(206d) Regrid: A Python Toolbox for Reservoir Simulation Reproducibility, Comparison, and Multiphysics Code Interoperability

Irons, T. - Presenter, University of Utah
Jia, W., University of Utah
McPherson, B., University of Utah
Geologic storage of CO2 remains a viable means by which global carbon emissions can be reduced at scales necessary for substantive impact. Geologic carbon storage projects exhibit varying concerns (risks) depending on lithology, simultaneous or subsequent utilization, and volumes injected. In most cases, multiphase flow simulations represent a critical management tool to forecast future behavior and also to evaluate reservoir properties.

However, reservoir models are often treated as “black box tools,” and modeling often involves as much art as science. As such it can be difficult to quantify uncertainty related to reservoir model simulations. Additionally, often coupled processes are of interest within a carbon storage project which cannot be readily modeled natively within a multiphase simulator. For these reasons, tools that allow for easier code comparisons, reproducibility between investigators, and incorporation of related simulations for coupled process modeling are extremely useful.

The Regrid project is an open source python module with these aims. Regrid allows for mesh conversion between popular simulators including Petrel/ECLIPSE, STOMP, SUTRA, MODFLOW, and TOUGH(2). Additionally, Regrid allows for models to be converted to VTK formats which can serve either as an intermediary format for multiphysics simulations as well as for post-processing and visualization. We present the use of Regrid with application at an active field site under investigation by the National Energy Technology Laboratory and its Southwest Regional Partnership on Carbon Sequestration. As an illustration, we demonstrate linking of reservoir models developed in Petrel with shallow USDW MODFLOW models. Additionally, injection scenarios are compared across simulators. Finally, multiphysics use cases will be shown alongside post-processing visualization.