A New Approach to Determine Parameter Sensitivities of Transfer Lines

In this paper, we develop a complete and practical method for efficient compulation of the sensitivity information of a transfer line throughput. Our approach, which is a combination of analysis and experimentation, is fundamentally different from the existing solution methods. Using this approach the gradient information of the line throughput with respect to various decision variables is made available by a single observation of the line production history plus computation based on the observed data. These calculations are simple enough to be implemented on microcomputers. After an analytical development, three applications and experimental verifications are presented, namely the optimization of the line throughput with respect to the buffer sizes, the cycle times and the repair times. The study provides useful insights to line performance improvement. For example, buffer sizes are but one set of decision variables, not necessarily the most important. Cycle time and repair time can be more significant parameters on the line performance optimization. In addition, the method provides a useful management support system in identifying the line critical points, where local improvement due to change in a parameter can result in most significant improvement in the overall line throughput performance.