Using Complex Event Processing for Continuous Process Control

Next week I'll be delivering a presentation on big data 101 for the American Institute of Chemical Engineers in Austin, TX.  In my experience, you can trot out TFLAs (Three and Four Letter Acronyms) all day long without improving someone's understanding of a topic.  What really helps learning is case studies and use cases, so I'll be using plenty of those.

One example is Schwering & Hasse, a company based in Germany that runs 400 production lines round the clock to manufacture enough copper magnet wire every day to wrap around the earth’s circumference three and a half times.  Copper magnet wire, coated with a thin yet precise layer of insulation, is a vital component in many electrical products, such as transformers and motors.  Made to fine tolerances, the wire is embedded within other components, so failure, product recalls etc. are very expensive for their customers to manage. Quality is everything - if quality is poor, they can get dropped as a supplier. 

This demand for high quality means monitoring the production process in real-time, about 20,000 measurements per second across the factory.  There are roughly 20 different feeds - oven temperatures, speed of the feed, rpm of the cooler, air monitoring - and so on. In addition, the quality of the insulation is checked once every inch.  All of this data is fed into a complex event processing engine (Software AG's Apama) to provide real-time process control and alerting.  This ensures that production quality remains high and has changed the way the factory works.  Without such rigorous real-time production monitoring, there was a possibility that a whole spool of wire might need to be scraped if it didn't meet the quality standards.  Continuous process control helps to reduce scrap, shipments are on-time, and service level agreements kept.