I've been considering a similar setup for my machine. I have an Antec Sonata (120mm exhaust L1A), and a Alpha 8942.
I've thought about using an 90deg bend 120mm diameter smooth-wall aluminum ducting to attach between the 120mm exhaust fan to a 120mm->80mm adapter mounted to the Alpha to draw case air across the heatsink (in suck mode) and exhaust the hot air out the back. This would eliminate the panaflo 80mm L1A I have on the Alpha now.
The only drawbacks to the design that I can forsee are these:
1) The majority of the case air would be drawn into the heatsink, thereby making it a coarse filter and catching loads of dust (if any).
2) Since the velocity of the air would increase with any decrease in duct diameter, the noise produced at the heatsink would be much greater than the two fans at present.
Example: 120mm L1A at 7V pulls approx 40CFM, that same volume of flowing air restricted to 80mm (50% reduction) would cause the velocity to increase proportionally, to approx 60 CFM (50% increase). 60 CFM flowing through the heatsink would cause lots of turbulence (noise).
Granted that the turbulent air would be at the heatsink and not at the leading-edge of the spining fan blades. The air entering the fan should be at the same turbulence level as without the duct.
I suppose if you wrapped the ducting in acoustical foam or something it shouldn't be too bad.
Somebody please correct me if my assumtions are wrong.
Turning my Alpha into a dust trap that needs cleaning every week isn't part of my plan.
The good thing about the design is that it would reduce the total number of fans inside the case and at the same time increase the velocity and volume of air flowing through the heatsink for greater cooling capacity.
Dell has come up with a similar plan in it's Optiplex tower series but they have the fan mounted halfway between the CPU and the rear opening.
Maybe if I can find some ducting locally I'll give it a shot and find out for myself.