Next Generation Parallel Computing for Large-Scale Reservoir Simulation

Abstract

Abstract This paper describes application of Project INTERSECT, a next generation highly scalable reservoir simulator on real large scale field models. High resolution reservoir simulation is required to better define and describe fluid flow and enable improved field development and tactical operational planning. Massively parallel computing techniques overcome limitations of problem size and space resolution.This paper demonstrates that large-scale simulation models can be performed on commodity hardware taking advantage of evolution in multi-cpu hardware architecture and software engineering. This allows both geologists and reservoir engineers to include more realistic geologic and engineering detail for better and more reliable production optimization. Intense computer simulation is essential for effective reservoir management. The advances in reservoir characterization techniques and the industry drive towards the ‘smart oilfield’ with rapid model updates will require more efficient model processing to achieve timely field operational decisions. Parallel reservoir simulators have the potential to solve larger, more realistic problems than previously possible. The size and application of reservoir simulation problems have been limited by the availability of computing hardware, reservoir simulation architecture and of solution methods for solving large-scale heterogeneous problems. The next generation reservoir simulator demonstrates that key modeling challenges has been overcome by a software architecture and capability to model more realistic subsurface and surface models. Applications of the new reservoir simulator illustrates how typical reservoir engineering options such as local grid refinement, local grid coarsening, multilateral wells and aquifer modeling affect the overall parallel performance and scalability using highly heterogeneous large-scale models.Application of new modeling techniques highlight increased accuracy of modeling results and more reliable field development planning and reservoir management decisions. Introduction To generate higher returns on capital employed, the oil and gas industry must follow a two-pronged strategy: reduce the cost of finding and developing new reservoirs while improving production performance for existing reservoirs. The evolution of hardware and software is increasing rapidly in the energy sector as personnel involved in field developments need to keep up with trends in order to make fit-for-purpose decisions for long-term operational designs and short-term tactical planning. Modern petroleum reservoir simulation requires simulating high resolution and detailed geological models. The advent of cluster computing relies on accurate and efficient model based computing, such modeling is primarily performed on detailed models representing flow in permeable media. Future production depends on large scale computational efficiency to enable enhanced reservoir characterization and adoption of new oil recovery technologies. Over the last twenty years high performance computing has had a significant impact on the evolution of numerical predictive methods throughout science and engineering. In particular, petroleum engineering applications has seen a significant enhancement in capabilities for reservoir simulation engineering. The complexity of geological and reservoir simulation models has led to computational requirements that have consistently challenged the fastest hardware platforms. Fig. 1 illustrates the general trend seen in simulation model grid resolution as seen by the oil and gas industry over the last 30 years. The increase in grid resolution is clearly linked to the advance in computer hardware technology and the price/performance of the overall hardware platforms. Early hardware platforms were largely based on mainframes that provided efficient processing; however, it only enabled coarse models. The emergence of workstations in the late 80s not only made computing hardware more accessible to the engineer, but also enabled more refined models that more closely resembled geological models. The evolution of workstations towards cluster computing emerged as reservoir characterization and up-scaling tools become more advanced and more easily accessible to the engineer. This enabled a step change in grid resolution as existing simulator technologies were migrated towards taking advantage of parallel processing.