Joint Shewhart control charts for location and scale monitoring in finite horizon processes

- Finite production horizon (FPH) processes allow companies to achieve a high flexibility.
- In FPH processes, SPC starts without estimates of the distribution parameters.
- We investigate joint control charts for monitoring a FPH process.
- Non parametric and robust statistics are considered to monitor position and scale.

Small production runs are becoming increasingly important in the manufacturing environment thanks to the technology advancements allowing products to be customized at competitive costs. Similarly, increasing flexibility in high volume production can allow for frequent and rapid changeovers from one part code to another to meet the lean principles. These manufacturing processes are characterized by finite production horizons. To assure high quality standards of products during a finite horizon production, implementing an efficient on-line process monitoring is a critical issue. In this paper we compare the performance of several control charts jointly monitoring location and scale for observations with a location-scale distribution in a finite horizon process where a limited number of inspections is scheduled. For an investigated set of symmetric distributions, our results show that the joint control charts implementing a signed-rank SR statistic and either the Downton's D estimator or the average absolute deviation MD from median generally perform the best. An example illustrates the implementation of the control charts on a simulated dataset.