Investigating Deferment of Maintenance Actions

Abstract When considering a single asset it is plausible to find an optimal solution for its replacement under a wide range of models. Ansell et al (2004) and Ansell and Archibald (2008) considered data driven models which included repair, refurbishment and replacement in the presence of covariates. The covariates may be aspects of the assets, such as age and demand, as well as environmental conditions. Archibald et al (2004) also explored the sensitivity of such solutions to parameter changes in the model. Yet often it is not possible to employ the optimal solutions in a practical context; particularly it may be a requirement to keep assets in operation beyond their repair, refurbishment or even replacement times. Ansell et al (2003) showed that it is possible to assess the impact of maintenance actions (replacement, refurbishment or repair), and by manipulation one can assess the impact of delaying actions. This was based on the assumptions of Cox and Lewis's NHPP model. It is plausible to consider alternative models with more severe degradation models, such as Crow's and others. In many circumstances, though, there is a need to maintain a specific level of availability which adds a constraint. The solution may be a trade-off between availability and timing of maintenance actions. The paper will discuss the modelling issues and through sensitivity analysis and simulation will provide some insights into the ‘best’ solution under differing modelling assumptions.