This article proposes a stochastic technique for determining the optimal maintenance policy for marine mechanical systems. The optimal maintenance policy output includes the average maintenance cost rate, maintenance interval and the performance thresholds for the three marine mechanical system classifications. The purpose of this study is to optimize maintenance, maintenance interval and performance thresholds based on maintenance and reliability data of the marine mechanical systems. Performance threshold and maintenance interval are used as the decision variables to determine the optimal maintenance policy. A stochastic model based on probability analysis is developed to trigger the maintenance action for mechanical systems. The model is later coded in MATLAB. Maintenance and failure data for a marine vehicle were statistically fitted using ReliaSoft, from which a three-parameter Weibull distribution best fitted all the mechanical system classifications. Model inputs were based on both the maintenance data and expert knowledge of the maintenance crew. Based on a 20-year marine vehicle life span, the optimal maintenance costs for plant and machinery are relatively the same. The model predicted annual total maintenance cost of US$183,029.24 is 11.11% more than the maintenance cost derived from experts’ threshold of US$164,726. Marine vehicle machinery presents a higher maintenance interval of 3.23 years compared to 2.92 years for marine vehicle plants. It was observed that for the performance thresholds greater than 84.54%, there is an insignificant difference between the plant and machinery maintenance costs. Sensitivity analysis results suggest there is little justification that changing maintenance costs will have an impact on the performance threshold [Formula: see text] and maintenance interval [Formula: see text]. A maintenance interval of 3 years results in a lower total annual maintenance cost deviation of 2.66% from the optimal total annual maintenance cost.
A Near-Optimal Maintenance Policy for Automated DR Devices