Applications for Response Surfaces in Reservoir Engineering

by
Keshav Narayanan, MSE

University of Texas at Austin, 1999
Supervisor: Christopher D. White

Pseudofunctions used in scaled up reservoir simulation models depend on the geology, recovery process and fluid velocities. Accurate pseudofunctions can be derived in every section of the reservoir model to be coarsened by expensive fine-scale simulations with appropriate boundary conditions. Experimental design and response surface modeling provide an accurate and much cheaper framework for calculating effective properties. Effective properties are calculated at a few chosen points (the design) by simulating flow with a fine grid. Best-fit polynomials can be computed using response surface methods. Models based on outcrop exposures of the Frewens Sandstone, a fluvial-deltaic deposit in Central Wyoming are upscaled using response surfaces with few fine scale simulations. Uncertainty in effective properties and their sensitivity to factors in the reservoir description can be accurately and inexpensively investigated using response surfaces. Probability distributions of the effective properties are derived by conducting Monte Carlo simulations with the response surfaces.

Back to theses index