Stochastic reliability models for maintenance optimization

Abstract

In this thesis, we considered and solved some problems which related to the maintenance of multi-component systems. From our previous work, we extend the basic unavailability models with calendar-based maintenance policy to the models considering age-based maintenance policy. There is a comparison between both policies approach, it is based on evaluating the effect of them to unavailability function U(t) of the systems which selected from both references and practice. The directed acyclic graph (DAG) is used to describe the functionality of systems and to compute their unavailability. Cost-optimization problem is solved where decision variables are changeable maintenance parameters that are optimally selected from a set of possible realistic maintenance models. Any system structure can be represented by means of a directed acyclic graph (DAG). In case it is a complex system, containing a lot of components, if an internal node has multiple dimensionality, i.e., a lot of input edges, resulting in an excessive summarizing combinations, we can expect computational difficulties performing unavailability calculations. For this reason, we improve the computing algorithm to reduce the combinations by multiple application of the pivot decomposition. The effectiveness of the process will be demonstrated on selective systems.

In this thesis, we considered and solved some problems which related to the maintenance of multi-component systems. From our previous work, we extend the basic unavailability models with calendar-based maintenance policy to the models considering age-based maintenance policy. There is a comparison between both policies approach, it is based on evaluating the effect of them to unavailability function U(t) of the systems which selected from both references and practice. The directed acyclic graph (DAG) is used to describe the functionality of systems and to compute their unavailability. Cost-optimization problem is solved where decision variables are changeable maintenance parameters that are optimally selected from a set of possible realistic maintenance models. Any system structure can be represented by means of a directed acyclic graph (DAG). In case it is a complex system, containing a lot of components, if an internal node has multiple dimensionality, i.e., a lot of input edges, resulting in an excessive summarizing combinations, we can expect computational difficulties performing unavailability calculations. For this reason, we improve the computing algorithm to reduce the combinations by multiple application of the pivot decomposition. The effectiveness of the process will be demonstrated on selective systems.