Jason;
As a fellow ME, I think this would be a good tset of your milling skills :-)
It only took me a few decades to finish mine....
The one item I see a small problem with is the watts you're calling for. 4.2 watts is quite a bit of wattage for LEDs normally the standard LED runs around 3.2Vdc and maxed out at 20ma which should come in ~ .06 watts.
Are you hoping to use a pre-built LED set?
Now to the motor setup. Go find a setpped on kids toys that is AA powered, make sure it is truely broken as in plastics not motor:-) otherwise the wrath will be ooooh:-(
Those little motors when spun by hand make a goodly little of DC and should be a good test point. The best motors would be one that is of the newer ones that have dual axles.
Grab your favorite drill and a handy DC meter, put the axle in the drill and if it's a variable type spin it up after connecting the two wires to the meter. Spin the axle up slowly while either watching or have someone else watching to note the voltages. Once you've gotten to a speed where the voltage does not go any higher make note of that. If you have one that goes above the normal 3.2Vdc then good, as this you can use at a higher speed to light the LEDs by use of resisitors.
You'll then need to design the impellers so that the flow is around both axles, however one should work too. The impeller should be the the same as hyro-"S" type so as to grab as much surfacant as possible , which will also help in slowing the flow.
By memory you'll need to figure an axle tip speed to outer diameter to be on a high order possibly 200:1 or more. I know sounds like a lot , but given the axles are usually measureable in the same as 16 - 14 ga wire this will not be a problem.
Vane propellers might work, but I can see problems with size and fatigue more so that the cup(S) type.
The little wires that will come from the motor will be more than enough to handle the micro-current you'll be pushing.
The Motor should of course be encapsulated by itself and then if possible a venturi effect encloseure for the impellers. This will do two things, venturi will cause enough turblance to slow the flow and enough back pressure to get your impellers moving with a constant output.
I no longer have the milling resources to do this as well, but with some good spare PVC you should be able to build one up with little to no cost (depending on your resourcefullness)
I am on a dial up and pushing this poor 233Mhz laptop to it's max even with a swap file (linux ) of 500 meg, so forgive any speeling misteaks:-)
Cheers
Bruce S