Sensors ensure smooth rail operations in Scandinavia, even in very harsh winters

Points that cannot be set due to icing or large snow accumulations lead to massive delays, sometimes even train cancellations. Heating them is not enough, as William Liu from the Royal Institute of Technology (https://www.kth.se/en ) has discovered. In Sweden alone, although almost all points are heated, there are around 1,000 cancellations every year, affecting hundreds of thousands of passengers. Sensors and artificial intelligence (AI) are expected to provide the solution.

Better despite cost reduction

The sensors record the weather conditions at each individual switch and measure temperature, wind speed and other data (https://link.springer.com/chapter/10.1007/978-3-032-04774-8_41) that are responsible for the fate of the switches. From this, AI software that has been trained with countless weather and associated icing data derives the necessary heating temperature.

Currently, this is a constant eight degrees Celsius in Sweden, which is not sufficient in certain weather conditions. Conversely, the temperature can be lowered if there is no risk of icing despite the extreme cold. The actual heat loss also depends on factors other than the outside temperature, namely wind speed, humidity and the presence of snow. Wind in particular is a critical factor because it cools the metal more.

Computer model for forecasting

Liu (https://www.kth.se/profile/zhendong) has developed a computer model (https://www.kth.se/en/om/nyheter/centrala-nyheter/rail-switch-data-may-reduce-train-delays-in-cold-climates-1. 1456986) to predict how a switch will heat up under different weather conditions, including factors such as air temperature, wind and precipitation. The forecasts correspond to the data actually measured on a test switch. It has been shown that short periods during which the switch is heated to well over eight degrees are far more effective in critical situations than continuous heating.