This one may have some interesting results. Do you only use the IEEE 1584 typical gap distances, i.e. 25 mm, 32 mm, etc. for arc flash calculations (most programs are based on this) or have you used actual measured gap distances?

I voted that we measured the gap only because we tried it once on a transformer. Of course the distances were quite large compared to IEEE 1584 so we ended up staying with the IEEE values.