a national resource for computational science education

HOME BWSIP Shodor Blue Waters

View Position
Position TitleHigh-Performance Multi-phase, Multi-component, Subsurface Fluid-Flow Simulator Development Using Fluid-Phase Equilibria Under Confinement
SummaryOur research group at The University of Oklahoma, Norman, is undertaking a challenging task of developing a multi-physics, multi-component, multi-phase simulator for subsurface fluid and heat flow, reactive transport, and stress simulation. The intent of this simulator will be to incorporate new physics at multiple scales that are currently not being implemented in available research or commercial simulators. Under the scope of Blue Waters Student Internship Program (BWSIP), we have identified a subset of tasks from this challenging undertaking as a suitable project. The goal of the BWSIP internship project will be to develop a prototype for high-performance, multi-component, multi-phase simulator for subsurface fluid flow.
Job DescriptionThe objectives/tasks of the BWSIP interns will include:

1. Building a prototype simulator using Matlab Reservoir Simulation Toolbox (MRST) from SINTEF. MRST is an open-source code for black-oil reservoir simulator.

2. Implementing a multi-component compositional fluid flow simulation module within MRST framework. Implement a high-performance scalable version of this multi-component simulator.

3. Developing a vapor-liquid equilibria calculation module under confinement incorporating pore-throat distribution-dependent capillary pressure within MRST framework. Implement a high-performance scalable version of this multi-component simulator.

4. Implementing a vapor-liquid-liquid equilibria (VLLE), vapor-liquid-liquid-solid equilibria (VLLSE) calculation module within MRST framework. This will be an extension of Steps 1-3 to multiphasic fluid-phase equilibria.

Computational Methodology

1. MATLAB Reservoir Simulation Toolbox (MRST) is developed by SINTEF ICT, and it is mainly built for rapid prototyping of single-phase and multi-phase flow in porous media and testing new mathematical models and simulation methods (Lie, 2016). For the BWSIP project, we believe MRST will be the ideal platform to implement the intended multi-component simulation modules.

2. MRST is mainly built for black-oil single and multi-phase flow simulation in porous media. The fluid models implemented in MRST rely on multitude of correlations for various thermo-physical properties. We will develop a multi-component compositional simulation module within MRST that relies on fluid phase equilibria of multicomponent mixtures minimizing global Gibbs Free Energy of the system. At equilibrium, the fugacity of each chemical component or pseudo-component (e.g, methane, ethane, propane, …) in any phase will be equal to that of the same component in the other co-existing phases. Naturally, compositional simulation requires huge computational effort and resource due to its rigor and complexity compared to black-oil simulation. Additionally, in recent years, numerical geological models being constructed contain multi-million cells. Computational effort for these large models becomes prohibitively expensive. Therefore, it is crucial to apply HPC in the compositional simulation to speed up the process.

3. In the third item, we plan to incorporate additional physics into the compositional simulator. Pore geometry of nano-scale porous media imposes additional complexity in fluid-phase equilibria determination. This additional physics will be captured by considering pore-throat dependent capillary pressure. Phase equilibrium can be substantially affected under capillary pressure (Sandoval et al, 2016 and Shapiro 2001). Including the capillary pressure requires an additional layer of computational complexity in phase-equilibria calculation. Naturally, we will seek to apply HPC to reduce the computational cost.

4. In the final step, we extend the work developed in Steps 1-3 from vapor-liquid phenomena to multiphasic (vapor-liquid-liquid and vapor-liquid-liquid-solid) phenomena. To capture the additional physics under the presence of more than two phases, the fluid-phase equilibria calculation module needs to be restructured. This will further increase the complexity and computational cost. HPC concepts will be needed to achieve reasonable simulation run-time for the multiphasic flow conditions.

BWSIP interns will assist in implementing various modules for the new multi-physics simulator.

Lie, K.-A., 2016. An introduction to reservoir simulation using MATLAB: User guide for the Matlab Reservoir Simulation Toolbox (MRST). SINTEF ICT, December 2016, Projectweb/MRST/Publications.

Michelsen, M. L., 1982. The isothermal flash problem. Part I. Stability. Fluid Phase Equilibria. 1982, 9, 1-19.

Shapiro, A. and E. H. Stenby, 2001. "Thermodynamics of the multicomponent vapor-liquid equilibrium under capillary pressure difference". Fluid Phase Equilibria. 2001, 178(1-2), 17-32.

Sandoval, D., W. Yan, M. L. Michelsen, and E. H. Stenby,2015. Phase Envelope Calculations for Reservoir Fluids in the Presence of Capillary Pressure. Society of Petroleum Engineers. 2015, September 28. doi:10.2118/175110-MS
Use of Blue WatersThe use of Blue Waters supercomputer will allow us to benchmark the performance of the new multi-phase, multi-component, subsurface fluid-flow simulator using fluid-phase equilibria under confinement. Scalability of various solvers in the simulator will be investigated.

We use here a simplistic example to illustrate the need for HPC in the multi-component compositional simulation. We considered 1-million grid cells in the entire reservoir model (which is quite common nowadays). Considering optimistic values for all factors required for computation, we obtain the simulation will require around 10^12 fluid-phase equilibria calculations. A complex realistic compositional simulation can take hours to days of run-time even in a 128-core high-performance server. Therefore, we will develop useful strategies and algorithms to reduce the number of phase-equilibria calculations needed and rely on efficient HPC techniques to further save computational cost.
Conditions/QualificationsInterns have already been identified.
Start Date06/01/2017
End Date05/31/2018
LocationNorman, Oklahoma, The University of Oklahoma,
Mewbourne School of Petroleum and Geological Engineering
Gage Russell
Justus Brown