Parallel Linear Solvers for Reservoir Simulation: A Generic Approach for Existing and Emerging Computer Architectures

TX 75083-3836, U.S.A., fax 01-972-952-9435. AbstractThis multi-disciplinary study involved the use of parallel simulation using the same grid size as the geological model. Various production technology options for the remaining wells in a field development were investigated and the benefits of gas cycling were examined. A geological model covering an area of 40 km 2 was simulated using two approaches; one using 3 levels of nested LGR's and the other using an areal upscaling, both had nearly the same number of active grid cells. The 12-component Equation-of-State ("EoS") was substituted by 5-component to optimise the run speed of the simulation, 12-component was used for investigating gas cycling. The final engineering model had 320,000 active cells and completed 1.5-year historymatching runs in 1 hr 27 min using 16 parallel 400MHz processors with 1GB RAM each.Both methods worked well and the history was matched with minor changes to the geological model. The primary match parameters were the relative permeability curves since relative permeabilities from special core analysis at reservoir conditions were not available. Correlations from the literature were used for the original relative permeability curves. These were modified to capture the effects of near-well velocity stripping by 'straightening' the curves and moving the end points. Through a combination of local grid refinement (LGR) and relative permeability modifications, the effects of the condensate dropout near wellbore and the transient effects were captured. These effects reduce the wells' productivity by 40% in the first weeks of production.In the prediction phase the full field model was used to examine, holistically, the effects and interaction of various options chosen by analytical screening. The main solutions compared were high angle (near horizontal) reservoir penetrations and hydraulically fractured wells. Hydraulic fractures were modelled using LGR to capture the behaviour of the condensate in the fractures. The high angle wells were surrounded by LGR. A large number of prediction cases were run, optimising the development scenario using stochastically derived drilling schedules. Considerations of capital allocation led to the final recommendation from the group of optimal solutions.

show abstract

“…Several tests helped to obtain the optimum parallel decomposition1 16,17,18 . An acceptable parallel efficiency (Fig.…”

Section: Dynamic Modelmentioning

confidence: 99%

Use of Parallel Compositional Simulation to Investigate Productivity Improvement Options for a Retrograde-Gas-Condensate Field: A Case Study

H.A

Logmo-Ngog

Sarkar

et al. 2001

Proceedings of SPE Reservoir Simulation Symposium

View full text Add to dashboard Cite

show abstract

“…Several publications in the literature have dealt with the application of parallel computing to petroleum reservoir simulation in sharedmemory parallel environments. Scott et al (1987) Bhogeswara and Killough (1993) and Cowsar et al (1991Cowsar et al ( , 1992 worked on the parallel solution of linearized finite-difference equations. Mayer (1989) solved the Buckley-Leverett problem on a Thinking-Machines CM-2 machine and showed that the machine can handle large problems as well.…”

Section: Parallel Computingmentioning

confidence: 99%

Parallel algorithms for modeling flow in permeable media. Annual report, February 15, 1995 - February 14, 1996

Pope¹,

Sephernoori²,

McKinney³

et al. 1996

View full text Add to dashboard Cite

“…irregular grids, and local grid refinement have only been addressed to a limited extent. 25 The effect of these non-regular data structures will be significant on load balancing as well as single node performance since vectorization will be reduced.…”

Section: Future Issues In Parallelizationmentioning

confidence: 99%

Will Parallel Computing Ever Be Practical?

*¹

1993

Proceedings of Middle East Oil Show

View full text Add to dashboard Cite

Recently the lure of scalable distributed memory parallel computing with hundreds or even thousands of inexpensive, but powerful, RISC-based processors has resulted in a significant effort in the development of software for petroleum engineering and seismic applications. Unfortunately, few if any applications have found their way into a production environment from the reservoir engineering perspective. Several hurdles remain to be overcome if parallel computing is to become commonplace as a tool for applications. This article addresses three areas of concern for a major area of petroleum engineering -reservoir simulation: load balancing, data structures, and linear equation solutions. Examples from real reservoir simulations and possible solutions for several situations are discussed.

show abstract

Parallel Linear Solvers for Reservoir Simulation: A Generic Approach for Existing and Emerging Computer Architectures

Cited by 5 publications

References 21 publications

Use of Parallel Compositional Simulation to Investigate Productivity Improvement Options for a Retrograde-Gas-Condensate Field: A Case Study

Use of Parallel Compositional Simulation to Investigate Productivity Improvement Options for a Retrograde-Gas-Condensate Field: A Case Study

Parallel algorithms for modeling flow in permeable media. Annual report, February 15, 1995 - February 14, 1996

Will Parallel Computing Ever Be Practical?

Contact Info

Product

Resources

About