Understanding the factors affecting wellbore stability is crucial for safe and cost-effective drilling and completion operations. Thick-wall cylinder (TWC) laboratory tests have been widely used to calibrate wellbore stability models. Although the TWC test is meant to simulate the wellbore geometry, the influence of the finite wall thickness makes the results difficult to interpret. To account for these problems, elastoplastic models were calibrated to laboratory tests using the Modified Cam Clay (MCC) material model. These calibrated models were then combined with finite element analysis to investigate failure mechanisms. To investigate the effect of wall thickness and rock properties on TWC failure mechanisms, a suite of core samples was tested using both multistage triaxial and TWC tests. The MCC model was calibrated to test results for the multistage triaxial and TWC tests for the three rock types (incipiently cemented Miocene Sandstone, Bentheimer Sandstone, and Austin Chalk). A dependence of TWC test results on wall thickness was observed. The ultimate strength at which the TWC collapses increases with increasing wall thickness. Failure mechanisms around the borehole were investigated using high resolution micro CT scanning. The three different failure mechanisms (spalling, borehole closing, and catastrophic) were all observed. The MCC model was used to further investigate stresses and deformation present interior to the wall of the TWC specimen. It was observed that failure occurs in thick-walled cylinder tests when a particular near cavity displacement gradient is achieved for each rock type, this displacement gradient is constant for each rock type independent of wall thickness.
Thick-wall cylinder (TWC) tests have been used for decades to estimate wellbore stability (Adams, 1912), (King, 1912), (Hoskins, 1969). The TWC test with axial loading and internal or external fluid pressure loading provides the closest analogy to the deformation mechanism observed in a wellbore. However, in many cases the laboratory measurements do not match the actual wellbore deformation. This in part could be due to the use of improper constitutive material model to describe near wellbore deformation (McLean & Addis, 1990). Elastic-brittle models do not accurately describe such behavior (Addis & Wu, 1993), and elastoplastic models could better describe rocks nonlinear plasticity. It is common practice to calibrate such models to triaxial test results, then check the model prediction against more complex geometry such as TWC (Cook, 1996). In this study, three different rock types (cored Miocene sand, Bentheimer, and Austin Chalk outcrop samples) mechanical properties were tested by multistage triaxial testing, and the Modified Cam-Clay (MCC) model was calibrated to the test results. The calibrated MCC model was used to fit the TWC tests results. The use of finite element modelling software ABAQUS™ allowed us to investigate the deformation inside the specimen and failure behavior near the cavity.