Data & Analytics

Data Initiative Aims To Get It All Out in the Open

Representatives from ConocoPhillips, Shell, Chevron, and BP came together onstage at the 2019 Professional Petroleum Data Expo in Houston to present the Open Subsurface Data Universe, “an open-source, data-driven, reference architecture for subsurface and well data in a cloud solution.”

osdu3.jpg

The largest oil companies in the world thrive off of competition with one another, and data are the fuel for this healthy competition. How they collect this data, however, is one area where the supermajors can work together.

urbano.jpg
Urbano

Representatives from ConocoPhillips, Shell, Chevron, and BP came together onstage at the 2019 Professional Petroleum Data Expo in Houston to present the Open Subsurface Data Universe (OSDU). The OSDU is, “an open-source, data-driven, reference architecture for subsurface and well data in a cloud solution,” said Paloma Urbano, information technology director, data analytics, at ConocoPhillips. The OSDU Forum, responsible for the creation and direction of the OSDU, is being managed and aided by the independent Open Group.

Certainly, the large corporations are capable of creating their own data architectures, but Phillip Jong, general manager at Shell Global Solutions, explained why his company chose to work with the Open Group on OSDU. “At Shell, at the highest level, we realized it’s the data that drives digitalization,” he said. “It’s not the other way around. … It’s not possible to deploy machine learning at scale without a data platform.”

“If we want to develop this data platform ourselves,” Jong said, “then we really need to ask ourselves whether that is the area where we really want to compete with ConocoPhillips, BP, and Chevron. We thought about that, and we came to the realization we do not want to compete with BP on how to access data. Nor do we want to compete with Chevron or ConocoPhillips. We’d rather compete with them on how to use the data to derive decisions. That’s the area we’d like to compete.”

jong.jpg
Jong

So far, 15 operators—Anadarko, BP, Chevron, ConocoPhillips, Devon Energy, Equinor, ExxonMobil, Hess, Marathon Oil, Pandion Energy, Petrobras, Reliance Industries, Shell, Total, and Woodside Energy—and 27 suppliers, along with the University of Oslo, have joined the OSDU Forum. “We are all part owners of that architecture as we are all contributors,” Urbano said.

“I have to say,” Urbano added, “I was very skeptical in the beginning when we joined the OSDU. This was going to be driven by the members and for the members … kind of like a crowd-sourcing type of approach. And I thought, well, how is that going to work? And to my surprise, it’s working extremely well.”

The concept for OSDU began in spring 2018, and a limited-scope demonstration is expected to be released this summer. “From the beginning, the forum founding members decided that we want to be agile,” Jong said. “The first demo release currently is planned for June 6. That is a P50 estimation, so it could be earlier, it could be later.”

The expected demonstration release will be limited, including only four data types from two data sets. The four types of data to be in the release are well logs, well trajectories, well markers, and documents, Jong said. These data will come from data released by the Netherlands and Equinor’s Volvedata set, which the company released last year. The Dutch data set has information on approximately 6,000 wells, while Volve has information on approximately 60 wells. While this may not sound like much, the plan is to build the OSDU architecture such that it can scale to handle data from millions of wells and many more types of data. The architecture is designed to handle legacy data, as well, allowing for data searching and extraction to facilitate machine learning.

“We want to produce a very limited-scope capability to get it into the hands of the members so that we can work together to define the next iterations,” Jong said, adding that a future iteration will include seismic data. The types of data used for the first iteration were chosen for simplicity, to keep the process agile, he said.

Once the demonstration is released, all members will have access to it. “You will be able to look at the public-domain Dutch data set and also Equinor’s Volve data set but only in the context of those four data types,” Jong said.

While the initial release will be small and limited in scope, the architecture it presents will be the foundation for an enormous undertaking. “The whole idea [of the OSDU Forum] was to define that reference architecture,” Urbano said.

Find more about the OSDU here.