A Bayesian model to assess rail track geometry degradation through its life-cycle

• Published in: Research in transportation economics Vol. 36; no. 1; pp. 1 - 8
• Main Authors: Andrade, António Ramos, Teixeira, Paulo Fonseca
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.09.2012
• Subjects: Bayesian model; Life-cycle; Maintenance costs; Rail track degradation; Uncertainty; Rail track degradation; Maintenance costs; Uncertainty; Bayesian model; Life-cycle
• ISSN: 0739-8859; 1875-7979
• DOI: 10.1016/j.retrec.2012.03.011

One of the major drawbacks in rail track investments is the high level of uncertainty in maintenance, renewal and unavailability costs for the Infrastructure Managers (IM) during the life-cycle of the infrastructure. Above all, rail track geometry degradation is responsible for the greatest part of railway infrastructure maintenance costs. Some approaches have been tried to control the uncertainty associated with rail track geometry degradation at the design stage, though little progress has improved the investors' confidence. Moreover, many studies on rail track life-cycle cost modelling tend to forget the dynamic perspective in uncertainty assessments and do not quantify the expected reduction of the uncertainty associated with degradation parameters as more inspection data is collected after operation starts. In this paper, a Bayesian model to assess rail track geometry degradation is put forward, building up a framework to update the uncertainty in rail track geometry degradation throughout its life-cycle. Using inspection data from Lisbon-Oporto line, prior probability distributions are fitted to the model parameters quantifying the associated uncertainty at the design stage, and then they are sequentially updated as more inspection data becomes available when operation starts. Uncertainty reduction in geometry degradation parameters is then assessed by computing their posterior probability distributions each time an inspection takes place. Finally, the results show that at the design stage, the uncertainty associated with degradation rates is very high, but it reduces drastically as more inspection data is collected. Significant impacts on the definition of maintenance cost allocation inside railway business models are discussed, especially for the case of Public and Private Partnerships. Moreover, potential impacts of this methodology in maintenance contracts are highlighted. For the case of a new infrastructure, it is proposed that maintenance costs assessments related to track geometry degradation are no longer assessed at the design stage based only on the prior probability distributions of the degradation model parameters, but renegotiated instead after a ‘warm-up’ period of operation based on their posterior probability distributions.