Unforeseen switch faults on the railways are a huge inconvenience for all parties involved. For passengers, these faults are a major cause of delays. For rail operators, they create a lot of extra costs, and in a context that sees operators increasingly adopting strategies for operational excellence that aim to do more with less money. But systematic cost cuts can lead to higher costs in the long run. That is why it is important that asset-intensive organizations continually recalibrate to optimize their business processes.
For rail operators, repairing faulty switches can be an expensive undertaking, so Arcadis and Asset Rail developed an advanced tool that makes it possible to predict switch faults. With the results generated by the tool, Asset Rail can service parts that are in need of maintenance, such as suspect switches or switches that are about to develop a fault. As a result, Asset Rail can prevent the faults that are a major cause of delays for train passengers.
Common switch faults can be predicted by measuring and analyzing electrical currents in the motor that controls the switch. To do so, a model that is 'trained' with legacy data was developed. This training data was sourced from a selected number of switches over an eight-year period. The data was then analyzed in conjunction with the technical data, using various data analysis and machine learning techniques. Together with Asset Rail, Arcadis used these fault profiles to create a predictive model that continuously compares the current profile with previously collected data.
On the ten switches on which the model has been tested, we are able to predict 40% of faults, a few days before they arise. This allows preventive maintenance to be carried out on the equipment in a timely manner. With properly functioning switches, there are far fewer delays on the tracks, which keeps passengers and transport operators happy.