A Hybrid of Sequential-Self Calibration and Genetic Algorithm Inversion Technique for Geostatistical Reservoir Modeling

Abstract

Geostatistical modeling is a widely used approach to model the heterogeneity of reservoir petrophysical properties. This paper investigates a geostatistical-based inversion technique, a hybrid of sequential-self calibration (SSC) and genetic algorithm (GA), to model reservoir permeability. In this method, a GA is used to search the optimal master point locations, as well as the associated optimal permeability. These permeability values are then propagated to the entire reservoir using Kriging algorithm to match the dynamic production data. We demonstrate that GA is easy to implement and the results are robust. Additionally, we experimented with various numbers of master points, including a linked-list genotype which permits a flexible number of master points. The results show that GA is able to find various numbers of master points and their locations that are suitable for the reservoir field we studied. These numbers are within a small range and are sufficient to capture the heterogeneity of the reservoir permeability to match the production data. The ability of the SSC-GA method to model reservoir permeability by simultaneously optimizing the number of master points, the locations of these master points and the associated permeability in this case study suggests that the technique might be effective with other larger fields.