PEGrid gets down to business
Although Petroleum Engineering Grid officially started up in late 2009, it is already a classic example of how research projects can have unexpected benefits.
PEGrid came into existence using techniques and tools created by TIGRE (Texas Internet Grid for Research and Education), a state-funded project based on the use of federally funded middleware such as Open Science Grid's Virtual Data Toolkit.
When TIGRE ended in 2007, it left behind a grid software stack, three completed demonstration projects, and budding relationships with contacts in the air quality, biosciences, and petroleum engineering industries.
"What we did is to go to industry and ask what they needed," said Alan Sill, a Texas Tech University researcher who led the TIGRE development team. Because the petroleum industry is well-funded, they decided to take their next step by creating PEGrid.
"What we're trying to do with PEGrid is, using the technology that we have, to allow the best practices in industry and academics to be shared," Sill said. "What we've found is that this isn't being done."
The most basic problem they uncovered is that the software that petroleum engineers use is extremely expensive and requires significant computational resources. Schools cannot afford to purchase licenses for recent versions of this expensive software, and so students are learning their trade using programs that are more than a decade out of date.
The companies that will eventually hire these students are also losing out. Their entry-level hires can't just hit the ground running; they need to be trained on industry-standard software first.
In response to those needs, Texas Tech researchers Ravi Vadapalli, Sill, Shameem Siddiqui, and Lloyd Heinze co-founded PEGrid in collaboration with three other Texas universities, four software companies, and four oil companies. Together, the industry partners have donated $45 million in software donations and private three-year grants (the latter in the form of student internships).
PEGrid came up with an elegant way of ensuring that the donated software licenses benefit as many students and researchers as possible. "You're probably not going to use the licenses all the time," Vadapalli pointed out. "Why not just grid-enable these applications?"
By sharing licenses between students, faculty, and institutions, PEGrid has enough licenses to meet the educational needs of students while providing faculty access to cutting-edge software for their research. And by working closely with industry, the PEGrid team has learned more about why their partners have not implemented any grid technologies.
In academia, budgetary constraints make it difficult to acquire enough computational resources. But in cash-rich industries, that's not the problem at all.
"The number of processors is never a limitation," Sill said. "What stops most people is they are not familiar with techniques for scientific computing."
In the petroleum industry in particular, they found that data security was a major concern that was stopping companies from exploring grid technologies.
"The whole idea of running their data on multiple sites is troubling to them," Sill said. "But we discovered quite quickly that the reason it is so troubling to them is that they are one to two generations behind in some areas of security, compared to the best practices in grid computing."
Thanks to this revelation, entirely new horizons are opening up for petroleum companies. "They do run very big models. Once they get comfortable with these techniques, we expect them to start more widely implementing grid techniques," Sill said.
In the meantime, as PEGrid gets up and running, students are learning to use the new software. Eventually, they will start running simulations and analyses that will place demands on PEGrid's large-scale clusters. Researchers will continue to learn more about oil reservoirs using these new resources. And industry partners will continue to learn more about cutting-edge grid technology.
What next? The people behind PEGrid are already pursuing opportunities in other fields. Vadapalli, for example, is interested in applying PEGrid's tools to other areas of energy research. There have also been conversations following up on TIGRE's initial demonstration projects.
"We've actually had a couple years of fairly deep discussions with people involved in cancer radiation therapy," Sill said. "The dosimetry calculations for radiation treatment with proton beams, for example, are very large-scale problems."