Modelisation de la rupture lors d'une mesure au doorstopper modifie dans une roche synthetique elastoplastique

Stress measurement techniques performed in rock media can be fundamentally grouped under three categories: hydraulic fracturing, relief and strain recovery.; Amongst the variety of equipment available for strain recovery methods, the Doorstopper is notable for its simplicity of use and the remarkable short length of recovered core required with its modified version. Interpretation of the results and subsequent stress calculations rely on several hypotheses, the most prominent of which is the elastic behavior of the volume of material involved.; However, failure may occur during a measurement: such was the case of experiments carried out with the modified Doorstopper on a synthetic rock in the rock mechanics laboratory at Ecole Polytechnique de Montreal. In some instances, core disking was observed.; In light of these phenomena, questions arise as to the effect of partial damage or failure in the core stub while performing a strain recovery procedure with the modified Doorstopper. Is the physical process duplicable with a numerical model? If so, how can one distinguish a measurement with failure from one without? In either case, how does one interpret data collected at the end of the strain recovery operation? Some answers are provided through comparison of strain recovery curves generated by the physical and numerical models. In addition, when using the RPR method to interpret measures from a single borehole, the RPR ratio is established with the peak and final strain invariants. Thus how well the shape of numerical curves follows the physical cases' will be highlighted throughout this study.; An axisymmetric model was built with the software FLAC to simulate the strain recovery process; the parameters used in the model are obtained as follows: (1) A series of results from uniaxial and triaxial compression tests are imported into a data fitting program; the regression parameters are taken as preliminary strength parameters for the synthetic rock. (2) A single-element model is used to test the preliminary parameters; these are adjusted so as to produce similar stress-strain curves yielded by the experimental tests with three failure criteria: Hoek and Brown, Mohr-Coulomb and double yield. The latter is considered due to the rock's substantial porosity. (3) A model at the scale of the cylinder tested is used to simulate a strain recovery procedure with the modified Doorstopper. Parameters obtained for each law in the previous step are adjusted so as to obtain similar strain recovery curves yielded by the physical model.; With the second model, the various steps of a strain recovery procedure are simulated: boring of the pilot hole, applying initial stress conditions at the model boundaries, relieving stress by coring along the diameter of the pilot hole. FLAC's dynamic graphical interface makes the numerical simulation appealing for its almost real-time visual rendering of the damage or failure progress within the core.; The outcome of this study is a quantified estimation of the effect of partial damage or localized failure on strain recovery curves generated by the modified Doorstopper technique.