Machine learning is responsible for a variety of changes across industries — from the way people work to how they interact with organizations. The oil and gas industry is no different, as was highlighted at the recent Machine Learning in Oil and Gas event in Houston, Texas.
At the event, presentations centered on how machine learning is and will continue to change the oil and gas industry over the course of the next decade, including in areas such as:
- Augmenting the existing workforce
- Improving safety
- Improving execution based on data
- Simplifying complex processes
When it comes to data in particular, there are tremendous opportunities for improvement across the industry. Without machine learning, many data scientist teams are still spending 70 to 80 percent of their time just getting data ready to analyze. Some of the presentations at the conference highlighted both the ways in which machine learning can solve oil and gas data issues, as well as some of the challenges organizations face in implementing machine learning.
Tying Machine Learning to Business Drivers
At many organizations across different industries, machine learning is a much-discussed topic — but actual implementation of tools that leverage machine learning can be challenging. As we have seen at other recent conferences, technology decisions are no longer just IT projects. Increasingly, technology implementations need to be tied to a key business driver in order to be successful.
Implementing machine learning in the oil and gas industry, therefore, is about showing the efficiencies and benefits that the technology can bring to the overall business. Afshean Talasaz, Sr Data Scientist at Chesapeake Energy Corporation, discussed how he addressed this issue by taking his team out to the field so they could understand the field worker experience. He then showed where machine learning could provide efficiencies in analyzing well performance, helping to increase both the top and bottom line. In his presentation, Afshean compared machine learning to a Ferrari — it’s fun, cool, and everyone wants to talk about it (or drive it). Data, on the other hand, are the roads the Ferrari has to drive on. No one wants to fix the roads and do the dirty work of preparing the data — but without clean, good quality data, the machine learning algorithms won’t be nearly as effective or accurate.
Machine Learning for Wells Mastering
One area where we have seen great potential for machine learning in the oil and gas industry is in mastering wells data. Graeme Gordon of Hess Corporation gave a presentation about how his team successfully used machine learning to reduce the amount of time spent on data cleansing, and unify data sources to ensure up-to-date wells information that leads to better analytical insights.
Working with Tamr, the Hess team was able to discover relationships between wells attributes across sources and master wells data across multiple vendor datasets to identify wells that were of high interest. In less than one day, Hess and the Tamr team loaded and associated all of the datasets for 19 vendors. Then, over the course of a week, subject matter experts at Hess trained the machine learning models by reviewing the software’s suggested matches. The result is a unified view of wells across vendors that can easily incorporate new datasets with a high level of accuracy. Further, data at Hess is now treated as a live feed versus a previously stagnant report as a result of Tamr’s unification efforts. The team is able to generate daily reports that data consumers at all levels within the organization can use to gain insights and answer business critical questions.
It’s clear from the Hess example and other presentations at the Machine Learning in Oil and Gas conference that there are many ways in which machine learning can help change the industry for the better. When it comes to data, there is a great opportunity to help companies unify their data to drive efficiencies and enable better decision-making. The challenge for many organizations will be tying machine learning initiatives to greater business drivers in order to show how the technology can have an overall positive impact on the business bottom line.