Phase-Change Material based coolants

[cprobertson1]

Good morning folks,

   Ran into an interesting idea this morning: I have several projects on the go just now, all in need of cooling to varying extents - everything from simple stepper motors with a heat-sink strapped to the casing, to oil-immersed resistors used in a 1kW dummy load.

   Air-cooling is all fine and dandy for most applications: push enough air across a large enough heatsink and you'll be fine for a good while.

   On the other hand we have my liquid-cooled 1kW RF dummy load: or "some resistors in a paint tin immersed in mineral oil - the project". While filling it with oil this morning I started wondering what would happen if I used solid paraffin wax as a coolant, relying on the phase change from solid > liquid to absorb energy but keep the temperature the same.

   Has anybody here used phase-change materials (PCMs) as a coolant before? I've seen some papers on it but I've not really investigated much (yet!)

   I shall be running various tests in the near future comparing the relative merits of solid paraffin, paraffin-oil emulsions and oil and will post the data here.

   I'll also post some figures here later in the day with a proposed test-setup and some ballpark numbers It's been a while since I've done any real SCIENCE^TM!!!

--edit-- NOTE: paraffin is a poor choice in this case as it is flammable and will boil if overheated - if this is done in a sealed system it can easily explode, covering the surroundings in molten paraffin: and depending on what pressure it was at when it burst, it may even cause a Boiling Liquid Expanding Vapour Explosion (BLEVE) which will ruin anybody's day.

 I will be addressing this and other safety matters, as well as alternative materials in the followup posts

[mikeselectricstuff]

This firefighters' thermal imaging camera uses a phase-change material for cooling.

[cprobertson1]

Draft procedure.

The below is a draft of the test procedure I intend to use to quantify the various materials for cooling applications. Procedure A aims at getting a rough idea of how the material responds, while Procedure B aims at actually quantifying the heat-input against the rise in temperature of the coolant - but will take longer and is unlikely to reach as-high a temperature as Procedure A (as I will be heating it with an immersed nichrome element and a constant current source rather than a 1.5kW hot-plate )

Note this is just a draft and is liable to change.

Intended test-mixtures are:
-deionised water*
-mineral oil (engine coolant)
-paraffin wax/mineral oil emulsion (25%w/v)**
-paraffin wax/mineral oil emulsion (50%w/v)**
-paraffin wax/mineral oil emulsion (75%w/v)**
-paraffin wax (solid)**

*possible electrocution hazard: special handling precautions required
**flammable: special handling precautions are required

Special handling requirements will be specified in final procedure

Aim:
To quantify the differences in efficacy of various materials in passive cooling: specifically the differences between oils and PCMs (Phase-Change Materials).
 
Hypothesis:
A conventional oil-based coolant will increase linearly with constant thermal input: while a PCM will increase linearly to the phase transition point and then continue to absorb heat while not actually increasing in temperature until the phase change is complete, allowing for more effective temperature control over certain ranges or greater total thermal capacity than an equivalent volume of conventional coolant.
 
Procedure
The experiment will be arranged into two sections:
 
Procedure A:
1.       A beaker will be coated with aluminium foil and 100g of material placed therein. A logging-thermometer and electrical stirring rod will be installed to maintain homogeneity of the mixture.
2.       The beaker and contents will be heated on a hot-plate at low heat until equilibrium is reached, or a temperature of 250^oC is reached. The experiment must be halted if the boiling temperature is reached.
3.       Temperature measurements shall be recorded electronically at 1-second intervals.
4.       Thermometer and stirring rod will be removed and the contents allowed to cool before disposal.
5.       Procedure will be repeated with alternative materials.
 
 
Procedure B:
1.       A beaker will be coated with aluminium foil and 100g of material placed therein. A logging-thermometer and electrical stirring rod will be installed to maintain homogeneity of the mixture.
2.       A controlled heating system consisting of a constant-current source and a coil of nichrome wire will be used to heat the mixture until equilibrium is reached, or a temperature of 250oC is reached.
3.       Entire apparatus will be placed in insulating material (ceramic wool) to reduce heat loss to the environment.
4.       Temperature measurements shall be recorded electronically at 1-second intervals.
5.       Thermometer, stirring rod and heating unit will be removed and the contents allowed to cool before disposal.
6.       Procedure will be repeated with alternative materials.
 

 
Expected results:

Procedure A results:
As the heat input is unknown but constant within definable error limits, a linear rise in temperature is expected in single-phase coolants with a plateau occurring once equilibrium is reached between heat-loss and heat-input. This plateau region may not be reached due to the experimental constraints on temperature.

Conversely, PCMs will show a small plateau during phase-transition, with little to no temperature increase (provided head distribution is homogenous across the test-material: which is unlikely.)

 
Procedure B results:
The results will be superficially similar to procedure A: however the heat-input will be known, allowing for finer control over the speed of heating and plotting of temperature vs energy input.

[salbayeng]

Some decades ago, we made the ultimate phase-change cooling system.
This was used to keep the electronics cool inside a sonde we dropped into a blast furnace, and was designed to last 8hrs at 1200C exterior temperature.

So inside it we had the electronics package, a PCB and 4 x AA cells, this was encapsulated in paraffin wax to make a 100mm ball. Then we had two hemisphere shells made of 1/8" steel, and 400mm diameter approx, so we mixed a real sloppy refractory insulation and poured into one half and positioned the paraffin ball in the middle, and quickly slopped the other hemisphere on top, filled with cement. A few quick welds around the edges and then left to sit in a tub of water.  Stainless steel rope antennae poked through ceramic spacers through the shell, so they looked like anti-submarine mines.

 So what happens is the water boils out of the refactory material, and as the steam goes outward, it pushes back the heat flux trying to get in. The refractory as it drys provides good thermal insulation as well. The paraffin mostly keeps water vapor out, the phase change effect is not as much as water.

We did a couple of test runs in a lab furnace, and actually got the 8hrs target, and the thermocouple readings were a close match to what we had modelled with an extremely primitive computer program.

We never got to use it in a real blast furnace, as we had all manner of difficulty trying to get the receive antennae through the gaps in the staves of the blast furnace walls .

[wolf32d]

Some useful

Specific latent FUSION heat of paraffin: 200 kj /Kg
Specific latent EVAPORATION heat of water 2200 kj/kg (which is huge compared to many other substances)

If keeping the heating elements at a constant 100*C temperature wasn't a problem I'd have a go with water:
Each kWh (=3600 kj) boils away just 1.6 liters of ultra-cheap tap water!

No toxic fumes, no fire hazard, just some kettle-grade insulation  (and maybe a couple of dehumidifiers or a condenser).

[cprobertson1]

Some useful

Specific latent FUSION heat of paraffin: 200 kj /Kg
Specific latent EVAPORATION heat of water 2200 kj/kg (which is huge compared to many other substances)

If keeping the heating elements at a constant 100*C temperature wasn't a problem I'd have a go with water:
Each kWh boils away just 1.6 liters of ultra-cheap tap water!

No toxic fumes, no fire hazard, just some kettle-grade insulation  .

Well that is actually the basis behind reflux heating in the lab: you boil a solvent (containing your reaction mixture): but because it's boiling the temperature remains constant. The solvent is then reclaimed with a condenser and drips back into the reaction mixture.

As I mentioned, the idea came to me when I was making a sealed 1kW RF dummy load and I had no mineral oil to hand - but I did have paraffin. I quickly said "um, no, that's going to cause a fire" and went for a walk to the shops to get some engine oil and I started wondering how the phase-change would affect temperature rise.

Obviously a PCM-based coolant would be for a niche application: such as an entirely sealed system or a system where you don't have a ready supply of coolant like air or water: or where you want to do something peculiar like bathing the entire board in the coolant, in which case regular tapwater, containing ions, will be slightly conductive and could cause various problems from corrosion to shorting the board out.

Tbh, I'm just experimenting for the fun of it, I am (literally) a scientist after all (though the last thing I actually scienced was a jar of diseases (that's a technical term, honest!))

 Conventional coolants for conventional situations: weird coolants for weird situations!

8hrs at 1200C exterior temperature

Oooft! Just a wee bit on the hot side then

Now that sounds like a fun project!


---
It just occured to me that procedure A is just measuring the melting point (in a very poor manner). It makes the assumption that heat input is constant within a certain range: meaning you can get a vague idea of the thermal capacity from heating it for a certain time and measuring the temperature rise.

I'll probably revise procedure B to use a TO-220-style resistor with a heatsink screwed to it and a second thermocouple mounted right beside the interface between them to try and get an estimate of the interface temperature - that way I'm measuring how much a part heats up in coolant - rather than how much the coolant heats up when the part heats up

[cprobertson1]

I've revised the procedure and will proceed at the weekend.

Going to use a thick film (or maybe a ceramic wirewound resistor) strapped to a heatsink with an attached thermocouple to measure the temperature of the part itself in order to estimate the efficacy of the coolants.

I'll the resistor from the bench supply in serial mode which can deliver up to up to 6A at 30V (Or 3A at 60V in parallel mode: either way that's 180W That doesn't seem like a lot when it comes to heating actually: we shall see if it's enough!)

I'll be doing this at the weekend once I get my boiler put in so that the ambient temperature isn't 5^oC (no seriously, there was frost on the INSIDE of the double glazing earlier in the week! Who'd have thought the world would seem so backwards when your boiler dies!)

--EDIT--

Aim:
To quantify the differences in efficacy of various materials in passive cooling: specifically the differences between oils and PCMs (Phase-Change Materials).


Hypothesis:
A conventional oil-based coolant will increase linearly with constant thermal input: while a PCM will increase linearly to the phase transition point and then continue to absorb heat while not actually increasing in temperature until the phase change is complete, allowing for a higher total thermal capacity than an equivalent volume of liquid-phase material.


Procedure:

Preparation:
Wax/Solid Coolant:
1.   Place borosilicate beaker on soapstone plate or hotplate as per procedure
2.   Place thermocouples as per Figure 1
3.   Weigh 100g of test material into test setup using additive measuring.
4.   Heat mixture to liquid while stirring to ensure homogeneity
5.   Allow to cool to room temperature
6.   Continue as per procedure

Oil/Liquid Coolant:
1.   Place borosilicate beaker on soapstone plate or hotplate as per procedure
2.   Place thermocouples as per Figure 1
3.   Weigh 100g of test material into test setup using additive measuring.
4.   Allow to cool to room temperature
5.   Continue as per procedure

Emulsion Coolant:
1.   Place borosilicate beaker on hotplate as per procedure
2.   Weigh fluid component (oil) of test mixture as per %w/w of intended emulsion using additive measurement: +10%
3.   Weigh solid component (wax) of test mixture as per %w/w of intended emulsion using additive measurement: +10%
4.   Ensure the mass of added material is 110g
5.   Heat fluid to liquid while stirring to ensure homogeneity
6.   Place borosilicate beaker on soapstone plate or hotplate as per procedure
7.   Place thermocouples as per Figure 1
8.   Weigh 100g of heated test material into test setup using additive measuring.
9.   Allow to cool to room temperature while stirring to ensure homogeneity
10.   Continue as per procedure


Procedure A: aims to measure a temperature profile for the material under test. 100g of material will be placed in a beaker and heated via hotplate. A thermocouple will be used to measure the temperature near the centre of the beaker during heat transfer, and a second thermocouple will be used to estimate the temperature of the hotplate.
1.   Carry out preparation specific to test material
2.   Ensure ambient temperature of mixture
3.   Begin temperature logging
4.   Begin heating mixture on medium heat using hotplate
5.   Continue heating until a temperature of 250^oC is reached OR the mixture begins to boil
6.   Remove heating and allow to cool to ambient temperature


Procedure B: aims to measure the efficacy of the material under test in dissipating heat from a part. 100g of material will be placed in a beaker and heated via a heatsink-coupled thick-film resistor in a TO-220 style package. A thermocouple will be used to measure the temperature of the part while a second thermocouple is used to estimate the temperature of the coolant. A constant current source will be used to provide a constant power input to the solution.
1.   Carry out preparation specific to test material
2.   Ensure ambient temperature of mixture
3.   Place non-flammable insulating wool (ceramic) over test setup
4.   Begin temperature logging
5.   Set voltage/current source
6.   Begin voltage/current logging
7.   Apply power to resistive element
8.   Continue heating until a temperature of 250^oC is reached OR the mixture begins to boil
9.   Remove heating and allow to cool to ambient temperature


Interpretation of results:
Procedure A Results:
Plot temperature (mixture) and temperature (hotplate) against time: estimate thermal capacity and note any transition points.

Procedure B Results:
Plot temperature (mixture) and temperature (heating element) against time: estimate temperature rise against input power.

[tautech]

Check out Transformer oils.
While their redeeming feature is dielectric insulation they do have thermal transfer properties too.

[cprobertson1]

Check out Transformer oils.
While their redeeming feature is dielectric insulation they do have thermal transfer properties too.

Not sure I can afford much just for the purpose of testing (I have no "use" of any of these materials - I'm just interested to see how they fare against each other!) - the one thing I actually need a coolant for I resorted to mineral-engine-oil, which *can* be electrically conductive depending on the impurities present; in my case my multimeter reads >8 Megohm  when I try to measure it's resistance across a 1cm cuvette)

As I mentioned, there will be niche applications for these sorts of things: somebody out there will want a coolant that lingers at a certain temperature during heating or something peculiar that I've not thought of

I would quite happily say that fluid coolants including air/oils would suffice for pretty much every application I can think of - it's the applications I can't think of that are interesting

Anyway, will report results this weekend hopefully

[IanB]

This is a long thread, too long to read, but the most important thing to think about is how long do you want the cooling effect to last?

For instance, if you rely on melting to provide cooling, then when the coolant has all melted the cooling effect will cease. So you have a definite time limit to consider. (Compare with putting ice packs in a picnic cooler, or even dry ice. It won't last forever.)

Secondly, you need to consider the effectiveness of heat transfer to the cooling medium. If everything is standing still such as with melting solid, the heat transfer will not be very good: the part being heated will get hot and the cooling medium will remain cool.

To obtain good heat transfer, effective systems make use of flowing materials, for example stirring or circulating. If you want to use phase change, then boiling is good as it produces very good heat transfer. You can circulate liquid refrigerant into the system, let it boil to absorb the heat, then pipe it away to be cooled and condensed back to a liquid.

[cprobertson1]

That's the results for 3x runs of paraffin with the figures averaged at ~220^oC on a hot plate as per the aforementioned procedure.

The approximate rise in temperature is 1 degree per minute: but varies depending on, well, everything, it's an open system

You can see the transition point between solid and liquid phases quite clearly on the chart (which I'm ashamed to say was made on excel: my minitab license has expired now that I'm no longer at uni )

Put simply, the cooling effect is "extended" by the length of the transition point (which lasted nearly ten minutes in the paraffin example) - which is in line with what was expected.

Paraffin has a specific heat capacity of 3260 J/kG.K while (pure) water has a specific heat of 4182 J/kG.K - which of course makes water a much better coolant (until it becomes conductive from ions dissolving in it)

Mineral oil has a specific heat of 2170 J/kG.K: so combining it with paraffin will result in a drop in overall melting point, boiling point elevation, and a specific heat capacity between 2170 and 3260 J/kG.K. Using that you should be able to tailor a mixture to suit a particular transition point.

This is a long thread, too long to read, but the most important thing to think about is how long do you want the cooling effect to last?

For instance, if you rely on melting to provide cooling, then when the coolant has all melted the cooling effect will cease. So you have a definite time limit to consider. (Compare with putting ice packs in a picnic cooler, or even dry ice. It won't last forever.)

Secondly, you need to consider the effectiveness of heat transfer to the cooling medium. If everything is standing still such as with melting solid, the heat transfer will not be very good: the part being heated will get hot and the cooling medium will remain cool.

To obtain good heat transfer, effective systems make use of flowing materials, for example stirring or circulating. If you want to use phase change, then boiling is good as it produces very good heat transfer. You can circulate liquid refrigerant into the system, let it boil to absorb the heat, then pipe it away to be cooled and condensed back to a liquid.

It really is getting long!

I'll sum it all up in an impromptu paper when I'm done

Long story short - a PCM based coolant may have a niche application somewhere: but for almost every other application an conventional coolant will suffice (i.e forcing air or water over a heatsink)

Anyway, got to go, will post more results once I've got them!

[T3sl4co1l]

Hmm, nearly eutectic, neat. ^

Mind that paraffin also expands considerably when heated.  Automotive thermostats use this expansion to actuate a pin which controls the flow of coolant through the radiator.

Melting phase change doesn't save you anything on power dissipation, only energy.  So I don't see any point for most electronic applications.

Evaporating phase change can do wonderful things, but you need a carefully sealed device to harness it, and some means of recirculating the condensed liquid.  More commonly known as heat pipes, or with a pump to force the fluid: refrigerators.

Tim

[cprobertson1]

Hmm, nearly eutectic, neat. ^

Mind that paraffin also expands considerably when heated.  Automotive thermostats use this expansion to actuate a pin which controls the flow of coolant through the radiator.

Melting phase change doesn't save you anything on power dissipation, only energy.  So I don't see any point for most electronic applications.

Evaporating phase change can do wonderful things, but you need a carefully sealed device to harness it, and some means of recirculating the condensed liquid.  More commonly known as heat pipes, or with a pump to force the fluid: refrigerators.

Tim

Aye, the paraffin increased in volume by a good few mm during heating: I didn't take exact recordings though. It was most obvious looking at the cooled paraffin afterwards, the surface was nearly concave due to the contraction as it cooled.

I was actually impressed at how little the temperature changed during the transition period - you can see it rises slightly across the entire transition period, followed by a sudden (but small) rise in temperature immediately after that: I believe this was due to the mixture not being homogeneous - with it melting at the bottom of the beaker, next to the thermocouple rather than equally throughout the mixture.

Mind that's an average of three repetitions of the experiment - looking at the raw data it seems to still be across about a 2-degree range though, which is pretty good - if I recall correctly is suggests there aren't too many impurities in the paraffin - BUT I'd be inclined to err on the side of caution and assume it's an environmental factor with the experimental setup.

Anyway, I'll probably get the other graphs up later in the week - or next week as I have visitors this weekend. Will repeat with paraffin as well to see if there are any differences.

I'll probably be fairly quiet post-wise until I get all the data together, and then I'll summarise in an impromptu paper. xD

[SeanB]

Well, this was used on the Apollo lunar rovers to cool the electronics, as there would be plenty of non run time for the system to cool, but it would get too hot running, and they had no real way to radiate the heat away in the volume envelope, so the phase change with a very pure wax was used.

Just remember that this transition temperature will vary between each batch of wax you buy, as it is not a controlled part of the production, but depends on the particular refinery incoming blend at the time they drew that blend off. Might only vary a few degrees either way, but might be important.

If you want a cheap insulating oil get some refrigerant mineral oil ( must be mineral oil, not POE or PAG  synthetics, as there react with water and are conductive) and use it.

[jwm_]

Heat pipes use the gas-liquid phase change of a coolant in order to do their magic.  The coolent condenses on the cold end and flows as liquid down to the hot side where it evaporates and the cycle repeats. They are extremely good at transfering heat relative to solid copper or plain convection. They can be found in higher end CPU coolers among other things.

[salbayeng]

The gentle slope you observe over 2deg is called a "glide" by refrigeration engineers, it happens with refrigerant mixtures which are not eutectic (same happens with non eutectic solder, that's why 63/37 is the best to solder with). In your case your "paraffin wax" is a mixture of different molecular weight hydrocarbons, after all it just comes out of a tap on a fractionating column, so you will have an assortment of hydrocarbons whose boiling point (under vacuum) is spread over maybe 2degrees.
Also we have used vegetable oil to fill submersible sensor probes that are lowered several km under the surface, it is necessary so the pressure equalises inside and out. The geophysicists who did this bought a dozen or so different oils and sent samples to a lab to pick the best. The more refined the oil, the better it will be, in general if it is clear and tasteless it probably doesn't have many other impurities either.
If you are bothered by trace moisture content, toss in some molecular sieve granules, any residual water will diffuse into these (heating the oil to 160C will also boil out any trace moisture). Another critical aspect of oils is the acid content, refrigerant oils(especially) and motor oils have been processed to remove acid content.
Also note the valve radio power transformers of 5 decades ago were impregnated with beeswax, these operated with 600v centretapped secondaries, and operated perfectly well with all the contaminants in beeswax.