Application of an Optimization Algorithm for Calibrating Soil Bounding Surface Plasticity Models for Cyclic Loading

Abstract

Advanced constitutive models require several input parameters, some of which represent intrinsic soil properties and some mathematical fitting parameters. Calibrating these modeling parameters is often a challenging component of using a constitutive model in numerical analyses, especially when models are to be calibrated against several data sets simultaneously and/or when there is no conventional geotechnical relation for some parameters. The calibration of constitutive models for cyclic loading is an even more challenging task compared with monotonic loading due to the nonlinearity of soil behavior with stress/strain reversals, amplitude, and the number of loading cycles. In this study, the efficiency of the Gauss–Newton trust-region optimization (GNO) algorithm for calibrating constitutive models for cyclic behavior is evaluated by applying it to three recently developed advanced bounding surface plasticity constitutive models for clays and sands and comparing the outcomes with laboratory test results. The GNO algorithm is shown to be an accurate and time-efficient alternative tool for calibrating the cyclic constitutive models studied.