09122-11

Primary Performer
Board of Regents of the University of Oklahoma

Additional Participants
BP
Chesapeake
Exco
Newfield
Total
Computer Modeling Group, Inc.

Abstract
A valid reservoir simulator is a key technology to plan, model and predict the results of production operations. A proper simulator must provide for the appropriate pore geometry complexity, and model the processes with valid physical assumptions. Shale gas reservoir models require both high complexity and modified physics. Production of such a simulator is the prime objective of this project.

In the last couple of years as shale reservoir rocks have been imaged using ion-milled samples and SEMs with nano-meter resolution, it has become clear that some portion of the pore geometry is contained in the organic material. This material is hydrophobic, gas wetting and probably never contained brine. The organics also store gas by adsorption. The shale also contains brine filled pores that are most likely water wet, and gas may also be present in non-organic pores of undetermined wettability. This complex matrix porosity is overlain by a natural fracture system, probably water wet, and the complex fracture system created by the simulation, which may be fractionally wet. There is as yet no clear understanding on how these pore systems are connected. The reported observations are that many shale gas reservoir rocks imbibe oil-based mud, and also that some do not imbibe water-based mud speaks to the complex wettability and pore geometry structure of the system. NMR measurements show strong surface relaxation for both oil and water implying a heterogeneous wettability system. Any valid reservoir model must be able to accommodate the four porosity elements potentially present and provide for the ability to have arbitrary connectivity between them.

Because of their very low permeability and potentially large capillary forces, standard commercial simulators make approximations that are inappropriate for porous media with very small pore sizes such as shale gas reservoir rocks. The most serious approximations are: 1) the assumption of instantaneous capillary equilibrium, 2) that transport can be completely defined by viscous flow (Darcy’s law), and 3) that relative permeability is not flow rate dependent. These assumptions result in simulators that: do not correctly predict the amount of produced water; do not properly handle the changes in gas transport rates with time, so probably can not correctly model gas production; and do not correctly predict the deposition of stimulation water especially during re-stimulation. The last has serious implications on predicting the possible development of water blocks. The project deliverable will be a simulator with the correct physics and possible pore geometries so that modeling and history matching provide valid information about the reservoir geometry its production rates and its ultimate recovery that are not corrupted by the approximations implemented in the simulator. It will further provide an essential component to any coupled stimulation simulation modeling system, and a starting point for simulators that can model wet gas shale reservoirs.

This project will first develop the algorithms necessary to implement the correct physical principles and pore geometry in a simulator. These formulations will then be tested on simple problems using a prototype simulator that is currently being developed for this purpose. After this testing stage the new algorithms will be implemented as modules in selected commercial simulators and tested on simple cases. The final stage of the project will be performing simulations on producing reservoirs using the modified simulators and the original commercial simulators to determine the practical effects of using a simulator that incorporates the better physical approximations and pore geometry.

Cost share mainly will be provided by BP, Chesapeake, EXCO, Newfield, and Total, who have partnered with the University of Oklahoma on this project. The University of Oklahoma will wave some of the graduate students tuition. Additional cost share will be provided by the waving of the user fees at the OU Super Computing Center, and using without fee commercial simulators from Schlumberger and CMG. Also CMG has agreed to provide expert help in implementing the new algorithms on their simulator.

Principal Investigator: Richard F. Sigal