by wkuhns » Mon Sep 21, 2009 8:11 pm
It's pretty hard to do well. I couldn't get any mathematical model that I believed in. Suffice to say that you WILL get flow if the layout is suitable and there are no bubbles. The flow rate will vary with the temperature difference. I get a peak of around 0.5gpm with mine.
One factor that makes flow rate less critical: if the flow rate drops, the delta T increases. Therefore, the net heat transfer rate doesn't change much. It works really well, and is great in that there are no pumps, no valves, and no controls.
If I were designing houses, I'd plan a set of thermisiphon panels in every one. In many cases, all you'd have to do is plan to put the DHW tank on an upper floor.