Stability and strength analysis in geotechnical engineering can be carried out in terms of either effective or total stresses. Given fundamental knowledge of soil mechanics and clear understandings of numerical modelling, the numerical simulations should result in consistent outcomes from both approaches. Undrained excavations were modelled in Abaqus, a software application for finite element analysis. The Extended Modified Cam Clay model was used to characterise the soil behaviour in the Effective Stress Analysis (ESA) and the Tresca model was used in the Total Stress Analysis (TSA).
For the comparison between ESA and TSA to be valid, it is critical for both analyses to represent identical soil conditions and characteristics. Therefore, the fundamental part of the procedure was to derive the values of total stress parameters from the effective stress parameters and numerical outputs from ESA. In order to confirm the precise match of soil conditions between ESA and TSA, initial stress distributions and initial values of K0 were compared.
The shape of yield surface in ESA was modified to minimise the difference in the yield surface between ESA and TSA. The values of su were also adjusted to reflect the shear strength in the plane strain problem. While those modifications improved results, most of the numerical outputs showed inconsistencies between ESA and TSA. By comparing the maximum values of forces and moments of structural elements, neither method produced results that were consistently greater than the other method throughout all excavation scenarios. It was justified that the differences in the structural forces and moments were mainly due to the differences in the passive stress on the retaining wall between ESA and TSA. The observations on the stress paths of passive soil elements revealed that the passive soil for all ESAs did not reach the critical failure state, and for TSAs the soil reached the failure for cantilever problems, but not for propped excavations.