A comparative study of time-based maintenance and condition-based maintenance for multi-component systems

In this paper, we investigate the benefit of Condition-Based Maintenance (CBM) over Time-Based Maintenance (TBM) in a multi-component system. We consider a series system, where the components deteriorate according to a gamma process, with stochastic dependence between components modeled with a Clayton-Lévy copula function. Economic dependence between components is modeled as a joint setup cost for replacing components. The CBM and TBM policies are obtained by formulating the two cases as separate Markov decision processes and using dynamic programming for optimization. We compare varying system parameter configurations using the relative decrease in cost rate as a measure of how much the CBM policy outperforms the TBM policy. From the results of our numerical experiments the largest relative improvement when using CBM instead of TBM is 45%, which occurs in the single-component system. In the multi-component system, the cost rate improvement increases when the degree of stochastic dependence increases. Furthermore, when the number of components increases the improvement becomes less sensitive to the severity of failures. Finally, the results indicate that when the difference in the system components’ mean time to failure increases, so does the relative improvement in the cost rate. • Multi-component series system with economic and stochastic dependence. • Continuous-time and continuous-space optimization problem. • Near-optimal policies computed using dynamic programming. • CBM is up to 45% better than TBM.