Thoughts on "Herding Cats"

Oct 27, 2011 4:22:00 AM | Posted by Jim Ford

I applaud Dr. Skogestad’s real-world, practical approach to solving plant control problems in the latest edition of Control Magazine. In particular, his recognition of the importance of the control hierarchy and its relevance to how control solutions should be designed. It’s a refreshing change from the typical academician’s approach (matrix arithmetic, partial differential equations, Laplace transforms, i.e., boring and incomprehensible). I wish more university professors slanted their work this way. You’d see a lot more cooperation between industry and universities if that were the case.

Dr. Skogestad claims that a centralized, plant-wide control solution will never be applied to a normal-sized chemical plant because “one can achieve acceptable control with simple structures”. While certainly not disagreeing, I would propose a much more fundamental, cause-and-effect explanation, namely, that the real world (in this case, the chemical plant) is much too complex, with too many unmeasured disturbance variables, too many unknown multi-variable interactions, too much measurement inaccuracy, too much noise, and too many random process and equipment upsets. Pursuit of a plant-wide solution is futile. Dr. Skogestag will still be working on it in 2025, and will be no closer to the answer.

Instead, I suggest that he and your readers re-focus their efforts on, and I’m quoting Allan Kern in his latest HPIN Control article in Hydrocarbon Processing magazine, “high-performance base-layer controls with automation intelligence embedded throughout.” This is the real solution for process stability, safety, reliability, and performance. A return to fundamentals (and getting away from large, “detuned”, ineffective MPC’s) will restore management’s confidence in modern process control as a productivity-improvement technology.

Topics: Process Expertise