Validation of numerical predictions for liquefaction phenomenon – Lateral spreading in clean sands

Abstract

During the last few decades, important efforts and developments in the computational modeling of geo-materials have contributed to an increase in the accuracy of the predictions of the dynamic responses of soil systems. Due to their catastrophic consequences, special emphasis has been placed on liquefaction-induced ground failures. However, the numerical tools for liquefaction modeling need to be continually assessed and validated in order to enhance their reliability and enable them to be included in design practices.Within that context, the main objective of this paper was to present a complete validation exercise that explores the capabilities of the numerical predictions to simulate the lateral spreading phenomenon in clean sands under a diverse range of densities and input motions. The validation exercise used the “Strain Space Multiple Mechanism Model” to simulate the lateral spreading phenomenon (although the methodology presented here might be applicable for the validation of other numerical tools as well), and was based on multiple, cross-checked, and high-quality physical models (centrifuge models) and element tests (hollow cylinder cyclic shear tests).Special focus was placed on the quantification of the median response and the associated variability of both physical and numerical models, including an analysis of the importance of the proper selection of validation metrics.The comparison showed that the numerical model is able to predict the displacements for the median trend and the 95% probability confidence bounds for PGA < 0.25 g.