Modeling the impact of prognostic errors on CBM effectiveness using discrete-event simulation

The kernel of implementing condition based maintenance (CBM) is the selection of optimal maintenance opportunity which is estimated by prognostic tool. Without considering the randomness of prognostic result, the unnecessary preventive maintenance and unnecessary system failures cannot be avoided. In order to improve the system performance under CBM policy, the impact of prognostic error on CBM efficiency must be assessed immediately. So this paper attempts to address this concern through the evaluation and comparison a simple system performance under three maintenance policies including CBM, run-to-failure maintenance and scheduled preventive maintenance. For each of maintenance policy, a discrete-event simulation model is built to obtain two estimator that is mean time between missions accomplishment and average cost for component replacement because these estimator can directly reflect the requirement of system user. After a set of numerical experiments under various operating condition is completed, simulate results suggest that condition-based maintenance can improve system performance as much as 10% to 15% over scheduled preventive maintenance in summary. However, as the prognostic error increases, the effectiveness of CBM will be inferior to scheduled preventive maintenance and run-to-failure maintenance sequentially.