Back analysis of geomechanical parameters by optimisation of a 3D model of an underground structure

One of the major difficulties for geotechnical engineers during project phase is to estimate the geomechanical parameters values of the adopted constitutive model in a reliable way. In project phase, they are normally evaluated by laboratory and in situ tests and, in the specific case of rock masses, by the application of empirical classification systems. However, all methodologies lead to uncertainties due to factors like local heterogeneities, representativeness of the tests, etc. In order to reduce these uncertainties, geotechnical engineers can use inverse analysis during construction, using monitoring data to identify the parameters of the involved formations. This paper shows the back analysis of geomechanical parameters by the optimisation of a 3D numerical model of the hydroelectric powerhouse cavern of Venda Nova II built in Portugal. For this purpose, two optimisation techniques were considered: one classical optimisation algorithm and an evolutionary optimisation algorithm. In the optimisation process, displacements measured by extensometers during excavation were used to identify rock mass parameters, namely the deformability modulus (E) and the stress ratio (K-0). Efficiency of both algorithms is evaluated and compared. Both approaches allowed obtaining the optimal set of parameters and provided a better insight about the involved rock formation properties. (C) 2011 Elsevier Ltd. All rights reserved.