A modified optimization algorithm for back analysis of properties for coupled stress-seepage field problems

Abstract

Abstract The study is aimed at developing a new algorithm for back analysis of rock mass parameters based on the observed displacements of tunnel excavations under coupled stress-seepage field problems which are generally encountered in hydraulic tunnel projects. The back-analysis algorithm is developed by incorporating the Levenberg-Marquardt optimization technique with complex-variable differentiation method. Additional auxiliary technique is also incorporated to enhance the convergence and stability of the proposed algorithm during the back-estimation of multiple rock mass parameters. A hypothetical hydraulic tunnel case was used for testing and validating the proposed algorithm by incorporating the method in a finite element code in which multiple rock mass parameters (such as modulus, permeability, and in situ stress) were treated as target parameters. Results show that the multiple rock mass parameters can be accurately and efficiently estimated by back analysis using a newly developed algorithm for coupled stress and seepage fields encountered in tunneling. The proposed algorithm can be used for predicting excavation behavior, particularly, the stress-induced deformations at subsequent stages of tunnel excavation under coupled multiple fields (e.g. stress and seepage fields).