May 8, 2017
The soaring productivity of oil and gas from hydraulic fracturing wells depends in large part on advancements in the “digital oil field,” according to a recent Wall Street Journal article that said computational technology is driving the output per shale drilling rig to rise by more than 20 percent a year.
“‘The cloud’ will be just as much of an economic accelerant for shale as it has been for other complex and distributed industries,” wrote Mark Mills, a senior fellow at the domestic policy think tank the Manhattan Institute in an editorial for the paper.
The Center for Subsurface Modeling (CSM) at the Institute for Computational Engineering and Sciences (ICES) is leading the way in modeling the complex physics that controls fracture propagation, geomechanics, and fluid flow during the hydraulic fracturing process. Research at the center is improving the scientific foundation that oil and gas software packages are built upon, and getting a better idea of the uncertainty present in the solutions. Work at the center is also applying "big data" analytics together with high performance computing to tackle the problem of fractured subsurface characterization by using advanced computation approaches on big subsurface seismic data sets. This improves the fidelity of the computational models for research and industry alike.
The cube shows a phase field model of fluid-flow driven fracture propagation. The Phase Field Approach, a new method developed at CSM, models a fracture as a volume in space with a route that’s determined by energy minimization—a scientific concept that selects the route with the lowest energy requirements when multiple options are possible. Previous approaches have treated a fracture like a line or flat surface moving through space, with the fracture’s path determined by ad hoc phenomenological rules. Recent reports show Dr. Wheeler's kind of computational technology is driving the output per shale drilling rig to rise by more than 20 percent a year.
"Many big data research teams focus only on data. Here at the Center for Subsurface Modeling, we emphasize physical models that incorporate data assimilation both at the well bore, production, and geophysics," said Mary Wheeler, the director of CSM.
The subsurface formations where hydraulic fracturing takes place are composed of layers of porous rock together with natural and hydraulic fractures. Wheeler has been at the helm of honing the mathematics that describes locally conservative fluid flow through heterogeneous porous media for decades. Throughout her career, she has co-authored hundreds of papers on the numerical simulation and analysis of multiphase flow, reactive transport, and geomechanics.
As hydraulic fracturing has become a more prominent method for recovering energy resources across the world, researchers at the center have zeroed in on simulating the interactions between the fracture, rock, and fluid.
"Understanding the behavior of the fractures in a porous media is key to making informed decisions about where and how to drill," said Sanghyun Lee, a research associate at the center.
The entire CSM team is focusing on developing new methods for modeling how fractures propagate through rock from the single point where they begin. The status quo treats a fracture like a line or flat surface moving through space, with the fracture’s path determined by ad hoc phenomenological rules. In contrast, the Phase Field Approach for porous media, a new method developed at CSM, models a fracture as a volume in space with a route that’s determined by energy minimization—a scientific concept that selects the route with the lowest energy requirements when multiple options are possible. This work extended previous studies done in solid mechanics to poro-mechanics with mathematical analysis and numerical computations.
“The crack propagation path and joining and branching is automatically determined by energy minimization,” Lee said. “This is one of the major advantages for the phase field only. For other methods, you really have to compute how fractures are joining and where they’re branching at the crack tip.”
Another important feature of the fracture propagation algorithm is its coupling to other physics, such as fluid flow. A variety of different fluids—such as proppant, oil, and gas—can pass through fractures, with different fluid types often intermingling together at the same time. The CSM team is working on integrating the mathematics behind this “multiphase” fluid flow as well as the non-Newtonian fluid flow of proppant on its own. Wheeler recently received a $1.5 million “big data” grant from the National Science Foundation to develop computational methods to implement these kinds of algorithms.
"We are currently undertaking ambitious plans to couple fracture propagation techniques with compositional multiphase flow in our in-house reservoir simulator IPARS," says Ben Ganis, a CSM research associate and principal researcher on the NSF grant. "We are combining adaptive mesh refinement, fully-coupled solvers, and computer science techniques to achieve this goal."
According to Wheeler, software for modeling hydraulic fracturing oftentimes treats the coupled physical processes independently, which may cause important interactions to be neglected. However, she notes, it’s difficult to know what mathematics exactly are governing the solutions produced by commercial software, because the codes are proprietary.
Obtaining actual data is an ongoing effort for the center. The center’s industrial affiliates program, which includes some of the world’s largest oil and gas companies, helps foster data exchange, as well as helps guide the research toward problems that are of mutual interest to industry and research.
The field data that the center has received so far has fostered the development of an in-house reservoir simulator and many additional computational tools. Wheeler said that making the “digital oil field” as realistic as possible is the key to improving efficiency of both research and industry moving forward. She’s working to gather more data, and find more places to put it to the test.
“I personally think that this is probably the best scheme that’s available,” Wheeler said of methods being developed at the center. “We’ve matched it with both benchmarks and experiments, but right now coupling it, and being able to use it in field experiments is what we’re in the process of doing.”
