September-October 2017 NPJ

22 NuclearPlantJournal.com Nuclear Plant Journal, September-October 2017 Predicting the... ( Continued from page 21) Lighthouse. It is a predictor of the level of risk to an adverse outcome at a plant. GE and Exelon developed an analytical model that ingests performance indicator data. Every nuclear plant collects and reports performance indicators once a month. Those performance indicators are things like open staffing positions, work order backlog, open control room work orders, for example. Many of those indicators are ingested into a model that has been trained to predict how likely you are to have an adverse event, based on the information that’s been given. Adverse events we are currently focused on predicting are forced downpowers, SCRAMs, or a human performance event like a station clock reset. We can help site leadership understand that there’s a level of risk presented in various areas, and here are the symptoms that the model is saying are the largest drivers that are creating that level of risk. We now have a working app, and we are working with the site leaders to gather feedback. This project tends to garner interest fromothers because lots of people are doing analytics on equipment, but almost no one is doing analytics on organizational outcomes. To help understand how it works, the model will take data in from March 2018, and it will predict relative levels of risk in a window of time in the future. March 2018 data will give you a prediction for June to September 2018. You have a few months’ buffer to determine how to avoid that adverse outcome. 7. How do you benefit from innovations in other organizations worldwide? Beyond nuclear for innovation support in Exelon, there is a corporate innovation group, and they are constantly on the lookout for new technologies and developing trends in the world, looking for how this could help Exelon. Data analytics and the concept of Predix was presented to us in Generation from them. We went through a fairly substantial series of research steps and time to understand what the Predix platform would provide for all of Generation. 8. Is GE providing you both hardware and software? We are using some hardware from GE. We are placing a device at every site that will connect downstream of the process computer to securely send the data to the cloud and to the Predix platform. The majority of the data currently transmitted to Predix originates in our data historian (typically OSI PI). The majority of the data we will utilize will be coming out of an existing computer system or existing database. In parallel, we are pursuing a couple of different options for wireless sensors to be able to add additional equipment sensors where it makes sense. 9. Are wireless sensors easy to implement? Nothing is truly easy in a nuclear plant. There’s still issues to be resolved on having that wireless sensor send information to our wireless, internal, intranet. In parallel, we’re using a distributed antenna system, which is essentially cellular data, at Nine Mile Point. That is a separate network that I can send data to a little more easily. We are trying to find a couple different paths to get us additional sensor capability, where we would be able to get more information to help us reduce our overall costs. 10. Concluding comments. My job is more than the digital transformation of nuclear processes. I also oversee our employee innovative idea pipeline. Our employees are encouraged to put in any innovative ideas they have into a platform. We review them utilizing a cross functional team. We determine if they’re connected to things that are already in progress or if they’re things we think could be built upon, turned into a concept and tried. We find ways to support those employees with high value ideas, and then we track the value that those will create at some point for the company. The ideas are quite varied, which keeps it interesting. This can be a challenging thing to do because you want to make sure people keep giving you their ideas. You want to make sure people know what’s next in the process. You want to make sure that you can support the high value ideas with resources and funding as appropriate. And you want to make sure they produce value. All of those things need to happen in order for people to continue to be motivated to be part of this innovation effort. They still have a day job, but they’re passionate about something. So you want to be able to harness that passion to create future value for the company. Some of the basis for why we selected GE and Predix is the ability to co-innovate to create value for both Exelon and GE by going to market with a co-developed product. We think that both Lighthouse and the things we are doing in the outage planning and execution area--still in the conceptual phases--are really going to be a differentiator for us going forward. Exelon is still leading the industry in outage performance. Others are catching up. We think this is an opportunity for us to take another step change up in our performance by using advanced tools like data analytics and the Predix platform. Contact: Joan Knight, Exelon Generation, 4300 Winfield Road, Warrenville, IL 60555; telephone: (779) 231-3100, email: Joan.knight@ exeloncorp.com. http://digitaleditions.nuclearplantjournal.com