A Simulation Study of the Integrated Model of Preventive Maintenance and CUSUM Control chart

• Main Authors: Chia-Wei Shiu, 許家偉
• Other Authors: Yun-Shiow Chen
• Format: Others
• Language: zh-TW
• Online Access: http://ndltd.ncl.edu.tw/handle/26653472338300448263

碩士 === 元智大學 === 工業工程與管理學系 === 103 === The paper applies simulation analysis to discuss the integration of preventive maintenance and CUSUM control chart. In consideration of the assignable case in process that could cause the process average to shift. Assume the particular time of assignable case in process fellow exponential distribution and when the assignable case occurs, the amplitude of process shift is micro or tiny change. The study uses CUSUM control chart to monitor the process and adds the mechanism for warning limit and preventive maintenance to analyze the particular time for executing preventive maintenance. Apart from executing preventive maintenance during scheduled time, in case the statistics appear in non-random model, such preventive maintenance will also be executed early. The study is proposed through runs testing principle and if the statistics fall on the multiple consecutive points between the warning limit and control limit, it is regarded as a non-random model. The process will renew after performing maintenance. Due to the inaccessibility to the optimal solution between the process cycle time and costs, the study decides to apply Monte Carlo Simulation to discuss process ARL, average number of executing preventive maintenance in every cycle, average costs of each unit time, and compare the difference between integrated model and non-integrated model of ARL and average costs. The simulation results illustrate that the statistics fall on the consecutive points between the warning limits and control limits are deemed as non-random model (r), the number of sampling prior to executing preventive maintenance (m), and the exponential distribution parameter (λ) are all inversely proportional to the ARL and the average number of weekly execution of preventive maintenance, whereas r,λ, and the average cost per unit time is proportional, while m is inversely proportion to the average costs of each unit time. In general, the ARL of integration model is significantly longer than the ARL of non-integration model while the average costs per unit time also declines. It is suggested that the higher proportion of preventive and corrective maintenance costs, the more suitable it is to consider the mechanism for preventive maintenance.