Willis solarsyphon

[Ivan]

Wookey,

In your situation, where you're differential controller is referencing temperatures low down on tank with solar panel, I can't work out the answer to this problem:

If low tank sensor reads say 30C and panel reads 40C (lets say your DTO is 10C), then the pump turns on, heating the heat exchanger to 40C. However, if the top of the tank is say 60C, how would the heat be transferred to the tank? If it isn't the pump would run continuously, dissipating the solar heat in the pipework, or if pipework dissipation power is lower than solar input, then temperature would rise until sufficient to overcome the temperature at the top of the cylinder.

It doesn't look like this is what's happening from your graphs - so where am I going wrong?

[wookey]

(This probably ought to be dscussed in the PHE control thread rather than this one really, but as you asked)

Well, just because the top of the tank is 60, doesn't mean that the average tank temp is above 40 - it's the whole water column that has to be considered in terms of whether thermosiphoning will occur. But clearly you could have a situation, as you describe, where the thermosiphon wouldn't do anything much - that's a disadvantage of having it only connected at the top (and why a stratifier would be better). In practice it does seem that so long at the panel is 10-20C more than the bottom of the tank, circulation does occur, and heat does get transferred. (there never seems to be more than about 30C difference from top to bottom in my small tank).

I also decided after thinking about it a bit that it didn't matter if the water going in was cooler than the top of the tank so long as it was hotter than 'useful' (i.e ~40C).

Clearly a perfectly controlled system would pump both sides of the PHE (see the Rebus design, for example).

[Ivan]

OK, so you're saying that the PHE column being averagely hotter than the tank column is more buoyant, thus pushes water at 40C into the top of the tank, where it will descend to its rightful level, mixing to a greater extent on it's way down.

I think you are better off with an unpumped tank-PHE exchanger - otherwise the pump will destroy the stratification in the tank. The scenario above wouldn't cause any great problem, because circulation would be much slower and so you wouldn't lose much of the stratification.

My argument for placing the sensor on the PHE is that if the thermosyphon action is slow, the panel heats up more, thus heating PHE more, thus improving the thermosyphon speed, and increasing heat transfer in the process. Would be really interesting for you to try this and to give us your view on whether it is significantly better.

[wookey]

That all sounds about right. You can see for yourself whether it looks like it would make any difference by comparing bottom of tank temp with bottom of PHE temp. As I say they are mostly the same. I'll give it a try anyway (but not in next 2 weeks as I'm off on hols).

[mick]

Finally got my diagrams done - still can't get them to show here so they are attached.

I was trying to work out what happened in the situation several have posed (including myself) - cylinder hot from previous input, bottom cold from incoming cold feed, weak winter sun, S2 on lower cylinder so pump starts when T diff above set value. (You could pose same question with S2 on Heat exchanger, etc).

Note: Diag 1 which shows my theory that the side arm and cylinder, in both Wookey's PHE version and with a Willis, are thermodynamically one vessel so eventually temperatures should level out across both 'sides' - diags will show what I mean.

Also note: Willis has hot flow from panel going to the bottom of the heat exchange element, exiting from the top of it.

Diag 4. is this month's competition. Answers on a post card or, easier still, posted here. Karma points may be awarded for best and worst!

Phew!

Mick.

 

[Brandon]

we tried placing the s2 sensor in various places, inside the willis jacket, in the normal place of s3, none of which provided stellar performance.

So far the results look pretty good with customers saying that they are impressed with amount of solar hot water.

Impressed with what? what are they comparing it to. willis' do produce solar hot water, as do OSO cylinders... 

One other consideration is to fit an injection tee to the cold feed between the pressure control valve and the cylinder on the Megaflow otherwise all you will draw off is cold water, as it will short circuit the system (I know because I found out the hard way and had to reposition the valve!) 
Rob

Doh! they will also draw cold through the short circuit on vented systems.

[JohnS]

My view, for diagram 4, is that once thermal syphon has started, it will continue even if the top of the cylinder is at a higher temperature.

This is analogous to a normal syphon flowing uphill once it has started.  As Mick shows on diagram 3, the head of cold water pushes down sufficiently to keep the syphon going.

[colinstone]

A couple of real noddy Qs/thoughts.

Like the idea of a Willis but what PHE is being used as an alternative??

Megaflow cylinder is used because.....??  What is so good about them??  I've had custom hot water cylinders made for my boat at a really good price and would be tempted to go to them again for a solar set up.

Thinking about a system with vac panel and solar powered elec pump to circulate, then a Willis/or PHE.

[martin]

From an installer's point of view, I can't think of one good thing about Megaflos, apart from anything else they can cost a sale of a system ("well, I've just had a new Megaflo installed, I don't want to rip it out") - which is where a Willis soda syphon could come in useful - punter keeps grot Megaflo, no tank change, useable, affordable solar install!

[wookey]

Mick - I agree with your analysis re how this works - best to think of both tank and side am as 'one funny-shaped vessel'. NIce diags - we really should fix you inabilty to post them as pics.

Colin - 'Any old PHE' is the PHE being used. As found in any combi boiler. The only real difference is the fittings needed to attach them.

Advantage of 'Megaflo' (unvented cyclinder) is you can get mains water without the small temp drop associated with a thermal store. I suspect one of their main advantages is the removal of the top vent pipe, which is a terrible heat-loss feature.

[mick]

JohnS,

My view, for diagram 4, is that once thermal syphon has started, it will continue even if the top of the cylinder is at a higher temperature.

This is analogous to a normal syphon flowing uphill once it has started.  As Mick shows on diagram 3, the head of cold water pushes down sufficiently to keep the syphon going.

Not sure about that. Think about the 'other side' of the dynamic, the rising force of hotter water which in Diag 1 is equal and opposite to the downward force of the cooler water. It could well be that Diag 4 (the next stage), the slug of warm water in the Willis just sits there. If, as per Ivan's .....

If low tank sensor reads say 30C and panel reads 40C (lets say your DTO is 10C), then the pump turns on, heating the heat exchanger to 40C. However, if the top of the tank is say 60C, how would the heat be transferred to the tank? If it isn't the pump would run continuously, dissipating the solar heat in the pipework, or if pipework dissipation power is lower than solar input, then temperature would rise until sufficient to overcome the temperature at the top of the cylinder.

...pipe loss and panel gain reach equilibrium then it will tick over all day with no gain.


Two things from that. First, the movement of heat rather than of water should perhaps also be considered here - the slug of warm water in the Willis will 'lose' heat to the cooler water below and 'gain' heat from the hotter water above. The heat will still have gone in.

Second. The dynamic may be different if there was more hot than cool in the cylinder, and vice versa.

Whilst doing this I have cottoned on to a difference between Wookey's PHE and the Willis - the Willis heats a small, tallish tank of water (like a parallel cylinder) - whether that is significant I don't know but....

continuing from my first thought above ...."will 'lose' heat to the cooler water below and 'gain' heat from the hotter water above. The heat will still have gone in." 

.....this would leave the lower part of the internal heat exchanger in cooler water. As the internal heat exchanger arrangement has the panel flow going directly to this lower part, a heat differential is re-established allowing for the transfer of more heat so back to Diag 3. Etc.

It may be cleverer than thought, and certainly different in this respect from either a Wookey style thermosiphon PHE or a DPS style pumped one. I've gone from an enthusiast to a sceptic to open minded to intrigued........we need some quality field observations and, preferably, data.

Over to you.

Mick.

[Ivan]

Well what Wookey and I think is that the buoyancy of the column of lukewarm in the side arm can overcome the buoyancy of the water in the tank, and the warmer water will cascade into the top of the cylinder, and then fall further down inside the tank to its correct level.

I'd need to see some proof before I'd accept this is definitely what happens, however, what might alternatively be happening is zero thermosyphoning, followed by increased panel temperature, increased PHE temperature until sufficient temperature is reached to overcome buoyancy of tank.

The easiest way to visualise what is happening is.....make a tank with many sensors, put a side arm eg Willis Solasyphon. Put lots of sensors on sidearm. Plot the data (preferably as an image rather than as a graph) and watch what happens. Navitron are onto it - we've started constructing the hardware, and working on the software...

[djh]

Navitron are onto it - we've started constructing the hardware, and working on the software...

Good on yer! What the Danish (I think it was the Danish) did was to make the tank out of transparent material so they could see what went on. I forget whether it was plastic or glass. They had some highly sophisticated gadget to measure what was happening but I'd guess that injecting some dye would work.

[Ivan]

Wookey posted some very interesting papers which used dyes in transparent tanks but 1)transparent materials don't conduct heat like copper does 2)the dye won't tell you very much once it has dispersed, so it's a batch experiment rather than a process experiment.

I certainly wouldn't call ours perfect (far from it) - Wyleu and I were recently discussing how the multitude of sensors one on top of the other, and their copper pockets would perturb the system they are trying to measure, but unless we're writing a thesis on designing the experiment, we have to start somewhere.

What I like about this is that we are going to be looking at a dynamic system - as it heats up and cools down, as water is run off etc etc. Hopefully, fairly representative of what happens in real life.

[mick]

Ivan,

Sounds exactly what is needed - excellent plan.

Another thing you might investigate - fair bit of plumbing ahead! - is John Willis's positioning of the junction with the HW supply. He says 500mm or so above the top of the cylinder. Wookey has 120mm. Someone else  (sorry not to be keeping up!) with a side arm is introducing the solar hot via a sort of surrey flange set-up that goes some way down into the cylinder - maybe use the shower boss (or secondary return on an unvented) as the top junction.

Good luck to you and your team of co-conspiritors - can't wait for the results!

Mick.