Thanks!

I've exchanged a few emails with a solar-thermal company here in the UK over the last 24 hours or so to get a rough idea of a system to cover all our DHW needs except in the depths of mid-winter.  In this case a pressurised system with ~8msq of collector total on our east and west facing roof surfaces.  Even then I might have to exchange my daily hot bath for a quick shower in December!  This system is not looking cheap, but it is looking fairly routine in terms of well-tested technology.

That arrangement would have us rejecting 80% of the collected energy, some 40kWh/day solar energy wasted, in June.

Now, to cover the bulk of our over-night electricity consumption we'd really only need somewhere between 50W and at most 200W for lights and the TV, and my Internet-facing Web server.  So bleeding out the energy slowly all night might be good and require smaller and quieter and cheaper equipment.

(I'm not attempting to handle the load spikes from the kettle, dishwasher, etc...  We might as well take advantage of the Grid for that since it's there!)

It looks as if we could relatively easily accommodate a 500l DHW tank.

Supposing we stored excess energy by letting that tank reach 90C, and given that we only need DHW at ~50C (achieved with a thermostatic mixing valve for safety), the stored energy not needed for the DHW itself (and possibly having allowed a weekly excursion above 70C to kill Legionella, etc) is 40Cx500l = 40*500*4.2kJ = 84MJ = 23kWh which is about half of the excess energy that we'd otherwise reject in mid-summer, which means that we could just about discard that portion overnight and the remainder during the day.

We could discard that stored heat at about 2kW (ie 2kWh per hour) to keep up in summer, and at an efficiency of (say) 10% for a Stirling/Rankine/similar engine, that would be an electrical output of 200W, which matches the night-time loads described above.  We might want to bleed off power at as constant a rate as possible overnight based on the length of night for the time of year, tank temperature, etc.

I'm wondering that if at 200W or less a solid-state (Seebeck) device might be possible.  I'm only expecting/needing a conversion efficiency of 10% or so anyway.  I still don't know about a grid-tie inverter, but maybe a Seebeck device might behave enough like solar PV (stable DC voltage) to use a small one of those.

In summer the otherwise-rejected 40kWh/day of energy might contribute as much 4kW/day of electricity to the Grid, a fair match to the 1.29kWp of solar PV of similar surface area already installed.

Someone tell me if I counted my fingers and toes wrongly again!  B^>

Rgds