This is the Brief story of Data Science in Energy

A Brief History of Time: Data Science in Energy Edition

Stephen Hawking passed away on March 14th, 2018; it was a sad day in the world but manly in science circles. Although there is a massive array of accomplishments, too many to name, Mr. Hawking is very well known for his book “A Brief History of Time”; one of the most influential books of our time. In his book he was able to explain things in a way that we could all understand.

In the spirit of this book and Mr. Hawking, I will attempt to describe the history of Data Science in Energy from the time I got in the field in 2011 to now; further, what the future holds. Through my early adoption of Data Science in Energy and my affiliation with Houston Energy Data Science, I have been lucky enough to take the first seat on the first row of the evolution of Data Science in the Energy Industry and the impact it has made in the past few years. I got to meet some of the most interesting people and the brightest minds in Data that live in Houston. My journey is still ongoing and growing, but I wanted to share just a few of the stories and events that have shaped Data Science in Energy.

Back in the day

Back in winter of 2011, when I first heard about Data Science, I wondered what everyone in the Energy Industry was doing in Data Science: to my dismay, there was no one I found with this title working for Energy Companies; based on LinkedIn titles. Oh, I am sure folks will say that they were doing algorithms, regression, etc. at that particular time, I do not dispute that, but they were not actually doing data science. I am talking about real data science; the actual process of gathering empirical evidence, transforming data, documenting these transformations, testing several algorithms and models, picking the best model based on cross referencing, and then deploying the model based on both empirical evidence and business decision. In this darkness, there was one shining light; Judson Jacobs 2010 presentation at CERAweek, where he described that the use of analytics could help improve efficiencies in operations of Oil and Gas assets. Mr. Judson outline how other industries, such as railroads had adopted predictive analytics for preventive maintenance and thought that the Energy industry should adopt such techniques to prevent failures on the oil patch.

In 2012…Not much had changed.

Kaggle gets in the mix

It was 2013 and I get an unexpected email from someone in my LinkedIn Group (analytics group) named Anthony Goldstein. Anthony was the CEO of a new startup for Machine Learning competitions named Kaggle (In later years, it will be bought by Google). He wanted to see if I wanted to meet for dinner while he was in town to meet potential clients to talk about Data Science in Energy. We met at a nice Korean place on the north side in the Greenspoint area. He was a tall skinny gentleman with a thick Australian accent which at times made him difficult to understand (I am sure it was vis a versa for him). We shared one common theme; disrupting the Energy Industry with data science. His company was already doing this; he had procured contracts with oil and gas companies. I was pleasantly surprised at the results; they were able to predict more accurate Expected Ultimate Recovery numbers and Production numbers by using a technique called Ensemble Methods. They were so successful at prediction that they were contemplating building a platform to be used in future consulting gigs. Kaggle was the first data science consulting company to ever concentrate their efforts in the energy industry and successfully deploy predictive analytics models. The good times did not last long; as we all know the market tanked and so did the contracts for Kaggle. So Kaggle decided to leave oil and gas and concentrated competitions but not before making a huge impact on data science in energy.

TERR and the start of HEDS

In the summer on 2014 I got an invitation by TIBCO to attend a Meetup event they were sponsoring at Maggiano’s. Their Chief Data Scientists, Michael O’Connell, was going to present a new product to their already successful Data Viz software Spotfire called TERR. It was free food and wine so why not! I was also interested in TERR, because it was the first time I had seen R being written in a Data Viz product and regression code run outside of R Studio. The Meetup was run by the R users group and there were about 100 people there. It was the first time I had seen so many data people together in Houston; granted there were very few data scientists in the group. This Meetup gave me the idea of opening a Meetup of my own that would bond data science/ data people together with Energy professionals, thus Houston Energy Data Science (HEDS) was started.

In November of 2014, HEDS held its first Meetup with a whopping 10 people at the Mays School of Business at City Center. The Texas A&M MS in Analytics program got wind of my Meetup and wanted me to host my Meetups at their location, at the time the MS in Analytics was the only program in Houston that was offering Data Science training and it was at its infancy. I was more than happy to promote their program to our members. Currently, HEDS brings in between 60 to 120+ people at our Meetups and has been sponsored by companies like IBM, Microsoft, SAS, TIBCO, QLIK, NVIDIA, Hortonworks, to name a few. HEDS has helped Energy Industry professionals understand the potential of Data Science while it has introduced many Data Scientists to the exciting world of the Energy Industry.

Early Data Science teams take form

By 2014, a few companies in Energy had taken the plunge into Data Science and they were starting to form what we would call, Data Science departments. The most notable ones were Chevron, ConocoPhillips, and Devon. In services, Schlumberger took the lead when it started to hire Data Scientists in late 2014 to early 2015. Hands down, the leader of the pack was Devon. Devon had bought all the bells and whistles that SAS had to offer and built a strong analytics department, almost all organically, with very strong training and full support by upper management. Some of the items they were starting to predict were remarkable, but their most remarkable achievement in the early days was the integration of all their data silos into one. The connections made, and insight found, just by putting several types of data systems together made all the difference in creating predictive models. Devon is still a leader in the use of Data Science but now the Majors such as ExxonMobil, Shell, Chevron, etc. all have Data Science initiatives that are being pushed through.

The Future is Here!

In the future, Data Scientist in Energy will need to learn how to create products with data, such as Artificial Intelligence models that learn through data and Virtual Reality environments, while also learning the new way to transfer data in the Blockchain.

Deep Learning and AI

Nothing garners more buzz these days than artificial intelligence (AI); but what is it? AI does not stand for Allen Iverson, the great Hall of Fame basketball player, but rather for something else that is also great; computers doing things that were thought to be only possible by humans. For example, reading books and making sense of sentences (Natural Language Processing), looking at pictures and its pixels; recognizing patters and being able to identify objects, voice recognition and being able to execute verbal commands, and making sense of seismic data better than a human. Yes, the last statement is where the energy industry is headed, but before we go into AI we must first be able to have the proper hardware to process massive amounts of data. If you are an energy company and you are excited about AI reading seismic and letting you know where the sweet spots are then you must be prepared to make in investment in some serious hardware like GPU’s

Blockchain

The Blockchain will change the future as we know it: but what is it? The Blockchain is the distributed database that is behind cryptocurrencies like Bitcoin and can now be used to transfer other valuable things like contracts, information, code, money, etc. in a secure way, creating a trustless system. In English, because the transactions in a blockchain are encrypted, the ledger can be stored on any computer and updated every few minutes to create a block of transactions, this block is then tied (chained) to the previous block, which is tied to the block before and so on. All the computers holding the blockchain must agree (in code) who will win and be the builder of the next block, which is copied to everyone’s computer (mining), the winner will be rewarded. With the Ethereum blockchain, you can now build your own “smart contract” which is an application on top of Ethereum and do not have to worry about mining or building a blockchain infrastructure. Smart Contracts will allow you to build encrypted transactions and use Ethereum as you backend blockchain for these transactions. Hyperledgers work similarly but are owned by companies like IBM and Microsoft. In Energy, you are starting to see Blockchain used in our supply chain where a procured product can be tracked in a secured way through its end destination. The biggest impact will be in trading commodities. Shell and BP have started a project to build a blockchain just for the purpose to a trustless system to build a trading platform {1}.

Virtual Reality

Not a lot is known about Virtual Reality (VR) because it is such a new space, but it is about to hit everyone in the face like a northern wind in the middle of January: why? New VR environments are being created to simulate actual scenarios that are faced in the field, including oil and gas operations in both offshore and onshore. Imagine training people not on the field but in a simulated space where they can face several scenarios and are able to train on those scenarios and get immediate feedback. Training on the field will improve tremendously and training time will be reduced with simulated environments. Another area that has been very promising is the simulation of an environment when you actually get a description of sockets, knobs, meters, hookups, etc. so instead of guessing what anything is or does, you can refer to the simulated environment to give you a description; improved safety and reduction of errors is the target of this VR environment.

By no means is this a full taxonomy of Data Science in Energy; there are a lot more data stories to be told and more events that shaped this field in the past years. My hope was to briefly present some very important events that changed Data Science in Energy and the future technologies that will continue to lead change in Energy.

1)”BP, Shell lead plan for blockchain-based platform for energy trading” https://www.reuters.com/article/us-energy-blockchain/bp-shell-lead-plan-for-blockchain-based-platform-for-energy-trading-idUSKBN1D612I