Post #1952645

@csk I suppose the obvious choice is to take something whose output in kW varies according to some statistical distribution, and then calculate its variance. But that's boring!