Simulation-based performance comparison between dynamic cells, classical cells and job shops: a case study

Abstract

This paper presents the results of a simulation-based performance comparison between dynamic cellular (DC) manufacturing systems and two other well-known systems, namely classical cells (CC) and job shop systems (JS). The performance comparison is made at different levels of turbulence. The experiment contains 13 independent variables, most of them related to demand turbulence, and 17 dependant variables related to performance measure. A stochastic simulation model developed on Microsoft Visual FoxPro 5.0 was used in combination with LINDO (a linear/integer programming software) to obtain the initial results. In view of the large number of variables and the time required to run each experiment, a Taguchi plan was used to optimize the model. The results obtained from the analysis of variance indicate that dynamic cellular manufacturing systems are generally more efficient than classical cellular systems or job shop systems, especially with respect to the average and maximum throughput time, mean and maximum work-in-process, mean and maximum tardiness, and the total marginal cost for a given horizon.

Keywords:

• Simulation
• Dynamic cell
• Classical cell
• Job shop
• JIT
• SMED
• DCMS
• SME
• Virtual cell
• Model
• Experimental design
• Factors
• Experimental plan
• Priority rules
• Schedule
• Interaction

Notes

†In the model, the inter-zone transfer batch size can be determined from the cumulative operation time, the number of components or the volume occupied. The value can be parameterised for each policy (e.g. 20 components for a policy based on quantity).

‡For these factors, the distribution mean and range are based on experimental data.

§The increasing and decreasing portions of the life cycle for a given product follows a beta-type distribution curve, where α = 4 and β = 3, to give a profile similar to the goods life cycle curve: marketing, growth, maturity and decline. The average order size for a product is the average of all orders for that product.