Optimal operation and control of process systems

Methods are developed for the rapid determination and implementation of near-optimal operating policies for process systems. In the case of vapor-liquid reaction processes with separation and recycle, processes are classified and as signed to an operating policy based on the process chemistry. For isothermal reactor operation, processes can be classified into one of two groups. For so-termed bounded chemistries it is optimal to operate with the reactor completely full at all times. For so-termed non-bounded chemistries it is optimal to scale the reactor holdup and recycle flow rate linearly with the production rate, subject to constraints. If reactor temperature is available as an additional degree of freedom, then processes can be further classified according to the relative magnitude of the activation energies.; When the optimal operating policy has been determined, this result can be used to guide the selection of a plantwide control methodology. In many cases the optimal operating policy can be implemented using only decentralized feedback control. In this case, this methodology suggests which pairings between controlled and manipulated variables are appropriate. In other cases a centralized controller is required to achieve the near-optimal operating policy. In this case the theory suggests a low-order process model to use in the design and implementation of the centralized controller. Steady-state and dynamic results are illustrated with two case studies: a process to produce chlorinated benzene and a process to produce high molecular weight ethers for gasoline blending.; Finally, the methodology is applied to the optimization of seeded batch crystallizers. Determination of the optimal temperature (or supersaturation) trajectory for a seeded batch crystallizer is among the most studied continuous optimization problems in chemical engineering. The majority of researchers find that a supersaturation trajectory that gradually increases over the course of the batch is optimal, but a significant minority find that a decreasing supersaturation trajectory is optimal. This apparent conflict is resolved in this dissertation. It is shown to result from the use of different objective functions by different authors. Furthermore, the qualitative nature of the optimal temperature (or supersaturation) trajectory can now be predicted for any given objective function.