Back-Analysis and Parameter Identification for Deep Excavation Based on Pareto Multiobjective Optimization

Abstract

In this paper, a back-analysis method of deep excavation based on the Pareto multiobjective optimization is proposed and a multiobjective optimization algorithm multialgorithm genetically adaptive multiobjective method (AMALGAM) is implemented in a commercial FEM software to identify soil parameters based on multiple types of field observations. The proposed method is applied to a well-instrumented deep excavation, i.e., the Taipei National Enterprise Center (TNEC) project. The observed wall deflection and ground surface settlement at Stage 3 of the excavation are simultaneously used to estimate the nine soil parameters of the modified Cam-clay (MCC) model for three clay layers. The Pareto front in the biobjective space exhibits a rectangular shape, which implies that the simultaneous minimization of both objectives can be achieved. The back-analyzed soil parameters of the compromise solution from the biobjective back-analysis can reasonably simulate both the wall deflection and ground surface settlement for Stage 3. The differences of the predictions and the actual observations for Stages 4 to 7 using the back-analyzed soil parameters from Stage 3 are mainly because the inadequacy of the MCC model to simulate the small strain soil behaviors of excavation.