I've done a few tests with some very interesting results.
I measured the facing angle of a turning unit in 0.001 sec. intervals. I used 4 different values of Turn Rate (0.1, 0.3, 0.5 and 1.0) and put the results in a graph:
Note that although the curves in the graph are continuous, facing angle updates only every 0.03 seconds.
The speed isn't constant, but rather slowly raises and then drops to make the turn more smooth. The second graph shows how angular speed depends on time (basically, it's the numerical derivative of the values in the first graph):
A few interesting facts to note:
- Turn Rate values from about 0.5 above result in virtually the same unit movement (the curves for TR 0.5 and TR 1.0 overlap)
- angular speed is capped at 6 rad/s, which corresponds to 0.18 radians per 0.03 seconds
- for Turn Rates lower than 0.18, the facing angle change per 0.03 sec. peaks at the value of the Turn Rate rather than at 0.18 (in the second graph, angular speed for TR=0.1 approaches 3.33 rad/s, which corresponds to 0.1 radians per 0.03 seconds)
To add on to my first post in this thread, the facing angle can indeed change by at most Turn Rate radians per 0.03 seconds, but the cap is 0.18 and the change is not constant.