I've got a pond I'm rebuilding after a dam collapse (See my "rebuilding a dam" thread in the Building a Dam forum). A few years before the dam collapsed there was a massive fish die-off. The local ag extension agent said it was caused by an algae bloom and subsequent "water turnover". As I'm rebuilding I'm trying to take all the past problems into account and correct them. Since we're in Central Texas and usually have very hot summers and the pond gets fairly shallow during that period I'd like to aerate the deepest end of the pond. I've looked at several (rather costly) professional solutions then took a look at resources I have on hand and came up with an idea I'd like to run by you guys.

My mother has always loved windmills. A couple of years ago a friend of my grandmothers was tearing down an old house on some farmland and gave my parents the windmill from the old water well. The windmill has a steel tower and a gear-head made to drive one of those old piston water pumps. When the blades are really spinning there's quite a bit of down force on that rod and it has about a foot and a half of travel.

What I'm considering is making a 4-cylinder air pump out of thick-walled PVC pipe around 6" in diameter. The cylinders will be designed so that they take advantage of all the rod's travel. 2 cylinders will be set directly under the rod and will be driven with the rod. The other two cylinders will sit off-center some what and be driven by a rocker arm so that they operate opposite the motion of the rod. This way 2 cylinders are pumping air on both the up and down strokes of the drive rod. Check valves will be used to make sure the air keeps flowing in the right direction.

I have all the materials necessary to machine the parts to build the pumps. I'm also planning on enclosing the system to prevent UV light from degrading the cylinders.

I found several ready-made windmill systems, but none that would retro-fit to an old-style windmill. Any thoughts on this system?

I like things simple. A larger single double acting piston could move as much air as multiple pistons but have less friction. I would design it to use mechanically operated valves as check valves often aren't very reliable

Well, I'm using "check valve" as a generic term for a one-way valve. Not sure how I would set up a mechanically operated valve without adding major complication to the system.

I considered a double-acting cylinder but I'm not 100% sure how I'd seal the shaft pass-through. Maybe with a recessed o-ring or something. Will have to put some thought into that.

Maybe UHMW Polyethelene? It's super slick, machineable, and highly wear resistant. It's not very flexible so I suppose the PVC would have to be very symetrical for it to work efficiently.

The valveing would work just like a steam engine. http://www.howstuffworks.com/steam1.htm

Same concept but instead of using steam to drive the piston that drives the rod, you're applying power to the rod to drive the piston.