Complex analysis of shear box tests with explicit consideration of interaction between test device and sample

The prior aim of lab testing is the determination of parameters. However, the measurements do not only reflect the rock or soil properties, but contain also, at least to some extent, the influence of the test equipment. Even if measurements are designed to minimize the influence of the test equipment often some measurement values are effected by an unknown amount. Therefore, it is proposed to build detailed numerical models of the test device including the specimen and to duplicate the lab test completely. This paper presents experimental and numerical tests performed with a shear box device. Multi-stage direct shear tests were conducted on smooth plane joints in artificial rock-like material under constant normal loading (CNL) conditions. A 3-dimensional numerical model of the shear box device including the sample was built and the laboratory tests were simulated. The numerical model comprises the loading frame, the hydraulic pistons, the shear boxes and the sample and considers the measuring devices. The numerical model was able to reproduce the non-uniform stress distribution at the joint connected with sample rotation, which in turn produces unwanted frictional forces at the upper shear box. Consequently, only 91-94% of the applied and measured normal force really acts at the joint. Moreover, numerical simulations also shows, to what extend the different parts of the loading frame deform.