Parallel visualization of reservoir models with ParaView

Interactive visualization of subterranean formations models is essential for reservoir engineering. With the growing resolution of such models, visualization also becomes more challenging. To get the image in reasonable time, visualization calculations should run in parallel. ParaView is a platform that supports parallelization out of the box. Moreover, it can run in client–server mode, where heavy processing and rendering are executed on a powerful multi-node cluster. Custom data format support can be implemented in ParaView as an extension by a plug-in or a script for embedded Python-shell. This in particular allows implementing visualization for GRDECL/EGRID—the most common reservoir data format. Partition of the grid between parallel processes should be defined explicitly in the logic of such custom extensions. It was identified that the way of splitting the grid between processes can significantly impact on computational efficiency. Current study is mainly focused on the class of reservoir models. Reservoir geometries are known to have horizontal sizes much greater than vertical size. For such grids, partitioning into vertical blocks has proved to be much more efficient for parallel visualization, than split into horizontal layers. Not only it requires less memory, processing and rendering are also faster. These important findings can help with faster visualization and increase the capabilities of the existing computational resources.