Paper No.

PILTA

PILES IN LATERAL SPREADING: A SIMPLE METHOD VERSUS CENTRIFUGE EXPERIMENTS Panagiota TASIOPOULOU 1, Nikos GEROLYMOS2, George GAZETAS3 ABSTRACT The paper presents a new physically simplified methodology for computing displacements and structural forces on piles under conditions of lateral spreading. The main method avoids the associated empirical selection of stiffnessreduction factors and does not involve the use of p-y curves. Instead, it introduces an elastic pseudo static numerical analysis of a horizontal slice from the liquefied layer including the piles, in order to estimate the soil-pile interaction under soil flow conditions in quantitative terms. This interaction is determined as a function of the relative stiffness between the pile and the liquefied soil. The pile stiffness is assumed to remain linear -elastic throughout the analysis. The predictions of the method compare well with results from centrifuge tests. Keywords: Lateral spreading; pile; liquefaction; simplified method

INTRODUCTION Numerous numerical and experimental studies have been performed during the last decade in order to figure out the mechanism of soil-pile interaction under soil flow triggered by extensive liquefaction. The moving soil mass provides the driving force displacing the pile a certain amount, which depends on the relative stiffnesses between the pile and the liquefied soil (Boulanger et al., 2003). Thus, the magnitude of the soil movement, the lateral load of the surficial non-liquefiable soil layer and the stiffness degradation in the liquefied zone are the key parameters which need to be taken into account when evaluating the pile response due to soil flow (Cubrinovsky et al., 2004). In engineering practice, several methods have been formulated. In general, they can be classified into three categories: (a) the force methods, including the Japanese Road Associate Method (JRA, 1996), the limit equilibrium method (Dobry and Abdoun, 2000), and the viscous fluid method (Hamada 2000, Yasuda et al., 2001); (b) the displacement methods, or else known as pseudo-static beam on nonlinear Winkler foundation methods according to which the free field soil displacement is imposed to the pile through empirical p-y springs (Boulanger et al., 2003); (c) the hybrid forcedisplacement methods, which are a combination of the first two (Cubrinovsky and Ishihara, 2004). These methods are mainly single pile analyses dependent on soil profile and problem geometry. Moreover, many assumptions are made regarding the stiffness degradation in the liquefied layer and soilpile interaction issues, such as the appropriate direction of the force exerted on the pile by the upper nonliquefiable layer. Inevitably, considerable uncertainty is hidden behind all methods of post liquefaction analysis (Finn and Thavaraj, 2001).
1

Ph. D. Candidate, School of Civil Engineering, National Technical University, Athens, Greece, ptasiopoulou@gmail.com 2 Lecturer, School of Civil Engineering, National Technical University, Athens, Greece. 3 Professor, School of Civil Engineering, National Technical University, Athens, Greece.

5th International Conference on Earthquake Geotechnical Engineering January 2011, 10-13 Santiago, Chile In this study, a new simple physical method is presented, appropriate for all kinds of soil profile and types of pile configuration (single piles, pile groups). The method falls into the "displacement" induced category, but avoids the associated empirical selection of stiffnessreduction factors and does not involve the use of p-y curves. It is applied to three different centrifuge experiments (Tazoh et al., 2005, Sato et al., 2001, Abdoun & Dobry, 1998), reproducing the test results with satisfying engineering accuracy (Tasiopoulou et al., 2009). However, herein, only the application to the centrifuge test by Tazoh et al. (2005) involving a 2x2 pile group will be demonstrated.

A NEW PHYSICALLY SIMPLIFIED METHOD The analysis comprises three consecutive steps. In STEP 1, the soil response without the piles is determined as in Fig. 1. To this end, a 2D effective stress numerical analysis is performed of a vertical section of the problem and its soil profile. Ground inclination, topographic irregularities such as river banks or existing structural boundaries such as quay-walls are included in the model, but not the piles. Thus, the "free field" soil response is obtained at the location to be occupied by the pile foundation, in terms of: the distribution of the horizontal soil displacement with depth, the depth and the thickness of the liquefied zone, the strength degradation in the liquefied layer and the shear strain distribution with depth.

Although the dynamic numerical analysis provides the required results at all times, the residual values after the end of the shaking are of greater importance, due to the accumulative nature of the liquefaction induced soil flow.

Figure 1. Problem definition.

In STEP 2, a horizontal section through the middle of the liquefied layer including the piles is analysed in 2-D, subjected to the "free-field" unit displacement at the boundary. The ratio of the pile displacement over the soil displacement in the free-field (adequately away from the piles) is estimated, hereafter termed

5th International Conference on Earthquake Geotechnical Engineering January 2011, 10-13 Santiago, Chile ratio as depicted in Fig. 2 in a plan view. This ratio of displacements reflects the soil-pile interaction due to soil flow, in quantitative terms. An elastic pseudo-static plane strain analysis of the horizontal section is conducted (Fig. 3). The pile group configuration is well represented, as is the liquefied soil layer with uniform shear modulus, GL, surrounding the piles. The pile sections are simulated as rigid inclusions into the liquefied soil, GL. In order to attribute the appropriate horizontal resistance to the pile sections against the moving soil mass, an out of plane horizontal spring, K, is connected to each pile section. It is evident that the effectiveness of the numerical analysis of the horizontal slice at the mid-depth, zs depends on the appropriate calibration of the following "key" parameters: the horizontal stiffness of each pile section, K, and the shear modulus of the liquefied soil, GL. Horizontal stiffness, K of each pile section Every single pile of the pile group can be simulated as a vertical beam element, a column, with appropriate kinematic constraints at the boundaries (boundary conditions). The rotation at the top depends on the rigidity of the pile cap, the number of piles and their axial and lateral stiffness. Increasing any of these parameters tends to restrain the rotation at the pile head.

Figure 2. Definition of the ratio a as the ratio of the residual free-field soil displacement over the residual pile group displacement due to liquefaction induced soil flow (plan view).

5th International Conference on Earthquake Geotechnical Engineering January 2011, 10-13 Santiago, Chile

Figure 3. Numerical model of the horizontal slice. The pile sections sustain the same horizontal displacement due to the pilecap constraint.

The active length of the pile depends on the depth to fixity, below the liquefied layer. The depth to fixity is defined in literature as the length of an equivalent column of the same stiffness with the pile stiffness to produce the same displacements under lateral load. Depth-to-fixity values can be estimated from available charts in dimensionless form based on subgrade reaction modulus, pile diameter and rigidity, provided by several researchers (Priestley et al. 1996, Budek et al. 2000, Caltrans 1990). In general, the depth to fixity is measured from the ground surface. However, in the current case of liquefaction induced soil-flow, where the liquefied soil cannot provide significant support to the piles (while flowing around them), the depth to fixity can be measured from the end of the liquefied zone. Therefore, the active length of the pilecolumn can be estimated by adding the depth to fixity to the portion of the pile embedded in the liquefied zone and its overlying layers. In retrospect, every single pile is equivalent to a column of a certain length fixed at the bottom with a certain degree of rotational freedom at the top. The pile slice in the numerical model is just a section of the mid-depth of the liquefied layer. Thus, the horizontal stiffness, K, of each pile section, is defined as the point load that must be exerted on the pile-column at that depth, to cause a unit displacement at the same depth. Shear modulus, GL of the liquefied soil In the framework of an elastic analysis of the horizontal slice, an equivalent linear shear modulus of the liquefied soil can be estimated by the following equation: (1) where, HL is the thickness of the liquefied zone, is the residual shear stress and is the accumulated shear strain after the end of shaking, obtained from the numerical analysis of the free field (STEP 1). This is how the stiffness degradation of the soil due to liquefaction is taken into account. Herein, the elastic pseudo-static analysis of the horizontal slice has been conducted for three different pile configurations: single pile 2x2 pile group (Tazoh et al., 2005) and

5th International Conference on Earthquake Geotechnical Engineering January 2011, 10-13 Santiago, Chile 2x8 pile group (Sato et al., 2003). The ratio , is obtained as function of the relative stiffness between the pile and the liquefied soil, K/GL, shown in Fig. 4 for the abovementioned different pile configurations. When the relative stiffness tends to zero (K/GL 0), the ratio tends to unity, which means that the pile sections move just like the soil, as a rigid inclusion. On the contrary, when the relative stiffness tends to infinity (K/GL ), the ratio tends to zero. This is due to the fact that the soil has practically zero shear strength (GL 0) and flows around the piles without exerting any significant load on them. Moreover, Fig. 4 indicates that increasing the number of piles, the resistance of the foundation to the moving soil mass becomes stronger. It is worth mentioning that the numerical modeling of the pile-group section into the liquefied soil is based on the assumption that due to the pile-cap constraint the piles sustain the same horizontal displacement over all their depth.

Figure 4. The ratio as a function of the relative stiffness between the pile and the liquefied soil for three different pile configurations: single pile, 2x2 and 2x8 pile group, obtained from several parametric numerical analyses of the horizontal slice. In retrospect, as long as the relative stiffness, K/GL, is determined, the ratio can be estimated. Eventually, multiplying the ratio with the free-field soil displacement at the mid-depth of the liquefied zone, at the position which the piles occupy behind the quay-wall, as obtained from the free field analysis (STEP 1), the pile displacement can be calculated at the mid-depth of the liquefied zone. In STEP 3, the total pile deformation at all depths and mainly the pile displacement at the top, is evaluated. The deformation shape of each pile, simulated as a beam element, is defined primarily by its boundary conditions and only marginally by the load distribution along it. In other words, it has been found that the shape of the load distribution (uniform, p; triangular, 0 p; trapezoidal, 0.5p 1.5p) along the pile only slightly affects the response of the pile in terms of displacements. This observation simplifies the analysis of pile response: a potential load distribution (e.g. trapezoidal) is chosen to be imposed on the pile. We, thus, obtain its deformation shape as a function of the unknown p.

5th International Conference on Earthquake Geotechnical Engineering January 2011, 10-13 Santiago, Chile Using the known pile displacement at the mid-depth of the liquefied zone (from STEP 2), the shape function of the pile with depth is calibrated and thus, the unknown load intensity, p is estimated. Finally, the distribution of the pile displacements with depth, as well as the bending moments along the pile, are determined by imposing the already fully known load distribution on the pile-column. The parameters of the method are summarized in Table 1.

Table 1. The parameters of the proposed simple method.

Soil GL HL zs uff shear modulus of liquefied soil thickness of liquefied zone mid-depth of liquefied zone measured from the ground surface free-field horizontal soil displacement at the characteristic depth and at the location potentially occupied by the pile foundation Pile mid-depth of liquefied zone measured from the top of the pile active length of the pile horizontal stiffness of the pile-column at the characteristic depth Soil-Pile Interaction K/GL ratio p relative stiffness between the pile and the liquefied zone at the characteristic depth the ratio of horizontal pile displacement over the horizontal soil displacement at the characteristic depth load intensity describing the potential load distribution along the pile

zp

l
K

COMPARISON WITH CENTRIFUGE RESULTS Several series of dynamic centrifuge experiments were conducted at Shimizu's Institute of Technology, in Japan (Tazoh et al., 2005), to evaluate the response of pile-foundation systems triggered by liquefaction induced soil flow after quay-wall collapse. The test, chosen to be analysed with the proposed simple method, is illustrated in Fig. 5 in prototype scale. It includes a 2x2 pile group fixed at the bottom and placed 3m away from the floating quay-wall. The soil profile contains a liquefiable layer of relatively loose sand (HL = 3.6 m), underlying a surficial unsaturated non-liquefiable crust. The input base accelerogram is also shown at the base of the model in Fig. 5 in prototype scale. The first step of the method involves the centrifuge model, as depicted in Fig. 5, without the pile group; it is simulated and analysed numerically using the finite difference code FLAC. The numerical simulation involves the constitutive law by Byrne (1991) for pore pressure generation which is incorporated in the standard Mohr-Coulomb plasticity model. The deformed grid after the end of shaking is shown in Fig. 6, indicating a relative horizontal displacement of the top of the quay-wall equal to 1.35m. It is worth mentioning that the displacement of the quay-wall recorded during the centrifuge test was less than 0.8 m but with the pile group.

5th International Conference on Earthquake Geotechnical Engineering January 2011, 10-13 Santiago, Chile

Figure 5. Geometry and soil properties of the centrifuge model of test Case CD, side D into the laminar box, in prototype scale (Tazoh et al., 2005). The input motion is illustrated at the base of the model in prototype scale.

Figure 6. Initial finite difference mesh (above) and deformed grid (below) of the numerical model without the piles. The residual horizontal displacement at the top of the quay-wall is estimated 1.35 m after the end of shaking.

5th International Conference on Earthquake Geotechnical Engineering January 2011, 10-13 Santiago, Chile Liquefaction takes place within the loose sand layer, according to the excess pore pressure ratio time histories (Fig. 7), obtained numerically. Liquefaction has indeed been observed during the centrifuge experiment. It should be made clear, that direct rigorous comparison between the centrifuge and the numerical results cannot be performed at this stage due to the presence of piles in the centrifuge test and their absence in the numerical model. However, a relatively similar response of the soil is expected away enough from the piles. In this framework, a comparison of the acceleration time histories within the liquefied layer (3m below the ground surface) between experimental and numerical results has been attempted in Fig. 8. There are significant similarities between the two accelerograms, such as cut-off of the acceleration, due to the occurrence of liquefaction, increasing the validity of the numerical analysis of the free-field response. Admittedly, differences also exist, especially in the shape of the individual acceleration cycles.

Figure 7. Computed time histories of excess pore pressure ratio at depths of 4, 3.5, 5.5 and 6 m below the ground surface, in the free-field (without the presence of piles).

Figure 8. Above: acceleration time history recorded during the centrifuge experiment 3m below the ground surface away from the piles. Below: acceleration time history in the free field obtained numerically.

5th International Conference on Earthquake Geotechnical Engineering January 2011, 10-13 Santiago, Chile The distributions of the residual shear stresses and strains as well as of the horizontal soil displacements are obtained at a distance of 3m behind the quay-wall (at the location to be occupied by the pile group). These distributions are useful while proceeding to the second step where important parameters of the problem such as GL, uff and K/GL, ratio a and up are determined. The values of these parameters are shown in Table 2. Each pile of the pile group is considered to be equivalent to a beam element fixed at the bottom with no rotation at top.

Table 2. The values assigned to the parameters of the methodology during its application to the centrifuge experiment by Tazoh et al. (2005).
Tazoh et al. (2005) GL (kPa) Soil HL (m) zs (m) uff (m) zp (m) Pile 125 3.6 3.9 0.505 3.3 8.1 585 4.6 0.146 0.074 3.8 4.2 4

l (m)
K (kN/m) K/GL (m) ratio

Soil-Pile Interaction

up (m) puniform (kN/m) ptriangular (kN/m) ptrapezoidal (kN/m)

At the end of second step, the pile horizontal displacement at that depth has been estimated. To obtain estimates of the distributions of horizontal displacements and bending strains three different load distributions (uniform, triangular and trapezoidal), of yet unknown intensity p, are imposed on the pilebeam element. The deformation line of the pile is thus obtained as a function of p, which can be determined from one known point of the pile displacement at the mid-depth of the liquefied zone (zp,up). Then, the bending strains along the pile are obtained through double differentiation of the displacements. The displacements and bending strains along the pile are depicted in Fig. 9, and they are compared with the experimental recordings showing satisfying agreement. In particular, the trapezoidal distribution seems to give the most accurate prediction of both displacements and bending strains.

5th International Conference on Earthquake Geotechnical Engineering January 2011, 10-13 Santiago, Chile
Horizontal Pile Displacement (m)
0 0 1 2 3 0.02 0.04 0.06 0.08 0.1 0.12

Pile Bending Strains (10 )

-1500 -1000 -500 0 1 2 3 0 500 1000 1500

-6

Depth z (m)

4 5 6 7 8

Depth z (m)

4 5 6 7 8

Analysis (Uniform load) Prediction

Prediction Analysis (Trianglular load) Prediction Analysis (Trapezoidal Load) Centrifuge test (Tazoh et al. 2005)

Figure 9. Estimated range of the horizontal displacement and bending strains along the piles, as obtained from the simplified methodology, in comparison with the centrifuge results by Tazoh et al. (2005).

CONCLUDING REMARKS In this paper, a simple physically-motivated methodology is highlighted for the evaluation of pile response due to liquefactioninduced soil flow. The main characteristics of this methodology are: It avoids the associated empirical selection of stiffnessreduction factors and does not involve the use of p-y curves. It combines the results of analysis of a vertical 2D section of the geometry without the presence of piles, and of the pseudo static analysis of a horizontal 2D slice containing the piles. Interaction between soil and piles is determined as a function of the relative stiffness between pile and liquefied soil. The effectiveness of the new methodology in combination with suitable engineering judgment and reasonable assumptions can provide sufficient accuracy for designing pile groups against liquefactioninduced large soil displacements. AKNOWLEDGEMENTS The work of this paper was conducted for the project DARE, financed by a European Research Council (ERC) Advanced Grant under the Ideas Program in Support of Frontier Research [Grant Agreement 228254].

5th International Conference on Earthquake Geotechnical Engineering January 2011, 10-13 Santiago, Chile REFERENCES Abdoun T.H., and Dobry R. (1998). Seismically Induced Lateral Spreading of Two -layer Sand Deposit and its Effect on Pile Foundations, Proc. Intl. Conf. Centrifuge 98 (T. Kimura, O. Kosakabe and J. Takemura, eds.), Tokyo, Japan, Sept. 23-25, Vol. 1, pp. 321-326. Boulanger R. W., Kutter B. L., Brandenberg S. J., Singh P., and Chang P. (2003). Pile Foundations in Liquefied and Laterally Spreading Ground during Earthquakes: Centrifuge Experiments and Analyses, Report No. UCD/CGM-03/01, Department of Civil and Environmental Engineering, University of California at Davis. Budek, A.M., Priestley, M. J. N., and Benzoni, G. (2000). Inelastic seismic response of bridge drilled shaft RC pile/columns, Journal of Geotechnical and Geoenvironmental Engineering, 126(4), 510 517. Byrne, P. (1991). A Cyclic Shear-Volume Coupling and Pore-Pressure Model for Sand, Proceedings of Second International Conference on Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics, St. Louis, Missouri, Paper No. 1.24, pp. 47- 55. Caltrans (1990). Bridge Design Specifications / Seismic Design References, Caltrans, Sacramento, California. Cubrinovsky M., and Ishihara K. (2004). Simplified Method for Analysis of Piles undergoing Lateral Spreading of Liquefied Soils, Soils and Foundations, Vol. 44, No. 5, pp. 119-133. Cubrinovsky M., Kokusho T., and Ishihara K. (2004). Interpretation from large-scale shake table tests on piles subjected to spreading of liquefied soils, Proceedings of 11th Int. Conf. Soil Dynamics and Earthq. Engrg. / 3rd Int. Conf. Earthq. Geotech. Engrg., Berkeley, USA, Vol. 2, pp. 463-470. Dobry R., and Abdoun T. H. (1998). Post-Triggering Response of Liquefied Soil in The Free Field and Near Foundations, State-of-the-art paper, Proc. ASCE 1998 Specialty Conference on Geotechnical Earthquake Engineering and Soil Dynamics (P. Dakoulas, M. Yegian and R. D. Holtz, eds.), University of Washington, Seattle, Washington, August 3-6, Vol. 1, pp. 270 300. Dobry R., and Abdoun T. H. (2000). Recent studies on seismic centrifuge modeling of liquefaction and its effect on deep foundations, Proceedings of the fourth International Conference on Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics , San Diego. Dobry R., Abdoun T.H., and O'Rourke T. D. (1996). Evaluation of Pile Response Due to Liquefaction Induced Lateral Spreading of the Ground, Proc. 4th Caltrans Seismic Research Workshop, Sacramento, CA, July, 10 pages. Finn Liam W. D., and Thavaraj T. (2001). Deep Foundations in liquefiable soils: Case Histories , centrifuge tests and methods of analysis , Proceedings of the fourth International Conference on Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics, San Diego. Hamada M. (2000). "Performances of Foundations Against Liquefaction-Induced Permanent Ground Displacements", Proceedings of 12th World Conference Earthquake Engineering, pp1754-1761. Japan Road Association (1996). Specification for Highway Bridges. (in Japanese) Priestley, M. J. N., Seible, F., and Calvi, G. M. (1996). Seismic design and retrofit of bridges, WileyInterscience, New York. Sato M., Tazoh T. and Ogasawara M. (2001). Reproduction of Lateral Ground Displacement and Lateral Flow Earth Pressure Acting on Pile Foundations using Centrifuge Modeling, Proceedings of the Fourth International Conference on Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics, San Diego. Tasiopoulou P., Gerolymos N. and Gazetas G. (2009). Soil-Pile Interaction due to Liquefaction-Induced Soil Flow, after Large Displacements of Quay-wall, Report LSM.NTUA-09-03, School of Civil Engineering, National Technical University of Athens, Greece, Laboratory of Soil Mechanics. Tazoh T., Sato M., and Gazetas G. (2005). Centrifuge Tests on PileFoundationStructure Systems Affected by LiquefactionInduced Soil Flow after Quay Wall Failure, Proceedings of the 1st Greece

5th International Conference on Earthquake Geotechnical Engineering January 2011, 10-13 Santiago, Chile Japan Workshop: Seismic Design, Observation, and Retrofit of Foundations, Athens, Oct. 11-12 pp. 79-106. Yasuda S. (2004). " Evaluation of Liquefaction-induced Deformation of Structures", Recent Advances in Earthquake Geotechnical Engineering and Microzonation, Geotechnical, Geological, and Earthquake Engineering, 2004, Volume 1, pp. 199-230. Yasuda S, Abo H, Yoshida N, Kiku H, UdaM (2001). "Analyses of liquefaction-induced deformation of grounds and structures by a simple method", Proceedings of the 4th International Conference on Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics, Paper No. 4.34.