In seismic Ground Response Analysis (GRA), the effect of local geology is significantly

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 INTRODUCTION In seismic Ground Response Analysis (GRA), the effect of local geology is significantly important since the response of any structure, subjected to the vertically propagating horizontal shear waves, depends on the regional seismicity, source mechanism, geology and local soil conditions (Boore 1972; Kramer 1996; Nath et al. 2013; Kumar et al. 2014 a,b). The bedrock motions get amplified or de-amplified within the soil stratum and, therefore, the estimation of dynamic soil properties of the subsurface of any site, prior to conducting the GRA, is extremely important. GRA is one of the important step to foresee the potential consequences of earthquake motion prior to the earthquake occurrence. Several GRA methodologies (such as linear, equivalent linear and non-linear) are available to evaluate different response parameters of the site (Kramer, 1996). The response spectrum at various layers of a stratified soil deposit, obtained as an outcome of GRA, are useful for earthquake based design of geotechnical structures. Almost all methodologies require strain dependent dynamic properties of soils, in terms of the shear modulus reduction and damping ratio curves, as the essential input parameters (Kumar et al. 2017a). Several researchers have performed GRA for Indian cities using DEEPSOIL (Hashash et al. 2016) and SHAKE2000 (Ordonez, 2000), as summarized Table 1. GRA for Guwahati city has been reported by Kumar and Krishna (2013) and Basu et al. (2017) using existing material model proposed by Seed and Idriss (1970) for sandy soil, Vucetic and Dobry (1991) for clay soil, or Ishibashi and Zhang (1993) for sandy and clay soils. Since these soil models have been established for the soils of different regions and of different compositions, the direct adoption of the same may not be suitable for the regional soils considered in the present study. Although the existing soil models were used by Kumar and Krishna (2013), Basu et al. (2017) and Kumar et al. (2017b), the use of dynamic soil properties of regional soils were not reported in the earlier literatures. Shukla and Choudhury (2012) have recommended the use of actual 1

26 27 28 dynamic soil properties for more realistic outcomes from GRA. It can be stated that, in the absence of proper site-specific dynamic soil properties, use of existing soil models might lead to inaccurate estimation of the parameters involved in the earthquake resistant designs. 29 30 31 32 33 34 35 36 37 38 This paper highlights the importance of site-specific soil model in GRA studies. In this regard, one-dimensional (1-D) equivalent linear GRA has been carried out using experimentally evaluated strain dependent dynamic soil properties (i.e. modulus reduction and damping ratio). In order to achieve this, firstly, dynamic soil properties of locally available soils in Guwahati city were determined, using cyclic triaxial tests, which were further used to conduct GRA using DEEPSOIL. As a comparative, GRA has also been carried out using existing soil models as mentioned earlier. Guwahati city, being located in one of the most seismically active regions (as per IS: 1893-2002), has been chosen for the present study. The outcomes are presented in terms of PGA, Fourier amplification ratio (FAR) and Peak spectral acceleration (PSA) for entire Guwahati city as contour maps. 39 GEOTECHNICAL CHARACTERIZATIONS OF SOIL 40 41 42 43 44 45 46 47 48 The characteristics of geological layers, especially up to a depth of 30 m from the ground surface, plays a dominant role in earthquake geotechnical engineering, since the amplification or de-amplification of the ground motion are predominantly controlled by these geologically surficial layers (Nandy 2007). In the present study, extensive geotechnical site investigation data were collected for the Guwahati city and analyzed for their response by subjecting to various strong motions. The location map of boreholes is presented in Fig. 1. At these locations, both borehole investigation data and laboratory test data were collated to obtain the relevant information required for the ground response analysis, namely the index properties of soil deposits, thickness of subsoil strata and the Standard Penetration Test (SPT) N values. 2

49 50 The SPT-N values were further used to estimate the shear wave velocity profiles of the substrata using standard correlation proposed by Maheswari et al. (2010). 51 52 Figure 1. Location of boreholes used in GRA of Guwahati city 53 54 55 56 57 58 59 60 61 62 63 64 Some typical borehole profiles at few sites, namely Boragaon (BRGN), Satgaon (STGN), Alok Sixmile (ALKSX) and Sundarpur (SUND), are presented in Fig. 2. As per the geotechnical site investigation data, it was found that the soil deposits consist mostly of either sandy, silty and clayey soils, or their varying combinatorial mixtures. In most cases, SPT-N values have been collected at an interval of 1.5 m, up to a depth of 26-30 m, or as per the premature termination depth of the boreholes. The ground water table (GWT) is located at shallow depths from the ground level (GL), mostly within 2 m from the surface. The particle size distribution of local soils were determined (as per IS: 2720, part-iv), and based on the zone of liquefiable soils as proposed by various researchers (Tsuchida 1970; Ishihara 1980; Xenaki and Athanasopoulos 2003), it was identified that the soils considered for the present study are susceptible to liquefaction (Fig. 3). The index properties of cohesionless and cohesive soils are presented in Tables 2 and 3, respectively. 3

65 66 Figure 2. Typical borehole profiles at site (a) BRGN (b) STGN (c) ALKSX (d) SUND 4

67 68 Figure 3. PSD of prevalent cohesionless and cohesive soils used in the present study 69 70 71 72 73 74 75 76 77 78 79 80 DYNAMIC SOIL PROPERTIES Different type of soils is present at the aforementioned locations, and hence, the dynamic characterization of all soils is essentially required for realistic ground response analysis. The accumulated strains in the soil vary over a wide range during earthquakes. The response of soils at high strain levels (γ>0.01%) is substantially different than that obtained at low strain levels (γ<0.01%), primarily due to the nonlinear stress-strain behavior and damping characteristics at higher strains (Ishihara, 1996). Researchers have used different testing methodologies, such as resonant column tests and cyclic triaxial tests, to evaluate the dynamic properties of soil, thus integrating the dynamic properties obtained at different strain ranges. In ground response analysis, dynamic properties of soils, such as the modulus reduction and damping ratio curves, are commonly used input parameters to define the soil properties. Therefore, relying on any particular testing methodology to get complete variation 5

81 82 83 84 85 86 87 88 89 90 of strain-dependent dynamic properties would not be sufficient. Since evaluating the dynamic properties for all type of soils are tedious and time consuming job, the presents study utilized the experimentally evaluated dynamic soil properties of the typical cohesionless and cohesive soils prevalent in the Guwahati city. The dynamic properties of cohesionless soil, pertaining to the low strain range (γ = 0.0001% to 0.01%) obtained from resonant column tests by Dammala et al. (2017) was used in the analysis. For high shear strain range (γ = 0.01% to 5%), the dynamic properties of cohesionless soil obtained from cyclic triaxial tests by Kumar et al. (2017a) was used. The evaluated modulus reduction and damping ratio curves, for wide range of shear strain (γ = 0.0001% to 5%), for both cohesionless and cohesive soils, as presented in Fig. 4, were used in subsequent GRA studies. 91 92 Figure 4. Soil models (experimental and standard) used in GRA 93 94 95 96 97 In the absence of site-specific modulus reduction and damping ratio curves, it is a common practice to adopt standard models [Vucetic and Dobry (1991) for clays, Seed and Idriss (1970) for sands] in GRA studies (Hashash et al. 2016). It is worth mentioning that any standard curves, as shown in Fig. 4, are developed on the basis of specific experimentations on soils of particular composition, subjected to particular strain ranges (mostly within 1%). In 6

98 99 100 101 102 103 104 105 106 107 108 109 commercially available software such as DEEPSOIL, as used in the present study, based on the trends of modulus reduction and damping ratio curves below γ 1%, the same are extrapolated to the shear strains in the range of 1-10%, which is beyond the actual limits of the conducted experimentations. Hence, the usage of the standard curves, without paying due attention to the actual composition and dynamic response of the soil (especially at higher strain ranges), may lead to improper response. This study highlights the difference of the responses obtained using the standard curves in comparison to the same obtained using the experimentally obtained dynamic soil properties of the local soils. The material curves, as used in DEEPSOIL, depend on the parameters such as the effective vertical stress and plasticity index (PI) of soil (Hashash et al. 2016). In the present study, the values of the required parameters were chosen accordingly as determined from the laboratory investigation of the soil samples obtained from various depths during the borehole investigation survey. 110 111 112 113 114 115 116 117 118 119 120 121 GRA METHODOLOGY The present study utilizes the Equivalent-linear (EQL) approach to perform ground response analyses on the basis of the following assumptions: (a) all substrata boundaries are horizontal, and the soil and bedrock layers are extended infinitely in the horizontal direction (b) the soil profile is subjected to vertically propagating SH waves. DEEPSOIL, commercial software developed by University of Illinois, Urbana Champagne (UIUC), has been used to perform the one-dimensional ground response analyses (Hashash et al. 2016). The input parameters required to conduct the GRA are the material type with strata thickness, unit weight, shear wave velocity, material properties (shear modulus reduction and damping ratio curve) and the input strong motion data. For some typical soil profiles, Fig. 2 shows the variations in shear wave velocity (V s ) and unit weight along with the depth. In the absence of direct estimation of the shear-wave velocity (V s ) profiles, SPT-N values, obtained from the borehole survey, 7

122 123 were used to estimate the same using the correlation (Eqn. 1) provided by Maheswari et al. (2010). 124 Vs 0.301 95.64 N (1) 125 126 127 128 One-dimensional seismic ground response analysis was carried out to estimate the response of stratified soil profiles in terms of the variations in peak horizontal and peak ground accelerations (PHA and PGA), Fourier amplification ratio (FAR), response spectrum, stress ratio and strain variations along the depth of the soil profile. 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 Fundamentals of EQL Method The equivalent linear method of GRA was developed to analyze the non-linear response of soil using frequency domain analysis with the aid of linear transfer functions. This methodology approximates the non-linear behaviour of the soil (i.e., strain dependent shear modulus and damping characteristics) using an adaptive iterative procedure (Kramer 1996). The iterative procedure is governed by the target of finding a compatible shear modulus and damping ratio for a particular effective shear strain. Generally, the effective shear strain is considered to be 65% of the maximum shear strain developed in a particular layer (Kramer 1996). In the frequency domain analysis, the strain-time histories, obtained for each layer, are used to identify the maximum shear strain, which is further used in the estimation of effective shear strain. The computed effective shear strain for a given soil layer is further used to estimate the corresponding strain-compatible shear modulus and damping ratio based on an iterative technique. This process is repeated until a convergent solution is obtained. The equivalent linear analysis considers the soil layer to be represented as a linear visco-elastic material, in which the shear modulus and damping ratio is maintained constant to the corresponding strain-compatible values during each of the iteration cycles of the analysis. For the purpose of viscoelastic wave propagation, soil behaviour is approximated as a Kelvin- 8

146 147 148 149 150 Voigt solid (i.e., a material whose resistance to shearing deformation is the sum of elastic and viscous resistance) governed by the linear elastic shear modulus and viscous damping (Kramer, 1996). Furthermore, soil layers of non-uniform thickness, resting on an elastic halfspace (bedrock), was considered for GRA analysis, for which the background details are well described in Kramer (1996). 151 152 153 154 155 156 157 158 159 160 161 162 163 164 STRONG GROUND MOTIONS To observe the effect of strain dependent dynamic properties of soils, three different acceleration time histories namely Bhuj (2001; PGA = 0.103g), Tezpur (2012; Scaled PGA = 0.36g) and Kobe (1995; PGA = 0.834g) earthquake motions have been used for the ground response analysis. Strong motion record of Kobe earthquake was obtained from the database available in DEEPSOIL, while the data of Bhuj and Tezpur earthquake were obtained from www.cosmos.org and www.pesmos.org, respectively. Figure 5 represent the acceleration histories and their Fourier spectra. Frequency-domain representation indicates the variation of energy content over a frequency band. It is observed that the significant energy content of Bhuj and Kobe strong motions are congregated over a frequency band of 0.5-4Hz, while the same for Tezpur motion lies within 2-15 Hz. The characteristics of these ground motions like predominant period, mean period, bracketed duration and significant duration, derived using SEISMOSIGNAL program (www.seismosoft.com), are shown in Table 4. The mean period of strong ground motions are observed to be varying between 0.17s - 0.64s. 9

165 166 Figure 5. Acceleration time histories and frequency spectra of input motion 167 168 169 170 171 172 173 174 175 176 177 178 RESULTS AND DISCUSSIONS Ground Response Analyses (GRA) has been carried out for eighteen borehole locations at Guwahati city, using three different seismic time histories. The typical four soil sites namely, Boragaon (BRGN), Satgaon (STGN), Alok Sixmile (ALKSX) and Sundarpur (SUND) are presented in Fig. 2, highlighting the variation of shear wave velocity and unit weight along the depth. The following sections illustrate the influence of different strong motions and site conditions on the outcomes of the ground response analysis. Typical results are provided for the stated four sites in order highlight the difference arising from the adoption of standard and experimentally estimated dynamic properties of soils. The results are presented in terms of the variations in acceleration and strain along the depth, as well as the Fourier amplification ratio and peak spectral acceleration at top layer. Contour maps exhibiting the overall distribution of the various responses along entire Guwahati city are also reported. 179 180 181 182 Influence of local site effects on GRA This section presents the seismic response obtained at four sites (BRGN, STGN, ALKSX and SUND) subjected to Bhuj motion. The soil layers at both the sites were defined by the dynamic soil properties obtained experimentally (i.e. experimental soil model) as well as 10

183 184 185 186 187 188 189 190 191 192 193 194 195 according to the existing standard soil models (i.e. Vucetic-Dobry, VD, and Seed-Idriss, SI, models). Figure 6 illustrates the variations in peak horizontal acceleration (PHA) along the depth, obtained at four sites. In comparison to the adoption of VD-SI data, the PGA (PHA at the topmost layer) at BRGN site, obtained by the use of experimental data, is observed to be approximately 35% higher. It is observed that the topmost clayey strata at STGN site (2.5-7.5 m) is relatively softer in comparison to the BRGN site (Fig. 2). Thus, in this case, the influence of choosing the experimental dynamic properties is evident, wherein it can be observed that the adoption of standard dynamic properties leads to the underestimation of the PHA or PGA. Similarly, based on experimental data, STGN and SUND sites exhibit nearly 40% and 9% higher PHA, respectively. However, in comparison to standard curves, ALKSX site shows nearly two times lesser PHA, based on experimental data. This is attributed to the damping effect imparted by clay layer with low shear wave velocity present at the bottom of the profile. 196 197 198 Figure 6. Variation in PHA along the depth at BRGN, STGN, ALKSX and SUND sites subjected to Bhuj motion 11

199 200 201 202 203 204 205 206 207 208 Figure 7 illustrates the variations in shear strain obtained from ground response analysis at the aforementioned four sites subjected to Bhuj motion. The strains at BRGN and SUND sites were observed to be less than 0.1% (Fig. 7a), from both experimental as well as VD-SI data. However, STGN site showed strain of nearly 2% at a depth of 7 m upon using the experimental data (Fig. 7b). This is attributed to the degradation of damping ratio in the soft soil layer. ALKSX site shows nearly 0.5% shear strain at an approximate depth of 10 m while using VD-SI data (Fig. 7b), which is relatively lesser than that obtained using experimental data. In comparison to the outcomes from adopting the standard damping ratio curves, the degradation of soil damping beyond 1% of shear strain, as obtained from the experimental curves, is found to significantly affect the ground response. 209 210 211 212 213 214 215 216 217 Figure 7. Variations in strain at (a) BRGN and SUND (b) STGN and ALKSX sites subjected to Bhuj motion Figure 8 presents the variations in amplification ratio obtained at all four sites. It was seen that the amplification ratio obtained at BRGN and SUND sites, with experimentally obtained dynamic properties, is nearly 40% and 28%, respectively, higher than the same obtained from VD-SI model. However, the amplification ratio obtained from the experimental model at STGN site is nearly 20% lesser than that obtained from VD-SI model. Similar response has been obtained at ALKSX site. This is attributed to the presence of soft 12

218 219 220 221 222 soil layer at the bottom of the profile, represented by low shear wave velocity, as presented in Fig. 2. Figure 8 reinforces the observation that the amplification or de-amplification of the strong motion depends on the variation of the damping ratio of the soil over a wide frequency range, and that the adoption of standard non-degrading damping ratio with increase in strain will lead to improper results. 223 224 225 226 227 228 229 230 231 232 Figure 8. Variations in amplification ratio at BRGN, STGN, ALKSX and SUND sites subjected to Bhuj motion The Peak Spectral Acceleration (PSA) at ground level (considering 5% Rayleigh damping) at the aforementioned four sites are presented in Fig. 9. In comparison to VD-SI model, higher value of PSA was observed at BRGN and SUND sites using the experimental soil model. At STGN and ALKSX sites, however, lesser value of PSA is obtained from experimental soil model in comparison to VD-SI model. Similar outcomes have been determined for all the other test sites and the results are presented in contour plots shown in Fig. 10-12. 13

233 234 235 Figure 9. Variations in PSA at BRGN, STGN, ALKSX and SUND sites subjected to Bhuj motion 236 237 238 Figure 10. Variations in PGA in Guwahati city subjected to Bhuj motion (a) based on experimental data (b) based on VD-SI data 14

239 240 241 Figure 11. Variations in peak FAR at surface level in Guwahati city subjected to Bhuj motion (a) based on experimental data (b) based on VD-SI data 242 243 244 245 246 247 248 249 250 Figure 12. Variations in PSA at surface level in Guwahati city subjected to Bhuj motion (a) based on experimental data (b) based on VD-SI data Figure 10 shows the contour map of the variations in PGA for Guwahati city using soil properties obtained from present experiment as well as VD-SI data. In comparison to the use of VD-SI data, utilization of experimental data (Fig. 10a) exhibited higher ground shaking (reflected by higher PGA) at sites Amingaon, Chandmari, Bongaon, Boragaon, Ahomgaon, and GNRC (Fig. 10b). Figure 11 presents the Fourier amplification ratio (FAR) contours for Guwahati city due to Bhuj motion, using soil properties by experimental data and VD-SI 15

251 252 253 254 255 data,. Fourier amplification ratio (FAR) is defined as the ratio of PGA at ground surface to the Peak Bedrock Input Motion (PBRA), which provides the information about amplification or deamplification of bedrock motion at ground surface. In comparison to the use of VD-SI data, higher values of FAR was obtained on the use of experimental data (Fig 11a), at sites Amingaon, Chandmari, Bongaon, Boragaon, Ahomgaon, GNRC (Fig. 11b). 256 257 258 259 260 261 262 263 264 265 266 Figure 12 illustrates the variations in PSA at ground surface considering 5% damping ratio. PSA is significantly important in the design of earthquake-resistant structure, as it provides an important descriptive representation of the influence of a given earthquake on a structure for a specified damping ratio. The spectral acceleration at different frequencies provides the response spectrum. Response spectrum describes the maximum response of a single-degree-of-freedom (SDOF) system subjected to a particular input motion, as a function of the natural frequency and damping ratio of the SDOF system (Kramer 1996). The maximum response of a structure to a particular input motion is commonly used for the earthquake-resistant design. In comparison to the use of VD-SI data, higher values of PSA was obtained on the use of experimental data (Fig. 12a) at sites Amingaon, Chandmari, Bongaon, Boragaon, Ahomgaon, and GNRC (Fig. 12b). 267 268 269 270 271 272 273 274 Table 5 presents the summary of the results in terms of percentage difference in average PGA, PSA and FAR for Guwahati city subjected to Bhuj motion. It reflects that based on VD-SI data, the average PGA and PSA is nearly 6% and 3% higher, respectively, than that obtained with experimental data. However, in comparison, the average FAR is nearly 3% higher based on experimental data. Therefore, based on Table 5, it can be recommended that, GRA of any region should be conducted on the basis of the regional dynamic soil properties, and in all possible cases, the adoption of standard reference curves should be avoided in order to achieve realistic GRA results. 16

275 276 277 278 279 280 281 282 283 Influence of Strong Motions on GRA Figure 13 shows the variations in PHA along the depth at BRGN site, subjected to different strong motions. It illustrates that the Kobe motion get amplified at a depth of 5-7.5 m if the standard VD-SI model is used, while the usage of present experimental data illustrates deamplification at the stated depths; while Bhuj and Tezpur motion does not show much difference at similar locations. Thus, it can be stated that the amplification and deamplification of strong motions is significantly affected by the chosen soil strata characteristics (stiffness of soil layer), material properties (modulus degradation and damping behavior) and the strong motion characteristics itself. 284 285 Figure 13. Variations in PHA with depth at BRGN with Bhuj, Tezpur and Kobe motion 286 287 288 Figure 14 presents the variations in strain along the depth obtained from Bhuj, Tezpur and Kobe motions. It can be seen that the layer with low shear wave velocity, represented by low shear stiffness, when subjected to strong motion with high PBRA, shows significantly 17

289 290 higher value of strains, corresponding to the use of experimentally obtained dynamic properties. Such high value of strain (>15%) can lead to catastrophic damage. 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 Figure 14. Variations in strain with depth at BRGN subjected to (a) Bhuj and Tezpur (b) Kobe motion The variations in amplification ratio at surface layer, obtained with all three strong motions, are presented in Fig. 15. It is seen that the amplification ratio obtained at BRGN site using Bhuj, Tezpur and Kobe motions, considering the experimental soil model, is nearly 10 40% higher than the same obtained from VD-SI model. It is to be noted that the amplification ratio obtained from Kobe motion considering experimental soil model shows lower magnitude of FAR with frequency shift towards the lower values by nearly 2 Hz. This is due to the availability of soft soil layer, represented by low shear wave velocity, as shown in Fig. 2(a), as well as due to the degradation of damping ratio at higher strains. The lower value of FAR indicates that in comparison to the lower input PBRA motions, the higher input PBRA motions amplify by a lesser extent. The Peak Spectral Acceleration (PSA) at ground level at BRGN site is presented in Fig. 16, which shows that the obtained PSA is significantly higher with Kobe motion as compared to that obtained from Bhuj and Tezpur motions. The PSA obtained from Kobe motion shows nearly similar value from both experimental as well as VD SI model, whereas, the PSA obtained from Bhuj and Tezpur motions is nearly 40% higher with experimental soil model. 18

309 310 311 Figure 15. Variations in amplification ratio at surface level for BRGN subjected to Bhuj, Tezpur and Kobe motion 312 313 314 Figure 16. Variations in PSA at surface level for BRGN subjected to Bhuj, Tezpur and Kobe motion 19

315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 Results obtained for all eighteen boreholes subjected to Bhuj, Tezpur and Kobe motion are presented as contour plots, as shown in Fig. 17-19. Figure 17(a-c) presents the variations in PGA for Guwahati city based on the experimental data, whereas Fig. 17(d-f) presents the variations in PGA based on the VD-SI data. Based on Figs. 17(a, d), it is seen that, in comparison to the experimental data, application of VD-SI data provides approximately 6% higher PGA (average PGA) using Bhuj earthquake. Figs. 17(b, e) present the variations in PGA at all boreholes subjected to Tezpur motion. The average PGA obtained using experimental data was found to be 0.45g (Fig. 17b), whereas the same with VD-SI data is noted as 0.437g (Fig. 17e). Therefore, it can be stated that, in comparison to the VD-SI data, the average PGA obtained from experimental data is approximately 3% higher. Similar response of PGA was observed with Kobe motion shown in Fig. 17(c, f). It is seen that the average PGA with VD-SI data is 1.039g (Fig. 17c), whereas the same with experimental data is 0.75g (Fig. 17f), which is nearly 39% lesser than that obtained from the use of VD-SI data. It can also be observed that with the increase in PBRA of input motion at the bottom of the soil profiles, the ground shaking also increases (reflected by higher PGA at ground surface). Figure 18 shows the variations in FAR for Guwahati city, presented in contour plot, obtained from both experimental as well as VD-SI models. On the use of experimental data, the average FAR for Guwahati city was 6.58, 6.85 and 4.16, when soil profiles were subjected to Bhuj, Tezpur and Kobe motion, respectively, and the same are presented in Fig. 18a, Fig. 18b and Fig. 18c, sequentially. With VD-SI data, the average value of FAR for Guwahati city was found to be 6.39, 6.34 and 3.20, when subjected to Bhuj, Tezpur and Kobe motion, respectively and, are sequentially presented in Fig. 18d, Fig. 18e and Fig. 18f. It is seen that, FAR obtained from experimental data is higher with Bhuj and Tezpur motions, than that obtained with Kobe motion. Similar observations were reported from VD-SI model. Moreover, the FAR obtained from experimental model is higher than that from VD-SI model. 20

340 341 342 Figure 17. PGA contours based on Bhuj, Tezpur and Kobe motions (a-c) based on experimental data (d-f) based on VD-SI data 21

343 344 345 Figure 18. FAR contours based on Bhuj, Tezpur and Kobe motions (a-c) based on experimental data (d-f) based on VD-SI data 22

346 347 348 349 350 351 352 353 354 355 356 357 Figure 19 shows the variations in PSA for Guwahati city, presented in contour plot, obtained from both experimental soil model as well as the VD-SI soil model. It is seen that the average PSA with experimental data was 1.32, 1.79 and 2.60, when soil profiles were subjected to Bhuj, Tezpur and Kobe motion, respectively and the same are presented in Fig. 19a, Fig. 19b and Fig. 19c, sequentially. The average value of FAR with VD-SI data was found to be 1.358, 1.763 and 3.228, when subjected to Bhuj, Tezpur and Kobe motion, respectively and, are sequentially presented in Fig. 19d, Fig. 19e and Fig. 19f. Based on the observations, it can be stated that with the increase in PGA of input motion, PSA at ground level also increases. It can also be stated that the PGA, FAR and PSA at surface, obtained from any input motion, gets significantly affected by dynamic soil properties. In comparison to the VD-SI soil properties, the higher PGA of input motion shows significantly higher values of strains at soft soil layers when analysed with experimental soil properties. 358 359 360 361 362 363 Table 6 presents the summary of the results in terms of percentage difference in PGA, PSA and FAR for Guwahati city. It reflects that the average PGA and PSA at surface level, obtained using experimental data, was significantly lesser than that obtained with VD-SI data, whereas, FAR was higher with the usage of experimental data. Therefore, it can be stated, and recommended as well, that the actual dynamic soil properties would be good option for more realistic GRA results for any site. 364 23

365 366 367 368 369 Figure 19. PSA contours based on Bhuj, Tezpur and Kobe motions (a-c) based on experimental data (d-f) based on VD-SI data 24

370 371 372 373 374 375 376 CONCLUSIONS In order to emphasize the significance of realistic site-specific soil model in GRA, onedimensional equivalent linear GRA has been carried out using experimentally evaluated strain dependent soil properties (i.e. modulus reduction and damping ratio curves), and the results were compared with those obtained using existing soil models [Seed and Idriss (1970) model for sand and Vucetic and Dobry (1991) model for cohesive soil]. Based on the present study, the following conclusions are drawn: 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 GRA based on the experimental data shows significant different response in comparison to the same obtained using existing standard models. Input motion with higher PBRA results in significantly reduced acceleration at surface layer, when the experimental data were used to define the dynamic properties of the soil layer s properties. The deamplifiaction of higher PBRA input motion (e.g. Kobe motion) is significantly influenced by the presence of soft soil layer within the soil profile and its corresponding damping behaviour (degradation of damping ratio beyond 1% shear strain). The average variation of PGA and PSA of Guwahati city, subjected to Bhuj motion, using VD-SI data, are approximately 3-6% higher than that obtained using experimental data; whereas, FAR obtained from experimental data is 3% higher correspondingly to that obtained with VD-SI data. On the use of experimental data, Using the VD-SI model, Tezpur motion shows higher values of PGA, PSA and FAR by 3%, 7% and 8%, respectively. Higher PBRA of input motion (e.g. Kobe motion), obtained using experimental soil properties, reflects approximately 38% and 24% lower value of average PGA and PSA, respectively, in comparison to the same obtained using VD-SI data; whereas, FAR 25

394 395 obtained using experimental data is found to be approximately 30% higher than that obtained using VD-SI data. 396 397 398 399 400 401 Therefore, overall in a nutshell, it can be stated that the degradation of damping ratio at higher strains significantly affects the GRA, especially when the subsurface containing soft soil layers is subjected to strong motions. Based on the results and discussions, this article recommends the use of proper/actual site-specific dynamic soil properties in GRA studies, for more realistic evaluation of the parameters required for the design of earthquake resistant structures. 402 REFERENCES 403 404 405 Anbazhagan, P., Thingbaijam, K. K. S., Nath, S. K., Narendara Kumar, J. N., & Sitharam, T. G. (2010). Multi-criteria seismic hazard evaluation for Bangalore city, India. Journal of Asian Earth Science, 38, 186 198. 406 407 408 Basu, D., Dey, A., & Kumar, S. S. (2017). One-dimensional effective stress non-masing nonlinear ground response analysis of IIT Guwahati. International Journal Geotechnical Earthquake Engineering, 8, 1-27. 409 410 411 Boominathan, A., Dodagoudar, G. R., Suganthi, A., & Maheswari, R. U. (2008). Seismic hazard assessment of Chennai city considering local site effects. Journal Earth System and Science, 117(2), 853 863. 412 413 Boore, D.M. (1972). A note on the effect of simple topography on seismic SH waves. Bulletin of the Seismological Society of America, 62, 275-284. 26

414 415 416 Dammala PK, Krishna AM, Bhattacharya S, Nikitas G, Rouholamin M (2017) Dynamic soil properties for seismic ground response studies in Northeastern India. Soil Dyn Earthq Eng 100:357 370 417 418 Govindaraju, L., & Bhattacharya, S. (2011). Site-specific earthquake response study for hazard assessment in Kolkata city, India. Natural Hazards, 61(3), 943 965. 419 420 Hashash, Y. M. A., Groholski, D. R., Phillips, C. A., Park, D., & Musgrove, M. (2016). DEEPSOIL version 6.0, Tutorial and user Manual. 98p 421 422 IS: 1893 (Part-1) (2002). Criteria for earthquake resistant design of structures-general provisions and buildings. Bureau of Indian Standard, New Delhi. 423 424 Ishibashi, I., & Zhang, X. (1993). Unified dynamic shear moduli and damping ratios of sand and clay. Soils and Foundations, 33, 182-191. 425 Ishihara, K. (1996). Soil behaviour in earthquake geotechnics. Clarendon Press. 426 427 Ishihara, K., Troncoso, J., Kawase, Y., & Takahashi, Y. (1980). Cyclic strength characteristics of tailings materials. Soils and Foundations, 20, 127-142. 428 Kramer, S. L. (1996). Geotechnical earthquake engineering. Prentice Hall. New York. 429 430 Kumar, S. S. (2012). Site-specific seismic ground response analysis of Guwahati city. M. Tech. Thesis, Submitted to Indian Institute of Technology Guwahati, Assam. 431 432 433 Kumar, S. S., & Krishna, A. M. (2013). Seismic ground response analysis of some typical sites of Guwahati city. International Journal Geotechnical Earthquake Engineering, 4, 83-101. 27

434 435 436 Kumar, S. S., Dey, A., & Krishna, A. M. (2014a). Equivalent linear and nonlinear ground response analysis of two typical sites at Guwahati city. In proceedings of Indian Geotechnical Conference, Kakinada. 437 Kumar, S. S., Krishna, A. M., & Dey, A. (2014b). Nonlinear site-specific ground response 438 439 analysis: case study of Amingaon, Guwahati. 15 th Engineering, IIT Roorkee. Symposium on Earthquake 440 441 Kumar, S. S., Krishna, A. M., & Dey, A. (2017a). Evaluation of dynamic properties of sandy soil at high cyclic strains. Soil Dynamics and Earthquake Engineering, 99,157-167. 442 443 444 Kumar, A., Harinarayan, N.H., & Baro, O. (2017b). Effects of earthquake motion and overburden thickness on strain behavior of clay and sandy soils. In Proceedings of 16th World Conference on Earthquake Engineering, pp. 9-13. 445 446 447 Maheswari, R. U., Boominathan, A., & Dodagoudar, G. R. (2010). Use of surface waves in statistical correlations of shear wave velocity and penetration resistance of Chennai soils. Journal of Geotechnical and Geological Engineering, 28, 119 137. 448 449 Naik, N., & Choudhury, D. (2013). Site specific ground response analysis for typical sites in Panjim city, Goa. In proceedings of Indian geotechnical conference, Roorkee, India. 450 451 Nandy, D. R. (2007). Need for seismic microzonation of Kolkata megacity. In Proceedings of workshop on microzonation, Indian Institute of science, Bangalore, India, Vol. 2627. 452 453 454 455 Nath, S. K., Thingbaijam, K. K. S., Adhikari, M. D., Nayak, A., Devaraj, N., Ghosh, S. K., & Mahajan, A. K. (2013). Topographic gradient based site characterization in India complemented by strong ground-motion spectral attributes. Soil Dynamics and Earthquake Engineering, 55, 233-246. 28

456 457 Ordonez, G.A. (2000). A computer program for the 1D analysis of geotechnical earthquake engineering problems, SHAKE2000 Manual, pp. 266. 458 459 460 Phanikanth, V. S., Choudhury, D., & Reddy, G. R. (2011). Equivalent-linear seismic ground response analysis of some typical sites in Mumbai. Journal of Geotechnical and Geological Engineering, 29, 1109. 461 462 463 Ranjan, R. (2005). Seismic response analysis of Dehradun City, India. M.Sc Thesis, International Institute for Geo-Information Science and Earth Observations Enschede, Netherlands, p. 92. 464 465 466 Seed, H. B., & Idriss, I. M. (1970). Soil moduli and damping factors for dynamic response analysis. Report No EERC 70-10, Earthquake Engineering Research Center, University of California, Berkeley. 467 Seismosoft (2012). Seismosignal, version 5.00, www.seismosoft.com. 468 469 Shukla, J., & Choudhury, D. (2012). Seismic hazard and site-specific ground motion for typical ports of Gujarat. Natural Hazards, 60, 541-565. 470 471 472 Tsuchida, H. (1970). Prediction and counter measure against the liquefaction in sand deposits. Seminar in the Port and Harbour Research Institute, Ministry of Transport, pp.1-33. 473 474 Vucetic, M., & Dobry, R. (1991). Effect of soil plasticity on cyclic response. Journal of Geotechnical Engineering, ASCE, 117, 89 107. 475 476 477 Xenaki, V. C., & Athanasopoulos, G. A. (2003). Liquefaction resistance of sand-mixtures: an experimental investigation of the effect of fines. Soil Dynamics and Earthquake Engineering, 23, 183-194. 29

478 Table 1. Material curves commonly used in GRA studies Material curves Proposed by Used for GRA by Software Gravel Seed et al. (1986) Ranjan (2005) SHAKE2000 Rock fill Gazetas (1992) Ranjan (2005) SHAKE2000 Rock Schnabel (1973) Boominathan et al. (2008) SHAKE91 Ranjan (2005) SHAKE2000 Boominathan et al. (2008) SHAKE91 Anbazhagan et al. (2010) SHAKE2000 Govindaraju and Bhattacharya (2011) DEEPSOIL Clay Sand Sun et al. (1988) Seed and Idriss (1970) Phanikanth et al. (2011) Shukla and Choudhury (2012) Kumar (2012) DEEPSOIL SHAKE2000 DEEPSOIL, SHAKE2000 Kumar and Krishna (2013) DEEPSOIL Kumar et al. (2014a,b) DEEPSOIL Naik and Choudhury (2013) DEEPSOIL Kumar et al. (2017b) SHAKE2000 Clay and Sand Ishibashi and Zhang (1993) Basu et al. (2017) DEEPSOIL 479 30

480 481 482 Table 2. Physical properties of cohesionless soil used in the present study Physical properties Values Specific gravity (G s ) 2.7 Min. dry unit weight (kn/m 3 ) 13.85 Max. dry unit weight (kn/m 3 ) 16.84 Mean grain size, D 50 (mm) 0.21 Uniformity coefficient (C u ) 1.47 Coefficient of curvature (C c ) 1.09 Classification symbol SP 483 484 Table 3. Physical properties of cohesive soil used in the present study Physical properties values Specific gravity (G s ) 2.65 Liquid limit (%) 41.50 Plastic limit (%) 22.60 Plasticity index (%) 18.9 Maximum dry density (kn/m 3 ) 17.5 Optimum moisture content (%) 19.3 485 486 31

487 488 489 Table 4. Strong motion parameters for different earthquakes used for the analysis Strong motion parameters 2001 Bhuj 2012 Tezpur 2012 Tezpur (Scaled motion) 1995 Kobe Magnitude (M w ) 7.7 5.0 5.0 6.9 Station Ahmadabad TZP TZP KJMA Site Class B B B B Distance from source 238 km - - 0.6 km Max. PGA (g) 0.103 0.027 0.36 0.834 Predominant period (sec) 0.26 0.08 0.08 0.36 Mean period (sec) 0.59 0.167 0.167 0.64 Frequency (Hz) 1.69 5.99 5.99 1.56 Bracketed duration (sec) 55.8 28.30 28.30 21.90 Significant duration (sec) 14.84 14.38 14.38 8.38 Arias intensity (m/sec) 0.268 0.003 0.556 8.38 Specific energy density (cm 2 /sec) 438 0.17 33 7573 Cumulative velocity (cm/sec) absolute 470 47 653 2119 v max /a max (sec) 0.136 0.021 0.021 0.103 490 491 492 32

493 494 495 Table 5. Comparison of percentage difference in PGA, PSA and FAR obtained for Guwahati city using VD-SI and Experimental data Earthquake Average value VD-SI data Experimental data Difference (%) PGA 0.287 0.27 6.29 Bhuj PSA 1.358 1.32 2.89 FAR 6.389 6.58 2.99 496 497 498 Table 6. Comparison of percentage difference in PGA, PSA and FAR obtained for Guwahati city using VD-SI and Experimental data Earthquake Average value VD-SI data Experimental data Difference (%) Bhuj 0.287 0.27 6.29 Tezpur PGA 0.437 0.45 2.97 Kobe 1.039 0.75 38.53 Bhuj 1.358 1.32 2.88 Tezpur PSA 1.673 1.79 6.99 Kobe 3.228 2.60 24.15 Bhuj 6.389 6.58 2.99 Tezpur FAR 6.336 6.85 8.11 Kobe 3.202 4.16 29.92 499 500 33