[Al Skierkiewicz]

Fluffy,
As Jim pointed out, there is no hard and fast rule as to which thermal design is too little cooling. No Victor is running at 40 amps for an entire match. You could model an average match but it would only hold true for a particular game and a particular robot design. Here is what we do know, the FET(IRL3103) is speced to run in the temperature range -55 to +175 C. Each FET can handle 64 amps at 12 mohm ON resistance at 25 C and the package is designed for about 50 watts dissipation. There are two pairs of three FETs in parallel in series for each direction so that translates to 8 mohm total series resistance (12/3 +12/3). If a speed controller is feeding a Chalupa in stall then the current is about 130 amps and the entire controller is dissipating 130^2 x .008=135 watts. So in the worst case, the fans must get rid of 130 watts or the controller will start to heat. As the heat in each FET rises, the operation of the FET changes. The ON resistance rises and therefore the max allowable current falls. At 125 C the max current rating is about half that at 25 C. Now as to failure, the junction temperature (that temp right at the silicon junction inside the case) can rise rapidly without the case temperature being affected. If that is the case, then no amount of cooling fan will protect the FET from abuse.
The better question might be "Have you experienced any Victor failures attributable to heat or blocked fans?" I have not seen one Victor that I could absolutely attribute a death to heat rise. The majority of failures have come from foreign objects (metal shavings and dust) or flat out abuse (rapid forward and reverse motor command or lengthy operation in stall such as repeated turning with tank drive and sticky treads).