I work in the aerosol industry and they have tried to model nozzle design to improve efficiency, noise etc. All the time the gas and liquid is contained with a channel it all moves very predictably but as soon as it exits into the atmosphere it pretty much does what it likes as so many other factors come into play.
I imagine that it shares similarities with the air off the back of an F1 wing..... from a controlled environment to an open environment with a huge pressure drop....
..at the momentI bet McLaren thought their 2009 car was a real world beater when they first modeled it on a computer and then stuck in a wind tunnel. As soon as they put it on the track it turned out to be a pup. As Pyrope points out the effects of turbulent flow are very difficult to model, you add into that wind speed & direction and humidity (both of which can change during the two hours of a Grand Prix), not to mention changing fuel loads and changing track conditions due to the weather and rubber build up, there are too many variables to cope with.