New Frontiers in Large Scale Reservoir Simulation
- Ali H. Dogru (Saudi Aramco) | Larry Siu Kuen Fung (Saudi Aramco) | Usuf Middya (Saudi Aramco) | Tareq Mutlaq Al-Shaalan (Saudi Aramco) | Tom Byer (Saudi Aramco) | H. Hoy (Saudi Aramco) | Werner Artur Hahn (Saudi Aramco) | Nabil Al-Zamel (Saudi Aramco) | Jorge Alberto Pita (Saudi Aramco) | Kesavalu Hemanthkumar (Saudi Aramco) | M. Mezghani (Saudi Aramco) | Abdulrahman Al-Mana (Saudi Aramco) | J. Tan (Saudi Aramco) | W. Dreiman (Saudi Aramco) | A. Fugl (Saudi Aramco) | Abdulaziz Al-Baiz (Saudi Aramco)
- Document ID
- Society of Petroleum Engineers
- SPE Reservoir Simulation Symposium, 21-23 February, The Woodlands, Texas, USA
- Publication Date
- Document Type
- Conference Paper
- 2011. Society of Petroleum Engineers
- 2.2.2 Perforating, 5.5.8 History Matching, 4.1.2 Separation and Treating, 5.5 Reservoir Simulation, 5.2.1 Phase Behavior and PVT Measurements, 5.8.7 Carbonate Reservoir, 2 Well Completion, 4.3.4 Scale, 3 Production and Well Operations, 5.1.5 Geologic Modeling, 1.6.9 Coring, Fishing, 5.4.2 Gas Injection Methods, 5.4.3 Gas Cycling, 3.3 Well & Reservoir Surveillance and Monitoring
- 3 in the last 30 days
- 587 since 2007
- Show more detail
- View rights & permissions
|SPE Member Price:||USD 8.50|
|SPE Non-Member Price:||USD 25.00|
Giant reservoirs of Middle East contain substantial portion of the world?s total hydrocarbons. Accurate simulation of these reservoirs requires as many as billion cells. A billion-cell parallel reservoir simulator was first presented in 2009.
This paper discusses the progress in the past two years and future projections.
In addition to black oil models, paper presents a large full field compositional model involving over billion unknowns. Throughout the paper grid size effects, computational work distribution and parallel scalability are discussed and illustrated. Paper also presents futuristic technologies for building and analyzing large simulation models.
In an earlier SPE paper by Dogru et al.1, we discussed the many features in a giga-cell simulator to address the growing needs for detail analyses of the giant reservoirs in the Middle East. These reservoirs typically have a low annual withdrawal rate and some of them have been in production for decades. During this time, very detail field and laboratory measurements have been collected. These, together with a comprehensive on-going reservoir surveillance program, have provided us with invaluable data to construct credible geological models for reservoir performance analyses. Up until recently, despite the many advances in parallel computing, typical reservoir simulators are still limited to the million-cell range. This is because many components in a reservoir simulator are complex and difficult to parallelize. Old-time simulators which have a serial-computing origin are still the de facto industry standard in use today. These simulators have very limited parallel performance and typically do not scale beyond tens of computational cores. For a large reservoir, a ten-million-cell resolution will mean a 250 meter areal grid size with very limited vertical layering definition. We estimate that in order to have reasonable resolution, giga-cell simulation is required. This is so that we can have both sufficient numerical resolution as well as geological resolution to describe the flow dynamics and heterogeneity such as reservoir simulation can be meaningful. Thus, an initiative was started in 2005 to build a next-generation simulator GigaPOWERSTM to tackle the giga-cell challenge.
Interest in parallel reservoir simulation started over two decades ago in the oil industry. The earliest attempt was by John Wheeler2. This was followed by work of John Killough3, Shiralkar and Stevenson4. In 1992, because of a lack of simulation capability in handling the large reservoir models, Saudi Aramco initiated a program to develop the company?s first in-house simulator: Parallel Oil Water and Gas Reservoir Simulator, POWERS5. This simulator was developed on a parallel computer (Connection Machine CM-5) in 1996. Shortly thereafter, the first field simulation model using 1.3 million cells was successfully completed6. In 2000, POWERS parallel code was ported to symmetric parallel computers (IBM Nighthawk shared memory systems), and later in 2002 to distributed memory PC Clusters.
By 2002, the full field model of the world?s largest oil field, Ghawar, was run using ten-million active cells with sixty years of production history and thousands of wells. Experience proved that parallel reservoir simulation was successful for reservoir studies. Therefore, more features such as local grid refinement7, compositional, dual porosity/dual permeability8, 9 and coupled surface facilities10 were added to the standard black oil options to cover more areas of applications.
|File Size||1 MB||Number of Pages||12|