Putting the 'al' in oil patch

Putting the 'al' in oil patch

University of Regina researcher Christine Chan creates the first-ever artificial intelligence systems to boost performance in the oil and gas industry
September 1, 2007
When baby boomers start retiring from the petroleum industry in the next few years, employers won’t just lose staff, they’ll also lose experience. Retirees leave with years of expertise and on-the-job knowledge not found in training manuals. Many worry about the issue, but Christine Chan, a forward-thinking software engineer at the University of Regina, has come up with a way to circumvent it.

Chan is one of the few people in the Canadian oil patch addressing the issue of knowledge sharing and retention. She combines historical data with expert knowledge to better predict oil well capacity and behaviour, tutor new staff, and get the most from employees.

“ We are seeing a loss of experience and we haven’t developed an adequate strategy for dealing with it,” explains Malcolm Wilson, Director of the Office of Energy and Environment at the university. “ We need more artificial intelligence to capture existing experience. This is where Christine’s work comes in.”

It hasn’t been an easy sell. Chan does much of the initial legwork to get the petroleum industry involved. “ I initiate on my side,” she explains. “ I maintain industry contacts and talk to them on a regular basis to understand what systems would be beneficial.” After only a few projects, the industry sees the benefits of her work. And with increasing oil prices and retirement rates, the petroleum industry realizes its need for artificial intelligence (AI).

The oil and gas industry collects barrels of information. Wells are outfitted with on-site sensors that constantly relay production data. Added to this are records of drill samples and drill-hole results dating back to the 1940s. In this overwhelming mass of numbers, useful information is hard to extract. That’s why Jon Hromek, Chief Operating Officer with Prairie Hunter Exploration Corporation, turns to Chan to make sense of his company’s oil-well data. They discuss what type of analysis is required, and Chan and her students come up with a system to provide it.

In one of Chan and Hromek’s projects, they studied rock porosity values at 32 oil wells to predict production. The results helped identify under-producing wells and guide future exploration. “ If we have an exploratory drill core and we know its porosity, we can come up with an expected production,” Hromek explains. “ And the only way we can get that is with the systems Chan develops.”

Chan takes things a step further by adding expert experience. “ We take their knowledge and apply it to the data,” she says. A veteran operator can look at the data from a malfunctioning well, and have a good idea what the problem is and how to fix it. That experiential information is extracted through interviews and input into the software, creating an AI system.

“ The system becomes a helper,” Chan explains. “ It aids in decision making. For a novice, it’s like a tutor—someone to suggest a good way of remedying a situation.” The software analyzes incoming data and generates suggestions. The new guy on the job now has a veteran at his side all the time. AI improves efficiency in maintenance too, by allowing routine work to be prioritized. The software tells employees which wells are most in need of maintenance and what to expect when they get there. And that’s just scratching the surface of what AI could do for the industry.

Chan has a multidisciplinary background involving training at Stanford University, University of British Columbia, and Simon Fraser University. It was at the last institute where she obtained her Ph.D that she worked with a noted AI scholar. “ I got more and more interested in it,” she says. “ AI is highly interdisciplinary, which is important for tackling major problems.”

By solving problems no one else had attempted in the petroleum industry, Chan has established herself as a leading AI specialist. This was especially obvious to Wilson at a recent AI conference, when he realized Chan and her grad students “ were doing better work than guys from a major U.S. university who were giving the talk.”

Hromek can attest to that. Before collaborating with Chan, he searched for applications of AI in the petroleum industry and couldn’t find a single instance. “ She’s making a lot of progress in how the data can be used,” he affirms.

In an industry that generates $75 billion each year, and with oil prices continuing to climb, Chan’s data analysis has become attractive and essential. “ Oil is getting harder to get out of the ground,” she says. “ What I do offers one way of finding new techniques, and it improves the efficiency and effectiveness of what is already being done.” As the rigs dig deeper for oil, Chan will do the same with AI.