An adaptive pluri-Gaussian simulation model for geological uncertainty quantification

Abstract In the reservoir exploration phase, different types of information are gathered and used for a reliable geological description. Combining seismic data, well log analysis, statistical rock physics or even paleobathymetry ranges, several methods have been proposed to estimate a probability field for each facies type in the reservoir model. However, these probability fields are typically not conditioned to the reservoir production history. Once the reservoir starts production new information becomes available, and an update of the probability fields is needed. The work presented here introduces a new framework for simulation of facies fields in the context of plurigaussian simulation where the facies fields are conditioned to the prior probability fields provided. The methodology is based on the probability integral transform and the topological characteristics of the facies types (number of the facies type and relative position among facies types). The developed method generates an ensemble of facies fields that honor the facies distribution as described by a probability field. The proposed method can easily condition the facies field to hard data and preserve the facies field during history matching in an ensemble based framework. A demonstration with the adaptive Gaussian mixture filter is presented here.