First, I'm not an engineer, so, please excuse my lay terminology. Nonetheless, I thank you for any thoughts you can offer me, as I certainly need some help here. I'm working on a project where I need to spin a 4"-6" fan blade, enough to create about 250 (more is even better) cfm of airflow . (1000 rpm?) (I realize the cfm is impacted by the design of the blade, let's not address that right now.) I need this fairly constant airflow to last for about 3 hours, without interruption. This will be inside a box, with fairly static air. I do not need to control the fan direction (uni-direction is fine) nor the speed of the fan (constant is good.) There won't be any back-pressure or torque against the blade. [Let's refer to this unit as the 'Fan Device'.] This Fan Device shouldn't be much bigger than 5" x 5" x 5". Now, here's the hitch. I don't want to use ANY electrical components, electricity, magnetic components, thermal stuff, or the such in the Fan Device. What I do want, is to have a separate unit, let's call it the Winding Device, which can be inserted or connected to the Fan Device, and somehow wind, or otherwise transfer kinetic energy (is that the right use of the term?), to the Fan Device, and then have the Fan Device store this transferred energy (spring, or something else???), so that energy can be slowly released and turn the Fan Blade. (In other words, the Winding Device only stays connected for a short period, enough to juice up the Fan Device, if you will.) It's perfectly fine for this Winding Device to use electricity, or whatever. Ideally, I'd like this winding time (connection time) to take no more than, say, a few minutes (1-5?). So, guys and gals, any ideas? If for any reason you'd like to contact me back channel, please feel free to e-mail me at: firstname.lastname@example.org If you got this far, thanks for just reading. If you can offer assistance, that would be really great. Raymond.