Maintenance Policy Applied In The Economic Design Of Cusum Charts

Nowadays, the control of manufacture costs and lean manufacturinghas become the priority of manufactures. Past studies always assumed theproduction process to be perfect process, indicating all products to bequalified and the related costs are only inventory and ordering costs.However, in the reality, caused by fatigue of equipment or other inevitableinterruptions, products may not be qualified and thus end in quality losswhich would produce more costs. Quality control chart is applied toinspect the production process and execute planned maintenance orpreventive maintenance according to the manufacturing status, so that theproduction process can be adjusted. Perfect maintenance policy wouldmake the equipment as good as new and imperfect maintenance wouldn’tmake the system as good as new,but younger. As more demand is requiredon the precision of certain products, the classical Shewhart charts can’tdetect the failures when the mean shifts are relatively small. CUSUMcharts are thus proposed to be more sensitive to small shifts by amplifyingthe detected information and measure the cumulated shifts, which wouldbe more precise to control the quality of high demand products with smallmean shifts.In this paper, the concept of quality loss is considered and CUSUMcharts is applied to monitor the manufacture process applying themaintenance policy. The aim is to get the smallest total production costs,thus to optimize the maintenance policy and the CUSUM chart design. Thesampling size and sampling interval, control limit coefficient andparameters concerning the maintenance policy would be set as decisionvariables and would be optimized according to the model. The pattern search method in the toolbox of Matlab would be implemented as thealgorithm for this model. Sensitive analysis and comparison study wouldbe conducted to measure the model. The model is more applicable toreality compared to the classical CUSUM economic design and moresensitive to the small shifts than the Shewhart control chart. In theconclusion,we summarize the future research possibilities following thissubject.