Did you know that if you stacked all the world’s railway sleepers end to end, they would stretch to the Moon and back more than 7 times?
There are an estimated 2 billion sleepers (also known as ties) in use in the world. Their role is to keep the rails along the world’s ~1.3M km of track in their correct place at the correct gauge and to transfer the load of a train to the ballast which in turn spreads the load evenly over the formation layer.
With such a vast quantity of sleepers and given their important role, we might ask ourselves what information about sleepers should be collected? Do we need to catalogue each individual sleeper? If we do, how can we obtain and use such data to provide value to asset owners?
Even at the turn of the millennium, maintaining databases of such large quantities of assets was a challenge often not tackled. Efforts were mainly focussed on recording defective sleepers usually by means of a visual inspection. But with the progress of machine learning algorithms and improved processing power, detailing individual sleepers has become not only practical but also relatively straightforward.
Cataloguing the existence and type of each sleeper can help answer simple questions such as ‘are there sufficient sleepers per km to meet railway standards?’ or ’how many sleepers do I need to purchase for a sleeper replacement program?’
A classification algorithm can be used to separate wood, steel and concrete sleepers and understand their distribution on the network. This can be expanded to determine the specific sleeper type, the manufacturer and the year of manufacture by analysing the shape of the sleeper and autonomously reading visible logos. This adds a powerful layer of information to help plan maintenance, for instance by considering the age of sleepers or targeting sleeper batches with known above average defect rates.
The precise location of each sleeper and spacings between sleepers is also useful information. A single skewed sleeper is unlikely to be of concern but clusters of skewed sleepers might indicate insufficient track support. Sections of track with systematic large spacing may require remedial action.
We can also automatically determine the surface condition of the sleepers. Typical linescan camera resolutions (0.5 mm – 2 mm) can help identify many types of cracks and chips which can be measured and categorised. By recording multiple runs the progression of cracks can be tracked over time. 3D linescan camera technology provides depth information by using a laser for illumination. This can be used to determine the depth of cracks and to separate marks on the sleeper that might look like cracks from true damage, reducing the likelihood of false detections.
Technologies for scanning the interior of wooden and concrete sleepers which include X-rays, ultrasonics and ground penetrating radar, are at various stages of maturity but all currently have limitations.
The most value is gained when sleeper locations, sleeper types and surface condition information are used together to provide data driven insights. We can then begin to ask questions such as; ‘are certain sleeper types performing better than others?’ , ‘what is the optimum sleeper spacing to minimise ballast degradation?’, and ‘how do certain sleeper types perform in different conditions?’.
Maintaining detailed records of sleepers and ensuring this information is easily accessible to capital planning and maintenance teams, is practicable for all railroads, and will help optimise sleeper maintenance programmes, reduce cost and reduce waste.
Stay tuned to our blog series on the benefits of railway asset condition mapping.