Using AI, RPA & Machine Learning to Drive Operational Excellence at Koch Industries
At last year’s IQPC Operational Excellence in Oil & Gas Summit, you spoke about how Koch Industries is using new technologies such as robotic process automation (RPA), machine learning and robots and cobots to drive operational excellence. Can you provide us with an update on these projects and initiatives?
At Koch we are focused on unleashing the power of our employees across the organization to transform. As a foundational question, we’re asking individuals how they can they do their own work differently, and how can they contribute to larger process and company-wide transformations? This is consistent with one of the key themes I talked about last year- the concepts of transformation and new-to. I showed a video of one of our accountants using robotic process automation for her work. I also talked about the analytics solutions, machine learning, robots, and cobots that we use in our plants. In the case of the accountant, it was a new-to-me example. The other examples are new-to-industry or newto- our business. Since last November, we’ve been building on these new-to concepts and applying them across our businesses at Koch. A specific example of analytics and machine learning that has helped us gain extraordinary insight into some of our processing units is central control rooms. At Georgia-Pacific, we have a central control room where we have paired our papermaking facilities’ technical knowledge and skills with sensor technology. This enables us to understand how our machines are operating in a more granular way than our existing plant-wide information systems. After looking at the data, we’ve found that these sensor networks provide more detailed information about our machines and equipment, so we’re often able to see small failures and prevent unplanned events before they happen. With the addition of our analytical and machine learning capabilities, we are faster at detection, giving us higher availability than we had previously.
This year, you’re going to be speaking about Koch’s partnership with the Environmental Protection Agency’s (EPA) Office of Research and Development to create next-generation leak detection technology for the oil and gas industry. Can you tell us more about this?
Several years ago, our environmental team came up with an idea about leak detection and repair, which is an environmental regulatory requirement for many chemical processing and manufacturing sites, as well as for the oil and gas industry. Currently, we dispatch technicians to walk delineated routes in the plan to test valves, connectors, pumps and other equipment for leaks using hand-held VOC detection devices. Given the size and scope of petrochemical facilities and other similar operations, this is a significant undertaking. To provide context, there can be hundreds of thousands of leak detection points that need regular monitoring at a large facility it is a daunting job—often requiring individuals to climb towers, make their way through complex processing units, and deal with the elements to access equipment in remote locations. Our team wanted to make this job safer, more effective, and more predictive. We wanted to know, for example, how could we find the leaks before they started leaking? And importantly, what is causing the leaks so we can make changes designed to eliminate or significantly reduce them? In search of a better way, our environmental team at Flint Hills Resources teamed up with Molex, a Koch electronics company, and the Office of Research and Development at the EPA on a novel idea – to put an array of sensors in a processing unit and use those to detect leaks. The sensors would be deployed near components that are subject to the lead detection and repair requirements and would be used to detect leaks—versus sending a technician to do the detection manually. Once the network finds low-level detections, the system estimates a leak area, and then sends an alert dispatching [.....]
To continue to read this exclusive interview, click here.