Bayesian statistics and axle maintenance
Supervisor: Dr Adam Beagles
The maintenance of axles is one of the most costly aspects of railway operations. Each axle has to be regularly inspected to determine if it is safe to continue in service; a very time-consuming process with associated risks. This project is to develop probabilistic models of crack initiation, development, and inspection that can be fitted to data from service depots to provide robust support for changes to the inspection regime.

The models will be prepared using modern Markov Chain Monte Carlo software so initial work will involve familiarisation with these. Aspects of the inspection process will be considered to develop understanding and to determine any problem specific issues that need to be addressed. It is expected that the project will separately focus on the initiation, development, and inspection aspects before combining them into a holistic model; part of the project will be to develop a framework into which different models for the separate aspects can be incorporated.