In situ stress measurements based on the recovery principle, commonly called overcoring techniques, consist in calculating stresses under strain gauge rosettes using recovered strains at borehole walls or ends. These stresses are then combined through models which account for the stress-strain relationships of the rock and the influence of the borehole on the in situ stress tensor. Up until1994, only strains resulting from complete stress relief were used in the calculation of stresses using such techniques. The RPR method developed by Corthésy et al. (1994), for the modified doorstopper technique allowed using transient strains on the recovery curve to increase the number of equations available for obtaining more stress tensor components from a single measurement. In order to increase the confidence and quality of the stresses estimated from the modified doorstopper technique, a new data logger is presented along with a new approach for calculating the stresses. It is shown how using the complete strain recovery curve and estimating the far field stress tensor with the inverse problem allows to dramatically minimize the adverse effects of noise on the strain recovery curve in isotropic or anisotropic rocks.
After more than 35 years of stress measurement field experience in varied and difficult geomechanical settings all over the world, the authors have come to consider in situ stress measurements as a sampling problem, in which the sampled data are multivariate regionalized random variables. Because of the heterogeneities and discontinuities found in the rock mass, stress variations in both magnitude and direction are present at various scales which further complicates the sampling and data analysis process (Corthésy et al., 1991, 1993a). Although this problem is complex, the first requirement in any sampling procedure is to have access to samples which are minimally biased, as non-biased rock stress measurements is something which has not yet been achieved because of the complex nature of rocks and rock masses and the aggressive environment in which such measurements are taken. In the following sections, sources of error at certain stages of the stress measuring and data reduction procedures will be covered and means of alleviating these errors or biases will be presented for the modified doorstopper technique.
