When interpreting results from stress measurements using stress relief techniques, one fundamental assumption is that the rock behaves elastically. This assumption may not be verified if the stress level to which the recovered core is submitted surpasses its strength. Local shear or tensile yielding may occur prior or concurrently to the stress relief operation. In either case, means of identifying this yielding must be available in order for the measurement to be rejected or treated with caution. Using an axisymetric numerical model with a strain softening constitutive law, the drilling and stress relief drilling operations associated with the modified doorstopper stress measurement technique are simulated to generate strain recovery curves which are then analyzed in order to detect non elastic behavior. The simulated curves are compared to actual experimental curves obtained in the lab or in the field.
In a paper by Corthésy and Leite (2008), core discing and damage mechanisms were simulated using Flac2D with an elasto-plastic cohesion softening friction hardening model, which showed that for a given stress state, discing or core damage could involve tensile failure, a combination of shear and tensile failure or only shear failure, depending on the stress state and ratio of tensile to shear strength of the rock. The numerical model used was validated by replicating core discing occurrence and disk thicknesses observed by Obert and Stephenson (1965) under controlled laboratory conditions. Using the same model, stress measurements conducted using the modified doorstopper technique were simulated. This technique allows continuous monitoring of the strains caused by stress relief and using the strain recovery curves, it is shown how a rock behavior departing from elasticity can be identified in order for the measurement to be rejected or interpreted with caution.