Modeling of apparel production system and optimization of lot size scheduling in apparel industry

Abstract

In the apparel industry, various production departments are involved to fabricate the completion of garments. Among the numerous departments, production planning and scheduling in the sewing department are the most complicated as they involve a large range of manual sewing operations. In the current manufacturing environment, the apparel production scheduling system in the sewing department which concerns the type of items for production, their quantities, when they would be produced, and on which production line in order to satisfy customer demand still heavily relies on the experience and subjective assessment of production management. Consistent and optimal solutions are difficult to obtain under a dynamic manufacturing environment. In the face of increasingly fierce competition and fast changing customer demands today, there is the need for the involvement of apparel manufacturers in the production planning and scheduling process by using scientific and effective methods. The purpose of this research is therefore to develop simulation- and algorithm-based methodologies for the production planning and scheduling process of apparel manufacture. The selection of an appropriate production system is a key element in the apparel scheduling process and considered critical to business success. To improve the selection of the production system, manufacturers must understand how the production system responds to dynamic manufacturing environments. However, it is costly and time-consuming to study the system behaviors through observation or experimentation. A computer simulation model, which takes into consideration the various factors in dynamic manufacturing environments and analyzes the impacts of these factors on production system performance to evaluate the performance of apparel production systems, has been therefore developed in this research. To enhance the representativeness of the simulation analysis, three inadequacies in the existing literature and simulation methods are identified and refined before the simulation model is applied to explore the mentioned objective. They concern the inadequacies in simulating the realistic learning curves of operators, and incompetence in the approach of the validation methods and the simulation experimental analysis. A set of rules for the selection of apparel production systems in different manufacturing environments is generated from the simulation experimental results. In typical practices in the industry, production scheduling plans are most often formulated with constant production capabilities, thus leading to frequent changes of operator assignments and machines on the assembly floor as they need to cope with the changes in actual production capabilities. Although frequent changes to the lot size scheduling plans are very common in the apparel manufacturing industry regardless of the associated high costs, it should not be used as an excuse to accept the current level of prediction accuracy. A novel hybrid computer simulation-genetic algorithm (CS-GA) that takes into consideration the stochastic characteristics of apparel production systems is thus proposed in response, so as to address the dynamic manufacturing environments during the scheduling process. The proposed CS-GA approach includes a single-objective as well as a multi-objective based optimization model. The former imitates a typical goal in the industry, which is to generate scheduling plans that minimize setup frequency with consideration of due-date constraints. In the latter, the goal is to generate scheduling solutions which consider the impacts of the scheduling decisions on the performance of apparel production systems in addition to minimizing setup frequency and satisfying the due-date requirements of customers. The aim is to improve the performance of apparel production systems in terms of resource utilization and work-in-process inventory through scheduling decisions. The proposed approach will therefore eliminate the weaknesses of the manual scheduling procedures and generate optimal scheduling solutions in apparel manufacture. Based on production data from a real apparel manufacturing enterprise, experiments have been conducted to evaluate the performance of the proposed methodologies. The experimental results demonstrate the effectiveness of the proposed approach to model apparel production systems and production scheduling in the apparel manufacturing industry.