Part 1: Musings from a Data Science Convention
By: Matt Bell
Data, Data Everywhere
I recently had the pleasure of attending a data science convention organized by the Society of Petroleum Engineers’ Gulf Coast Section. Over 400 people filled the room with chatter about artificial intelligence, machine learning, and the promise both technologies hold at the big picture level. I was interested to hear presentations by data science leaders from several mega-corporations and listen to examples of how these technologies, proven so valuable in other industries, are finding real applications in the upstream oil and gas sector.
My business, Premier Oilfield Group, is not a data science company per se, but we are active on the front-end of the data science machine – generating, aggregating, and sharing data from rock and fluid samples. We fervently believe that the data-driven oil and gas industry is here.
The industry is making decisions in a fundamentally different way now compared to how those of us with 20+ years’ experience remember earlier in our careers. As Detlef Hohl, Chief Scientist for Computation and Data Science at Shell, described it when driving home the importance of data-driven approaches: “we desire to replace experimentation with computation.”
We take a deep interest in where the data we generate and share will be used, and how the end user will extract value from them. Understanding these dimensions helps us to produce relevant data and make it accessible in appropriate formats via the right platforms.
Myths and Realities
If our industry is to effectively capitalize on data science, the first challenge we must overcome is to understand what’s actually possible. Like the sea of hucksters pedaling tours and souvenirs outside a major tourist attraction, the hype and hyperbole arriving from data science providers can overwhelm.
Rob Kelly, Head of Upstream Digital at BP, talked about digitalization offering “$1.6 trillion of potential value” to the industry and the promise of eliminating “2.3 million tons of CO2 emissions.”
Data Science Manager, Sarah Karthigan, from ExxonMobil acknowledged that digital technologies are “redefining boundaries in value creation,” but sounded a cautionary note, stressing the importance of understanding “how to tell myth from reality.” I think many of us are indeed struggling to see the real trees amid the mythical forest.
Perhaps the most telling comment, which several speakers later echoed, came from Mojtaba Shahri, Senior Scientist for Drilling Analytics at Apache, who observed that “85% of our challenge lies in project definition and obtaining the relevant input data,” while the remaining 15% that actually relies on artificial intelligence or machine learning “is getting far too much of the focus.” Subsurface data, it seems, is not as readily “scienced” as we might like to think.
Want to keep reading? This is the first in a three-part series. To read part two, click here.