A multiple-choice genetic algorithm for a nonlinear cutting stock problem

Abstract

In the environment of intense competition facing many firms today, cutting costs from operations while meeting customer expectations is a prime objective. Industrial computing can help satisfy this objective by guiding many aspects of operations planning-for instance, mathematical models of operational problems can be developed and solved as part of information systems. Typical approaches come from the fields of operations research (mathematical programming), computer science (constraint satisfaction), or the interface (metaheuristics such as genetic algorithms, tabu search, and neural nets). To illustrate how to use computational models to gain a cost advantage in operations, let's consider an actual cutting stock problem faced by a Fortune 500 wire and cable manufacturer. Our team, consisting of a vice president, several IT professionals, and business unit managers, sought to reduce scrap and other costs with the help of an operations research consultant. I will describe two approaches to the problem: mathematical programming and genetic algorithms. Each approach has advantages and disadvantages in application.