DISPLACEMENT ANALYSIS OF SUBMARINE SLOPES USING ENHANCED NEWMARK METHOD N. ZANGENEH and R. POPESCU Faculty of Engineering & Applied Science, Memorial University, St. John’s, Newfoundland, Canada A1B 3X5 Abstract The Newmark method for predicting seismically induced displacement of slopes is enhanced by introducing of the yield (threshold) acceleration on the real soil strength, accounting for the effects of pore water pressure. Seismically induced pore water pressure build-up and its dissipation after the earthquake are calculated as a function of soil properties and ground motion characteristics, based on well-recognized practical methods. The proposed model provides more realistic predictions of slope displacements. It is validated based on centrifuge experimental results. Keywords: Newmark method, submarine slopes, pore water pressure build-up and dissipation, seismic loading. 1. Introduction Displacement analysis is a more rational alternative to pseudo-static seismic analysis of slope stability. Newmark (1965) introduced a limit-equilibrium-based displacement analysis method that predicts the displacements of an infinite slope during an earthquake based on a soil strength–dependent yield acceleration and purely kinematic criteria. As opposed to the pseudo-static method of slope stability analysis, which provides a factor of safety for some very short time instants during an earthquake, the Newmark method can provide a prediction of slope performance based on the total displacement at the end of shaking. In many applications of the Newmark method, the yield acceleration is assumed constant during the earthquake. When applying the method to saturated granular soils, however, due to the build-up of excess pore water pressure (EPWP), soil strength and consequently the yield acceleration will decrease. In addition, after the end of shaking the generated pore pressures start to dissipate with time, which results in increasing soil strength and yield acceleration. In this paper, the effects of EPWP build-up/dissipation are investigated, and a procedure for calculating permanent displacements of submarine slopes subjected to seismic loads is introduced. The method is based on the algorithm proposed by Newmark, and it uses state-ofpractice methods for estimating EPWP build-up and dissipation. The results of the analysis are also verified using centrifuge test results. 2. Analysis Procedure Non-cohesive soils may experience significant pore water pressure build-up due to cyclic or earthquake loading. In the limit, it can lead to a state of zero effective stress 193 194 Zangeneh and Popescu and soil liquefaction. Therefore, in case of non-cohesive deposits, a total stress analysis is not appropriate and may give highly over-conservative results. Instead, an effective stress approach should be used to consider the effects of EPWP and changes in soil shear strength. 2.1 YIELD ACCELERATION Figure 1. Pseudo-static analysis of an infinite submarine slope. For stability analysis of an infinite slope, the failure surface is assumed as a plane parallel to the slope (Fig. 1). The factor of safety ( FS ) is expressed by the ratio of available soil shear strength ( τ f ) to the shear stress developed on the failure plane ( τ ): FS = τf τ (1) in which, soil shear strength at failure is expressed in terms of effective parameters according to the Mohr-Coulomb failure criterion: τ f = c ′ + (σ − u ) tan φ ′ (2) where c ′ is the soil effective cohesion, φ ′ is the effective internal friction angle, σ is the total stress (normal to the failure surface), and u is the total (hydrostatic + excess) pore water pressure. Therefore, the factor of safety can be written as follows: FS = c ′ + (γ ′d cos 2 β − u e − kγd sin β cos β ) tan φ ′ γ ′d sin β cos β + kγd cos 2 β (3) γ ′ is the effective (or buoyant) unit weight of soil, d the depth of failure plane, β is the slope angle, u e is the EPWP (in excess of hydrostatic) generated due to earthquake, and k is the seismic coefficient defined as the where with reference to Figure 1, ratio between the horizontal earthquake acceleration and the gravitational acceleration (g). Displacement Analysis of Submarine Slopes Using Newmark Method 195 In this study, only fully saturated soils are taken into account. The buoyant (or effective) weight of the sliding block, W ′ , is used in Equation (3) to calculate the normal effective stress. However, because it is assumed that during a seismic event, the soil behaviour is mostly undrained, the inertial force of the earthquake is applied to both soil particles and pore water. Thus, the inertial force is equal to k × Wsat . By setting the factor of safety equal to 1, the yield acceleration coefficient at each time instant t for downslope sliding can be obtained as follows: k yd (t ) = or, with and c ′ + [γ ′d cos 2 β − u e (t )] tan φ ′ − γ ′d sin β cos β γd cos 2 β + γd sin β cos β tan φ ′ (4) ru = u e / σ v′0 (the ratio between EPWP and initial effective vertical stress) σ v′0 = γ ′d cos 2 β : k yd (t ) = c ′ + γ ′d cos 2 β [1 − ru (t )] tan φ ′ − γ ′d sin β cos β γd cos 2 β + γd sin β cos β tan φ ′ (5) The yield acceleration coefficient is defined here as k y = a y / g , where a y is the yield acceleration and g is the acceleration of gravity. For the case of very mild slopes, it may be worth considering also the possibility of seismically induced upslope sliding. The yield acceleration coefficient for upslope sliding is: k yu (t ) = − c′ − γ ′d cos 2 β [1 − ru (t )] tan φ ′ − γ ′d sin β cos β γd cos 2 β − γd sin β cos β tan φ ′ (6) The sliding block downslope a bd (t ) and upslope a bu (t ) accelerations can be calculated using the following equations: a bd (t ) = (k (t ) − k yd (t ) )g cos(φ ′ − β ) cos(φ ′) (7) a bu (t ) = (k (t ) − k yu (t ) )g cos(φ ′ + β ) cos(φ ′) (8) where k (t ) ⋅ g represents the seismic acceleration time history. Finally, the slope displacement can be computed by integrating twice the block acceleration based on the direction of motion. 2.2 ESTIMATION OF EXCESS PORE PRESSURE BUILD-UP According to Seed and Idriss (1982), the rate of pore pressure development in undrained cyclic simple shear tests on most granular soils, falls within a fairly narrow range when plotted in the normalized form shown in Figure 2. Zangeneh and Popescu EPWP Ratio (ru) 196 Average (α = 0.7) (Neq/NL) Figure 2. Rate of pore water pressure build-up in cyclic simple shear tests (Seedet al. 1975). Curves such as those shown in Figure2 can be expressed by the following relation: N eq u 2 ru = e = arcsin ′ σ v0 π NL where 1 2α (9) N eq is the number of equivalent stress cycles applied to the sample up to a certain moment, NL is the number of stress cycles required to produce liquefaction, and α is called the pore pressure build-up parameter. For a real acceleration time history, N eq , can be calculated based on a procedure introduced by Seed (1975). By varying the value of α , Equation (9) can fit a large the number of equivalent stress cycles, palette of undrained pore water pressure generation curves, as shown in Fig. 3. EPWP Ratio (ru) α= (Neq/NL) Figure 3. Rate of pore pressure generation for different values of α (Seed and Idriss 1982). The results of a typical Newmark analysis, as described before, which accounts for the decrease of yield acceleration due to EPWP build-up is shown in Figure 4. As the Displacement Analysis of Submarine Slopes Using Newmark Method 197 earthquake induces a gradual increase in pore pressure, the yield accelerations decrease gradually. In this particular example, one should note that if no reduction in the yield acceleration were considered, the permanent displacement would be much smaller, and therefore, non-conservative. Also note that in Figure 4, two different permanent displacements are calculated, one of which considers the possibility of upslope sliding that is reasonable for nearly flat submarine slopes. Displacement (without upslope) Displacement (with upslope) Effect of Build-up Downslope Yield Acceleration Earthquake Acceleration Upslope Yield Acceleration Figure 4. Slope displacements considering the effect of build-up. Downslope Yield Acceleration Indefinite Displacement Constant Negative Yield Acceleration Earthquake Acceleration Upslope Yield Acceleration Figure 5. Slope displacement ignoring the effect of excess pore water pressure dissipation. 198 Zangeneh and Popescu 2.3 ESTIMATION OF EXCESS PORE PRESSURE DISSIPATION After the end of shaking the excess pore water pressure starts dissipating. In cases where soil liquefies, the yield acceleration at the end of shaking is less than zero, which results in infinite post earthquake displacements (Figure 5). In reality, however, due to dissipation of pore pressure the soil regains part of its original shear strength and the yield acceleration increases and becomes positive (Figure 6). This results in limiting the displacements. Therefore, accounting for EPWP dissipation after the earthquake will provide better prediction of post-seismic displacements. Downslope Yield Acceleration Measured Displacement Effect of Dissipation Predicted Displacement Earthquake Acceleration Upslope Yield Acceleration Figure 6. Slope displacement considering the effect of excess pore water pressure dissipation. The one-dimensional consolidation theory can be applied to estimate the EPWP dissipation rate. In a soil layer with any distribution of the initial excess pore pressure with depth [u i ( z )] , the excess pore pressure at any time, u e (t ) at the depth d is (e.g. Craig 1992): n=∞ u e (t ) = ∑ [( n =1 n 2π 2 c v t 1 2d nπz nπz ( ) sin )(sin ) exp( − )] dz u z i 2d 2d 4d 2 d ∫0 (10) d is length of longest drainage path, and c v is the coefficient of consolidation: c v = k /( m v γ w ) where k is soil permeability , m v is the coefficient of volume compressibility and γ w is the unit weight of water. In this study, it is assumed that the soil layer above the failure plane liquefies during the earthquake, therefore, u i is a linear function of z , and Equation (10) becomes: where n=∞ u e = ∑[γ ′ cos2 β ( − n =1 n 2π 2Tv nπz 4d 4d cos nπ + 2 2 sin nπ )(sin ) exp(− )] (11) nπ nπ 2d 4 Displacement Analysis of Submarine Slopes Using Newmark Method 199 3. Calibration and Validation Using Centrifuge Test Results To calibrate and verify the analysis procedure described in the previous section, the results of VELACS (Verification of Liquefaction Analysis by Centrifuge Studies, Arulandan and Scott, 1993) centrifuge test for model 2 performed by RPI (Figure 7) have been used. This test simulates an infinite submarine slope with a depth of 10 m subject to an earthquake with maximum acceleration of about 0.2g. The soil is a uniform sand with relative density Dr = 40%. The geomechanical soil properties were inferred by Popescu and Prevost (1993), based on results of laboratory soil tests. 3.1 PORE PRESSURE BUILD-UP PARAMETER (α) 3.1.1 Using the acceleration at levels of each pore pressure transducer In this section, α is back calculated for two intermediate elevations where both EPWP and accelerations were recorded, namely (P6,AH4) and (P7,AH5) as shown in Figure 7. In Figures 8a and 8b, values of α equal to 0.5, 0.7, 2, and 4 are shown for points P6 and P7, respectively. The EPWP parameter α corresponding to the best curve fit is α = 4. The value of the number of cycles to liquefaction, N L , is directly obtained from the pore pressure records and the variation of the equivalent number of cycles of the input motion with time. 3.1.2 Using the acceleration of the box (measured at the base of the model) The curve-fit procedure has been repeated for the same points at the same levels but using the centrifuge box acceleration. In this case the best curve fit can be obtained for α = 2 to 4 (Figure 9), which is close to the values obtained in section 3.1.1. Therefore, in real-life analyses, one could use the base (bedrock) seismic acceleration and still obtain acceptable prediction. Figure 7. VELACS Model #2 Configuration. 200 Zangeneh and Popescu Recorded EPWP build-up Recorded EPWP build-up α=4 α=4 α=2 α=2 α=0.7 α=0.7 α=0.5 α=0.5 a) P6 b) P7 Figure 8. Calibration of α using the acceleration at each level. Recorded EPWP build-up α=4 Recorded EPWP build-up α=4 α=2 α=2 α=0.7 α=0.7 α=0.5 a) P6 α=0.5 b) P7 Figure 9. Calibration of a using the acceleration of the box. 3.2 CALIBRATING THE COEFFICIENT OF CONSOLIDATION The values of EPWP after dissipation calculated using Equation (11) as well as the values recorded in the centrifuge test at point P6 are shown in Figure 10. Predicted using cv=5.1 m2/s Measured Figure 10. The predicted and measured excess pore water pressure values during the buildup and dissipation phases (at transducer P6 – see Figure 7). Displacement Analysis of Submarine Slopes Using Newmark Method 201 The best-fit curve of post-earthquake EPWP dissipation was obtained for cv = 5.7 m/s2. A value of cv = 5.1 m/s2 was calculated using a value k = 3.3 × 10 −3 m / s for soil permeability (Popescu & Prevost 1993) and computing the coefficient of volume compressibility as m v = 1 / B . B is the low strain bulk modulus of the soil that is a function of the average effective confining stress during the dissipation phase. 3.3 SLOPE DISPLACEMENTS CONSIDERING DISSIPATION Finally, by applying all the previously described procedures the permanent displacement of the slope is calculated and shown in Figure 11. The predicted value is satisfactorily close to the measured value. Downslope Yield Acceleration Measured Displacement Predicted Displacement Earthquake Acceleration Upslope Yield Acceleration Figure 11. The predicted and measured permanent displacements considering EPWP buildup and dissipation effects (at transducer P6 – see Figure 7). 4. Model Limitations and Further Research This paper presents a study in progress that needs further addressing of a few aspects: (1) the acceleration applied at the base of the block was that recorded in the experiment at that location (or level), and (2) the number of cycles to liquefaction was estimated from the measurements using the time at which ru reached one in the centrifuge model. These two elements are unknown when making a prediction of slope displacements. They were used in the current study to check the proposed method for calculating excess pore water pressure build-up and dissipation. In the next phase, for a given seismic motion the number of cycles to liquefaction will be estimated based on information on soil strength and maximum seismic acceleration, and using a method compatible with the current guidelines for liquefaction strength assessment (Youd et al. 202 Zangeneh and Popescu 2000) using in-situ tests results (e.g. CPT). The Newmark model assumes that slope displacements are concentrated in a narrow band, i.e. below a moving rigid block. This is a good assumption for layered soils having a weaker layer sandwitched between more resistant soils. In homogeneous soils, such as that in the VELACS model #2, the downslope displacements are distributed with depth, with maximum values at the soil surface. For such situation, the rigid block assumption is a limitation of the Newmark model. This can be mitigated by either: (1) considering the flexibility of the moving block (e.g. Rathje and Bray, 2000) and accounting for continuous softening, or (2) considering a "stack of rigid blocks". 5. Summary and Conclusions The original Newmark model has been enhanced by applying state-of-practice methods of estimating the EPWP build-up during seismic events and dissipation after the earthquake to obtain more realistic predictions of permanent slope displacements. The results have been calibrated, and validated based on centrifuge test results. The results show that the proposed procedure is promising, especially for risk assessment, involving a large number of analyses and requiring a reliable and time effective algorithm. The method has some limitations that will be mitigated as discussed in section 4. 6. References Craig, R.F. 1992. Soil Mechanics, Chapman & Hall, London Newmark, N.M., 1965. Effects of earthquakes on dams and embankments. 5th. Rankine Lecture, Géotechnique, 15(2): 137-160. Popescu, R. and Prevost, J.H. 1993. Centrifuge validation of a numerical model for dynamic soil liquefaction. Soil dynamics and Earthquake Engineering, 12: 73-90. Rathje, E. M., and Bray, J. D. 2000. Nonlinear Coupled Seismic Sliding Analysis of Earth Structure. Journal of Geotechnical and Geoenvironmental Engineering, 126(11): 1002-1014. Seed, H.B. and Idriss, I.M. 1982. On the importance of dissipation effects in evaluating pore pressure changes due to cyclic loading. Soil Mechanics - Transient and Cyclic Loads, eds., Pande, N., Zienkiewics, O.C.: 53-70. Seed, H. B., Idriss, I. M., Makdisi, F., and Banerjee, N. 1975. Representation of irregular stress-time history by equivalent uniform stress series in liquefaction analyses. Report No. EERC 75-29, Earthquake Engineering Research Center, University of California, Berkeley. Youd, T.L., Idriss, I.M., Andrus, R.D., Arango, I., Castro, G., Christian, J.T., Dobry, R., Finn, W.D.L., Harder , L.F., Hynes, M.E., Ishihara, K., Koester, J.P., Liao, S.S.C., Marcuson, W.F., Martin, G.R., Mitchell, J.K., Moriwaki, Y., Power, M.S., Robertson, P.K., Seed, R.B., and Stokoe, K.H. 2000, Liquefaction Resistance of Soils: Summary report from the 1996 NCEER and 1998 NCEER/NSF workshops on Evaluation of liquefaction resistance of soils, Journal of Geotechnical and Geoenvironmental Engineering, Vol. 127(10): 817-833.