Asset intensive industries today are surrounded by gigabytes of industrial data, countless data sources and diverse data systems that continuously collect a wide variety, volume, and velocity of industrial data. However, most have not adopted the right technologies to mine intelligent insights. The Energy sector has traditionally been reluctant to take chances with new technologies. Given heavy regulations, the nuclear industry has been even more conservative than its counterparts. This has resulted in increasing operating expenses that the nuclear industry must now address.

Part of the Energy CIO and/or CDO role is to imagine the art of the possible in a new digital world where people, data, and machines integrate to do what was not possible just a few years ago. Exploring and sifting through business cases is both an art and science, and can achieve remarkable results.

There are myriad opportunities to harness the power of big data, from the trending and analysis of facilities energy consumption, to notification of quality and process problems, to predicting machine failures before they happen. Some industries have implemented a basic asset strategies involving real-time sensor readings that feed static solutions. Benefits are reaped initially, but diminish over time, as the original conditions/context becomes irrelevant due to changing equipment and external conditions. The results of unplanned downtime can be severe for every industrial company with real bottom line impacts. Every Energy CIO's near-term roadmap should include predictive and proactive maintenance using machine data and analytics to understand which and when equipment is likely to fail. 

At GE Hitachi Nuclear Energy, we have embarked on our digital journey. We are leveraging Predix, GE’s cloud-based PaaS(Platform as a Service), to achieve internal productivity and help our customers reduce their operating expenses. Our “Fast Works” approach includes working intimately with our business partners and customers, evaluating use cases, testing hypothesis with minimal spend, creating MVP’s (minimal viable products), and pivoting early, when necessary.

We have utilized this approach with Exelon, our largest customer, in a co-innovation pilot. The hypothesis was whether we could predict events at a plant (i.e, scrams, capacity factor, forced loss rate, etc.), using historical human performance data. We were able to identify the most relevant data points and created and trained predictive models using five years of historical data. It has been a resounding success: Our models accurately predicted the events, three months in advance. Based on these predictions, a risk score is calculated for each plant, and for the whole fleet. We continue to tune the models and create new ones, and are working with Exelon to take the product to market.

There are tremendous opportunities awaiting in the area of digital transformation. Now is the time to take the first step.