Reverse convection flow in wood-burner

[cbell]

Last December we fitted a fully boilered wood-burner to our central heating system. It's a really simple setup:

- Wood-burner heats store (210 litre direct copper cylinder) in roof by gravity convection.

- Thermostat on heat store says "I'm hot" when it reaches about 55 deg C.

- When the (fully pumped) CH system wants heat it opens the zone valve to the heat store if this is hot, and draws water from there; otherwise it opens the valve to the boiler, fires this up, and takes it from there.

It works very well except from one unexpected problem:

If the heat store is quite hot (say 70+ deg C), and there is no demand on it, and the wood-burner dies right down so that its boiler temperature is less than that of the heat store, then convection stops. So far, so good.

However if you load up the wood-burner so that convection starts again it can draw down water from the heat store that is actually hotter than that in the boiler, resulting in the flow direction reversing. Once this happens it seems to be impossible to stop until the wood-burner has gone out and the temperature of the pipework on both flow and return has reverted to ambient.

I think I've diagnosed the problem correctly since I have a dual channel digital thermometer connected to flow and return, and can measure their temperatures accurately. This, combined with the bumping noises as the boiling water can't escape from the wood-burner, makes me pretty confident that this is what is going on.


This has only happened twice and I can manage the problem by not letting the fire die right back, or by drawing some heat from the store (so that the temperature at its base drops) before refuelling a luke-warm stove. However I'd like to prevent this happening, and the only solution I can think of is to add a pump to the gravity circulation circuit between wood-burner and heat store which could be switched on manually to restore the "correct" flow direction.

However this is a bit of contradiction in terms: gravity circuits are - by definition - not pumped, and I'd like to preserve the gravity circulation aspect of the system for safety's sake in the event of a power or pump failure. So, finally, here is the question:

Is there an accepted way of fitting a part-time pump to a gravity system? I can imagine two solutions: one being a pump with a low resistance to flow simply plumbed in series, the other being to insert a section with a low resistance non-return valve, and run the pump in parallel with this.

All ideas welcome!

 

[cider]

Firstly you must not add any type of valve to a gravity / solid fuel system.
If you do choose to add a pump to a solid fuel circuit it should be ran from a pipe stat to switch on and off and prevent chilling of heat exchanger. However better prevention then cure! so personally I think the best way to sort your problem would be to raise the flow from the gravity circuit above the heat store as high as possible (with an open vent over header tank) and then drop back down to the flow connection to the store.
Also if possible you could drop the return below boiler height and then back up again.
Both above will encourage and assist natural gravity circulation.

 

[blondini]

Maybe the problem is due to incorrect system design. The heat store can only absorb so much heat if there is no demand on it. With no demand and the wood burner is still producing heat when the heat store is full, there will be nowhere for the excess heat to be absorbed or dissipated. Once the heat store gets as hot as the boiler, gravity circulation will fail and it'll start to boil. Pumping round the same circuit won't help, the system should be designed with a heat leak...

 

[cozzmic]

He is almost certainly on about a Gledhill thermal store and yes they do have a pump on the "gravity" circuit. Saw one recently but there is a wiring issue as the diagram that came from Gledhill was a hand written scrawl. Even the sparky couldn't figure it out and was clashing with the plumber over how it operates.

 

[Onetap]

I think I've diagnosed the problem correctly since I have a dual channel digital thermometer connected to flow and return, and can measure their temperatures accurately. This, combined with the bumping noises as the boiling water can't escape from the wood-burner, makes me pretty confident that this is what is going on.

I think your diagnosis is wrong. You will not get gravity circulation under the circumstances you describe.

First, you can't transfer heat from the stove to the store when the store is hotter.

Secondly, the gravity circulation is driven by the different densities of the hot flow and cold return water. You won't get hotter, less dense water descending and displacing colder, denser water upwards.

There is inadequate flow through the boiler and the bumping noises are localized boiling.

Why restart the boiler again when the heat store is full of heat?

 

[Agile]

I can envisage a situation where reverse gravity circulation could occur.

However he does not explain how the store gets to 70° when the fire is shut down when the store is at 55°.

Nor does he explain why he is starting a fire when the store is at 70° or how this 55° interlock has apparently been lifted!

Tony

 

[oilman]

I agree with onetap, diagnosis is wrong. A digital instrument is no more help than a couple of pipe thermometers with dials on, and may well serve to confuse.

He is almost certainly on about a Gledhill thermal store and yes they do have a pump on the "gravity" circuit. Saw one recently but there is a wiring issue as the diagram that came from Gledhill was a hand written scrawl.

Where is any reference to a Gledhill store?

Even the sparky couldn't figure it out .....

They rarely can.

.......and was clashing with the plumber over how it operates.

Blind leading the blind?

 

[cbell]

Firstly thanks to one and all for their responses, taking them in turn:

<From Cider> The stove is on the ground floor, and the heat store is in the roof, 2 floors higher, with its outlet about 6" under the tiles, so I can't increase the height. Also the floor is solid (onto rock!) so I can't go down. The pipe runs are all 28mm, with the straightest route we could manage, so I don't think there is much we can do on that front.

I take your point about chilling of the heat exchanger, but with an un-pumped solid fuel system the *only* variables you have are heat input (fuel) and output (demand). The former is me with the log basket, and the latter is managed by the "I'm hot" thermostat on the heat store, which switches off at about 50 deg C.

Were I to add a pump it would indeed be connected to a pipe stat on the stove outlet, probably set to around 70 deg C to improve circulation into an already hot store. (There is some circuitry to limit store temperature, see the next section.) However I would also want the manual override I mentioned to correct this reverse flow.


<From Blondini> I didn't describe the control system in detail, since I thought it was irrelevant to this problem.

The system also has an "overheat" thermostat at the base of the store, which comes on at about 70 deg C. This pumps unconditionally - ignoring timers and thermostats - round the selected CH circuits (there are 5 zones); and in practice it brings the temperature down very quickly since it can extract heat far faster than the stove can deliver it.

My original design included a heat leak radiator for safety's sake in the event of a power failure. But in the end we omitted this because:

- the stove dies down very fast if all air is shut off, reducing the heat input to not much more than the losses inherent in the system.

- a 210 litre store can absorb a *lot* of heat (17kWh from 20 to 90 deg C), so there is a big margin of safety in the system anyway.

- in the event of a protracted power cut we have a generator, and we would use that to pump water and keep the house warm!

If anyone is interested I can post more about the control system and heat calculations, but in summary the wood-burner has a max output of 20kW, of which 3.8kW goes to the room and the rest to the water. It's quite a big house and the CH system could easily absorb twice that amount of heat.

Run absolutely flat out it will raise the temperature of the water by 33 deg C, and will start to boil in the roof boiler under these conditions if the inlet temperature is above about 55 deg C - not surprising really!

However a wood-burner is managed manually, and you would have to be a loony to bring about these conditions deliberately since there are lots of warning noises from the boilers. I only did it (once) to test the overheat control system and to find the absolute limits of the setup. In practice shutting down all the air inputs damps down the fire within seconds, and everyday operation is completely drama-free.

Also, if push comes to shove, it's an open system so there is no possible danger of explosion. Yes, I know about the dangers of the expansion / feed tank melting (that baby was in Taunton, not far from here), but I've tested for reverse flow into there and - after a spot of replumbing - that categorically is not happening.


<cozzmic> mentions a Gledhill thermal store. I've never heard of this, but I'll check it out. However it sounds pretty amateur...


<Onetap> Disagrees with my diagnosis and, in theory, I'd agree with him.
However you *can* draw hot water down the return pipe as follows.

Flow and return initially have the same temp, ie no circulation current. Add heat in stove to start normal circulation, which will draw hot(ter) water from base of heat store into return pipe to stove. It gets about 1/2 way down, at which point its lighter density halts the circulation, and if it "overshot" slightly the circulation direction may reverse a little. In a wood-burner with side and roof boilers it is the side ones which deliver most heat near their bottoms, close to the grate, so even a small reversal of flow will tend to draw hot water from the base of the boilers back up the return pipe, reinforcing this reversed flow direction. Once the reverse flow is established it does not correct itself until the temperatures in the flow and return pipes cool down again and equalise.

Or at least that's my theory - the reverse flow is *definitely* happening even if the explanation is wrong!


<Agile> Asks how the store can be hotter than the stove. Simple: fire heats store, fire dies down. Any residual heat rises into store which is heavily lagged, fire cools down and is unlagged (also heating room), therefore it becomes cooler quite easily.

In fact I rely on this behaviour overnight: I heat the store last thing, letting the stove go out. The store then kicks off the heating the next morning until it runs out of energy, at which point the control circuit switches over to the oil-fired boiler.

As to why you would blaze up a cool fire when the store is hot. I don't currently have a remote indication of store temperature beyond that it is "hot" (meaning 50+ deg C). Also the stove heats the room it is in directly. Therefore a mixture of lack of detailed knowledge about the store's state, plus a desire to heat the room directly, leads to this situation. As I said in my original post I can (if I concentrate) manage the situation so that the problem doesn't arise.


<Onetap and Oilman> think my thermometer is misleading me. Well, it has two remote thermocouples attached tightly to the return and flow pipes, and normally registers very accurately how the system is behaving. This is confirmed by the behaviour of the (independent) thermostats.

In addition the fall-back test of touching the pipes confirms the diagnosis: when reverse flow occurs the return pipe at 70 deg C is too hot to touch, whereas the flow at 50 deg C is hot but touchable.

I may be stupid but, in this respect at least, I don't think I'm confused!



The conclusion so far seems to be that there is no accepted method of adding a pump to a gravity system. I was hoping that someone would chip in with "there's a special pump which has a low resistance to flow when not running, and doesn't impede gravity flow"; but maybe not.

Thanks for the answers, and any further input would be welcome.

 

[cider]

OK then, how about fitting pump on a bypass with a reverse acting (motor closed) diverter valve. When the pump has to run (manually or via stat) the valve motor closes, end switch starts pump and circulation correceted. If power cut or when not needed, valve de energised and opens and pump stops.
So in other words 28mm tee, branch then 90 elbow up to pump (with valves) tee back into flow, 28mm reverse acting zone valve in between tees or fit it on return if easier

 

[Agile]

You can have a pump as long as its not between the fire and the vent!

It can draw off heated water and you can pump it anywhere!

I can envisage a pump on a bypass with swept tees which would leave the 28 mm unobstructed for gravity flow but with the pump activated would produce a positive pumping action as a result of the orientation of the swept parts of the tee. Those with drawing experiene could produce a drawing.

Not Colin ex-RD are you?

Tony

 

[cider]

Or make up an oversized injector tee where the pump joins back in, that also would save having the zone valve

 

[cbell]

Cider wrote:

OK then, how about fitting pump on a bypass with a reverse acting (motor closed) diverter valve.

I thought about this, but my long-term fear is that the valve will eventually stick partially or fully closed. Also any sort of valve is going to be an impediment to gravity flow. But it's a possibility.


Agile wrote:

I can envisage a pump on a bypass with swept tees which would leave the 28 mm unobstructed for gravity flow but with the pump activated would produce a positive pumping action as a result of the orientation of the swept parts of the tee.

I think I understand this (sorry for my ignorance, I'm an engineer not a plumber). Is a swept tee a larger diameter pipe in which a smaller internal diameter nozzle points in the wanted flow direction? It sounds ideal: no impediment to gravity flow, and a boost from the pump when required.

And no, I'm not "Colin ex-RD". What's "RD"?


Cider wrote:

Or make up an oversized injector tee where the pump joins back in

I think this is essentially the same idea.


Can either of you point me to anywhere where I can find out more about injector tees? It sounds like the right answer, so thanks guys.

 

[oilman]

Can either of you point me to anywhere where I can find out more about injector tees? It sounds like the right answer, so thanks guys.

Try here near bottom of page.

 

[cbell]

Oilman wrote

Try here near bottom of page.

Thanks, that's more or less what I thought it would be. I think I can design and make something which will do that - probably open the 28mm flow out to 35mm around the Tee to preserve area, and inject with 15mm or so from the pump, taking the pump inlet from further back along the pipe.

The forwards velocity from the injection nozzle should stop the flow just circulating around the loop ... I hope. In fact I think I feel some experimentation with a plastic bottle, transparent pipe and inky water coming on before I cut any copper!

Injection tee layout

• cbell
• 16 Feb 2009
• 1

Sorry, rubbish image, but this is what I have in mind. Black is existing 28mm gravity circuit, red is new probably 15mm to increase speed.

 

[Onetap]

I still say that it is not possible for reverse circulation to occur as you describe. The gravity circulation is driven by the diference in the pressures and the pressure difference is created by the temperature differences.

If reverse ciculation, is occuring it is more likely to be caused by a pressure difference generated by the pump on the heating system.