Integrative analytics which incorporate Big Data, Artificial Intelligence, and Deep Learning, are increasingly important to geoscientists and engineers. Welcome to an interview with Patrick Ng, who heads AAPG’s Analytics Technical Interest Group (TIG).
What is your name and your experience in geology / analytics?
My name is Patrick Ng and my experience ranges from A to Z. 1) AVO diagnostics, azimuthal features extraction, to 2) inversion for reservoir parameters, and 3) prestack depth imaging / model building to map subsalt reservoirs, 4) time-lapsed 4D monitoring of oil movement within a field, and 5) drilling wells learning from the drill bit all the way to total depth (Z).
What do analytics mean to you as a geoscientist?
Attributes come to mind, more like what measurable features we choose to describe what we see and make decisions on. Not all have to be numbers, sometimes we want binary decisions like Yes / No (e.g., refrac hunting); other situations, prefer graded scale A, B, C, D, E for opportunity screening.
A hybrid approach to well economics. Oil and Gas Financial Journal, July 2016.
What are some of the analytics that you see being used on a daily basis?
Examples include 1) AVO - gradient, intercept, and difference between near and far offsets. The really basic stuff. 2) subsalt - something like changes in velocity model from the output of tomography, to maybe more sophisticated (and time consuming) illumination quality of seismic amplitudes. More recently, 3) focus on return and risk, as well as payback time and production volume.
In terms of new technologies, what are 2 or 3 of your favorite tools?
1) Python - a programming language helps bridge the gap between fundamental analysis and machine learning (ML), 2) Alexa - voice computing as the natural front end to AI, 3) Microsoft Azure ML Studio - nice visual UI and really good for facilitating collaboration on workflow without getting deep into coding.
Operating profitably with $50 oil, Oil and Gas Financial Journal, December 2016.
What would you like to see for the future?
Be aspirational with tying everything together. 1) First principle - develop and get better understanding the risk in the business, 2) Leverage tech - one platform to enable geoscientists, engineers and financiers quickly test assumptions and get answers to what-ifs, better understand / quantify the impact and benefit from decisions they make. So within a common framework, we can collaborate and make prudent investment decisions in the drilling cycle. 3) Transfer learning - here I’d draw a parallel between medicine and what we do. Earth, like us, has memory (just orders of magnitude difference in time frame). Yes the diagnostics using the right measurements (analytics) and the protocol to eliminate possibilities, leading to better and more cost-effective treatment someday. Both industries face hard problems and daunting challenges, so we may arrive at practical solution faster by taking lessons learned from medicine.
Are you working on any new technologies or applications?
1) Tech - experiment with voice computing front end to AI in oil & gas, e.g., Alexa Well Skills is first take. 2) Analytics - how we make decisions on (risk, return) and translate / triangulate ML output, first principle (say G&G) and financial constraints (capital) into actionable wisdom (that of Warren Buffett).
Recommendations for reading:
1) First principle “The Business of Petroleum Exploration” AAPG treatise on petroleum geology, edited by Richard Steinmetz, my handy go-to reference, 2) Book “The Master Algorithm" by Pedro Domingos, Professor of computer science at the University of Washington, Just came out in paperback. Make ML readable in plain English, not infected by jargons. For those who wish to explore further, 3) Online course "Machine Learning" by Andrew Ng, Co-Founder of Coursera and Adjunct Professor at Stanford University. Get a good feel for what ML can do, and how we go about applying AI, together with first principle in our own work.