Some Results on the Accuracy of a Classical Upscaling Technique Using an Intuitive Multilevel Preprocessor for Smart Simulation

Resumo

The progress in geology has allowed the creation of high resolution reservoir models, however hard-
ware limitations turn multiple simulations using these models unfeasible. To overcome this problem, upscaling

techniques were developed allowing fine scale information to be projected onto a coarse scale discrete domain,
through homogenization of petro-physical parameters, decreasing the amount of processing and storage needed,
and so turning the simulation feasible. Different upscaling techniques were developed in order to improve the
tradeoff between accuracy and computational performance for different boundary conditions, so studies to identify
appropriated approaches and coarsening ratios to certain types of problems are important to industry, raising up the
necessity of suitable tools to perform these tests. In this context, the IMPRESS (Intuitive Multilevel Preprocessor

for Smart Simulation) has been developed to simplify multilevel simulations, such as those in the family of upscal-
ing techniques. In this article, we demonstrate an application of the IMPRESS that integrates a classical upscaling

technique with a commercial petroleum reservoir simulator, the IMEX. Results are presented on the second model
of the 10th SPE Comparative Solution Project (10th CSP-SPE).