Over the past several decades, many advances have been made in optimizing the maintenance practices of capital assets. Condition-based maintenance strategies in particular analyse real data and system measurements to facilitate evidence-based asset management. In traditional condition-based maintenance practices, regular inspections are considered a given requirement. Most measurements, however confirm that the machine in question is in good working order and that no maintenance is necessary.

The central thesis of this dissertation is that under some circumstances, non-periodic inspection scheduling can result in significant savings by eliminating superfluous inspections, increasing the inspection rate as the asset ages, resulting in better use of the remaining life of the asset. The cost and time of inspections are incorporated into our decision policy that produces an inspect-or-replace result jointly with an optimal time for the next inspection. We use a Weibull proportional hazards model to compute the system reliability, combining the age of the system along with the condition monitoring data.

Through our case studies, we compare our models against a periodic inspection policy, and the classical age policy. We study some preconditions for our model, under which our model performs better than classical age and periodic models.

We further develop the model to allow for amortized payment of the initial installation cost. Using the amortized model, we integrate maintenance decisions with economic decisions, resulting in a more realistic model. We find that the amortized model provides costs savings above and beyond our first model.

In our final model, we introduce failure costs and repair times that are dependent on the age and state of the system at failure. By adding this component, we reflect the reality of varying types of failure, and the variability of the resulting repair time. Our numerical results indicate that allowing the cost and repair time to be a function of the age and state of the system at failure produces better results than the model with fixed cost and repair time.

By these three models, this dissertation demonstrates the significant cost saving that can be made with appropriate strategies to optimally schedule inspections, without sacrificing system reliability.