It’s a Mad, Mad, Mad MADLab

Duke Energy’s MADLab (machine learning, artificial intelligence and deep learning) initiative is an important driver of innovation culture.

The film starts with an ex-convict in a car crash near Palm Desert, California, U.S. Five motorists stop to help the crash victim. Before the convict dies, he shares that he’s buried $350,000 in Santa Rosita State Park under a “big W.” The five good Samaritans talk about sharing the money if they work together, but it quickly evolves to “Everyone for themselves.”

Today, analytics projects, process teams, and various initiatives are sprouting up within utilities. And too often each group is using different tools and platforms while often attempting to solve similar problems. This creates a mad, mad, mad utility quagmire. But recently many utilities have been working on solving this exact situation.

I just returned from the Utility Analytics Institute’s 8th Annual Utility Analytics Summit conference hosted by Duke Energy. The event had nearly 300 utilities attending from more than 90 organizations coming together to share their successes, failures and challenges in their digitization journey. Duke Energy is one of the leading utilities that has been working on analytics for quite some time. One of it’s new initiatives is called MADLab (Machine learning, Artificial intelligence and Deep learning).

Brian Savoy, the senior vice president of Business Transformation and Technology at Duke Energy Corp., kicked off day two of the Summit with his keynote on how MADLab came into being to support Duke’s innovation transformation.

Cultural Change

To drive this massive cultural change, Brian took several C-level leaders for a road trip out to Silicon Valley to meet with Bay area companies known for their innovation capabilities. The purpose was to understand how this organization created a culture of innovation and speed. After that first trip, the Duke leadership understood the need to shift its culture to enable the utility to break the fossilized processes and procedures of a 100-plus year-old utility and to speed innovation.

This road trip was so inspiring, Brian took another group of leaders back out to California to expose them to the innovation culture of Silicon Valley. One of the results of the trips was the concept of the MADLab. The MADLab team is located in a completely new space in a new building that is separate from any existing Duke Energy facilities. The layout, the design, the culture, and the entire experience at this building is specifically designed to create the right environment to drive innovation and change at a pace unrealized before within this utility.

Two of the co-founders of the MADLab are Brian Wilkerson, senior manager of Big Data and Analytics, and Norv Clontz, director of Data Science Innovation at Duke Energy (The folk at Duke calls him Super Norv. Search for him on Linkedin you will see his Superman pose!).

So what is happening at the lab? Is Duke seeing any success? The answer is an absolute yes! High-level examples revolve around Duke Energy building its internal team’s analytics muscles. It is now delivering projects in emerging areas including deep learning for video and image, analytics from data collected from drones, digital kiosk video analytics, neural networks, natural language processing and object character recognition.

Some specific examples include using drones to inspect boilers at power plants. The results speak for themselves. Instead of a week-long project, where the utility workers would set up scaffolding, do the inspections, capture the information, take the scaffolding down and upload the data, the MADLab decided to try using drones to do the exact same task in hours resulting in savings of tens of thousands of dollars. Also, the safety aspect cannot be overstated.

Another drone application is flying UAVs over solar farms and using machine learning algorithms to identify faulty cells within the solar arrays at a speed and cost reduction that is exponential when compared to sending out a crew to the solar farm to manually test all of the arrays. Once other teams within the organization saw the art of the possible with advanced analytics, they quickly realized they had similar problems that could be solved with drones and analytics.

Success Story

One critical success story of the MADLab has been a massive increase in how quickly new solutions are being rolled out into the company. The team uses agile design to quickly test project/product ideas and readjust the solution in two-week sprints. It recently rolled out a new iPhone app for the field team at Piedmont Natural Gas, replacing an old hand-held technology. At the event Brian played a video testimonial from the manager of that group stating how amazed he and his team were with the successful roll-out of this new tool: the speed, the cooperation, the results.

One success example was the ability of field engineers to make suggestions to improve the product. Several of those suggestions were acted on in one day and delivered in the app the next day. This is happening at a utility!

At the end of the day the MADLab is not only the heart of analytics efforts at Duke Energy it is also one of the most important drivers on changing and creating an innovation culture.

Mark Johnson is the Managing Director at Utility Analytics Institute and T&D World.