The oil and gas industry is seeing incremental gains in the first phase of digital technology adoption, but the biggest disruption is yet to come. That change will be people-driven and will require collaboration, panelists said during a May 2 session at the Offshore Technology Conferencein Houston.
The industry has looked to Big Data as one way to cut costs in recent years, and data analytics for such use will continue. Year-over-year, said Preston Cody, head of analytics solutions at Wood Mackenzie, the firm has tracked “some pretty dramatic cost savings,” made by industry. While a portion is attributed to high-grading and other factors, digital technology has played a role in doing more with less. Working with an expected structural cost savings of 10% from digital technology adoption, Cody said, the industry could see a savings of $150 billion/year in value delivered.
That number could rise, but it’s going to take organizational change.
“It’s people that drive the machine,” said Francesco Menapace, an exploration geologist at Shell. We need to find a way to collaborate to tackle integration, he said. On an industry level, where is the line at which companies need not compete so we don’t “solve the same problem 20 times individually?” To win as an industry, we need a stronger ecosystem, not islands, he said.
On an individual company level, companies need to embrace a system of reverse mentoring. Within every organization exists several generations containing a wealth of industry experience and natively high technological literacy. Companies must find a way to cross the generational divide. Knowledge from both sides must be transferred.
Simon Sheather, academic director of the MS (Analytics) Program at Texas A&M University and interim director for the Texas A&M Institute for Data Science, noted that it’s not good enough to be a data expert—one must also be a subject-matter expert. Oil and gas companies are in the data business as much as they are in the hydrocarbon business. Machine learning implicitly makes a series of assumptions about the data and the possible set of models, he said. “Machines are only as smart as the instructions given.”
Contact Mikaila Adams at [email protected].
