Heuristic Based Scheduling for Toothpaste Filling Problem

Abstract

Abstract This research focuses on the development of a heuristic method for the Toothpaste Filling Problem (TFP) whose complication lies on a vast variety of products (formulae and sizes), together with specific restrictions on filling machines and storage tanks. More specifically, filling machines are non-identical, each of which has different speeds and capacities for filling toothpaste of various sizes. As unloading requires storage tanks, each with much larger capacity when compared to one particular production order, a tank will be then matched with several production orders and be repositioned among filling machines to avoid unnecessary changeovers. We solve the TFP by means of a two-phase method, where the initial solution is constructed by an Earliest Due Date (EDD) dispatching rule. Once completed, the initial solution is then iteratively improved by a series of improvement operators, mimicking the concept of Variable Neighbourhood Search (VNS), until no improvement could be found. We test our heuristic on a set of sample instances acquired from one of the biggest consumer products manufacturers in Thailand. Impressively, when compared to the current practice, our proposed heuristic could help reduce makespan by 32% on average, or equivalently an annual saving of $0.6 million, mainly from the reduction on overtime payment and changeovers.