実験計画法を用いた貯留層の不確実性評価

Abstract

Systematic approaches to uncertainty analysis in undeveloped oil or gas fields are everlasting challenges in the oil industry. Probabilistic estimations of recoverable reserves, and plateau period are far from being simple because reservoir parameters often have so-called the interacting effects. Further difficulty stems from the fact that they need many flow-simulation runs before running the Monte Carlo simulations. Solutions to reduce the number of necessary simulation runs without losing the accuracy of estimation have been studied.In this decade, experimental design (ED) approach is focused on instead of the conventional one-factor-at-a-time (OFAAT) method. One of the advantages of ED is that much less flow-simulation runs are needed to obtain relatively unbiased probabilistic estimates, while the OFAAT method needs more flow simulation runs, and tends to yield a narrow range of probabilistic distribution.This report reviews the ED approach in the oil industry and the results of application to our field development case.