Evolutionary Algorithms for Scheduling of Crude Oil Preheating Process Under Linear Fouling

Abstract

The crude oil preheating process in refineries is required to be scheduled in a way to minimize the processing cost involved with it, subject to the satisfaction of various process related constraints. The process forms a mixed-integer optimization problem as the scheduling of the processing units involves binary variables, while the discharges from the running units are real valued. The two parts of such problems are usually handled by two different algorithms, where the optimum scheduling obtained by one algorithm is fed to another algorithm for optimizing its discharge process. In the present work, formulating the crude oil preheating process under the effect of linear fouling as a mixed-integer nonlinear programming (MINLP) model, three binary-real coded evolutionary algorithms (EAs) are investigated in order to demonstrate that a single EA can successfully tackle its both binary and real parts. Further, the statistical analysis of the performances of the EAs are also presented through their application to a benchmark instance of the problem.