic potential Behavior domains if r k < r hys if r hys < r k < 1 if r hys < r k < 1 Isotropic mechanism 67

68 II Theory and Numerical modeling Employed and developed Tools 3D Axisymmetric formulation of the ECP multimechanisms model: 3D CV shear test (axisymmetric conditions) τ s * z y x * τ s shear stress 68

69 II Theory and Numerical modeling Numerical Validation GEFDYN FEM code for pile applications Soil : Mohr-Coulomb model Interface: Mohr-Coulomb model FEM codes: ABAQUS and CESAR (LCPC) Case 1 Case 2 ABAQUS GEFDYN Q T Q s CESAR GEFDYN Q b Trochanics [1991] Neves et al. [2001] 69

70 III Interface behavior Simulation of Direct Shear Test Boundary conditions and model us σ n =cst us us K=cst Case 1: CNL Case 2: CV Case3 : CNS 70

71 III Interface behavior Simulation of Direct Shear Test 71

72 III Interface behavior CNL test Simulation of Direct Shear Test influence of interface thickness (t) 72

73 III Interface behavior Simulation of Direct Shear Test Parameters determination: CNL test influence of interface thickness (t) λ (1+e0) λ(t) (1+e0) 73

74 III Interface behavior Simulation of Direct Shear Test Parameters determination: CNL test influence of interface thickness (t) G*=G soil t 74

75 IV Non-Displacement Piles Parametric studies Influence of : Soil initial state Pile-soil surface roughness Pile length In : Shaft resistance Base resistance Average load transfer coefficient: β avg Short β avg = Q s /(As.σ vo avg ) Smooth interface : No soil-pile interaction TS24, 40, 93 Long rough smooth 75

76 IV Non-Displacement Piles Pile length Parametric studies Local load transfer coefficient: β β= τ/σ vo 76

77 IV Non-Displacement Piles Cyclic Load transfer Mechanisms Increasing number of cycles for mechanism 2 Reduction of the maximum resistance Total Load 77

78 IV Non-Displacement Piles Cyclic Load transfer Mechanisms Increasing number of cycles for mechanism 3 2 depths stress path N=10 monotonic N=5 N=10 No stress (σ 3 =0) 78

79 IV Non-Displacement Piles Cyclic Load transfer Mechanisms Increasing number of cycles for mechanism 1 (N=12, Q T =200 kn) 79

80 V ISC 2 Static Load Tests Compared performance of bored and CFA piles (E9 and T1): Measured resistance Bored E9 CFA T1 Q T Q s Q b Simulated base resistance Q b - Base resistance 80

81 V ISC 2 Static Load Tests Estimation of residual loads: CFA pile Two steps modeling : 1) Base pre-loading with s/d=5% 2) Load removal residual loads locked-in soil state modification Calculation 1: load+unload s Q=0 Q sres Q bres

82 V ISC 2 Static Load Tests Estimation of residual loads: CFA pile Residual load =76 kn Experimental Static load test Simulated: load + unload Simulated: reload after load +unload Q b - Base resistance 82

83 V ISC 2 Dynamic Load test V(t) Drop height: h t L Wave propagation x L c t wave refraction F, Force v, velocity Z, impendency 83

84 V ISC 2 Independent blows simulation Comparison between static and dynamic load tests Resistance mobilization load settlement results Q s - shaft resistance Q b - base resistance Q T = Q s + Q b 84

85 V ISC 2 Sequentially applied blows Complete load history ( pre-blow + blow 1,2,3 and 4) Load settlement response Q s - shaft resistance Q b - base resistance Q T = Q s + Q b 85

86 V ISC 2 Resistance after dynamic test Influence of the soil state and loading history: application of dynamic load modifies the pile static resistance Simulation 1: Virgin SLT Static load test from initial stress field STATIC LOAD TEST Simulation 2: After DLT Static reload from stress field generated in the DLT DYNAMIC LOAD TEST + Load settlement response STATIC LOAD TEST 86