Validation And Testing Of Coupled Flow-Thermal-Mechanical Hydrate Reservoir Models

The second International Gas Hydrate Code Comparison Study (IGHCCS2) is an international effort to compare the simulators used to assess gas hydrate production and evaluate their strengths, weaknesses, and validity. As part of the IGHCCS2, LBNL scientists designed a radially symmetric, 1-D coupled flow-geomechanics problem to evaluate depressurization and production from hydrate reservoirs using a single vertical well. LBNL also used this opportunity to perform a code validation exercise, examining the numerical methods used, as such tests have been rare in the literature concerning hydrate simulator development. During the formulation of hydrate problems and meshing for reservoir simulation, and when evaluating earlier hydrate studies (performed using smaller meshes and computationally slower codes), we have often wondered whether the correct discretization is being used—that is, are the grid elements sized as to correctly resolve the problem solution. Previous studies evaluating hydrate reservoir simulators have not addressed this issue, and it has been clear that the selection of mesh element dimensions has often been driven by computational limitations or mere convenience.

The tests involved two steps. First, we tested the LBNL TOUGH+/Millstone code against an analytical solution (Rudnicki, 1986) for a 1-D single-phase depressurization problem (flow-geomechanics only). Second, we performed a mesh convergence study on the flow-geomechanical (no hydrate) and the flow-thermal-mechanical (with hydrate) problem, varying the discretization from very coarse to very fine and evaluating the changes in the simulator response and convergence toward the analytical solution.

Our tests show that, with the proper choice of parameters, the TOUGH+/Millstone flow-geomechancial formation closely matches the derived analytical solution for a 1-D problem. Our results also show that mesh convergence occurs for both flow-geomechanical and flow-thermal-geomechanical problems, with and without evolving solid hydrate phases. For hydrate dissociation cases, mesh discretization determines whether hydrate lensing behavior appears and to what degree, although it is unclear whether net gas and water production are strongly affected, at least in the short-term evolution of the system.