Abstract
This paper presents the parameter identification of an elasto-plastic constitutive model for geomaterials and a risk evaluation for neighboring work based on a data assimilation using the particle filter (PF) and incorporating a soil-water coupled finite element analysis with an elasto-plastic model. After the numerical implementation of input data for the soil-water coupled FEM is examined, based on an in situ investigation, its applicability to an actual excavation work is discussed. A risk assessment for an alternative design, using the probability density which the PF computes and updates on the basis of observations, is also proposed. It is revealed that, for a practical example, the FEM results using the parameters identified by PF show a good agreement with the observations and that a quantitative and reliable risk evaluation can be achieved using the proposed method.
