| This thesis addresses the development of efficient optimization strategies for the optimal synthesis and flowsheet integration of chemical reactor networks. Despite the impact of the reaction system and reactor design on the character of the flowsheet, little research has dealt with the synthesis of reactor networks. In the first part of this thesis, we develop bounds, or targets, on the performance index of an isothermal reacting system. A multiple compartment targeting model is first developed, which is based on optimizing flows between different reacting environments. It turns out that the segregated flow limit of this model is often a linear programming formulation, and sufficient conditions for global optimality of this simplification are derived. Moreover, even when these conditions do not hold, tests formulated as nonlinear programs can be applied to evaluate the suitability and enhancement of the target obtained by the segregated flow model. This approach leads to an iterative process (each iteration corresponds to the addition of a reactor) for the targeting problem, where we solve simpler models and verify their sufficiency, and consider more complex models only when required. The success on a battery of example problems demonstrates the effectiveness of the targeting methodology. The constructive targeting approach is then extended to include nonisothermal reactor networks.; With an efficient scheme for reactor network synthesis developed in the first part of the thesis, we extend our synthesis approach to include the optimal energy integration of reactor networks. Here, we combine our non-isothermal reactor targeting formulation within an energy targeting framework, to derive a general framework for bounding energy costs while optimally synthesizing the reactor network within flowsheet constraints. The combined synthesis model is a nondifferentiable nonlinear program and solution strategies are presented. On a typical flowsheet example, significant differences and savings in the reactor network are observed for simultaneous synthesis, as compared to a sequential reactor design and utility minimization.; The third part of this dissertation addresses the development of a unified formalism for the synthesis of reaction-separation systems, while ensuring optimal energy management. Here, we show that by postulating a species dependent residence time distribution function, one can arrive at a general representation for a reaction-separation network. The synthesis model is formulated as a mixed integer optimal control problem, where the integer variables account for the fixed costs of separation. The control profiles include the temperature, the separation profile, and some mixing functions defined for the network. The approach is integrated within the constructive targeting framework, and results on typical flowsheet problems indicate that depending on the relative costs of raw materials and by-products, the conventional scheme of complete reaction followed by separation can be suboptimal. Instead, this integrated approach leads to a new scheme for overall synthesis, where, given a well defined reaction system, one could assess the relative importance of mixing, separation and temperature profiles at the preliminary design stage. |
