Data Analytics in Batch Operations
“There is just too much stuff to get my arms around! My batches run in a complicated multi-train facility and there are always slight variations in raw material feedstocks, cooling tower temperatures, and agitation configurations. How can anyone make sense of all this and achieve repeatable results?”
Clearly making a complicated batch chemical processing facility operate at optimum conditions all the time is a challenge. That challenge was taken up by Lubrizol’s Bob Wojewodka and Emerson’s Terry Blevins who set out to provide process operators with an indication that a batch process was either on a path for optimum conditions or straying off into the weeds. They wrote up their process and results in an article that appears in Control magazine titled, “Data Analytics in Batch Operations (333KB PDF).” They developed a statistical model that determined if operations were running well or not. They were able to use principal component analysis and a partial least squares analysis to analyze batch data from the Lubrizol production facility in Rouen, France. They used this to generate a set of parameters that operators could monitor to keep their operations working smoothly. If you have difficult batch processing issues their work may point the direction out of the muddle.
