The availability of multi-core CPUs for personal computers makes desktop parallel computing a reality. Parallel computing for reservoir simulation creates the opportunity for significant run time reduction but also introduces additional technical challenges, particularly in the areas of load balancing and general algorithmic robustness.

In this paper we will discuss our experience with the Adaptive Implicit Method (AIM) in parallel desktop computing environments, with ExxonMobil's numerical reservoir simulator, EMpowerTM, focusing on the following areas:

  • A highly efficient and robust parallel partition of the fully unstructured grids for modeling large and complex reservoirs

  • Adaptive Implicit Method (AIM) for a high-level of numerical stability with significant run-time speed-up and memory footprint reduction

  • Unique load balancing challenges arising from integration between shared-memory based parallel computing, unstructured grids, and adaptive implicit methods

We will also describe experiences with real models and discuss mechanisms for addressing problems encountered and demonstrate that we can achieve a 3~4 time speedup over serial runs without AIM on real field cases using quad-core desktop machines.

You can access this article if you purchase or spend a download.