You know, I don't know if this has been discussed, but the whole pressure differential thing got me thinking of something that I'm surprised I hadn't thought of earlier.  Since the draw of air via low pressure on the output side seems to be key, I wonder if an eductor type of scenario would work.

Back in my Navy days, we used to use a device called an in-line eductor for dewatering flooded spaces.  Similar to a venturi, more or less, you pumped water through it, it created a suction, and it was designed to suck as much water through it as you put into it.  100 gallons per minute input would dewater at 100 gallons per minute with 200 gallons per minute flowing through the output.  A representative device is here (perhaps not for fluid use, but the concept is similar):

http://www.1877eductors.com/eductor_gas_dimensions.htm

I presume a similar approach could be taken with a radiator setup, would you think?  I imagine it would work well for a center mounted radiator with a centerline, below-the-nose scoop, so one could utilize the cheek inlets to provide source air for the outlet side.

~Chris