Large-scale scheduling of casting sequences using a customized genetic algorithm

Abstract

Scheduling a sequence for molding a number of castings each having different weights is an important large-scale optimization problem often encountered in foundries. In this paper, we attempt to solve this complex, multi-variable, and multi-constraint optimization problem using different implementations of genetic algorithms (GAs). In comparison to a mixed-integer linear programming solver, GAs with problem-specific operators are found to provide faster (with a sub-quadratic computational time complexity) and more reliable solutions to very large-sized (over one million integer variables) casting sequence optimization problems. In addition to solving the particular problem, the study demonstrates how problem-specific information can be introduced in a GA for solving large-sized real-world problems efficiently.