Stochastic modeling of SPT N-value and evaluation of probability of liquefaction at Guwahati City

Standard penetration test (SPT) reveals the spatial complexity of standard penetration resistance (N-value) with depth. In this paper, a 1D stochastic characterization of spatial complexity of N-values is developed by considering data obtained from sixty two boreholes in Guwahati City. The N-value profile is modeled as the sum of deterministic part and a stochastic component. The deterministic part which characterizes the non-stationary mean of the data is determined by linear regression analysis. The remaining error is modeled as a spatial random field. The characterization of error heterogeneity as a homogeneous Gaussian random field successfully captures the observed auto-correlation function. The proposed stochastic model is used to compute the probability of factor of safety against liquefaction by Monte-Carlo simulation. The results obtained are presented in form of fragility surfaces, expressing the probability of liquefaction as a function of magnitude of the earthquake and epicentral distance. It is observed that the probability of liquefaction at Guwahati city due to strong earthquakes occurring even at large distances is very high. © 2008 World Scientific Publishing Company.