Study On Maintenance Effect Assessment And Maintenance Decision Of Repairable System

Manufacturing system is usually subjected to preventive maintenance, which is recommended by the manufacturers, for the sake of improving the operation reliability, economy and safety. As a result of the effects caused by the operating environment, maintenance operations and conservative maintenance system, inappropriate maintenance maybe introduced. Thereupon a theoretically and practically more significant method should be made decision of preventive maintenance based on the operating status. However, the cost and risk of obtaining system condition information through condition monitoring are very high. And consequently the operational and failure data recorded by the management information system should provide information support for reliability analysis and preventive maintenance decisions. Then this dissertation carries out a thorough study about the maintenance effect evaluation and preventive maintenance decisions based on the field data. The primary research contributions and innovative outcomes are summarized as follows.Understanding the system maintenance performance firstly can promote the maintenance management. This dissertation presents a complete modeling method and procedure to analyze the corrective maintenance and preventive maintenance performance based on field failure data. The analysis models are the proportional intensity model and two proposed composite models. Then the concept of “maintenance performance scorecard” is introduced to get the snapshot of the corrective maintenance and preventive maintenance performance. Through the case study, many factors have a significant influence on the bus operational reliability. And consequently these factors should be considered when formulating and adjusting the maintenance policy.Maintenance performance analysis is a qualitative description, and quantitative description is more meaningful to carry out maintenance optimization. The existing maintenance effect models cannot be used to solve all of the five maintenance states. Besides, very few of these models have worked on the joint planned corrective maintenance and preventive maintenance process. Therefore, this dissertation proposed a failure-counting-based maintenance effect assessment model. Specifically, this model assume that the cumulative failure number can be reduced or be aggravated a quantity after a maintenance activity, which can be proportional to the difference between the inherent failure path and the last failure path. This proportion is thus used to represent the maintenance effect. As a result of the proportion can be randomized from negative to positive, the proposed model can be used to deal with all of the five situations of the effectiveness of maintenance activities. And the proposed model is very flexible.System maintenance performance is linked closely with maintenance inputs(such as downtime). Then the downtime will be restricted for the production requirement. The most appropriate description of the system availability should be the mission availability. But in the previous literatures, very little literatures introduce the availability analysis of the bounded-cumulative downtime system. Meanwhile, the cumulative downtime and cumulative uptime are seldom considered simultaneously. In the existing studies, the alternating renewal process is widely used to analyze the mission availability. But this solution is not suitable for computational work since there is no closed expression form. A mathematics model is thus proposed to describe the mission availability for bounded-cumulative-downtime system. Numerical simulations are presented to illustrate the effectiveness of the proposed model. Demonstrated by numerical studies and examples, the evaluation result of this proposed model is very accuracy. And the model is very easy to be calculated and applied since the model has a closed form expression.Inappropriate maintenance will be introduced due to the effects of the operational environment and maintenance activities. The maintenance decision should be optimized once the inappropriate maintenance is found. A complex repairable system often comprises a number of subsystems, and the reliability of each subsystem is usually different. In this dissertation, all failures are classified into three categories: minor, moderate and major failures according to two factors, the downtime and repair cost. Different failure categories are introduced to optimize different grades preventive maintenance. The relations between the condition variables and operating time can be represented by a random failure point process model to the setup of maintenance decision making optimization models. According to the data mining technology and methods, the proposed model is extended to multi-grade preventive maintenance decision-making optimization. Demonstrated by numerical studies and examples, the maintenance policy obtained using the proposed method is more effective than that is recommended by manufacturers.Considering the maintenance effect and finite usage simultaneously are very meaningful. Nevertheless, very little literatures introduced the maintenance effect and finite usage simultaneously when carrying out maintenance optimization. Then the preventive maintenance policy presented in this dissertation considers the maintenance effect and finite usage simultaneously. When carrying out the model conformation, the alternation between downtime and uptime is considered. In addition, the preventive maintenance cost is developed to consider the maintenance effect as a function of allocated maintenance cost. And this method provides much more flexibility in dynamic optimization of preventive maintenance. Demonstrated by numerical studies and examples, the proposed method is better than the existing methods in literature, which can reduce the maintenance cost and improve the system operational reliability.