Stress calculation methods for overcoring techniques in heterogeneous rocks

Rock stress determination with overcoring techniques such as the CSIR triaxial cell, CSIRO hollow inclusion or SSPB’s Borre Probe require in situ recovered strains at a circular pilot hole wall caused by stress relief to be converted into stresses by using the deformability parameters at the measuring point. Ideally, these parameters are obtained by reloading the recovered overcore to which the strain sensing cell is attached, assuming the stress and strain fields match the theoretical distribution one would find in an isotropic linear-elastic homogeneous thick-walled cylinder. The paper focuses on the influence that some of these assumptions have on the determination of the deformability parameters and consequently on the in situ stress tensor obtained with the CSIR triaxial cell. Stochastic numerical analyses are performed using the software FLAC3D to obtain the recovered strains under the CSIR triaxial cell strain gauges for different far-field stress tensors in continuous rock with varying degrees of heterogeneity in terms of elastic moduli. For each applied stress tensor, the overcore reloading process is numerically simulated and the deformability parameters are calculated by using three different approaches. One of these, called the LA method, assumes that the rock material under each strain gauge rosette can show a local anisotropy, while the thick rock cylinder is considered globally isotropic. The influence of these approaches on the calculated stress tensors are analysed by comparing their magnitudes and orientations to those applied to the model boundaries and by evaluating the scatter of the results obtained with each approach. It is shown that the LA method lends much more accurate principal stress orientations and much less scatter for both orientations and magnitudes.