Adel Mhamdi, Willi Geffers, Folko Flehmig, Wolfgang Marquardt:
On-line optimization of MSF desalination plants
In: Encyclopedia of Desalination and Water Resources (DESWARE), EOLSS Publishers Co. Ltd., Oxford, UK, Springer Verlag, 2000
In this paper, the real-time (or on-line) optimization (RTO) of the control system setpoints for MSF desalination plants is addressed. The recomputation of setpoints is necessary to counter slowly varying disturbances which move the process from optimal operating conditions. Typically, RTO is implemented using a modular structure. Since RTO is based on steady state models, its results are only reliable if steady state data are used. Here, a method based on wavelets is used for steady-state detection. To improve the quality of the steady state data, data reconciliation and gross error detection are used. The reconciliation is generally formulated as a constrained minimization problem. Once the data are validated, new setpoints are computed such that an economic objective is optimized. In this work, RTO has been implemented in a simulation environment for MSF plants using powerful commercial tools, i.e. SPEEDUP for data reconciliation and optimization and Matlab for steady-state detection. The coordination and data exchange module between the different tasks of the optimization scheme has been implemented based on the communication mechanism EDI (External Data Interface) of SPEEDUP. A distributed client and server architecture has been chosen, to map the partial problems of the full RTO problem to independent and self-sustained software processes. The communication between client and server processes is based on remote procedure calls.