I didn't see this posted anywhere in the air cooling forum or anyone mentioning it here or on any other website but I thought I'd throw in a few pennies to the topic FWIW.
One of my former employers was a heavy R&D outfit. A production process we had was using a 8' x 10' oven, ramping the temps up to 750 degrees F for an hour and a half. Cooling it down as rapidly as possible became critical because of the material and process involved but pricing was an issue.
Solution, using two fans of the same speed behind one another to inject almost ice cold air. Two fans side by side overall moved a higher quantity of air (volume) but the temps would only come down at a certain rate (plateau). Using two fans in a series produced much colder air using essentially the same energy and delivered temps that were almost 20 degrees F cooler.
Actual cool down time of an inkanel/hastaloy mandrel went from 1.75 hours (room temp) to a little over 38 mins (room temp), almost 1/4 +/- the time.
Somethings of note. The fans were 14" on pedestals (4 ft high) and about 1 foot from each other. I would assume the fan in the rear used the same amount of energy as it normally would at it's current speed setting (no difference). The fan in the front however was actually running faster (pushed) by the turbulence of the fan in the rear causing it to run slightly faster. The air, even at room temp coming from the front of the forward fan was actually cold, not slightly colder or cooler I mean cold.
The engineer I was working with didn't believe me until I proved it to him. Because of the physical properties of the material involved (not the metal), the rapid cooldown with minimal investment was imperative for what we were doing. I was curious if the same thing would work for processors and if it would be more cost effective than running a water pump given their life short expectancies?
Just a thought.