A Successive Mixed Integer Linear Programming approach to the grade change optimization problem for continuous chemical processes

摘要

To enable flexible, market-oriented operation of continuous chemical processes the control of transitions between different operating conditions (hereafter called: grades) should be enhanced. Dynamic optimization is believed to be a major enabling technology in this respect. In this paper, we formulate the grade change problem as an economic optimization problem using a finite horizon evaluation of the added value of the processing plant. The resulting objective function is discontinuous due to the transitions between grade-regions which makes standard, gradient-based optimization methods unsuited for solving the problem. A new, Successive Mixed Integer Linear Programming approach is proposed and its potential is demonstrated on an example: the optimization of transitions in a binary distillation column.