MIT GeoNumerics Group

The primary research objective of MIT Geonumerics is to develop next generation oil reservoir simulation technologies which are able to enhance our knowledge of reservoir physics and provide a means by which to enhance oil recovery.

Conventional simulation tools used for reservoir modeling generally decouple the multi-physics nature of the environment and model phenomena such as fluid dynamics, rock mechanics, surface chemistry, electro-chemical physics and acoustic response using separate, specialized modeling tools. The unique approach of MIT Geonumerics has been to recognize that many of these physical reservoir attributes have complex interdependencies and isolated modeling of any one trait may, in fact, be missing features critical to an accurate simulation result.

Correspondingly, it has been our goal to develop a simulation framework capable of simulating multi-physics in an integrated way. Due to the high complexity of such an endeavor, high performance computing is integral to our approach.

While still in the early stages of this multi-disciplinary project, initial results strongly support the idea that such an integrated simulation framework can be developed.
History

In an oil reservoir, 20-40% of the oil can be recovered by primary development techniques. The rest remains trapped in the rock pores. Enhanced Oil Recovery (EOR) techniques, such as water flooding, gas injection, chemical injection and thermal stimulation, optimistically recover an additional 10-20% of the oil. This still leaves almost half of the oil trapped in the rock pores. The Department of Energy (DOE) has estimated that if "next generation" EOR is applied, the United States could generate an additional 240 billion barrels of recoverable oil resources - over 30 years supply at the present US consumption rate of 20 million barrels per day. For comparison, the Middle East holds an estimated 685 billion barrels that are recoverable and the tar sands of Alberta 300 billion recoverable barrels of "heavy" oil, with over a trillion barrels potentially recoverable using enhanced methods.

Designing new EOR technologies is inhibited by our poor understanding of the fundamental physical processes within a reservoir, particularly at the pore scale. Some of the modeling challenges include:

(*) Predicting multi-phase fluid flow of oil, water and gas through complex pore networks,
(*) Predicting the properties of the rock and fluids from indirect down-hole measurement techniques, such as electrical resistivity and acoustic wave propagation,
(*) Understanding the hydro-fracturing of rocks,
(*) Understanding the mechanisms of sand production and borehole collapse,
(*) Mastering the computational challenges, which stem from reasoning about the positions and contacts of large numbers of objects, and
(*) Architecting integrated multi-physics software systems that can run on multi-core/multi-machine architectures.

With the huge potential benefit of next generation oil reservoir simulation tools, MIT Geonumerics was founded with the commencement of the “Canonical Rock” project in the fall of 2007. Canonical Rock is a collaborative effort between MIT and the Schlumberger Doll Research Center. After successful implementation of a new multi-core high performance computing framework and a powerful Smooth Particle Hydrodynamics simulator for multi-phase fluid flow simulations, Saudi Aramco joined the collaboration early in 2009. Canonical Rock is viewed as being a 10 year research effort at MIT.

Future Directions -- Within the bounds of the Canonical Rock collaboration, short term future research goals include implementing an integrated, deformable rock model into the fluid flow simulator; incorporating particulate flow into the fluids to research the behavior of nano-particles used for reservoir mapping; incorporation of solute transport and precipitation to model phenomena such as stress induced grain erosion; and integration with chemical and electrical simulation algorithms currently being developed.

Additionally, MIT Geonumerics is investigating the potential of using the developed simulation tools for modeling other environmentally significant geological problems such as carbon sequestration, hydrates recovery (i.e. mining subsea methane deposits) and geothermal energy production.

MIT Geonumerics is a cross-disciplinary collaboration between academics and researchers from multiple MIT departments. These departments include the Departments of Earth, Atmosphere, and Planetary Science; Civil and Environmental Engineering; and Engineering Systems. By using a cross-disciplinary approach to oil reservoir research it is our aim to develop a “Canonical Rock Model” which represents the true multi-physics nature of the problem within a single simulation framework.