5 Takeaways from the Big Data, IoT, & Machine Learning in Oil & Gas Conference

March 2, 2017 David Poole
  1. Exciting changes are still on the horizon, but tangible, pilot case studies are now taking hold, solving real business problems

Even a year ago, tangible IoT, Big Data and Machine Learning projects were few and far between. Now, the desire is there to support the widespread use of these technologies. For North American Oil & Gas companies to be competitive at the current oil price, they need to think creatively by using technology to drive down the cost of production and improve operational efficiency. Edge computing provides an affordable solution to allow distributed assets to act on sensor data in a timelier manner.

Forward-thinking companies such as Beaver Drilling, Halliburton and Devon Energy are already seeing early successes from their IoT projects on the production side of the business.

Those who have succeeded have done so with small pilot projects that address real business problems that they can later scale. Many of these projects improve worker safety as a byproduct. Remote monitoring means fewer workers have to go into the field in hazardous conditions.

  1. Big picture thinking will be required to get the tech right

Getting IoT, machine learning, and analytics in EHS right will mean making some big decisions early on. It is one thing to find a scalable project, but it is another to find a scalable technology stack that will work for you. Stuart Payne of Gibson Energy gave a fantastic presentation of how they had solved this.  There are all kinds of considerations to be made, including your existing technologies, security, your most pressing business problems, and your budget.

Many companies are going their own way and employing a siloed approach to IoT. While this may solve some problems – for example, getting something done quickly without going through the bureaucracy of approvals – it is not a recipe for long-term success. There is still no gold standard that will be used for IoT, but a consensus is emerging around SensorThings API – an open standard that builds on web protocols and the OGC Sensor Web Enablement Standards. It also applies an easy-to-use REST-like style according to Dr. Steve Liang at University of Calgary. Take time now to ensure you pick a tech stack that will be future-proof as well as scalable.

  1. Subject Matter Expertise will not be replaced by machines any time soon

There was a telling quote at the beginning of this conference that quickly poured water on some of the claims that integrating with Watson (or any other out-of-the-box ML provider) is not a human vs machine scenario:

“We teach Watson to think like an engineer and Watson teaches us to think like a thousand engineers,” - Mike Biddle, Evok Innovations

Subject matter expertise will not be replaced any time soon, but ML provides us with a new set of complementary possibilities. Like so many other things, these technologies are a tool to validate our professional judgment as EHS professionals rather than replacing it altogether.

  1. It’s unacceptable not to think about your dark data

We are now living in an age where most things can be tracked. It’s just a case of finding the business problem you want to solve and the right mix of technology to do so.

Dark data is the data we could be tracking but we aren’t. Spend some time thinking about it as an EHS department. Imagine what you would track in an ideal world and why. Chances are you’ll be able to find a way to do it if the business case makes sense.

  1. Sometimes the data isn’t big enough

Halliburton chief data scientist Dr. Satyam Priyadarshy contended that many times when organizations talk big data, it isn’t big at all. They also spend a lot of time cleaning up the data so they can compare datasets before starting their analysis. He warned organizations that looking at the ‘dirty’ data is part of the analysis. What is missing can be just as telling as what is not.

Many companies discussed how for many datasets they really didn’t have enough data to drive predictive insights.  Perhaps instead of focusing on their own data, they should be thinking about their peers’ data. We should encourage collaboration when it comes to data – this could be the best way to find scale.

SaaS companies like Medgate are able to offer insights from a cross-organizational dataset. This helps clients achieve the scale they are looking for, allowing them to achieve predictive and prescriptive insights.

About the Author

David Poole

David Poole is the Product Marketing Manager at Medgate. Before joining Medgate in 2014, David was the National Director at the ESOP Centre in London, The UK.

More Content by David Poole
Previous Article
How to Build the Case for an Environmental Management Information System
How to Build the Case for an Environmental Management Information System

Learn how to build a business case and achieve corporate buy-in for an EMIS. Find out which stakeholders to...

Next Tweet
Verdantix

What trends does @MedgateSoftware's acquisition of regAction reflect about the #EHS software market? https:...