Application of Dynamic Upscaling for Thermal Reservoir Simulation
Abstract
Reservoir simulation is inherently an imperfect tool for forecasting. However, given sufficient analysis and post-processing, the areas of uncertainty can be quantified and effort can be made to mitigate their impact and improve the confidence in the prediction. The focus of the research documented here is to analyze the extent to which grid size definition and the location and quantity of reservoir heterogeneities impact the performance of a simulation-based recovery process. In analysing two different sets of models (binary and facies-based), a new methodology was developed and applied to mitigate the observed difference in the forecasted result. When the observations are coupled, it suggests that an increase in near wellbore reservoir heterogeneity, as indicated by a reduction in gridblock connectivity, has an increased impact upon the simulation results when compared to the impact of strictly length scale definition of heterogeneity. Additionally, the impact of length scales can be normalized when the focus is upon connectivity within the reservoir model.
