Probabilistic Nonlinear Ground Response Analysis of Newtown Suburb, Kolkata, India

Ground response analysis (GRA) helps immensely in the assessment of seismic hazards on the surface of any region, as it captures the influence of local soil conditions on the ground surface motion.In this study, a probabilistic one-dimensional (1-D) nonlinear GRA (NGRA) is performed for Newtown, Kolkata, using DEEPSOIL software. The methodology incorporates soil nonlinearity, variations in soil properties, and water table fluctuations to achieve more realistic seismic response than conventional equivalentlinear GRA (ELRA).Although ELRA methods are widely used due to simplicity, they may underestimate soil behavior under strong shaking or liquefactionprone conditions. Thus, a comparison of results from ELRA and NGRA reveals significant differences in predicted peak ground acceleration (PGA), highlighting the need to capture nonlinear soil behavior for robust seismic design and risk assessment.However, the nonlinear approach being the most appropriate method for GRA, its performance largely depends on the input parameters used in the analysis.For lower discrepancy and faster convergence,a quasi-Monte Carlo Simulation (QMCS) technique is adopted to sample uncertain input parameters, including shear wave velocity, shear strength, shear modulus, damping ratio, and soil density, based on their probabilistic distributions and this uncertainty is subsequently reflected in the probabilistic ground response analysis (PGRA). Eleven spectrum-compatible bedrock motions with PGA varying between 0.16g to 0.24g are used as seismic input at the base of the soil column, representing a range of earthquake scenarios.Dynamic soil properties, including modulus reduction and damping curves, are calibrated considering the effect of overburden pressure, plasticity index and geologic age against local field and laboratory data to ensure reliable analysis.The variation of PGA across the soil profiles is captured and the amplification ratio of surface PGA and bedrock PGA in Newtown is in the range of 1.5-2.5. The PGA at surface level obtained from the PGRA corresponding to a bedrock PGA of 0.17g is 0.356 ± 0.03g and 0.5 ± 0.035g from nonlinear and equivalent linear approaches respectively. The developed hazard curves will enable the practicing engineers to establish the probability-based response spectra for seismic response analysis of structures and to make risk-informed decisions in seismically active regions.

Keywords: Nonlinear ground response analysis, DEEPSOIL, Uncertainties, Quasi-Monte Carlo simulation, PGA, Probability distribution.