Advanced controls for multi-stage flash (MSF) desalination plants and fluid catalytic cracking (FCC) units

This dissertation deals with a systematic analysis of the control problem of industrial units. The control analysis begins with a degrees of freedom analysis to determine the number of variables that can be set independently. This gives the number of manipulated variables in the system. From the knowledge of the process and the operational objectives, the variables that need to be controlled are selected. Operational objectives such as maximizing feed through-put, or production, or minimizing energy consumption can lead to sizable savings in the plant operation costs. Multivariable constrained model predictive control technology is a powerful tool which can be used to realize the economic benefits. It is important to have a good control structure before implementing the multivariable control technology. For this a sound understanding of the process and the operational objectives are essential. Two industrial examples are studied in detail.; One is a multi-stage flash (MSF) desalination plant. Currently, the MSF plants all around the world are operated under the command of PID-type controllers. Also, the operation procedure for making shifts in the production is ad-hoc and it does not guarantee the desired production. A new MSF control formulation is proposed which significantly improves the plant operation. The interaction analysis of the MSF control system points to an interacting problem. Also, there are bounds on various controlled and manipulated variables that need to be obeyed while operating the plant. Thus, the MSF control problem is re-formulated as a multivariable control problem. Constraint model predictive control is applied on a dynamic mathematical model of the MSF plant and operational objectives such as maximization of distillate production, minimizing of the performance ratio while obeying the constraints on the process variables are demonstrated.; The other industrial example is a fluid catalytic cracking (FCC) unit. A detailed dynamic mathematical model of a short contact time FCC unit is developed using the open literature information. The dynamic model is used to conduct the detailed control analysis of the FCC unit. There is significant incentive to study the on-line optimization of the FCC unit. Even a small improvement in the amount of feed processed through the unit or an improvement in the gasoline yield can lead to a sizable savings. Constraint model predictive control with on-line optimization capabilities is demonstrated using a dynamic mathematical model of the FCC. The results of feed throughput maximization and gasoline production maximization are demonstrated.