What's new in exploration

Fig. 1. Exploration sunset: “Red in the morning, sailors take warning, red at night, sailors delight.” Photo by author.

In 1984, oil was at $28/bbl, and Marathon Oil V.P. for International Exploration Fred Spindle told me, “clearly,” that I was to only look for hydrocarbons offshore, or from a friendly shore, within reach of a market that could pay. More than once, he poked his cigar in my face to remind me that he knew everything I was doing, everywhere. With his photographic memory, and shall I say outward personality, I was a believer. When I asked about technology, he instructed me to make maps, stating that if he thought I was stupid, he would not have hired me. In no uncertain terms, my “job was to make him look good.” Let me explain, now that I am more mature. 

Is an exploration sunset near (Fig. 1)? Here are exploration’s greatest 2020 challenges, in order:

1. Market: Ships find places to sell product, not limited by existing trains, pipelines, politics, or mega capital cost. The Exxon Valdez lesson was learned, due to environmental interdictors preventing a safe pipeline. Oil companies became independent of their fleets, and the Jones Act, rewarding the expanding access to independent buyers. No sales? Reserves don’t matter.
2. Cost: Efficiency is always in the forefront, but only as a driver for quantity, not quality. Budgets cannot define quality, only availability. I have looked for oil in awful places at $4/bbl [1974] and still found a way to survive. No money, no fun.
3. 

   Fig. 2. Elephant photo at 70 ft, Kruger National Park, South Africa, 2007. Photo by author.
   Government: The “ELEPHANT”-controlling location, Fig. 2. Government regulation, ideology, taxes and objectives are often created out of sight, but always remain present. Policy today, when threatened, is immediately destructive to safe exploration. Impediment or enabler?
4. ROI: Stupid, we are, when claiming return is 7 or 10 to one, when ROI for ExxonMobil/Chevron hovers near 3% annually, and less than 3% on average for all businesses. Positive cash flow over any quarter is good in this industry. New money?
5. Personnel: Skilled explorationists evolving with technological improvements are in demand, and are only somewhat replaceable. While substitutes are sought with AI, its benefit is absorption and reformatting data into analysis, quickly, especially in replacing routine, manual tasks. Seeking machine autonomy is a distraction from physics advancements that lower exploration risk. Besides, computers will never replace all people, because management refuses to be replaced! 
6. Technology: More science is present in computer technology than in one person or 1,000 books. However, tech has limits. Example: PCs can only converge approximations of math integrals. Integrals look at what occurs from, say, minus infinity to infinity or an infinite number of possibilities within a finite range of a math relationship. For a machine to appear to “solve” the impossible, an approximation is created. A sample of finite calculations is conducted, converging [if-statements, delimiter loops, etc.] to a single number. Think of an area under a bell curve, all positive y axis. Divide that area into small rectangles at the correct heights, and calculate each area and sum. Make more rectangles, reducing the x width dimension, y centered on your curve. At some programmed incremental limit, you have an approximate answer. You can keep going to 30 decimal points, but why? Humans developed the science determining what to sample; humans engineered the sampling. 
7. Data analytics: Analytics can be used to predict a form of what might happen, if enough relevant data is available, best from multi-sources. Optimal use of AI, proper machine training, and maximizing output are claimed to be near, but are oh so far. Machines can be taught to make assumptions to assesses [converge or convolve] data, but they cannot calculate an infinite set of possibilities. All AI systems make assumptions of probable finite limits. Testing and learning autonomously proceed, only within parameters set by a human. We can never know what we do not know. Hybrid AI is the actual value of the tool, maybe indicating what we should know. 

The Leading Edge, 2017, March Vol. 36, #3. Special Section: Data analytics and machine learning; it’s a complex read, and is again in 2019 and 2020. World Oil and the SEG present authored seminars and webinars on this topic at all levels. Advancements persist, but we remain in the experiment-to-toy stage of most AI

About the Authors

William (Bill) Head

Contributing Editor

William (Bill) Head is a technologist with over 40 years of experience in U.S. and international exploration.