2D High Performing Compositional Petroleum Reservoir Simulation in Conjunction with Unstructured Meshes
Ivens da Costa Menezes Lima1; Vanílson Nogueira Avedo1; José Cláudio Batista de Oliveira1; João Henrique Bessa Gomes1; Joaquim Bento Cavalcante Neto1; Kamy Sepehrnoori2; Francisco Marcondes1
1 UFC; 2 The University of Texas at Austin
doi:10.20906/CPS/COB-2015-2537
Resumo
Petroleum reservoir simulators are indispensable tools for enhanced oil recovery (EOR), optimization of well configurations, history matching, among others. In order to be efficient tools for the oil industry, these simulators need to produce accurate results with small CPU time. There are many ways to improve the speed of a simulator, but the main focus of this work is on the parallelization, i.e. give the simulator the ability to reduce the CPU time by using various processors. In this work, we used some open source libraries like FMDB (Flexible Distributed Mesh Database), ParMetis (Parallel Graph Partitioning and Fill-reducing Matrix Ordering), and PETSc (Portable, Extensible Toolkit for Scientific Computation). The numerical approach is based on Element based Finite Volume Method (EbFVM). The FMDB and ParMetis libraries are used to divide and manage the grid transfer between the processes, and PETSc for solving the linear systems arising from the discretization. In fact, the main challenge of this work is to organize the grid, fluid, and reservoir data set in such a way that the communication between processes is reduced. The difficulty to transfer the information between the processors is mainly due to the use of unstructured grids. In the current implementation, we use the UTCOMP simulator. UTCOMP simulator is a compositional, multiphase/multicomponent simulator designed to handle several hydrocarbon recovery processes that was developed at The University of Texas at Austin. The results of this work are presented in terms of oil and gas production, and CPU time for various case studies.
Palavras-chave: parallelization; UTCOMP; EbFVM; 2D unstructured grids