Better railway engineering through statistical model checking

Abstract

Maintenance is essential to ensuring the dependability of a technical system. Periodic inspections, repairs, and renewals can prevent failures and extend a system’s lifespan. At the same time, maintenance incurs cost and planned downtime. It is therefore important to find a maintenance policy that balances cost and dependability. This paper presents a framework, fault maintenance trees (FMTs), integrating maintenance into the industry-standard formalism of fault trees. By translating FMTs to priced timed automata and applying statistical model checking, we can obtain system dependability metrics such as system reliability and mean time to failure, as well as costs of maintenance and failures over time, for different maintenance policies. Our framework is flexible and can be extended to include effects specific to the system being analysed. We demonstrate that our framework can be used in practice using two case studies from the railway industry: electrically insulated joints, and pneumatic compressors.