Integrated Effective Railway Maintenance

Railway infrastructure is excellent for providing transport capacity and move things and people. This transport capacity cannot be provided during interventions. Performing interventions requires time, money and machines, personal resources; and (partially) blocks access to the track, requiring modification to traffic and timetables, impacting passengers. Typical interventions include monitoring infrastructure; small/large maintenance; expansion. Typically small interventions, based on the concept of interval management, are done during time of little demand usage, spread over a long time frame; or grouped together, for a smaller unavailable time, with a stronger effect to travel chains and demand. 

The crucial question we address is: How to take decisions, especially when, where, and how should interventions be grouped or separated, how railway timetables should planned, given joint uncertainty over maintenance costs, timetable feasibility, and passenger impact? How to find holistically good system-wide decisions, taking into account the existing interdependencies?

This question is particularly challenging because effects and factors are difficult to quantify; uncertain; decision making power is distributed and partially following guidelines, partially following contingent choices, at different time horizons.  

We tackle the strong interdependence between the problems by studying in detail the decision making process, and improving it by quantitative estimations of relevant aspects, described with the associated estimated uncertainty. This requires fundamentally new approaches on maintenance process (estimating amount of resources, grouping, total time required, costs, under uncertainty); timetable planning and adjustment (with given and partially given boundary constraints); and estimating passengers reactions to any intervention plan (additional travel time, time/route/mode choice/not travelling, long term effects), which leads to loss of revenues from reduced ticket sales, and reduction of support from taxpayers. We test through qualitative analysis and field experiments how the additional values we can now quantify, and which process to learn from the past allows most effective outcomes.