Department of Mathematics
 Search | Help | Login | pdf version | printable version

Math @ Duke



Publications [#58648] of John A. Trangenstein

Papers Published

  1. Trangenstein, John A. and Bi, Zhuoxin, Multi-Scale Iterative Techniques and Adaptive Mesh Refinement for Miscible Displacement Simulation, Proceedings - SPE Symposium on Improved Oil Recovery (2002), pp. 924 - 936, Tulsa, OK, United States
    (last updated on 2007/04/09)

    Many enhanced oil recovery processes in reservoir engineering involve localized phenomena that could be due to several features, such as injection fronts, wells or reservoir heterogeneity. In order to reach sufficient accuracy in field-scale simulation, the localized phenomena need to be resolved and modeled in appropriate scale-dependent ways. Our approach to treating the localized phenomena is to use high-resolution discretization methods in combination with dynamically adaptive mesh refinement (AMR). The purpose of adaptive mesh refinement is to concentrate the computational work near the regions of interest in the displacement processes, which may evolve constantly in space. Adaptive mesh refinement requires appropriate techniques for data communication in a hierarchy of dynamically adaptive mesh. The selection of appropriate scaling rules as well as computationally efficient data structures is essential to the success of the overall method. We have exploited the object-oriented features of C++ for the AMR program structure and data management, while numerically intensive routines are implemented in FORTRAN. It turned out that adaptive mesh refinement can significantly reduce the computational cost required to obtain a desired level of accuracy in the simulation. However, the emphasis here is on the development of efficient techniques for solving linear systems that arise in the numerical discretization of an elliptic equation for the incompressible pressure field. We use a conjugate gradient algorithm preconditioned by multiplicative domain decomposition between refinement levels, in which additive domain decomposition and incomplete Cholesky factorization were employed as "smoothers". In this paper, the combined adaptive mesh refinement technique has been applied to a single-phase tracer transport model for miscible flooding. Numerical results demonstrating the effectiveness of the method are presented and discussed.

    Petroleum reservoirs;Solubility;Object oriented programming;Radioactive tracers;Aquifers;Vectors;Linear systems;Computer simulation;
ph: 919.660.2800
fax: 919.660.2821

Mathematics Department
Duke University, Box 90320
Durham, NC 27708-0320