A Characterization of Shale Gas Reservoir Using Fast Marching Method Combined With Model Selection Approach

Abstract

Shale gas reservoirs show various production profiles depending on methods of hydraulic fracturing. Characterization for given shale gas reservoir is essentially included to predict future performances. Clustering of various reservoir models by their similarities and selecting of a cluster similar to the production history data are introduced in the process of characterization. The conventional clustering method using static properties such as fracture half-lengths and mean permeability has shown limited capability for realistic characteristics of a shale gas reservoir. In this study, the Fast Marching Method (FMM) combined with a model selection approach is proposed to develop reservoir models showing similar production profiles with the history data. The stimulated reservoir volume (SRV) is obtained from FMM as a value of the similarity in order to reflect the dynamic connectivity. The method is applied to various shale gas reservoirs with different fracture geometries for verification. It effectively gathers models with similar production profiles and fracture distributions. The accuracy of Estimated Ultimate Recovery (EUR) prediction is improved up to 7-14%p compared to the conventional method.