Dynamic optimization of fluid catalytic cracking unit using a nonconvex sensitivity-based generalized Benders decomposition

Abstract

Fluid catalytic cracking unit, whose batch operations are operated in a multirate mode, is a typical continuous process with batch operations. The integrated optimization of this problem can be formulated as a hybrid parametric dynamic optimization. To obtain a high-quality solution, adaptive direct methods are usually required to solve the problem iteratively. By exploiting the decomposable structure, a novel framework is proposed in this paper, which can obtain an equivalent or better precision solution with relatively coarse discretization. In detail, by designating the batch operations as complicating variables, an optimal solution and sensitivity information about batch operations are obtained by a nonconvex sensitivity-based generalized Benders decomposition algorithm. Then the optimal continuous operations are implemented as extra closed-loop controllers by tracking the necessary conditions of optimality, while the optimal batch operations are improved by a line search method. The practical potential of the framework is demonstrated with several operation modes.