2019 Machine Learning conference: Tapping into oil and gas data
HOUSTON -- The 4th Annual Machine Learning in Oil and Gas conference kicked off Wednesday, April 17, in Houston. The first day focused on the different ways that IT providers and various service companies are collaborating to implement software solutions, to extract value from the vast amounts of untapped data that the industry has accumulated. The show had a decidedly different mix of vendors and clients than seen at traditional upstream events. The digital transformation initiative has caused principal actors to hire and cross-train more data scientists, programmers and systems engineers, then apply their unique skill-sets to solve a multitude of new challenges.
The technical presentations started with a keynote address by Thor Schueler, innovation lead, Avanade. Mr. Schueler claims the focused effort on digital transformation has launched a fourth industrial revolution, which he labeled “an era of intelligence,” where enterprises need to transform to take advantage of massive advancements in technology, including: 1) cyber physical systems; 2) internet of things; 3) networks; 4) intelligent agents; and 5) machine learning (ML) and artificial intelligence (AI).
Schueler then used an interesting quote by Arthur Samuel, published in 1959, for time reference and perspective: “machine learning gives computers the ability to learn without being explicitly programmed.” Going forward, the combination of ML/AI technologies should think and reason like people, to determine the meaning of data and form value-driven conclusions. To accomplish these goals, Accenture and Microsoft have partnered to form the JV, Avanade. The new venture will strive to expand on the latest advancements in machine learning and artificial intelligence to enable oil and gas companies to increase efficiencies, reduce costs and improve safety. The company also plans to investigate robotic process automation, intelligent agents and how to use AI in engineering and HSE. “By the end of 2019, digital transformation spending will reach $1.7 trillion worldwide,” Schueler concluded.
Another presentation, by Yang Cong, principal data scientist, Wood Mackenzie, demonstrated how utilizing ML on well performance can identify sweet spots in shale plays to predict how much frac sand/fluid is needed to obtain the best production performance. The service, called GeoFactor, was presented in a Bakken case study that mapped individual well production, which was overlain with a similar map of frac intensity and formation brittleness (the force required to break the rock). The service requires multiple operators to contribute data in overlapping areas for best results.
