Big Data Drives Oil Exploration

Project Kaleidoscope (Image credit: Repsol)

While depressed oil prices have sharply reduced the oil industry’s investment in exploration, Repsol aims to make every well count thanks to its decade-old head start in innovating exploration technology. According to Francisco Ortigosa, Repsol’s Director of Geophysics:

“The business of exploration is the business of risk, and we want the best possibility of success.”

“We had a long tradition of using technology to improve competitiveness, but in the early 2000s, our president and board decided to make a long-term commitment to differentiate on exploration,” added Fernando Temprano, Repsol’s Research Director.

One of the first places Repsol looked were the Gulf of Mexico and the coast of Brazil, which were believed to possess significant deposits under the ocean surface but, literally, were difficult to “see.” Developed in conjunction with the IBM computing center in Barcelona, it built a system that correlates seismic imagery from proven sites with potential locations, using proprietary algorithms that crunches data 50 times faster than previously possible.

Called Kaleidoscope, it enabled Repsol to find deposits 15 times faster than the industry average, and contributed directly to its discovery of Seat, a well located in block BM-C-33, one the more prolific areas of hydrocarbons in the Brazilian pre-salt. The system has been constantly improved and evolved since 2007, using more computing power and looking at more data points, including sound waves.

“Our goal is to have 100% certainty in our subsurface models,” said Santiago Quesada, Repsol’s Exploration and Production Technology Director.

Sherlock followed in 2009, using a deep understanding of the geology surrounding existing deposits as a model for locating new ones, thereby increasing success rates up to 38% while reducing the risk of its activities. Excalibur, a predictive math tool to increase success rates, came in early 2014, and then HEADS (a real-time analytic tool to assess safety) and Pegasus (the first global cognitive tech program for the oil industry, also in collaboration with IBM).

Its latest innovation, called BOLT, was announced a few months ago, is the result of a three-year cooperation with RSI (Rock Solid Images), and could disrupt the way oil is located by combining seismic imagery with electromagnetic data.

“The future of exploration is integration of more and more data,” explained Ortigosa. “Now we’re adding electromagnetics to the program, and next could be gravity or integrating macro data from satellites.”

“We’re constantly innovating new and better ways to see what our competitors can’t.”

The ways Repsol has pursued these ideas is perhaps as innovative as the projects themselves, considering it allows the company to maintain its investment in R&D while meeting its profit and other performance requirements.

“It’s crucial that we combine our interest in leading-edge tech with a focus on early deployment, so every idea gets analyzed and developed in a cost/benefit model that must yield short, mid, and long-term profits,” explained Temprano. “Many of those ideas come from our own people, who we encourage to contribute and get involved, both in ideation, and also directly in what we call ‘key tech’ projects.”

“Our goal is to build a better company while building great projects.”

“We also utilize partners to bring additional skills to the table, which gives us flexibility to act on our vision,” said Quesada, with Temprano noting that the company has over 110 such collaborative agreements and, while it once brought ideas to potential partners, now it receives regular inquiries from interested parties.

“Exploring the earth is like looking at the night time sky,” Ortigosa. “It’s the same for everybody, but if you use binoculars, you see more. We’ve built something like the Hubble Telescope, and now we’re improving it.”

Can Repsol’s innovation tell us when oil prices will rise again, or if its competitors will commit more resources to discovery technology?

“No model in 2004 suggested today’s prices, and I can’t tell what they’ll be a few years from now,” Quesada said. “But how we search and develop are variables we can control. They’re the future we don’t have to predict, because we’re inventing it.”

Then he admitted that the company is exploring AI with enough intelligence for model future market conditions, too.