Simulação paralela composicional de reservatórios de petróleo em conjunto com malhas não-estruturadas

Abstract

In petroleum engineering, numerical simulation has proven itself a fundamental tool to the improvement of methods applied during exploration and production stages, such as history matching and enhanced oil recovery. The need for a reliable and low time-consuming response drives the search for faster simulators. There are many ways of enhancing the efficiency of these computational tools, but the present work focuses on parallelization, i.e. give the simulator the capacity to distribute the simulation activities among various processors, aiming to reduce the wall clock simulation time. In this work, the parallelization implementation was done in UTCOMP, a multiphase and multicomponent compositional simulator designed to handle several hydrocarbon recovery processes that was developed at The University of Texas at Austin. Some open source code libraries were used, such as FMDB (Flexible Distributed Mesh Database) and ParMETIS (Parallel Graph Partitioning and Fill-reducing Matrix Ordering), to divide the grid and manage grid data transfer, and PETSc (Portable, Extensible Toolkit for Scientific Computation), to solve the linear systems resulting from the discretization of the governing equations. The numerical approach used is the Element-based Finite Volume Method (EbFVM) coupled with an IMPEC (Implicit Pressure, Explicit Compositions) formulation. Three case studies consisting of gas injection processes in hypothetical reservoirs of different geometries were carried out. The results in this work are expressed in oil and gas production curves and in a comparison between wall clock simulation times before and after parallelization. These results show that the parallel simulator produced the same response as the original simulator and the speed-up was considerable for every case studied.