I have a duct design, but I need to make one more tweak. The clearance between the hot end and the inner hole was too close and it melted.
The tough thing about fan cooling is strength and bed temperature. With the fan on, the bed doesn't get much more than 70-80, so unsticking is a hazard. Also if the plastic is cooled too quickly as it is extruded, it doesn't bond as well. I had legs that snapped off with minimal force. I was printing at 190 so the plastic wouldn't have as far to cool, but I'm thinking that was a mistake. I want to try printing at 200, and hope the fan still cools it enough to get the layer rigid.
Right now the fan has to be on all the time, but there is a way to hook it up to the board for gcode control. Unfortunately it involves soldering headers to the board, and soldering a small electronic kit to go between the fan and the board. It would be worth it though, because it would let you turn on the fan only for small layers and bridging.
My duct is sized for a 40mm case fan. To power it, I took a wall wart from a dead router and soldlered a connector to its wires. I'll post the STL after I finish the non-melting version.