Home

Quick Links

Search

 
A Parallel Streamline Simulator for Integrated Reservoir Modelling on a Desktop (SPE 154484)Normal access

Authors: M. Gaupaas, F. Bratvedt, S.K. Khataniar, A. Primera and F.P. Ruan
Event name: 74th EAGE Conference and Exhibition incorporating EUROPEC 2012
Session: Reservoir Simulation II - Numerical (Europec)
Publication date: 04 June 2012
Organisations: SPE, EAGE
Language: English
Info: Extended abstract, PDF ( 1.08Mb )
Price: € 20

Summary:
Desktop computing is undergoing a revolution with parallel processing on multi-core workstations. Parallel streamline simulators have been developed for shared memory architecture systems, using off-the-shelf compilers with an application programming interface for parallel programming. Here, we discuss the implementation and the performance analysis of a parallel streamline simulator based on native threading technology for both WindowsCB. and LinuxCB. operating systems. Although parts of the general streamline simulation algorithm are relatively straightforward to parallelize, there are several challenges that require special attention to avoid computing bottlenecks and inconsistent results across different computing environments. An efficient load balancing algorithm to avoid idle processors has been implemented, combined with a data-accumulation scheduling algorithm to ensure consistent results independent of the platform and the number of processing units. The combined performance of a multicore computer and a parallel streamline simulator offers significant opportunities for reservoir management applications. It can also increase the use of 64-bit desktop workstations that are commonly used for 3D geological modelling and the creation of applications that integrate the geosciences. Parallel scalability analysis for various model characteristics and simulator options is also analyzed. For a variety of models, we have observed an almost linear scalability for as many cores as available on a typical shared memory high-performance computer.


Back to the article list