I finally got around to reading this. Reasonable results I thought. It seems like there were at least a couple of ways to improve the performance of the willis.
Firstly, they didn't account for the heat loss from all the extra plumbing, so in some ways of looking at it the willis performed better than they reported. But the insulation on the willis itself as well as the pipework could all be increased with useful results I suspect.
Secondly, there was a large gap between the temperature of the heated water above the willis and the hottest that the water in the tank got. Some difference must be necessary for the syphon to work but I suspect there's scope to optimise that part of the circuit.
I was also interested by the results for the coil in tank, which didn't destratify the top of the tank as much as I expected from previous reading.
