Metric Space Mapping of Oil Sands Reservoirs Using Streamline Simulation

Abstract

ABSTRACT Oil sands reservoir models have high uncertainty because of complex shale distribution and insufficient geophysical data. Therefore, reservoir behavior is predicted stochastically using lots of generation and flow simulation of feasible reservoir models. However, it takes huge time and cost to simulate all of reservoir models in order to understand its behaviors. The objective of this study is to map the relationship between oil sands reservoir models in metric space using streamline simulation. In order to understand structure of reservoir models visually, 100 reservoir models are generated by sequential indicator simulation. Before mapping, a distance, as a dissimilarity index is defined between models using streamline simulation. Since the production behavior of an oil sands reservoir is heavily affected by injected steam, streamline simulation is applied to predict the trend of the steam flow in an oil sands reservoir. After streamline simulation, the distance is calculated by comparing the results of streamline simulation. Each distance estimated is represented as a geometric distance in metric space map by using multi-dimensional scaling (MDS). From MDS analysis, reservoir models, which have similar production character, are gathered in metric space.