Assembling Technique Based on the Statistical Feed-Forward Control Model for Low Precision Manufacturing Processes

This thesis embraces the challenge of developing an assembling technique to deal with the problem of producing assemblies that have low dimensional variation by means of mating components that have high dimensional variation. The main objectives of the proposed model are the reduction of the resulting variation, the reduction of the scrap levels and the improvement of the process capability indices. The innovative approach proposes the dynamic management of specifications, target and tolerance, using the Statistical Dynamic Specifications Method (SDSM) and the assembling of complementing groups using the Statistical Feed-Forward Control Model (SFFCM). While SDSM can be seen as a statistical tool that helps determine the right specification adjustments using reduced samples taken from groups of items produced consecutively during a short-time interval; SFFCM can be seen as a monitoring tool that helps counter the effect of an eventual ôdetectableö long-term component present in the variation. With the help of proprietary software, the production of lots of one thousand assemblies made of two components coming from processes characterized for producing items that present high dimensional variation was simulated. For the analyzed conditions, in comparison to a fully randomized assembling, simulation results revealed an average reduction by 89% of the mean shift, an average reduction by 14% of the standard deviation, an average improvement of the actual capability index of the assembling process by 16%, an average improvement of the potential capability index of the assembling process by 101% and an average reduction of the assemblies out of tolerance by 100%. In conclusion, the proposed SFFCM-based assembling technique effectively helped achieve the mayor objectives of this thesis: reduce the process variation, reduce the scrap level and improve the process capability indices.