Stream of Variation Testing in order to Fault Diagnosis of Multistage Manufacturing Processes

One of the most popular problems in quality improvement of multistage manufacturing processes is the variation of quality characteristics. Stream of variation methodology models variation propagation of dimensional deviation that comes from key quality characteristics. This research describes new procedure for source of variation identification in variation propagation modeling. At the firs, discriminant function used to separate variations and then future deviations assign to model but detected source may not be real source. In order to find real source, stream of variation testing is used. The applicability and performance of source of variation identification investigate with an illustrative case study in connecting rod production line.