Oil Refinery Example Problem
Processor Simulations: Each process unit type must be simulated with as many control setting combinations as possible to see if it agrees with reality. Each control needs testing at least at three points: maximum, normal/average, and minimum. Given that each process has k-control valves thats a lot of testing. When satisfied move on to finding an optimal productivity for each type of processor.
Processor Optimizations: Convert each process simulation into an optimization by adding an objective (function) to each model. For example, process 123 may want to
- maximize products 321, 432, & 543;
- minimize input waste;
- maximize profit from input; or,
- optimize ???
Once an optimal value is found you need to verify that points around this optimum are higher or lower as they should be. Move away from optimal control settings and see if simulations prove this is truly an optimal point. Find objectives for all processes and test all optimal solutions to ensure they are optimal.
Combine Optimizations: Those processing units that have common interest (e.g. same Project) should be nested into one optimization problem with an objective function whenever possible. Here one is trying to insure that common parameters will work together properly. Once this is done its time to pass these simulations / optimizations onto your Total site optimization group.