This article presents a stochastic dynamic programming model for a condition-based maintenance application. The approach seeks to determine the most opportune moment to inspect and execute preventive maintenance over each component of a nonredundant system, where the number of inspections to be performed simultaneously during each period is limited due to resource constraints. The model minimizes the total maintenance cost per unit of time, considering failure, maintenance, and inspection costs. Unlike most related literature, the model proposed herein allows nonperiodic inspections; it does not require to predefine a maintenance threshold and does not necessarily connect inspections to maintenance actions. Also, the criticality of each component is not static through time, or defined beforehand, but dynamically determined according to the available resources and the risk of failure. A numerical example illustrates the performance of the proposed model in comparison to three traditional maintenance models, namely corrective maintenance, age-based maintenance, and condition-based maintenance with periodic inspections. Results suggest that the proposed model yields the best solution among the studied policies and is more efficient, with a significant reduction of 90% in inspection resources.

Failure causes of production systems need to be readily identified, for effective recovery solutions. Root cause analysis can be used for this, integrated by a method for identifying coherent and sustainable solutions. In this article, root cause analysis for failure cause identification is coupled with a method of problem solving to guide the search for effective solutions, called theory of inventive problem solving. The integration of these methodologies is implemented and demonstrated on a real case concerning the solution to the potential failure of the equipment used in the torque test performed on truck engines used in high-tonnage mining operation. The incorporation of a key performance indicator analysis allows evaluating realistically the results obtained and determining the success of the solution.