In the coming machines, clear bubble-free compact ice is produced at the surface of already existing ice, hence by its cold, but the existing ice itself is cooled from its growing face, thus speed isn't limited by the thickness.
One machine raises the existing ice against a cold plate (not too rigid) for a few seconds, then sinks it less than 1mm under the water level for a few seconds (possibly aided by a wave), so all the new water layer freezes (important to stabilize an even surface). One cycle gains about 0.5mm in 20s, so the thickness gains 90mm per hour. Less thickness per cycle would increase the rate.
If the ice can be grown around a mandrel, then existing ice is cooled by gas or a liquid bath or jets on a fraction of a turn, and the proper small amount of additional water is brought (spray, bath, brush, sponge...) during the rest of the turn. Cooling and water can alternate several times a turn.
Or build a large head that moves (for instance with small circular oscillations) over the growing ice face. It carries many nozzles blowing cold gas and many pipettes, in an alternating pattern like white and black squares on a chessboard for instance.
In a continuous process, the thickening bar passes under alternated gates blowing cold gas and adding a bit of water.
These machines are energy-efficient, as they need little more cold than the latent heat of fusion, and at a temperature not much colder than 0°C. The three last ones can be nicely fast.
Man, am I good.
Incidentally, it could be interesting to have a look at how normal people produce ice. After all, it has been done for quite a few decades, so methods must exist, and not necessarily bad ones. Sometimes inventions that I thought mines existed before, or even better ones - how insolent - but checking it would take me too much time. Other people are welcome to check it.
Marc Schaefer, aka Enthalpy