The oil and gas industry is changing rapidly, driving a need for fast and scalable high-performance implementations of reservoir simulations. Reservoir simulation models are used to predict the flow of oil or gas within porous media. The development of new fields can be simulated in order to assist investment decisions, for example about the position and number of wells required, the optimal completion of wells, or the present and future needs for artificial lift, etc. The models become more sophisticated and simulating them is a computationally highly complex task.

This white paper gives an overview of the state-of-the-art in reservoir modelling and simulation for the oil and gas industry and outlines where the computational complexities lie. Opportunities for software optimisations are identified and practical examples are given. Further, the paper analyses the suitability of accelerator processors such as GPUs as target platforms, used in combination with traditional CPUs.

  • State-of-the-art review in reservoir modelling and simulation
  • Best use of cache and NUMA memory hierarchies
  • Vectorisation techniques for reservoir simulations
  • Mapping to Nvidia GPUs
  • Combining CPUs with accelerators
  • Hiding of data transfer latencies in a cluster
top