Introduction
Did you know that the oil and gas industry is currently only using close to 1% of the data it generates? A mind-boggling figure, and not one we usually think about when talking about artificial intelligence and machine learning applications.
In episode #13 of the DataHack Radio podcast, we are joined by Yogendra Narayan Pandey, Ph.D, as he takes us on a knowledge-rich journey in the world of oil and gas.
This is not a field that grabs a lot of headlines in the AI and ML community, but as you’ll find out in this podcast, the potential applications are vast. The amount of data collected in a typical oil and gas exploration and production process is staggeringly high, and that in turn spawns multiple use cases where machine learning techniques like regression, clustering and neural networks can be applied.
Yogendra has done a phenomenal job of condensing the end-to-end oil and gas life-cycle into byte-sized knowledge for you and me to capture. It’s well worth spending your time listening to this podcast and broadening your horizons. Happy listening!
Subscribe to DataHack Radio today and listen to this, as well as all previous episodes, on any of the below platforms:
Yogendra Pandey’s Background
Yogendra is the Founder and Managing Consultant at PRABUDDHA, an organization that provides AI solutions for the oil and energy industry. He is a chemical engineer from IIT-Varanasi and successfully completed his Ph.D. from the University of Houston in the same field (his dissertation topic was “A Simulation Approach to Thermodynamics in Interfacial Phenomena”).
In his professional career, Yogendra has worked for organizations like Halliburton, Innowatts, and W.D. Von Gonten Laboratories. His role in all these organizations has been in the capacity of a data scientist. His passion for the oil and gas industry has driven him to pursue and make a mark in this field.
In the initial part of the podcast, Yogendra has described his work in this fascinating space following his Ph.D. Anyone with an interest in data science and the energy field will really appreciate this episode!
Applications of AI and ML in the Oil and Gas Industry
Oil and gas is a high-risk industry, so this makes the validation phase longer than usual. Decision makers have to be far more cautious, and this is one of the primary reasons why AI and ML have seen a slow adoption rate in this domain. But as Yogendra mentioned, this scenario is starting to change as technological advancements gather pace.
One of the most important use cases of AI in oil and gas are predictive maintenance and equipment failure analytics. Another application is around autonomous drilling rigs, which means designing an end-to-end fully automatic drilling system. This system is smart enough to understand where to drill (optimal well path), how to drill, and the optimal duration required to finish the job. Like most AI applications these days, these autonomous rigs aim to augment the manual effort workers put in, rather than replace their jobs.
The Oil and Gas Pipeline, and the role of AI and ML in this Industry
To give you a very high-level overview, we can broadly divide the end-to-end oil and gas life-cycle (starting from a drop of oil found thousands of feet beneath the surface) into three major segments:
- Upstream
- Midstream
- Downstream
For drilling operations, a large setup offshore can generate up to 1-2 terabytes of data everyday. The same goes for a large downstream refinery – it can generate up to 1 terabyte of data per day. So if you were wondering where and how much data this industry can come up, this is a pretty good place to start!
Each segment mentioned above has been explained eloquently and in detail with multiple examples by Yogendra in the podcast and trust me, the entire process is incredibly enthralling. My favorite part was about how a model can tell you whether a certain region has oil in it or not with a remarkably high accuracy rate (a probabilistic model). This helps the organization(s) decide whether it’s worth drilling in that region. Unsupervised learning techniques like clustering are heavily leveraged in this process.
Other algorithms used by data scientists in this domain for forecasting include regression, Hidden Markov Models for time series, Recurrent Neural Networks, Gated Recurrent Units (GRUs), and Long Short Term Memory (LSTMs), among others.
A Couple of Resources for Keeping up with the Latest in Oil and Gas
- Journal of Petroleum Technology: They cover the latest happenings in the oil and gas industry. They have covered in detail the digital transformations in this domain
- Companies like McKinsey and Accenture (and other forums) publish high-level and authoritative reports that tell us the current state of digital transformation in this space
End Notes
Before hearing this podcast, I honestly had a very vague idea of the oil and gas industry, and how AI and ML are being used to transform traditional processes and drive revenue. It took just 50-odd minutes for me to get a good idea of the entire oil and gas pipeline, from identifying the drilling regions and starting the drilling process to getting the oil to petrol pumps, etc.
I was fully immersed in this podcast and I’m sure you will be as well. Make sure you listen to this and let us know your thoughts on how else AI and ML can drive change in the oil and gas industry.
Senior Editor at Analytics Vidhya.Data visualization practitioner who loves reading and delving deeper into the data science and machine learning arts. Always looking for new ways to improve processes using ML and AI.
Its really very informative information, Could you please post any case study based on real life situation in oil and gas industry, how can we use AI and ML?