Train Aerodynamics (part 1)

The metro trains experience an interesting set of aerodynamic effects in the tunnels. The train itself is used to circulate air through the tunnels and ventilation systems (hence its shape is not very aerodynamic at all, not that it matters at these low speeds). Of course, the main aerodynamic effect on the train is the drag force. On long sloped down sections of track (where the train moves without any power input) the aerodynamic drag can provide about 5 km/h of speed difference as train accelerates from 30 km/h to 80 km/h.

For Subtransit, I’m working on an aerodynamic model that would account for both drag produced by the train as it moves through the tunnel as well as additional torque force produced when train passes by a ventilation shaft. Currently, the CFD results are only preliminary, although there are a few interesting results already.

Currently only drag coefficient Cd is used in-game, although its value is set pretty arbitrarily. I also want to add Cmy coefficient (yawing moment/torque) which depends on relative position of the ventilation shaft. The drag coefficient slows the train down, while the yaw moment twists the train around its vertical axis, producing a small disturbance as train is passing by a ventilation shaft.

To make a better estimate of the aerodynamic coefficients I’m planning to mostly use a CFD combined with some real world data that we can obtain. The first iteration of CFD revealed an interesting effect. In addition to compressing air heavily ahead of itself, the train will regardless of the speed (for the normal tunnel configuration) have a really strong airflow around itself as it moves through the tunnel. The CFD shows that airflow speed around the train may be almost double of its actual velocity (nearly 140-160 km/h for 80 km/h train velocity) – although this result doesn’t fully account for the fact that train would displace and accelerate air along the tunnel.

Pressure distribution (normalized):

Velocity distribution (normalized):

I’m not entirely sure if this effect is real or no, though it does make sense – the wide tunnel transitions to a relatively narrow gap between the train itself and the tunnel. So far I’ve only tried a handful of obvious boundary conditions, but they all result in the same kind of behavior. Even simulating circulation around the train doesn’t seem to change the situation!

There is one potential solution which may explain this kind of behavior. If the train would ‘drag along’ the airflow, the actual effective airspeed would be slightly reduced (as train is moving along with the air itself). This is something that I will be still testing a lot with the CFD, but preliminary result shows that the train may cause air to drag along at up to 40 km/h (when train is moving at 80 km/h). And this effect does cause a significant reduction in Cd.

And also another well known aerodynamic effect is the “sucking” effect between the top and bottom of the train in the area between the wagons, which can be seen at the picture below (to some extent). Some of the air above the train is flowing between the wagons and down underneath the train (where there is a lower pressure area). I wonder if it would be possible to install aerodynamic sensors on the real train to mesure real dynamic pressure outside of the train at various points.

The color of the lines in pictures below indicates velocity along the airstream: