This study was conducted to investigate the performance of the FMS under different WIP levels and scheduling rules. The mean tardiness, throughput and machine utilization were used to evaluate this system. The scheduling rules utilized for this study were input sequencing rule, next station selection rule, AGV dispatching rule, and part dispatching rule. This study was conducted by simulating an FMS consisting of four machining centers, one load station, one unload station and three AGV's for various WIP levels. SIMAN was used to model this system. The following scheduling rules were evaluated:

Input sequencing: SPT, LPT, and FIFO Work station selection: NINQ and WINQ Part dispatching: SPT, LIFO, and FIFO AGV dispatching: NV/RW and NV/MROQS

Due dates were assigned to jobs upon entering the queue in front of the load station. A triangular distribution was used to assign processing times at the stations. Job inter arrival times were assigned by an exponential distribution.

Previous research had indicated that results are dependent on conditions of the FMS. In this study however, the FMS was studied under several WIP levels. Furthermore, several conditions were altered to evaluate the system under all possible extreme conditions. For this purpose the system was analyzed after changes were made in the following conditions:

1. Routing flexibility
2. Arrival rate
3. Processing times
4. Product mix

These scenarios were evaluated for different WIP levels in order to evaluate the effect of those levels on FMS performance. Resulting in a total of 633 runs. Analysis of results indicated that WIP had a significant impact on system utilization and throughput. This study indicated that the best part dispatching rule was the SPT rule when the due-date was the measure of performance. This has been observed when decision points were studied separately. Input sequencing did impact the throughput and machine utilization with the LPT rule being the best rule when the WIP level was a constraint. However, when WIP became less of a constraint the input sequencing rules studied did not influence throughput and average machine utilization. Next station selection rules studied did not seem to influence the performance measures. This could be due to the fact that there is a correlation between number of jobs and total waiting time in the queues in front of the machines. Increased routing flexibility reduced the demand for WIP space. Therefore, allowing it to maximize its throughput and machine utilization with a lower WIP level than when the system is less flexible. Previous research has indicated that not one rule will be best for all performance measures which has been the case in this study as well. Researchers had indicated that AGV dispatching rules might affect system performance. In this study, AGV dispatching rules only affected throughput and machine utilization when the WIP is low and there were bottlenecks in the system.

Previous studies have not made an attempt to investigate the effect of WIP in an FMS. Furthermore, they have studied a fixed set of parameters and indicated that the results were valid only under those circumstances and that results may change if the system parameters would differ. However, in this study all possible decision points were integrated into the system. This system was furthermore studied under various extreme conditions with different WIP levels to see whether system parameters would alter results. Therefore making the findings more generic.