Ensemble size investigation in adaptive ES-MDA reservoir history matching

Abstract

In this work, we study the ensemble size influence on an adaptive ensemble-based methodology for history matching of petroleum reservoirs. The assimilation scheme used is an adaptive ensemble smoother with multiple data assimilation (ES-MDA) in which both the total number of assimilations and the inflation factor of each iteration are defined automatically by the algorithm. This fact leads to the assumption that the predefined algorithm parameters may have influence in the total number of assimilations and the inflation factors. One main parameter that can be investigated is the number of ensemble members used in the assimilation, also called ensemble size. The ensemble size influence was analyzed by applying the adaptive ES-MDA in a synthetic large-scale reservoir model. As a result of the investigation, the ensemble size showed influence on the reduction in the uncertainty of the posterior models, but it did not show any influence on the total number of assimilations and on the inflation factor selection.