Transformator Oil & high flow?

[Frank M]

Hi,

As we all know, oil has a lot of benefits (non-corrosive, non-conductive, high
viscosity, lubricant, no stuff buildup, etc), but a big backlash: takes up and
gives off heat much harder. Could this be negated by using higher flow? Or
would it be even worse?

Just curious about what you think.
Thanks.

[Graystar]

Contrary to popular belief, increased flow rate, in of itself, does not increase the heat transfer rate.

People believe that it does because of a misinterpretation of the Fourier’s heat transfer formula. In the formula, if you increase the “rate” then you increase the value of Q, or heat transferred. So, supposedly, you’re transferring more heat because of the higher rate. But what everyone misses is that Q is actually a constant...it is the heat that must be transferred from the processor, and it has nothing to do with the flow rate. So when you change flow rate, Q stays the same and something else must change...the water temperature. Strangely, people in the “know” understand the temperature drop and constant Q, yet still seem to cling to some phantom relationship between speed and heat transfer. Velocity is distance over time, and there is no distance over time component in any heat transfer formula that I’ve seen.

But wait...life’s not this simple. Increased flow rate increases turbulence in the water. Increased turbulence affects something called the Convection Heat Transfer Coefficient. If you increase the Heat Transfer Coefficient, then you can increase the heat transfer rate of the turbulent fluid. This is another reason why those in the know believe that increased flow increases heat transfer.

Unfortunately turbulence is very low at the rate in which we pump water in a water-cooling loop. The flow of water through the latest generation of radiators is barely turbulent. The flow through the Storm waterblock (arguably the most turbulent block of all) is likewise not that turbulent. The water flow through 1/2“ tubing is actually more turbulent than either one.

Oil’s high viscosity would need a much stronger pump to even match the rate in which we pump water, and that means more heat from pumping. This factor alone is reason enough not to use oil.

[Frank M]

Hmm, I see. Thanks for the clarification!

[[XC] MarioMaster]

you might want to also resize your sig, it's a bit large

[Bun-Bun]

Contrary to popular belief, increased flow rate, in of itself, does not increase the heat transfer rate.

People believe that it does because of a misinterpretation of the Fourier’s heat transfer formula. In the formula, if you increase the “rate” then you increase the value of Q, or heat transferred. So, supposedly, you’re transferring more heat because of the higher rate. But what everyone misses is that Q is actually a constant...it is the heat that must be transferred from the processor, and it has nothing to do with the flow rate. So when you change flow rate, Q stays the same and something else must change...the water temperature. Strangely, people in the “know” understand the temperature drop and constant Q, yet still seem to cling to some phantom relationship between speed and heat transfer. Velocity is distance over time, and there is no distance over time component in any heat transfer formula that I’ve seen.

But wait...life’s not this simple. Increased flow rate increases turbulence in the water. Increased turbulence affects something called the Convection Heat Transfer Coefficient. If you increase the Heat Transfer Coefficient, then you can increase the heat transfer rate of the turbulent fluid. This is another reason why those in the know believe that increased flow increases heat transfer.

Unfortunately turbulence is very low at the rate in which we pump water in a water-cooling loop. The flow of water through the latest generation of radiators is barely turbulent. The flow through the Storm waterblock (arguably the most turbulent block of all) is likewise not that turbulent. The water flow through 1/2“ tubing is actually more turbulent than either one.

Oil’s high viscosity would need a much stronger pump to even match the rate in which we pump water, and that means more heat from pumping. This factor alone is reason enough not to use oil.

First off I am going to agree with everything you said except... the part about the Storm. Before going off and saying that the storm is hardly turbulent at all (the princible behind its design is turbulence...) then I am going to have to ask you to run the calulation and post it here and proove to me and everyone on here that the Reynolds number for the flow through a storm is indeed not very turbulent.

Also increased flow has many benefits if not directly affecting the "speed" at which heat is transfered. I just think you should clarify this.

[Frank M]

you might want to also resize your sig, it's a bit large

I tried resizing, but that makes the punch-line unreadable.
I think some other sigs that just show brand preference are just as big - mine
is at least funny

[Graystar]

Before going off and saying that the storm is hardly turbulent at all (the princible behind its design is turbulence...) then I am going to have to ask you to run the calulation and post it here and proove to me and everyone on here that the Reynolds number for the flow through a storm is indeed not very turbulent.

I think it’s pretty obvious that the focus of this discussion is centered on what happening within the Storm cup. But before we get to that lets look at what’s happening everywhere else.
Reynolds number calculations were performed using the engineeringtoolbox calculator.
http://www.engineeringtoolbox.com/re...ber-d_237.html

Given:
Black Ice GT Stealth 240 & Black Ice Xtreme XFlow
Tube size 16mm x 2mm x 0.2mm thick wall
(Source Cooling-Masters.com, GTS 240 review, 1/9/2006)

inside tube size = 15.6mm x 1.6mm converted = 0.051181102 ft x 0.005249344 ft

Storm
Microjet inner diameter - 0.029”
Microjet outer diameter - 0.0625”
Cup diameter – 0.123”
Microjet extends into cup by 0.072”
(Source – Systemcooling.com, Swiftech Storm review, 7/24/2005)

At this time I do not know the exact depth of the cup, but I guess 0.125”



So the question we need to ponder is...will the Reynolds number of the water churning in the Storm cup exceed the calculated numbers?

My personal answer is that I don’t know. I can only guess.

But here’s what I do know:
Eddies start to dissipate immediately after the conditions that created them are altered.
Water, like any fluid, will flow towards areas of low pressure.
In a closed system the velocity at any given point is a function of flow rate and cross-sectional area.

Knowing these things, I theorize that the water exiting the Storm microjet is slowed immediately due to the changed conditions and counter flow of the exiting water towards the low pressure region. We know that the conditions cannot maintain a Reynolds number of 9501 (assuming 2 GPM,) so it is my guesstimate that the true Reynolds number is somewhere between 9501 and 2240 (assuming 2 GPM).

That’s my theory. I would have to learn more about turbulence calculations in a mixing situation in order to get a better idea of what’s going on. In any case, I can’t see the Reynolds number in the cup reaching that of 1/2" tubing, which is 19287 (assuming 2 GPM.)

It is interesting to note that the Storm is, according to SystemCooling.com, optimized for flows of 1 GPM and higher. It is at 1 GPM where the flow down the microjet becomes fully turbulent.

Also increased flow has many benefits...

Many? Like what?

[Bun-Bun]

I think it’s pretty obvious that the focus of this discussion is centered on what happening within the Storm cup. But before we get to that lets look at what’s happening everywhere else.
Reynolds number calculations were performed using the engineeringtoolbox calculator.
http://www.engineeringtoolbox.com/re...ber-d_237.html

Given:
Black Ice GT Stealth 240 & Black Ice Xtreme XFlow
Tube size 16mm x 2mm x 0.2mm thick wall
(Source Cooling-Masters.com, GTS 240 review, 1/9/2006)

inside tube size = 15.6mm x 1.6mm converted = 0.051181102 ft x 0.005249344 ft

Storm
Microjet inner diameter - 0.029”
Microjet outer diameter - 0.0625”
Cup diameter – 0.123”
Microjet extends into cup by 0.072”
(Source – Systemcooling.com, Swiftech Storm review, 7/24/2005)

At this time I do not know the exact depth of the cup, but I guess 0.125”



So the question we need to ponder is...will the Reynolds number of the water churning in the Storm cup exceed the calculated numbers?

My personal answer is that I don’t know. I can only guess.

But here’s what I do know:
Eddies start to dissipate immediately after the conditions that created them are altered.
Water, like any fluid, will flow towards areas of low pressure.
In a closed system the velocity at any given point is a function of flow rate and cross-sectional area.

Knowing these things, I theorize that the water exiting the Storm microjet is slowed immediately due to the changed conditions and counter flow of the exiting water towards the low pressure region. We know that the conditions cannot maintain a Reynolds number of 9501 (assuming 2 GPM,) so it is my guesstimate that the true Reynolds number is somewhere between 9501 and 2240 (assuming 2 GPM).

That’s my theory. I would have to learn more about turbulence calculations in a mixing situation in order to get a better idea of what’s going on. In any case, I can’t see the Reynolds number in the cup reaching that of 1/2" tubing, which is 19287 (assuming 2 GPM.)

It is interesting to note that the Storm is, according to SystemCooling.com, optimized for flows of 1 GPM and higher. It is at 1 GPM where the flow down the microjet becomes fully turbulent.


Many? Like what?

First off what units are your area's in? Those are bogus area's for 1/2" tubing as there is no way reynolds number is that much higher for tubing then inside of the storm. Just think about it... the velocity of water in tubing is going to be no where near that of inside a storm. I calculate that 1/2" ID tubing will have a cross sectional area of 0.1963 in^2

As for your theory of the water becoming that much less turbulent between teh jets and cups... it's interesting and I have never thought of it before. I will run my own calculation of reynolds number for a storm and 1/2" tubing later to confirm these numbers.

And as far as flow goes, I have not figured out the scientific reasons yet but am currently discussing with others to figure it out for myself but it is proven that with blocks and radiators their thermal resistance goes down with increased flow rate. And if I was just to guess I would say it has to do with the amount of water that is flowing through them and not so much the speed as you previously said there is no distance over time involved in the thermal equations. But more water is able to absorb heat and diapate heat in a given time. Thats just a theory though I am in the process of figuring it out.

[Graystar]

First off what units are your area's in? Those are bogus area's for 1/2" tubing

All diameters are in feet. All areas are in square feet. All time components are in seconds. If my volume is in feet, and my velocity is in feet, then it’s reasonable to presume that my area is in feet as well. It is puzzling that you would make such a strong condemnation of my calculation without at least attempting to see if the numbers fit some other reasonable unit of measure.

As you stated...
1/2" diameter tubing has a radius of .25”. Area = Pi*.25^2 = 0.19634954”

OnlineConversion.com says that:
0.196 349 54 square inch = 0.001 363 538 square foot

there is no way reynolds number is that much higher for tubing then inside of the storm. Just think about it...

Don’t have to think about it...it’s been calculated. The flow of water through the jet is approximately 8.5 times faster than through 1/2" tubing...obviously, a good deal faster. However, the hydraulic diameter is much, much smaller. For any given velocity, as diameter decreases turbulence will decrease. That’s why the Reynolds number isn’t that high.

but it is proven that with blocks and radiators their thermal resistance goes down with increased flow rate.

I believe that there isn’t a single test you can point to that demonstrates this. If you do know of such test results, I would appreciate seeing them. Please note that I am fully aware of all tests at SystemCooling, ProCooling, Cooling Masters, and all of Bill Adams’ radiator tests results at Swiftech and Thermal Management. Even David at Cooling Masters noted in his GTS 240 review:

“Attention, we do not work at constant dissipated power, but with constant water-air variation, it is different!” (translated by Google)

C/W is a very conditions-dependent calculation, and in these tests the conditions are so foreign to what water-cooling loops actually do that any application of the resulting C/W values to a live system is foolhardy. Those test results are perfectly valid for comparison purposes, and that’s about it.

But as I said, if you do know of some other tests I would be interested in seeing those results.

[Bun-Bun]

All diameters are in feet. All areas are in square feet. All time components are in seconds. If my volume is in feet, and my velocity is in feet, then it’s reasonable to presume that my area is in feet as well. It is puzzling that you would make such a strong condemnation of my calculation without at least attempting to see if the numbers fit some other reasonable unit of measure.

As you stated...
1/2" diameter tubing has a radius of .25”. Area = Pi*.25^2 = 0.19634954”

OnlineConversion.com says that:
0.196 349 54 square inch = 0.001 363 538 square foot


Don’t have to think about it...it’s been calculated. The flow of water through the jet is approximately 8.5 times faster than through 1/2" tubing...obviously, a good deal faster. However, the hydraulic diameter is much, much smaller. For any given velocity, as diameter decreases turbulence will decrease. That’s why the Reynolds number isn’t that high.


I believe that there isn’t a single test you can point to that demonstrates this. If you do know of such test results, I would appreciate seeing them. Please note that I am fully aware of all tests at SystemCooling, ProCooling, Cooling Masters, and all of Bill Adams’ radiator tests results at Swiftech and Thermal Management. Even David at Cooling Masters noted in his GTS 240 review:

“Attention, we do not work at constant dissipated power, but with constant water-air variation, it is different!” (translated by Google)

C/W is a very conditions-dependent calculation, and in these tests the conditions are so foreign to what water-cooling loops actually do that any application of the resulting C/W values to a live system is foolhardy. Those test results are perfectly valid for comparison purposes, and that’s about it.

But as I said, if you do know of some other tests I would be interested in seeing those results.

After running numbers myself I figured out your areas were in ft^2. However why are you using hydraulic radius for circular areas? You only need to concern yourself with hydraulic radius for non circular cross sections such as square tubing.

But as such is I ran the numbers myself manually using nothing more then a calculator and came up with these results.
Nr=reynolds number

For Storm @ water temp. 70°F flowing @ 2 GPM
Nr=6387
For Storm @ water temp. 80°F flowing @ 2 GPM
Nr=7330
For tubing @ water temp. 70°F flowing @ 2 GPM
Nr= 12955
For tubing @ water temp. 80°F flowing @ 2 GPM
Nr= 14866

So I actually proved it less turbulent then what you came up with. Funny how that works sometimes. Thats weird I would have thought the storm to be more turbulent then that but even as that stands at 1.5GPM flow rate (recomended for storm by swiftech) the storm is still > 4000 reynolds number and thus is still turbulent. I am not sure how much your theory holds about what happens after the water jet though although it does seem like it would be affected somehow by the currents going out of the storm. The distance between the jets and cups is small so I would think the stream would stay turbulent even though I do not have the knowledge to prove as such.

I cant beleive the water in our tubing is that turbulent...heh

And I was refereing to tests by Billa and others that you mentioned. You are correct they are a controlled enviorment and don't replicate what our loops do inside the computer, however there is indeed an increase in efficiency with increase of flow. As far as thermal resistance is concerned anyway.

And increased flow will always mean more turbulence which does help heat transfer... even if it does cause more frictional losses.

Although that being said I beleive there is a balance point. Ive seen loops suffere due to poor flow and loops suffer due to too much flow, the former usually being more extreme. There are many factors that affect this though so it is difficult to just say what will work. A bit of tweaking and trial and error is required.

I would like to see a storm made out of acrylic so that you could look at how the jet stream is behaving... would be very interesting to see.

[Graystar]

However why are you using hydraulic radius for circular areas?

I’m not sure where you’re referring to when you say hydraulic radius, as there are no hydraulic radiuses in my chart.

The Reynolds calculator asks for Characteristic Length or Hydraulic Diameter. For a round pipe the hydraulic diameter equals the pipe diameter. The hydraulic diameters in the chart are the inside diameters of the given component, converted to feet.

The Characteristic Length of the rectangular radiator tubes was calculated using:
dh = 2 a b / (a + b) (4)

...and confirmed using the calculator available at Engineeringtoolbox.com.

And I was refereing to tests by Billa and others that you mentioned. You are correct they are a controlled enviorment and don't replicate what our loops do inside the computer, however there is indeed an increase in efficiency with increase of flow. As far as thermal resistance is concerned anyway.

No, there is no increase in efficiency with increased flow. There is simply and increased in heat dissipated only because, under the test conditions, there is more energy being delivered to the radiators.

You turn up the flame on a simmering pot and it begins to boil. Do you now say that the efficiency of the pot as increased? That it’s performing better? Or is it simply passing the larger amount of heat?

In these radiator tests, when flow is doubled but temperature over ambient is kept constant, the amount of energy carried by the water is also doubled. It is exactly like turning up the flame on the radiator. The radiator passes more heat because it’s being given more energy to pass, not because it’s performing better.

And increased flow will always mean more turbulence which does help heat transfer... even if it does cause more frictional losses.

That's correct, but the true effect and necessity of turbulence is the big unknown. For example, it’s clear that the latest generation of radiators are built around the idea of expanded surface area, while keeping flow less turbulent. So, at least in radiator design, the role of turbulence is being diminished.

Yeah, an acrylic Storm base would be really interesting, along with some high-speed photography of the flow.

[Bun-Bun]

I’m not sure where you’re referring to when you say hydraulic radius, as there are no hydraulic radiuses in my chart.

The Reynolds calculator asks for Characteristic Length or Hydraulic Diameter. For a round pipe the hydraulic diameter equals the pipe diameter. The hydraulic diameters in the chart are the inside diameters of the given component, converted to feet.

The Characteristic Length of the rectangular radiator tubes was calculated using:
dh = 2 a b / (a + b) (4)

...and confirmed using the calculator available at Engineeringtoolbox.com.


No, there is no increase in efficiency with increased flow. There is simply and increased in heat dissipated only because, under the test conditions, there is more energy being delivered to the radiators.

You turn up the flame on a simmering pot and it begins to boil. Do you now say that the efficiency of the pot as increased? That it’s performing better? Or is it simply passing the larger amount of heat?

In these radiator tests, when flow is doubled but temperature over ambient is kept constant, the amount of energy carried by the water is also doubled. It is exactly like turning up the flame on the radiator. The radiator passes more heat because it’s being given more energy to pass, not because it’s performing better.


That's correct, but the true effect and necessity of turbulence is the big unknown. For example, it’s clear that the latest generation of radiators are built around the idea of expanded surface area, while keeping flow less turbulent. So, at least in radiator design, the role of turbulence is being diminished.

Yeah, an acrylic Storm base would be really interesting, along with some high-speed photography of the flow.

I meant hydraulic diameter. And now that I look at it again it is the same diameter of the tubing thus nulifying what I said...

Ok so efficiency was the wrong word. But the c/w is lower with increase flow for the reasons you said above and to me thats more efficient or at least "better" then lower flow. Esspecially since those doing the designing and testing (cathar, marci...etc) themselves use c/w to compare at different flow rates then I am going to trust them until I can prove it myself in my quest for ultimate knowledge.

Cathar if you read this... could you make an acrylic G4 for the sole purpose of creating a high-speed video of the water flow through the jets at different flow rates

[Graystar]

...use c/w to compare...

Yes, exactly. As I said earlier, as a comparative measure against other items being tested under the same conditions, those values are perfectly valid, and I would recommend using them to determine your next purchase.

Just don't think that you'll be able to use those C/W numbers to estimate loop behavior because you'll be off by 5280 feet.

[Bun-Bun]

Just don't think that you'll be able to use those C/W numbers to estimate loop behavior because you'll be off by 5280 feet.

I never would there would be no way of calculating exactly whats going to happen because the head load, flow rate, water temp, water/air delta... etc is all different and varying in different ways.

But in a loop it would still mean more flow mores more energy but as I said before there needs to be a balance as can been seen by some loops that suffer with low flow and others with high flow. Gotta find the happy medium.

[Graystar]

But in a loop it would still mean more flow mores more energy

No, it doesn't. That's what I'm trying to tell you. The energy contained in the water is set by the CPU. It's fixed at whatever point the CPU happens to be operating at. You can't pull more energy out of the CPU simply by increasing flow. That means that something else has to change to keep the balance. That something else is the water temperature...or more accurately, the delta between water in and out of the waterblock.

If you double the flow, the delta drops in half. This drops the average temperature of the block, which then drops the temperature of the CPU. This is why increasing the flow has such a dramatic effect on systems that were running on a trickle of water. On such systems, the delta could easily have been 6 -10 C or more. So increasing the flow, even a little, makes big changes. But once you get up to 1 or 1.5 GPM, then your delta is really low...usually in the low tenths of a degree. Increasing flow doesn’t get you anything measurable.

[Bun-Bun]

No, it doesn't. That's what I'm trying to tell you. The energy contained in the water is set by the CPU. It's fixed at whatever point the CPU happens to be operating at. You can't pull more energy out of the CPU simply by increasing flow. That means that something else has to change to keep the balance. That something else is the water temperature...or more accurately, the delta between water in and out of the waterblock.

If you double the flow, the delta drops in half. This drops the average temperature of the block, which then drops the temperature of the CPU. This is why increasing the flow has such a dramatic effect on systems that were running on a trickle of water. On such systems, the delta could easily have been 6 -10 C or more. So increasing the flow, even a little, makes big changes. But once you get up to 1 or 1.5 GPM, then your delta is really low...usually in the low tenths of a degree. Increasing flow doesn’t get you anything measurable.

Did I say anything about going above 1.5 GPM? I am not going to pretend that I know about the thermodynamics involved in water loops (my studies have only led to me phase change thus far) but I do know you need a certain amount of flow and to a point it shows an improvment and beyond that it can even hinder performance.

[epion2985]

Its a pretty retarded idea to use oil as a coolant. The thermal properties are horrible, horrible conduction/capacity. The mechanical properties are horrible too, very viscous so puts heavy strain on the pump, hurting the flow further, making the pump run hotter and dumping more heat in to the loop.

I mean this is as reasonable as using mud as a coolant... If you are trying to find the worst performing coolant, then sure this is not a bad candidate. But why would you want to use a poor performing one is beyond me.

[Graystar]

Its a pretty retarded idea to use oil as a coolant.

Oil is used as a coolant in metal fabrication and in other applications.

Please don't be so disparaging and dismissive of others’ ideas. None of us has knowledge so absolute that we can’t be wrong in our suggestions...or our retorts.

[Bun-Bun]

Its a pretty retarded idea to use oil as a coolant. The thermal properties are horrible, horrible conduction/capacity. The mechanical properties are horrible too, very viscous so puts heavy strain on the pump, hurting the flow further, making the pump run hotter and dumping more heat in to the loop.

I mean this is as reasonable as using mud as a coolant... If you are trying to find the worst performing coolant, then sure this is not a bad candidate. But why would you want to use a poor performing one is beyond me.

Actually some of its thermal, mechanical, and chemical properties are favorable depending on the application. As graystar mentioned oil is used in metal fabrication and other applications as a cooling method. Look at transmissions in our cars.

I'm sure someone somewhere would be more concerned with the non conductive properties of oil (as apposed to maximum cooling) and design a proper loop for cooling with. You never know, it could happen.

[Xeon th MG Pony]

Epion, just what do you think they use to cool hydro transformers with? I'll give ya a hint it starts with an O; Computer loops for most average person it wouldn't be the good choice but for some of the stuff I've built I seriusly looked at it and there are plenty of other apps where it is well apt to do the job.

[epion2985]

Oil is used as a coolant in metal fabrication and in other applications.

Please don't be so disparaging and dismissive of others’ ideas. None of us has knowledge so absolute that we can’t be wrong in our suggestions...or our retorts.

Indeed it is used but only where cooling performance needs are minimal and where it has to touch the actual component, like a transformer.

Actually some of its thermal, mechanical, and chemical properties are favorable depending on the application. As graystar mentioned oil is used in metal fabrication and other applications as a cooling method. Look at transmissions in our cars.

I'm sure someone somewhere would be more concerned with the non conductive properties of oil (as apposed to maximum cooling) and design a proper loop for cooling with. You never know, it could happen.

Everything is used somewhere. What I said was it would be a retardation to use oil as a coolant in pc coolant, which still stands. Places that use it are usually more concerned with its dielectric property ie cooling a transformer, because its immersed in the oil. Which is not the case with computers.

I don't know anyone who would rather their computer ran hotter then it would on poor air just to have dielectric coolant. Especially since there is no need for a dielectric coolant as it never contacts any circuitry, while there is a need for cood thermal conduction, heat capacity and low viscosity.

Epion, just what do you think they use to cool hydro transformers with? I'll give ya a hint it starts with an O; Computer loops for most average person it wouldn't be the good choice but for some of the stuff I've built I seriusly looked at it and there are plenty of other apps where it is well apt to do the job.

Re-read my post, thats exactly what i said, "for computers".

Like I said above, for some things it is used because its contact cooling, this is not the case, end of discussion.

Computer loops for most average person it wouldn't be the good choice