Railway track asset management modelling

Railways are an important type of transport infrastructure but can be expensive to run with the UK railway costing \pounds 1.525~billion to maintain in 2018/19. To reduce the cost without reducing the quality of the infrastructure improved asset management is required. To enable the impact of possible decisions to be understood, and the optimum ones chosen, a railway track asset management model, such as the one developed in this thesis, is required. This thesis first studies the degradation and maintenance of railway track. The UK railways network actual geometry and maintenance data was used to understand track degradation. This data covered 8 years of the whole UK rail network, a much greater length of time and track than previous research has considered. Many aspects were shown to have a significant impact on the rate such as track speed, sleeper type and maintenance history. The methodology of singling out factors showed that rail types have less of an impact on track geometry than previous research had shown. Weibull distributions were then used to characterise the rates of degradation of separate combinations of these significant aspects. The improvement in geometry from maintenance is also explored, with the effectiveness reducing with each further maintenance action. The improvement in geometry from maintenance is modelled using a linear fit with a stochastic element added to model the effectiveness variability. Maintenance output rate, which has previously not been considered in literature, has also been analysed and modelled, utilising Weibull distributions, allowing working window lengths for maintenance to be incorporated within models. The likelihood of different rail faults occurring was also explored using data from the UK railway. The analysis showed that the rail type, joint and age were not only linked to the rate of faults but also the track geometry. This link has been mentioned in literature but has previously not been proved or quantified. The link between rail faults and track geometry shows how the railway tracks assets are interlinked and hence need to be modelled as such. Saving money by reducing the amount of track geometry maintenance will increase the quantity and hence the cost of rail faults. Rail faults have been modelled using probabilities within this thesis, with the fault rates of each fault type related to the track geometry. The second part of this thesis develops a Colour Petri-Net (CPN) model which incorporates the analysis and models developed. A CPN has been used in a novel way of acting as a decision framework, joining the separate degradation and maintenance models together and allowing them to interact. This removes some shortcomings of state based modelling techniques such as requiring discrete states of degradation. The model predicts, over any given line of track and time period, the number of inspections, maintenance actions, track quality and number of speed restrictions. Utilising the model the user can assess the impact of decisions such as maintenance thresholds, asset upgrades and traffic changes. Additional aspects such as opportunistic maintenance, as well as maintenance productivity and work windows lengths are considered. This allows aspects like opportunistic maintenance thresholds and varying the maintenance window length to be analysed, which previous models in literature have not. Different scenarios can be run through the model and the outputs compared to enable evidence based asset management decisions to be made