The overall objective of reservoir modeling is to reduce the uncertainty in production forecasts by utilizing all available data to construct a calibrated reservoir model. Geological heterogeneities have a fundamental impact on the growth of a steam chamber and the performance of a SAGD (steam assisted gravity drainage) process. The objective of this work is to incorporate geological heterogeneities into the history matching process using a probability perturbation method (PPM) to preserve the geological consistency of a reservoir model.
A PPM is a geological data integration framework which employs a multiple-point geostatistics (MPS) algorithm. The heart of this method is to systematically perturb the underlying probabilities used to generate the reservoir facies. A PPM generally consists of two loops: an outer loop which is responsible for randomly generating a global configuration of the facies and an inner loop which systematically perturbs the generated facies to match the dynamic data. The combination of these two iterations creates a set of realizations that preserve the geological information.
In this paper, a training image is built based on a 3D outcrop description of a meandering channelized reservoir that is analogous to some of the Canadian heavy oil reservoirs. All other available data including reservoir properties at well locations, trends and production data are also incorporated into the PPM framework for this history matching process. The reservoir model is characterized by three facies: clean sands, medium-grained sandstones and silts, which have different porosity, horizontal permeability and vertical permeability. The SAGD performance is a function of steam chamber development, which depends on the level of heterogeneity in the reservoir. The results show that the heterogeneity distribution has a large impact on the fluid flow at different stages of production. The results show that such complexities can be well preserved during the history matching process using the PPM by generating the geological patterns depicted in a training image. The PPM is shown to be an efficient approach for history matching in presence of complex reservoir geology.