Data-driven predictive maintenance of rails

This cooperation with COWI seeks to develop better mathematical models for the prediction of wear, and subsequent need for maintenance, of railway rails. Better understanding of the causes of both net rail deterioration and the speed at which rails deteriorate can lead to more efficient planning of maintenance. Better planning of maintenance can not only save expense, but reduce the consumption of materials, which themselves are typically non-sustainable and generators of greenhouse emissions.

Have you ever arrived late with the train? Or did a small bump spill some of your coffee in the coupe? All that and much more are caused by rail defects posing a serious concern for railway infrastructure managers. Initiated rail cracks are not necessarily easily detectable nor predictable; they are expensive to repair and expensive not to repair, and dangerous to ignore. Performing maintenance operations before defect detection is emerging as a transformative approach with the potential of making the railway safer and more cost efficient. With this industrial PhD project, we develop data-informed mathematical computer models that can help realizing predictive maintenance in the railway business.