Modeling a Hydro


There’s no doubt hydrofracturing technology has keyed the boom in energy production in the United States, pulling oil and natural gas from underground formations previously considered impermeable, or at least too difficult and costly to exploit. Ironically, the great success came despite the inability of design engineers to accurately predict how and where the fractures form underground. The unpredictability is now becoming predictable because of high-speed computer modeling.

Researchers at Lawrence Livermore National Laboratory had been developing models that would predict how fluid moves through underground formations, partly in relation to underground waste disposal and nuclear testing, says Roger Aines, a senior scientist at LLNL. “The hydrofracking boom in the oil industry was happily coincidental to these kind of things,” he says. “It turns out no one has ever done this before. It is a very hard thing to compute.”

Best practices have been evolving as fast as the industry has developed. Producers have become more efficient in placing multiple wells on a site, as they gather data on the underground geology. That information, together with what is available in existing literature for a particular formation or “play,” is fed into the LLNL model, which will allow researchers or producers to actually design a fracture, says Aines.


“The question is, ‘What is actually occurring in the subsurface?’” says Randy Settgast, a Livermore staff scientist and code architect for the model. “Nobody really knows.”

Every shale formation is different, and the differences within a formation vary greatly over short distances, even from county to county, say engineers. Such conditions preclude a generic or wide-ranging type of model. But the formations share one basic quality: until recently, the source rock could not be tapped because “It’s got the permeability of a roof shingle,” says Settgast.

Fracking, which relies on horizontal drilling technology, solved that problem. Drilling crews drop vertical wells to a predetermined depth and then drill horizontally, fracturing the rock with a high-pressure stream of water, sand or chemicals. “You are creating a reservoir by cracking rock and creating surface area,” says Settgast. “The trick is how you make a fracture, where you want it to be ... and not just break the rock will nilly.”


Hydraulic fracturing technology is not new and has been used since the 1940s. George Mitchell of Mitchell Energy is widely considered the “father of fracking” because in 1997 he successfully applied the technique to economically extract shale gas from the Barnett Shale in north Texas. “There were very few design controls then and that is still true today,” says Settgast.

But producers are keen on dropping fewer wells while increasing recovery from the ones that are drilled, as well as limiting environmental impact. Environmental objections are the key impediment to the industry, as state regulators and local citizens’ group worry over groundwater contamination, industrial pollution and other issues. Using a tool such as the Livermore model could go a long way to not only accurately predicting fracturing, recovery and possible remediation, but in restoring trust, say researchers.

Livermore researchers have been developing the model for three years, seeking industrial partners to with whom to work and verify results. Drawing from the existing knowledge and literature of the Marcellus shale in western Pennsylvania and eastern Ohio, for instance, they predicted and showed self-propogating fractures that extended during production and pressure draw-down. “We set out to make the fractures more designable,” Settgast adds. “This is based on the physics of rocks.”

Computing Power

“No one has ever seen the result of a hydrofracture,” he notes. While there are animated simulations that illustrate the technique and process, “The difference in physics-based software requires an extreme amount of computational power,” he says. Energy companies, no matter how large, are just not geared up for supercomputing. “It takes the kind of resources we have at the national laboratory to make the model.”

“We have the ninth-largest computer in the world” with which to develop the model, says Aines. “And now we have relationships with a number of oil and gas firms, to teach them how to use and how to run the model.”

As private industry learns how to use and manipulate the model, researchers expect them to supplement it with what they bring back from the field, making the model more accurate. “We’re providing a tool to develop knowledge of what happens in the subsurface,” says Aines.

More information builds more capacity and the ability to run more complex scenarios, say the researchers. “For example, people would like a fracture to look like a tree,” says Settgast. “But hydraulic fractures tend to be more horizontal.”

He says even when you know that, it is fascinating to see the scenario play out on the model. The simulator, developed from a finite element/finite difference model, shows how a fracture could affect seismicity and where an incident may occur. “We try to simulate the process to fill in the blanks,” he says. “Once you see how thing are interacting, then you can push that knowledge further and develop strategies to optimize work. But at some point, you have to go in the field and see if you can match the computer simulations.”

It turns out no one has done this before. It is a very hard thing to compute.

Roger Aines,
Lawrence Livermore National


March 2014

by John Kosowatz, Senior Editor,