| Description |
Verification and validation (V&V) principles are applied to assess massively parallel simulations of hypervelocity perforation of a well bore casing and subsequent penetration into oil-laden sandstone via shaped-charge jet. This technique for liberating oil from geological formations has the potential to be optimized through development of robust and accurate computational frameworks. Accordingly, the overarching objective of this research is to systematically assess the accuracy of the numerical algorithms used (verification) and the appropriateness of those equations for this engineering purpose (validation). Automated methods for single-element verification of constitutive models under a variety of loading modes are developed. This modular test suite incorporates previously documented verification tests, both generally applicable to plasticity models reducing to von Mises plasticity, as well as model specific tests of the geomechanics model (Arenisca) under continual development. These tools are extended to extract the deformation histories of single particles from full-scale Material Point Method (MPM) simulations, which helps to analyze problematic loading modes of a larger simulation on a single-processor workstation. The velocity gradient in these single-element tests must be evaluated in a manner consistent with the underlying integration algorithm used in the source simulation, which is a nontrivial observation making this new capability novel. Testing capabilities are extended to provide arbitrary loading paths similar to those extracted from full-scale penetration problems to serve as robustness and verification tests in future regression testing. A key focus of the work was devising methods to automate the testing of Arenisca and its implementation. As MPM is a relatively new approach to modeling large deformation problems, novel visualization methods are developed along with supporting Python postprocessing scripts. Analytical penetration models in the literature, which have historically been developed for hydrodynamic flow of metals, are tested for their applicability to the penetration of sandstone. A representative sampling of simulation results (some using new methods to account for confining stress) are presented to illustrate how full-scale V&V trend testing often reveals issues not evident in smaller tests, thereby helping code developers better understand, and eliminate, undesired trends or anomalies in the results. |
