| The research presented in this dissertation provides a rigorous and systematic methodology for the optimal synthesis of isothermal and nonisothermal reactor networks, as well as for the synthesis problem of reactor separator recycle systems. In the case of isothermal reactor networks, a superstructure is postulated that has embedded all the desired structural alternatives and includes options for reactors in series, parallel, series-parallel or parallel-series, as well as options for bypasses and recycle streams. The proposed methodology can handle multiple feeds, side streams, as well as reactions of any type. Since reactors are usually coupled with a system of separation units, by studying the reactor network and the separator network separately the interactions between these two systems, as well as the trade-offs, are not well understood. Thus, in the second phase of the work, the reactor network is extended and coupled with a separation system so that a superstructure is generated that includes different types of reactors, different separation tasks and considers all the potential interconnections among the reactor and separator units.; The synthesis problem of nonisothermal reactor networks is also addressed with the focus being on developing an approach applicable to homogeneous exothermic or endothermic complex reactions and capable of handling both thermodynamic and economic objective functions. In addition to the alternatives of the isothermal operation, the synthesis structure embraces all different types of temperature control for the reactors and includes, apart from the pure adiabatic operation, options for perfectly controlled units and directly or indirectly intercooled or interheated reactors. In the last phase of this work, a systematic methodology is presented for the optimal design of stable systems. The methodology is based upon the formulation of a parametric problem that bounds the optimal stable solution to which it converges within a number of iterations. When extended to the optimal problem of reactor networks, the approach is able to couple the optimization problem with stability concerns even in cases where the number of reactors is large and the reaction mechanism is described by a general complex reaction scheme. |
