Using an Integrated Multidimensional Scaling and Clustering Method to Reduce the Number of Scenarios Based on Flow-Unit Models Under Geological Uncertainties

Abstract

Understanding the role of geological uncertainties on reservoir management decisions requires an ensemble of reservoir models that cover the uncertain space of parameters. However, in most cases, high computation time is needed for the flow simulation step, which can have a negative impact on a suitable assessment of flow behavior. Therefore, one important point is to choose a few scenarios from the ensemble of models while preserving the geological uncertainty range. In this study, we present a statistical solution to select the representative models (RMs) based on a novel scheme of measuring the similarity between 3D flow-unit models. The proposed method includes the integration of multidimensional scaling and cluster analysis (IMC). IMC can be applied to the models before the simulation process to save time and costs. To check the validity of the methodology, numerical simulation and then uncertainty analysis are carried out on the RMs and full set. We create an ensemble of 200 3D flow-unit models through the Latin Hypercube sampling method. The models indicate the geological uncertainty range for properties such as permeability, porosity, and net-to-gross. This method is applied to a synthetic benchmark model named UNISIM-II-D and proves to offer good performance in reducing the number of models so that only 9% of the models in the ensemble (18 selected models from 200 models) can be sufficient for the uncertainty quantification if appropriate similarity measures and clustering methods are used.