Condition-Based Release of Maintenance Jobs in a Decentralized Multi-Stage Production/Maintenance System

摘要

1 Problem Condition-based maintenance is analyzed for multi-stage production with a separate maintenance unit. Deterioration depends on multiple production parameters. After reaching a critical deterioration level a machine failure occurs and reactive maintenance is required. Preventive maintenance reduces the failure probability. In both cases the machine status is set up to "as good as new" (AGAN). The task of maintenance is to keep up agreed availabilities αi, of m production stages (i = 1,...,m), defined as the probability of no machine failure between two consecutive AGAN states. In this context the problem of determining optimal triggering conditions α'_i to release maintenance jobs at the right situation arises. Since maintenance jobs compete for limited maintenance capacity, scheduling is required. Due to division of labour, some problem specifics are relevant: (a) Information asymmetry between production stages and maintenance unit, (b) delays Wi between releasing and carrying out maintenance jobs, (c) production during W_i with modifiable production parameters, and (d) influences of scheduling and triggering onto production performance. Previous approaches of decentralized production/maintenance-coordination~1 assume instantaneous information exchange and immediate availability of maintenance resources. Since this is not always fulfilled in decentralized systems, we present a problem-specific continuous condition monitoring (CM). Thereby, we operationalize the components of waiting time, identify decision relevant effects of setting α'_i, develop a maintenance priority rule and a stochastic approach to determine α'i based on chance constraints and analyze it by simulations.