D.I. Hans H. Eder KG
ACT, Elsbach, Austria
There are cases where a test on the life process is simply not possible. An example is a delayed coker. We needed the process parameters for a MBC application, but the time to line out was estimated to be six to seven hours. However, every four hours a drum switch took place—a heavy disturbance for the plant. Luckily, we had a simulator and could run the tests there.
Even if we can make a plant test, estimate the process parameters, and then calculate the controller tuning, we should test the loop for disturbance behavior. Yet triggering a disturbance just to test one controller is typically not possible in the real plant, but can be easily done with simulations. We can also quickly check the effect of changes in the process behavior (e.g. longer deadtime at lower throughput) on the loop.
I have long ago given up doing level controller tuning and testing in the plant. Again, we can only make SP tests there. However, the SP never changes, and how the loop responds to disturbances is a different question.
One more point should be brought up here. More and more, dynamic simulations are used before start-up of a new plant for operator training. That is good news.
The bad news is that, after start-up, nobody is assigned to keep the simulation up-to-date. Thus, it loses its value. This is very sad, as it could really help to develop and test controls in much faster time and without disturbing the production.