| Περίληψη: | An important and challenging problem that Process Engineers frequently encounter is the determination of appropriate control structures that minimize the loss of process performance under the effect of uncertainties. This can be achieved by selecting subsets of controlled and manipulated variables and designing their interconnection (controller synthesis). This is known as the Control Structure Selection Problem (CSSP) In this Thesis, a systematic optimization methodology, based on the back-off concept proposed by Prof J. D. Perkins and co-workers, is presented for the CSSP. The proposed formulation offers the following improvements: a) improves the accuracy of calculations and b) reduces computational time and effort.
Specifically, the error involved in the approximation of the nonlinear constraint that defines the magnitude of the back-off vector (needed for control structure selection) is reduced by the introduction of a more accurate linear approximation. In addition, the methodology is able to track the effect of simultaneously occurring disturbances at the same time and estimate their worst impact on process economics. The reduction of computational time is achieved by eliminating the state variables from the final formulation. In this way, the number of equations needed for the CSSP solution is significantly reduced and allows the algorithm to locate the solution faster without affecting the performance.
The proposed methodology is firstly applied in a classical distillation column (medium scale) and in a complex and highly nonlinear reactive distillation column (large scale). The results obtained from these case studies made the way for the application of the methodology on the plantwide problem of the benchmark Vinyl Acetate monomer production plant in order to demonstrate the benefits of the proposed algorithm.
|
|---|
