B Why does hot water rise to the top of a water heater?

AI Thread Summary
Hot water rises to the top of a water heater because it expands when heated, becoming less dense than cooler water, which sinks to the bottom. The textbook used in a plumbing program confirms that this natural property of water allows for efficient circulation within the tank, ensuring a consistent supply of hot water. When heated from 50 to 100 degrees Fahrenheit, one gallon of water will occupy more than one gallon of space due to this expansion, provided there is room for it to do so. Water heaters typically include a pressure relief valve and an air gap to accommodate this expansion and prevent pressure buildup. Understanding these principles is essential for proper plumbing and water heater function.
  • #51
sevensages said:
Why would heating the water be too vigorous to allow stratification? Please explain this using simple language.
Stratification just means hot and cold (light and dense) separate. It's usually describing a static situation or the act of creating that static situation (they separate and stay separated, they don't circulate). Convection acts to disrupt stratification.
 
Physics news on Phys.org
  • #52
I am going to stand by what I've said - that sufficiently strong convection disrupts stratification - but admit an exception for a slow rate of heating that is incapable of making a strong overturning circulation. That would be unusual for most tank type hot water systems.

My thinking is based on observation of heating elements in open containers, pots of water on stoves, a basic understanding of convection... and reason.

A heating element (or coil or sleeve) will make an overturning circulation strong enough to break down any stratification and will mix the contents. This won't be true of a heating element near the top - but in practice almost all heating is done either near the bottom, sometimes midway and bottom, or up the sides or around a central fire tube (gas fired) and in ways that make an overturning circulation.

This is an advantage and desirable rather than a problem; the whole tank gets heated - maximising the amount of hot water it holds - and the result is it gets heated uniformly (or very close to it). Convection (strong enough) disrupts stratification. Stratification comes after the active heating and accompanying overturning circulation ceases and is (therefore) not produced by water rising from the heat source.

Call it surmise if people here insist but my reasoning says that subsequent stratification will be primarily from cold water downflows from imperfect insulation pooling in the bottom and - in a not-static situation - cold water inflows.

Stratification - or at least differentiation; a true thermocline will take time sitting still to develop - will happen in a "static" situation but I still surmise it will be due to cooling - around the fittings and tank body - from cooler water sinking and pooling, with little mixing. That will lack the flow rates necessary to make the full overturning convection loop that results in mixing. There is still convection but the upward flow will be in the form of the top of the pooled cool water rising as a layer. (Sort of - can nitpick about thermoclines and whether it is a layer).

I don't expect water of uniform temperature with perfect insulation to stratify - but am not so sure of that to claim it cannot. But I do expect (by reason) that heat conduction within still water will tend to maintain temperature homogeneity.

That differentiation/stratification in a hot water tank is useful, allowing influx of cold water in the lower part of the tank without mixing with and cooling the hotter water nearer the top and that gives longer times between heating cycles. (Ours is set up to run daytime only, to take advantage of rooftop solar).

@russ_watters re the air space - it is definitely there in our water heater tank - the outlet and relief valve are on the side, a few cm below the top - but thinking about it some more I don't know that it will persist over time rather than be displaced with water. I could be wrong that it does persist - which wouldn't be the first time I've got something wrong.

My thinking was it would moderate water hammer type stresses to the tank. I am not sure any of the materials used are especially subject to corrosion - but it does have a sacrificial anode. Ours is a one piece Chromagen heat pump model.
 
Last edited:
  • #53
Ken Fabian said:
And (duh!) cold water coming in near the bottom will be the main cold water source and will tend to stay there, without much mixing - it won't make the sustained convection cycling that a heating element does, that breaks down any stratification.
You haven't mentioned the Heat Source. At least a part of it should be very near the bottom . The nearer the better; it's part of the design of a good water heating system. You also need to consider how a hot water tank with a heat exchange coil works best. Iirc if the hot input of the coil is at the top the result will be 'very hot' water at the top level and water at the bottom would not be as hot. If hot water goes into the bottom then you would have stronger convection and a more uniform temperature throughout the water column. It's a matter of choice as to what you want; quicker start up with hot water at the top or plenty of warm water, right down to the bottom. This is more complicated than I imagined.
 
  • #54
sophiecentaur said:
You haven't mentioned the Heat Source.
Actually, yes I did; from my previous comment -
Ken Fabian said:
- in practice almost all heating is done either near the bottom, sometimes midway and bottom, or up the sides or around a central fire tube (gas fired) and in ways that make an overturning circulation.
My understanding of those (electric) heaters with midway AND bottom elements is the midway one is a "priority" heating element and will come on first in order to give the user hot water with less waiting; it doesn't make a full overturning circulation and won't mix in the coldest water at the bottom, just heat the upper portion. After the upper part of the tank is hot enough the bottom element switches on to complete the heating. (Not sure if the upper element remains on or switches off when the bottom heating element comes on.)
 
  • #55
Ken Fabian said:
Actually, yes I did; from my previous comment -
'Indirect ' heating of water in a tank is very common in some places (UK etc). The boiler heats water (same water that goes round the radiators) and passes through a spiral heat exchanger in a tank. That's not mentioned in your comments but is very relevant (in civilised parts of the world Lol ). US and UK argue all the time about domestic electrical wiring and I can see the same hot discussions about hot water, too.
I learned a lot from PF about that 'impossible' steam heating system in many US homes. It's said to be very suitable in some buildings and it's one of those comedic tropes when the noisy heating contributes to situations. It would scare me to death!
 
  • #56
sophiecentaur said:
'Indirect ' heating of water in a tank is very common in some places (UK etc). The boiler heats water (same water that goes round the radiators) and passes through a spiral heat exchanger in a tank.
Our heat pump hws has the heat exchange pipe wrapped around the tank (welded to the tank I think?) but could have been coiled inside it near the bottom or helically coiled from higher up down to bottom - or be a 'sleeve' - and I think those will be functionally equivalent to piped steam or very hot water from a boiler. Likewise an electric element too. I don't think those specific details will change anything; an overturning convection is desirable and easy to achieve.

The designers may prefer hot coming down a coiled heat exchanger pipe, working with rather than against the (separate) convection within the heating fluid - and some systems may be built to rely on that passive convection to cycle the fluid - which solar water heaters do. But if the fluid is pumped it can go either way and I think the flow direction won't make much difference to the heat exchange; once that overturning convection kicks in the water temperature tends to homogenise from mixing.
 
  • #57
Ken Fabian said:
Our heat pump hws has the heat exchange pipe wrapped around the tank (welded to the tank I think?) but could have been coiled inside it near the bottom or helically coiled from higher up down to bottom - or be a 'sleeve' - and I think those will be functionally equivalent to piped steam or very hot water from a boiler. Likewise an electric element too. I don't think those specific details will change anything; an overturning convection is desirable and easy to achieve.
Heat exchange and an electrical element are fundamentally different. For heat to pass from the flowing water in the coil to the water in the tank there has to be a temperature gradient, falling from coil input to output. If there were not, there would have been no heating effect. You can calculate the heat transferred by the drop in temperature of the circulating water and the rate of mass flow. If the pump rate is low then the exit water will be significantly cooler. If the flow rate is very high the ΔT will be low. In a domestic system, the flow rate will not be high so the ΔT will be significant.
An electric heater will have a small temperature gradient because of the ΔT across the pipe wall. .
 
  • #58
sophiecentaur said:
Heat exchange and an electrical element are fundamentally different. For heat to pass from the flowing water in the coil to the water in the tank there has to be a temperature gradient, falling from coil input to output. If there were not, there would have been no heating effect.
An electric heater will have a small temperature gradient because of the ΔT across the pipe wall. .
Or in another fundamental way, the heat in heat exchanger is carried as heat capacity of water inside the pipe and therefore temperature in an heat exchanger must fall along the direction of the flow. It must have lower temperature in the output than input - otherwise it is a cooler. Whereas in the electric heater, the energy is carried as electric field and converted into heat in the wire. Gradient is required between the wire and outside of the wire.
 
  • Like
Likes sophiecentaur
  • #59
Ken Fabian said:
But if the fluid is pumped it can go either way and I think the flow direction won't make much difference to the heat exchange;
But, as @snorkack says, the direction of flow of the input water will affect the power transferred. It's a common technique - even used in the humble Condenser Tumble drier. The exit (hot) air from the drum is passed over the (cold) inlet air from the room in the opposite direction on the other side of the heat exchanger. Condensation occurs all along the heat exchanger path but mostly where hot wet air is close to the coolest dry room air. Amazingly (at least in ours) the emerging air is 'dry' enough not to raise the humidity in the utility room appreciably; almost all the water ends up in the collecting tank.
 
  • #60
@sophiecentaur - I am not convinced it matters to whether it will make an overturning convection within the water tank or how well the water heats; yes there will be a difference in temperature between the heating fluid coil as it enters the water tank and as it leaves but it doesn't have to flow downwards to exchange heat. Typically the temperature of water to be heated will be colder than the coolest part (outlet) of the heating coil.

The reasons for making it a downward flow rather than the other direction will be about the convection and flows within the heating fluid side of the system, not the effectiveness of heat exchange within the water tank.

Our previous hot water system was a rooftop solar one, with hot water tank above the solar collectors. Heating fluid was glycol that flowed through a sleeve around the bottom third of the water tank and the flow of that heating fluid was entirely convection driven and that requires the inlet (hot) to be at the highest point of that sleeve. (Not a big height difference with the outlet - more horizontal flow than top to bottom.)

If it had a ground level hot water tank it could not rely on convection and would use a pump to circulate the heating fluid - and it could be made to flow either direction, but typically will be made to go with the convection flows rather than against them. The heating inside the water tank will be the same.
 
  • #61
Ken Fabian said:
The designers may prefer hot coming down a coiled heat exchanger pipe, working with rather than against the (separate) convection within the heating fluid
That configuration is the well-known Counterflow Heat Exchanger. Its advantage is that the hottest part of the hot side heats the hottest part of the cold side fluid.

If both fluids are flowing in the same direction thru the heat exchanger, then the cold fluid can typically get only as hot as the average of the hot and cold sources. (unless the hot side is grossly over-provisioned)

https://web.mit.edu/16.unified/www/FALL/thermodynamics/notes/node131.html

Cheers,
Tom
 
  • #62
Tom.G said:
That configuration is the well-known Counterflow Heat Exchanger. Its advantage is that the hottest part of the hot side heats the hottest part of the cold side fluid.

If both fluids are flowing in the same direction thru the heat exchanger, then the cold fluid can typically get only as hot as the average of the hot and cold sources. (unless the hot side is grossly over-provisioned)

https://web.mit.edu/16.unified/www/FALL/thermodynamics/notes/node131.html

Cheers,
Tom

A hot water system exchanges heat but a dedicated heat exchanger is designed differently.

Typically in a hot water system it is a tank - and the OP appears to be about hot water rising to the top in HW systems like that - and typically the hot side IS over-provisioned (and very quickly an overturning circulation is generated). Whilst that mixing is short of absolutely homogeneous there is very little vertical temperature differentiation left.

For most of the heating process the temperature difference between top and bottom within the tank is insignificant; the heating pipe will still have a temperature gradient between inlet and outlet but the water being heated will not be much hotter at the top than the bottom.
 
  • #63
Tom.G said:
That configuration is the well-known Counterflow Heat Exchanger. Its advantage is that the hottest part of the hot side heats the hottest part of the cold side fluid.

If both fluids are flowing in the same direction thru the heat exchanger, then the cold fluid can typically get only as hot as the average of the hot and cold sources. (unless the hot side is grossly over-provisioned)

https://web.mit.edu/16.unified/www/FALL/thermodynamics/notes/node131.html

Cheers,
Tom
The thing is, there is no flow in the cold side of the tank, just convection if any.
If you put an electric heating coil at the top of a cold water tank then the thin layer at the top of the tank will be heated to boiling while the bottom of the tank is heated by conduction only - very inefficiently.
The same will happen if you put a heat exchanger pipe on top only.
If you put a heat exchange pipe in a tank with inlet on top and outlet in bottom then the contents of the tank will be heated layer by layer from top down. There will be top layer of the tank where the heat exchange is inefficient because the tank contents are heated to almost the temperature of pipe inlet, middle layer where the active heat exchange occurs and bottom layer where the heat exchange is again inefficient because the pipe outlet has cooled to almost tank bottom temperature.
 
  • #64
Well, I seem to have been un-clear in my post about counterflow heat exchangers. (post #61, above. https://www.physicsforums.com/posts/7250242)

I was referring to post #60 by @Ken Fabian where he describes that exact configuration, hot heat exchanger inlet towards the top of the water tank and cold exchanger return closer to the bottom of the tank.
Ken Fabian said:
Our previous hot water system was a rooftop solar one, with hot water tank above the solar collectors. Heating fluid was glycol that flowed through a sleeve around the bottom third of the water tank and the flow of that heating fluid was entirely convection driven and that requires the inlet (hot) to be at the highest point of that sleeve. (Not a big height difference with the outlet - more horizontal flow than top to bottom.)
I was pointing out that the configuration he describes is the most efficient/effective for heating with a heat exchanger.

If you care to go thru the math, the link I provided in post #61 has the details.
(https://web.mit.edu/16.unified/www/FALL/thermodynamics/notes/node131.html)

I hope this helps clear up any ambiguity.

Cheers,
Tom
 
  • #65
@Tom.G - I still maintain it makes no significant difference to the efficiency of heat exchange; it is done like that to enable passive (no pump) convection driven circulation within the heat source side of the system.

Without a pump you would impede that circulation if the hottest point of the exchanger piping is at the bottom instead of the top. With a pumped system (solar HWS' with water tank below solar heat collectors use pumps) it could run in reverse - that convection is weak compared to a suitable pump.

The reason I think it doesn't matter much to the efficiency of heat exchange is convection driven overturning circulation in the water tank mixes the water and eliminates any vertical temperature gradient. The hottest part of the exchange pipe (where it enters the tank) is in contact with water of the same temperature as the coldest part (where it leaves the tank).

One more time - overturning convection mixes the water in the tank and eliminates a temperature gradient.

We can quibble about initial conditions - before that overturning circulation mixes the tank contents - and end state conditions where the contents approach the same temperature as the heating pipe. Or quibble that the temperature is not absolutely homogeneous even when that overturning circulation is dominant and quibble that the rising hotter water from a lower coil will flow past higher coils. I just don't think that will make significant difference to the heat exchange.

I have tried to be clear in my language that I am talking about most of time that heating is taking place and the difference between strong heating that will drive mixing and weak heating that will not.
 
  • #66
snorkack said:
The thing is, there is no flow in the cold side of the tank, just convection if any.
Draw a sketch of the convectional flow in the tank. For every gram of water that flows upwards there's a gram flowing downwards. Up past past the coil and down the outside skin; there's your flow. (In the case of the coil being outside , the flow is still upwards for the heated water by the input water).

You could ask why the tanks are always connected that way. Could it just be general ignorance?

Remember the old 'gravity systems? Hot water flowed up from the back boiler and dropped down through the hot tank; there was no way you could persuade it to ignore gravity and flow the other way. If you connected it so that hot water flowed into the bottom of the coil you could end up with a plug of hot water with nowhere to go. (Stratification?)
 
  • #67
sevensages said:
Why would heating the water be too vigorous to allow stratification? Please explain this using simple language.
You can only get stratification when there's no horizontal gradient of temperature. There will always be some gradient from the heating coil and the skin of the tank so the outer (virtual) column will drop
 
Back
Top