Ice Dragon Update

[relttem]

After a lot of iterations and taking into account what our reviewers have told us, we have finally made the nanofluid available for purchase on our website - www.icedragoncooling.com. We are working on the website too, so don't hammer us too badly on its design. We have coozies left over from CES that we will include in the order.

We are also waiting for a few more reviews to come out and will link them on our website. The reviews we have had done are linked already. We have a few reviews with people cooling multiple things (not just a CPU) that are turning out quite nicely and should be up there shortly. The nanofluid really shows its stuff when you start cooling multiple things.

We also worked on the chemical make-up as to inhibit air from getting entrained in the fluid - which is really nice. We are going to put a video of it in action on the website. As mentioned before, the nanofluid is a biocide on its own, so there is no need to add anything for that.

We are also looking at the thermal aspects of dead zones. A dead zone is where the fluid constantly rotates in a fixed manner, like in a corner. This would come into play with the nanofluid because of the particles. So, you would have a small area of rotating particles. We think that this would be better than just DI water spinning in a corner because of the thermal aspects. We are going to do some CFD and see what happens. I have noticed some settling in these corners of my square reservoir after I have drained my system, but the circular reservoirs don't have this issue.

I emailed mlwood37 to see if we could trade some nanofluid for some UV dye to see how that works. I, personally, don't see any problem with the dye, but we will still test it to make sure. I am waiting to hear back from him.

We are getting some bench tests made. Basically it will be a copper block with some heaters that will let us control the heat input to the block. Then, we will have a Tj thermocouple and some in the fluid of the WC block as well as the radiator. This will let us test two of the same rigs against each other
rather quickly - one with nanofluid and one without. The plan is to have it all webcammed too.

I want to thank Skinnee, HESmelaugh and LittleOwl for taking the time to run some tests and getting us involved with this website. Stop by if you are at CES in Jan..

Just an update..

rock on

[skinnee]

Any changes to the C2 formula I have to the retail formula?

[WhiteFireDragon]

yeah, what are the changes this time compared to the ones we saw 2 years ago?

[skinnee]

the C2 formula is newer than the initial tests done back in 2008. Tube staining was greatly reduced, but still there a tiny bit.

[wez]

How much liquid is it per bottle? Didn't find any figures in your shop.

[relttem]

Probably the biggest thing we did was the air inhibition. That is impressive and, as most of you know, any air in your system is bad. So, yes there are changes to C2. We also worked with our nanoparticle maker on a lot of small issues that were corrected. I thought the first stuff was good, but this stuff is better. Staining is still there, but I would call it more of a light coating. My experiences with it is that it is easily removed -pipe cleaner, rag, etc. I have removed it from my reservoir with hot water. I don't know if Skinnee tried to remove it or not. But, that is fluid dynamics - it is called 'law of the wall', which means there is zero velocity at the wall. That is true in any newtonian flow, and it is how you get velocity profiles. Still, that really isn't going to effect anything but the appearance..also, the fluid is white as is the staining..it is hard to see.

Skinnee, if you have time for a test let me know. As I mentioned in another thread I would love to see a comparison of all (most) of the available cooling fluids at different heat inputs..

Our next test will be our stuff in corsair WC v's what corsair has in their's already.

The price is listed. That is all inclusive in the USA..plus, a free coozie..

[Slowb00st]

Probably the biggest thing we did was the air inhibition. That is impressive and, as most of you know, any air in your system is bad. So, yes there are changes to C2. We also worked with our nanoparticle maker on a lot of small issues that were corrected. I thought the first stuff was good, but this stuff is better. Staining is still there, but I would call it more of a light coating. My experiences with it is that it is easily removed -pipe cleaner, rag, etc. I have removed it from my reservoir with hot water. I don't know if Skinnee tried to remove it or not. But, that is fluid dynamics - it is called 'law of the wall', which means there is zero velocity at the wall. That is true in any newtonian flow, and it is how you get velocity profiles. Still, that really isn't going to effect anything but the appearance..also, the fluid is white as is the staining..it is hard to see.

Skinnee, if you have time for a test let me know. As I mentioned in another thread I would love to see a comparison of all (most) of the available cooling fluids at different heat inputs..

Our next test will be our stuff in corsair WC v's what corsair has in their's already.

The price is listed. That is all inclusive in the USA..plus, a free coozie..

That fluid you cleaned how long was it in the reservoir and what temps did it see? Was it used on an actual computer or a test setup?

Another thing a bit off topic but should you charge sales tax even if shipping out of state?

[relttem]

Yes, it was in an actual computer system..that is actually still running with nanofluid in it. It is a big water (thermaltake) res. I moved it to another computer, so I had to clean it out. I think when it is being used to play games (WOW/AION) it is at 60C and maybe 40C idle (that is at 3.4GHz on 3.0 phenom chip). That system has been running nanofluid in it for over 2 years.

I think we are required by SC Business Law to charge sales tax..that was actually automatically applied when we set it up. We just had to back it out because we wanted an even number and not 39.99. At the end of the tax year we have to pay taxes on money we make regardless of where it goes. I would imagine that is why that is, but I am not a business major by any means. I'll ask and see why tho. I am curious now.

[OC Maximus]

Should only be charging tax on in state customers.

[skinnee]

yeah, if you can ship it out yet this week I can include it in the coolant testing. Just completed testing of 9 coolants including C2.

Didn't think there were changes to C2. oops.

[relttem]

We can send it tomorrow..will that work? The big difference will be the lack of air in the system. I am not sure how you fill a system, but I fill it up and turn the computer on. Then, when the nanofluid is warm (after the comp has been running a while) I cap it. Anyway, I'll be sure to get it out.. same address as before?

on a side note: Way back we sent samples to Skinnee and HESmelaugh. We sent two samples to each..I can't remember the combination, but I think Skinnee got B2 and C2 and HESmelaugh got A2 and B2. A2 was 20nm particles, B2 was 40nm, and C2 was 60nm. C2 had the most settling if left on the shelf due to the size of the particles and stokes settling velocity, but, as with the others, would mix back up if shaken. None of the three had the air inhibitor, which does improve the thermal abilities. The new stuff has 20nm particles and a much better distribution. We did tests using AMD overdrive on a machine overclocked by 400mhz and were a degree lower with the new stuff.

[skinnee]

Testing turn-around will be about 2 days, need 18 hours of pure test time. Finished with all the fluid I have, thats the reason I am pushing to ship. Yup, same address.

For filling, depends on the loop really. The bench running the coolant tests has an EK150 at the highest point, so I did the fill/pump on-off/fill routine and let the loop bleed for 20 minutes, topped the res off (filled to top threads) to where when I put the cap back on I would usually have a drip or two out the sides, pump off of course. Powered the system on and kicked off the test. Come time for draining, the reservoir level would drop by 5-6mm, never really more than that.

I have 4 bottles in total, but the B1 and C1 had the clumping problem real bad and did not remix entirely when shaken. C2 was the latest one sent over way back when, clumping problem resolved though. Yeah, tube staining isn't really the right word for it... slight coating is better. The coating from C2 did clean up after several flushes.

Loop consists of DDC3.2 with XSPC Top, Swiftech GTZ, DangerDen GTX280's in serial (thanks Conundrum), Koolance FM17 flow meter, Drain T, Temp T1, Temp T2, XSPC RX360 (Yate Lows @ 12v), Temp T3, Temp T4 and an EK 150 Res


i7 930 @ 3.8GHz (19x200) w/ 1.375v under the GTZ, 280's are at stock clocks with 1138mV and some UV (little things are bright) provided by the Gigabyte UD7. OCCT loads the CPU and Kombustor on all threads for the GPU's.

[Church]

skinnee: which things do you change for coolant tests? I'm guessing flow and heatload?

[skinnee]

Nope, straight load tests. This is the last coolant/fluid testing I'll do... no plans for any beyond this.

[Church]

skinnee: it's just that imho coolant relative effectiveness to water might be different at low or high flow scenarios or with higher or lower temps, so it might be worth to test at least corner cases to claim once for all, 'coolant X is better then Y at such setups by Z%, yet at such setup difference is minimal and not worth the extra price', just like TIMs that might have different effectiveness with different heatloads/temps, no?

[Vapor]

skinnee: it's just that imho coolant relative effectiveness to water might be different at low or high flow scenarios or with higher or lower temps, so it might be worth to test at least corner cases to claim once for all, 'coolant X is better then Y at such setups by Z%, yet at such setup difference is minimal and not worth the extra price', just like TIMs that might have different effectiveness with different heatloads/temps, no?

Eh, what?

Lower flowrates will differentiate the coolants more, it's pretty simple.

[skinnee]

Adding in those variables won't change the results. What that additional testing will do is only add time and provide data that states the exact same thing.

Ninja edit: beaten to the punch...

[Church]

Vapor: relative thermal transfer of TIM pastes / coolants won't change at different temps? There is no way for some to be more effective/optimised at some specific range and less effective at other? I don't know, that's why i'm partially guessing and asking you veterans.

[relttem]

The nanofluid will be at an advantage when the heat load increases to the point that the system can't handle the load. Meaning that when the liquid's heat capacity is nearing its peak the nanofluid will continue to cool. This is due to the increase in the convective coefficient, h, of the nanofluid to that of other fluid's convective coefficient. The value of 'h' is dependent on the flow rate. So, if the flow rate is constant each fluid will have its own 'h' value - and that will drive the temps. BUT, this will not be noticed until the upper limit of the cooling capacity of the fluid is met.

Skinnee, you should get the stuff on Saturday...

rock on

[neonatas]

Question.:

Are you gonna supply this in blue only? Or will colors like green and red be available aswell?

[Vapor]

Vapor: relative thermal transfer of TIM pastes / coolants won't change at different temps? There is no way for some to be more effective/optimised at some specific range and less effective at other? I don't know, that's why i'm partially guessing and asking you veterans.

For the most part, and as long as there is no change of phase, coolants should perform within .2% similarity at 25C vs. 50C. Maybe there's some fluid out there that behaves vastly differently that's useful as a coolant, but if it's water based, temperature doesn't matter much.

Heatload does matter, as does flowrate. From the coolant's perspective, lower flowrate = higher heatload, so all that will happen is that coolants will differentiate more, i.e., bad coolants will fall behind even more. If you push the test setup hard enough to one extreme (whether it's super low flowrates or super high heatloads) so that you get a lot of differentiation, you'll be at or past the testbed's point of failure (in terms of stability), and you also get results that don't mean much for the typical reader (since the testbed is no longer representative of realworld conditions and the impact of coolant becomes exaggerated).

Based on what I've seen from skinnee's data so far, there is such a thing as a "bad" coolant, coolant to air delta is limited by the air and therefore dictated by radiator power (air flow * air saturation) and not the coolant itself, water is extremely good as a coolant (and coolant base), there seems to be some surface-level interaction that has an impact on performance (maybe there's a correlation to coolant viscosity, I haven't looked that closely), and testing coolants on a PC is really hard (loading both a CPU and a GPU simultaneously is a bit of a nightmare for testing due to random inconsistencies--at least they're detectable though).

[Church]

Vapor: BTW, i start to think that using some specialised heat source with variable/settable heatload to which waterblocks can be mounted instead of real CPUs/GPUs would make some tests simplier/faster. At least mounting wise, ease of testing with different heatload wise (no more searching for especially hot cpus like in recent amd TIM test), easier to test with cooling performance for even wider spread corner cases then hardware on hands (so can be used for estimated heat output of future hw), cooling testing with one heatsource for any hardware (just few tests before to calibrate it to dissipate heat "like amd cpus o/c-ed @ XXXX MHz" or like "i5 o/c-ed @ YYYYMHz" and things like that).

[relttem]

Question.:

Are you gonna supply this in blue only? Or will colors like green and red be available aswell?

It is actually white..skinnee has a few pictures of it, but we are trying to get our hands on some of the UV dye to try that out. But, I think the color will be up to you. We will let you know what happens when we dye it, then you can decide if you want to. I would love to see it Orange, but I am probably the only one, so I will dye it myself.. At least white is a good place to start when dying.

[relttem]

Vapor: BTW, i start to think that using some specialised heat source with variable/settable heatload to which waterblocks can be mounted instead of real CPUs/GPUs would make some tests simplier/faster. At least mounting wise, ease of testing with different heatload wise (no more searching for especially hot cpus like in recent amd TIM test), easier to test with cooling performance for even wider spread corner cases then hardware on hands (so can be used for estimated heat output of future hw), cooling testing with one heatsource for any hardware (just few tests before to calibrate it to dissipate heat "like amd cpus o/c-ed @ XXXX MHz" or like "i5 o/c-ed @ YYYYMHz" and things like that).

we mentioned that in the very first post of this topic.

I am a little confused on some of Vapor's comments. So, I am hoping he can clarify some stuff for me so I can better understand his post.

coolants should perform within .2% similarity at 25C vs. 50C. - where did .2% come from?

Lower flow rate = high heat load ? What does that mean? Is your heat load not constant? Or do you mean lower flow rate = higher outlet temperature of the fluid - that makes sense based on Q = mdot*Cp*(dT).

Also, "coolant to air delta is limited by the air and therefore dictated by radiator power (air flow * air saturation) " - what is air saturation?

In forced convection heat transfer you have a few variables that are dictating how much heat you can pull out - h, Tin, and A. If I can make a fluid with a higher 'h' value than water, and keeping Tin and A constant my high 'h' will let me remove more heat. But, if Q stays constant then my Tout has to drop in order for the equation to balance. Your radiator's power is the same way. Air flow matters, as does the surface area, and the h value.

A good comment about viscosity. Viscosity definitely comes into play. That is because your Reynold's number is function of that. Then, from a Dittus-Boelter correlation you can get your available heat transfer. BUT, viscosity is also going to have detrimental effects on pumps. It is a trade off.

The testing until point of failure was dead on. This also can be seen when using a radiator capable of dissipating 500W and is from a Chevy. It is overkill.

[relttem]

skinnee, is the FluidXP nanofluid stuff included in your tests?