The problem with pumping water to form ice is that gear assemblies will freeze in place below a certain temperature.
I've never seen or heard of that, could you provide a movie?
It's easy enough to try for yourself. Try powering a screw pump with a windmill in a freezing biome. It doesn't have to pump anything, of course, but it won't be able to because the machinery is frozen, and it will display this when you 'q' over the buildings..
This sounded simple enough, so I embarked a mountainside tundra (both biomes described as "freezing") and gave it a quick shot. Windmill, gear assemblies, and pump functioned normally throughout my observation. Maybe my map only actually gets cold enough for this during the winter, or it takes a while for the frost to build up, or something.
Point being that it's not so simple, and I don't really have the patience to set up a permanent antarctic research station, so someone else may have to delve into this matter. It'd be nice if you just posted a screenshot or two of a setup in which this occurs for further researchers to observe and imitate. Given the possible differences in sites, a save file - or at least worldgen parameters - may also prove to be of use.
I think thats just the windmill freezing and locking the whole chain?
Possible set-up for testing: Set up four chains of gear assemblies. First two are powered by windmill, second two are powered by underground waterwheels. (Cave river, aquifer, perpetual motion machine, whatever.) Second and fourth chains of gear assemblies heated from below by magma pipeline.