Computational Horsepower: Kyle Spikes Harnesses High-End Computing For Exploration Geophysics
October 21, 2010
When it comes to heavy-duty computer modeling, there are the Armed Forces, the space program, and then rock physics experts, such as Kyle Spikes.
Spikes, an assistant professor at the Jackson School of Geosciences, studies rock physics — the exploration of the physical behavior and properties of rocks — using computer resources to analyze seismic data and lab samples. He completed his Ph.D. at Stanford University, which has been at the forefront of the field. After a brief postdoctoral appointment at a university in Norway, he arrived at the Jackson School in 2009, where he uses his background in rock physics to concentrate on exploration geophysics, the study of rock properties beneath the surface to detect conventional and unconventional energy reserves, ore minerals, and other geologic resources.
“My involvement focuses around making computer models of how fractured rocks behave when seismic waves go by and then comparing theoretical predictions to what we see in the real seismic data,” Spikes says. “Really what we want to know is how much space is there for fluids to sit in these rocks, what are the rocks made out of, how are they oriented, are they layered, are they homogeneous.”
“Teasing out those kinds of properties from seismic data is the world of what we call rock physics,” Spikes adds, “and that’s where my specialty really lies.”
That specialty places Spikes in front of computers most of the time, and his work includes developing computer codes to numerically simulate the seismic data and the behavior of the subsurface rock.
Endeavors in exploration geophysics date back decades, but high-power computation has allowed for major advances in recent years. Today, industry scientists, government researchers, and academics rely heavily on complex models and powerful computers needed to quantitatively analyze and integrate very large datasets.
“Technologically, to put it in perspective, the only people who use more computer resources now are the Department of Defense and NASA,” Spikes says. “We’ve come a long way technologically in 80, 90, 100 years, but to mimic the physics we have to make a lot of assumptions because we simply don’t know what’s going on at every point in the subsurface. We understand roughly what goes on down there, but it’s never a perfect picture.”
The challenges are also heightened because exploration geophysics has historically concentrated on oil and gas reserves, but new applied and academic inquiries have branched out to study unconventional resources, such as gas shales, and other pursuits, such as underground carbon sequestration and storage.
“The trick is that we have to venture a little bit away from techniques that people have used in the last 20 to 30 years, where we just take the images from the seismic data and we have some well known models that explain the physics of the rocks,” Spikes says. “Now, we’re having to make new ones, and having to look for different patterns in the seismic data.”
In the case of gas shales, it’s very difficult to identify promising locations, and fractures that show up may still be unsuitable to developing a well because of the presence of water or other factors. With underground carbon storage, computer models have to account for where and how fluids move in the subsurface in the presence of carbon dioxide. Currently, government regulations demand very precise measurements, which means Spikes and other scientists must constantly improve their techniques and instrumentation.
“They will no doubt evolve more rapidly in the next ten years than they have in the past, simply out of necessity,” Spikes says.
Spikes is looking forward to the coming decades and his contributions to computational modeling and the developments in rock physics and exploration geophysics.
“It’s a long-term goal to keep refining these models of the subsurface over and over until we are convinced that they’re correct,” Spikes says. “By the time that I think I’ll be done with all this — which hopefully is a long time — then we should probably be to a point where we’re satisfied with the accuracy of the models. I think within my career that should probably happen.”
For more information about the Jackson School contact J.B. Bird at email@example.com, 512-232-9623.