Simulation of petroleum reservoirs is becoming more and more complex due to increasing necessity to model heterogeneity of reservoirs for accurate reservoir performance prediction. With high oil prices and less easy oil, accurate reservoir management tools such as simulation models are in more demand than ever before. The aim is to capture and preserve reservoir heterogeneity when changing over from a detailed geocellular model to a flow simulation model, minimising errors when upscaling and preventing excessive numerical dispersion by employing variable and innovative grids, as well as improved computational algorithms.For accurate and efficient simulation of large-scale models there are essentially three choices: upscaling, which involves averaging of parameters for several blocks, resulting in a coarser model that executes faster; the use of streamline simulation, which uses a more optimal grid, combined with a different computational algorithm for increased efficiency; and, the use of parallel computing techniques, which use superior hardware configurations for efficiency gains. With uncertainty screening of various multiple geostatistical realisations and investigation of alternative development scenarios— now commonplace for determining reservoir performance—computational efficiency and accuracy in modelling are paramount. This paper summarises the main techniques and methodologies involved in considering geocellular models for flow simulation of reservoirs, commenting on advantages and disadvantages among the various possibilities. Starting with some historic comments, the three modes of simulation are reviewed and examples are given for illustrative purposes, including a case history for the Bayu-Undan Field, Timor Sea.
Tutorial: Parallel Computing of Simulation Models for Risk Analysis