Improved sustainability in cellular manufacturing systems: Sensitivity analysis of penalty function driven NSG-II

Abstract

Purpose: Research suggests that material handling costs account for 20-50% of production costs. Furthermore, these production cost could be reduced by 10-30% by dynamically changing the layout. We propose an integer programming model to sustainably solve plant layouts in a financially conservative, yet environmentally friendly way.

Design/methodology/approach: We propose a bi-objective Non-dominated Sorting Genetic Algorithm (NSGA-II) approach to optimizing Dynamic Cellular Manufacturing Systems (DCMS). The mathematical model’s first objective function minimizes economic cost, while the second objective function minimizes environmental emissions. The NSGA-II solver uses the penalty approach to handle constraints. The solver is customized beyond the traditional NSGA-II, such that constraint violating solutions are repaired, and unique solutions are prioritized to enhance population diversity and exploration.

Findings: Although a manufacturing plant layout may be optimal for a particular demand period, when the demand changes that system may not be optimal for the new demand period. Extensive simulation shows that our bi-objective model dominates the single objective model from literature. Adding an environmental second objective to DCMS reveals that the most economical solution is often the least environmentally friendly approach, and vice versa. A convex relationship is observed between the two objectives. A weighted compromise is required when setting up a sustainable production system.Research limitations/implications: Only carbon emissions were simulated. Hazardous liquid waste, Energy consumption, and water consumption were not considered. Practical implications: Manufactures and their contracted line builders will need to consider environmental implications when setting up a production line. Decision makers need to be aware that the most cost conservative approach may lead to significantly higher carbon emissions. Social implications: The social pillar of sustainability was incorporated via a set of constraints in the mathematical model of this study. The solution space showed that the model was not restricted by this objective. Originality/value: The value add of this work is presented in a case study comparison of our multi-objective model against a single objective model from literature. The multi-objective model dominates the literature model giving invaluable insights to possible improvements to previous research work.