EVALUATING THE IMPACT OF PETROPHYSICAL IMAGES PARAMETERIZATION IN DATA ASSIMILATION FOR UNCERTAINTY REDUCTION

Abstract

Parameterization is a crucial step during uncertainty reduction of reservoir properties using dynamic data. It establishes the search space based on prior knowledge of the model and can have a significant influence on the final response. A less-appropriate parameterization might fail to have a reasonable representation of the reservoir and lead to models unable to predict the correct reservoir characteristics. Parameterization of petrophysical images (as facies, porosities, and permeabilities) plays an essential role during data assimilation processes due to the strong influence in fluid flow in the porous media. This work shows how important the parameterization of petrophysical images is and how a less-appropriate parameterization can affect history-matching and uncertainty reduction process. Using a benchmark case, we compare two parameterization techniques, one capable of treating all blocks in the model (distance-dependent covariance localization), which is considered more appropriate, and one that considers a group of blocks under the same update rule (zonation) (less-appropriate). Results show that parameterization of petrophysical images has a high impact on the final response, and a less-appropriate parameterization, as the zonation, can generate higher data mismatches and fail to represent the real reservoir response. The analysis carried in this work quantifies and qualifies the impact of the parameterization of the petrophysical images in the data assimilation for the uncertainty reduction process.