Nanofluid: Dual Core Tests and Review

[relttem]

Tapwater + copper block = really bad

[relttem]

For these conditions, is the 1 to 2 degrees difference statistically significant? If so, with what confidence interval?

And you still wouldn't want this stuff in a mix Al/Cu loop. You'd still get corrosion, unless you had an all Al loop, but then that wouldn't be mixed.

Its tough to answer that question for me..maybe the testers can chime in. I can say that the research project we did for the Air Force contained so much data that any anomalies in the data would be filtered out due to the amount of data. Also, when performing an error analysis on the equations, devices used to for data collection etc, the error is a +/- sort of thing..thus, the bar that goes above and below your data point.
Also, that is why we sent out samples to numerous people for testing. But, so far all the tests are falling into the same realm, so I would lean towards pretty good confidence that the data is correct.

[skinnee]

For these conditions, is the 1 to 2 degrees difference statistically significant? If so, with what confidence interval?

And you still wouldn't want this stuff in a mix Al/Cu loop. You'd still get corrosion, unless you had an all Al loop, but then that wouldn't be mixed.

I did not have a standard deviation or confidence interval on this set of test runs, I only ran each test once. For the upcoming quad tests I will be able to provide metrics for both of those.

[T_Flight]

Can we get more info on where this fluid came from, and what type it is, and it's makeup? I've been doing a bit of research and it seems there are different types of this stuff, and they have different nanoparticles in them.

I'd also like to know where to obtain some, the cost, and whether I can find out what it's makeup is so I don't get anything with any Al particles in it. I won't put anything in my loop unless I know what's in it.

I'd like to try some of this, because I'm one of those scientiffic types, and this is just plain cool, but I want to be sure what I'm getting, and what it's makeup is before I attempt to run it through brand new hardware...very expensive high end brand new hardware I might add.

[TNTSunshine]

I did not have a standard deviation or confidence interval on this set of test runs, I only ran each test once. For the upcoming quad tests I will be able to provide metrics for both of those.

Ah, ok. I remember reading about how you were going to go onto the quad core and do more trials to hopefully see the greatest difference. Good luck! I know its a lot of work, but we all appreciate your hard work

As to what the powder's phase composition is, a quick XRD analysis of a dried powder sample shouldn't be too hard to do. Though, I guess I'd have to wait until some of this stuff became commercially available. I'm just wondering how different it is from typical colloidal polishing suspensions with alumina, or silica... maybe its even titania. Then lastly, I'm curious about the particle size and shape distribution and how "nano" it really is. SEM, though not ideal, should give some idea.

[relttem]

we can't say what is in it yet, but if we get to the point where we are selling it we will. I think it is the law that we have to anyway. Also, the guys on here that are doing the testing haven't reported on any adverse effects. I believe there has even been some leaks with nothing bad happening other than the leak.

[relttem]

we did all the SEM and TEM.. If I remember correctly, 40-60nm was our measured size.

[TNTSunshine]

I don't think it'd be a problem in the short run for PC users, unless they went way overboard. In general, when the stuff drys, you get the powder left behind, which can be a health hazard when its substantial enough, like in a metallography lab that isn't cleaned regularly. Though, this is only really an issue at sub 0.3 micron powder sizes, which is technically "nano." Again that's only in large quantities that it would be an issue, such as when you're mixing your own suspensions. So I'm not worried about that for PC use.

About the contents, yeah most companies have to disclose the main ingredients, but they don't typically give stuff like the particle size/shape distribution stuff, or quantitative compositional info on ingredients. Good stuff though.

[NaeKuh]

For these conditions, is the 1 to 2 degrees difference statistically significant? If so, with what confidence interval?

okey for those of you guys who are still having problems understanding this...

I have no idea about how the coolant works, but i know the theory behind the coolant so let me explain it to you guys.

The coolant is designed to be heavier, meaning the amount of molecules it can accept is widen.

Remember when we said water can only pick up X molecules and release it, its tweeking this statement by picking up X+1.

How does this help us? Because more can be picked up at one pass. However it comes with a big price.

First off i dont recomend injectors, would be VERY BAD as your sand blasting particles now, meaning yeah, your really sand blasting. -however i heard the particles are small about 1000 micron size, so im not too sure, either avoid high pressure injector systems.

The cost of the fluid is one, the second one is the viscosity, and how people are saying its a milky texture gives it away.

So the creators statement saying that it helps people with low flow is correct, however it will hurt you people with CRAP pumps. Because the fluid is more viscous now. However i dont know what the exact gains would be on such a crap pump because remember X+1.

There... now that we understand the theory,

WHO DO I NEED TO KILL TO GET A SAMPLE.

[TNTSunshine]

I wasn't asking "how it works." I was asking about the whether or not the difference between using straight DI vs the new fluid is statistically significant:

http://en.wikipedia.org/wiki/Statistical_significance

Though there are other formulas for determining this than those given on the wiki page. And the answer is we don't know for the dual core b/c there was a sample size of 1 for each condition. But skinnee said he plans to do this for the quad core, so we'll see how it turns out for those conditions.

As a general rule though, one cannot simply say A is bigger than B, therefore A is better than B (or whatever). One has to first demonstrate that the difference between your two conditions is statistically significant, perhaps through ANOVA. I know that that's a pain in the arse, but I was really getting at that though means are nice, variance is just as important. Usually a standard deviation is just fine for characterizing the distribution of a sample population (assuming its normal).

[relttem]

good stuff..alright, the nanofluid is not any 'heavier' than water. Its density is the same, which is a measure of its weight. The viscosity doesn't change either..as best I can remember - I will find out today. The particles that you get in your tap water are many times larger than the nanoparticles. 1000 micron is 100,000 times as large as a nanoparticle. That is quite a difference. You won't see any sandblasting effects with our stuff. Now, if the concentration of nanoparticles were high enough that might be an issue. In the lab we were flowing the nanofluid past a copper plate at a Reynolds number of over 400,000..that is cruising. The copper plate was painted black because we were shining a laser on it, and the nanofluid did nothing to the paint. What is going on is the convective coefficient is increasing...as that increases, so does the heat transfer. Remember to PM if you want the paper..you can read all about it and get a better understanding. It even has SEM and TEM pics in it

[relttem]

this was posted in WCG section. I am not sure if he has compared it to a water system yet or not, so the point might be mute..still, its good news

EDIT: back OT, I just finished a loaded 8 hour run at 3.45g on this old B3 quad, with the nanofluid. I have to check my records, but I think this might be a new high for this machine....Still testing points.

[Kayin]

OK, but we keep asking-where can we get it?

[iandh]

I don't think it'd be a problem in the short run for PC users, unless they went way overboard. In general, when the stuff drys, you get the powder left behind, which can be a health hazard when its substantial enough, like in a metallography lab that isn't cleaned regularly. Though, this is only really an issue at sub 0.3 micron powder sizes, which is technically "nano." Again that's only in large quantities that it would be an issue, such as when you're mixing your own suspensions. So I'm not worried about that for PC use.

About the contents, yeah most companies have to disclose the main ingredients, but they don't typically give stuff like the particle size/shape distribution stuff, or quantitative compositional info on ingredients. Good stuff though.

FYI

OSHA says <10 micron = no-no

We have trouble buying sandblast grits lower than 12 micron from some places, we use for surface prep on some thin film coatings that we do.


Kinda silly to me because I live in the desert and on a windy day I am breathing in all kinds of stuff smaller than that, but whatever.

[T_Flight]

Where and How Can we get it to be more exact. I think this is the neatest stuff...very cool.

[cegras]

40-60nm = 0.040 - 0.060 microns.

You won't be noticing anything related to a 'thick' solution or a solution of suspended macroparticles. This is truly 'nano' fluid.

[relttem]

I am not sure, but I think OSHA is dealing with the powder form. Ours is already in the liquid, so it would be difficult to breathe. That is one thing that we are looking into now - Regulations. Since the nanotechnology is so new not many regulations are there yet.

[T_Flight]

3 times is a charm for me. I'm calling this unobtanium. I am assumning that no answer means that it's not available.

[BlueAqua]

3 times is a charm for me. I'm calling this unobtanium. I am assumning that no answer means that it's not available.

I don't think this was meant to be a retail release right now. I think we're pretty lucky to test it and at least gain some knowledge on it.

Take it for what it's worth, learn from it. If it comes to the market then great, if not, oh well.

[relttem]

I put this in WCG:

We are thinking about sending out one more sample for testing, but we want someone to run it thru an off-the-shelf WC system cooling a quad. If anyone on here has something like that send me a PM and let me know about your set-up. A few things need to be monitored (temps etc). I go thru them in the order received - if I receive any...

thanks.

[Deneteus]

What does this stuff do to a radiator after it has dried?

Did anybody finish the testing?

[Serpentarius]

so with these tests, does that means lower flow is better?

maybe becoz the vicosity comes play ... higher vicosity (thicker,richer) means more molecules per mol ..
the closer the molecules .. when the molecules vibrate (due to heat), the closer the molecules, the faster it heats up by banging each other
so the heat absorption is faster?????? it is something like this ??? i cant understand it


PS: you know that milky texture, it reminds me of the fluid being fortified with calcium .. lol

[relttem]

actually, with our tests the resulting increase was almost linear in terms of increase over DI water with velocity - just a small increase in heat transfer as the velocity increased.
Viscosity does come into play, but that doesn't change as you speed up the fluid. Tho, your fluid will heat up as it goes fast the change in temperature is not large enough to have a impact of the viscosity.

You always have vibration of the particles - called Brownian motion. But, this will be overcome by the turbulence of the fluid. If the fluid was still there might be some measurable Brownian motion effects.

What we have found is that your convective coefficient, h, increases.

So, you can start with this equation

Nu = .332*Re^.5*Pr^.333

Nu = nusselt number
Re = Reynolds number
Pr = Prandtl number

The Reynolds number takes into account the viscosity and velocity of the fluid - Re = (density*Velocity*L)/(viscosity)

What we did is measure the viscosity at different temperatures, we used a few different instruments to measure the velocity (doppler and pitot tubes), and we knew the distance, L, at which we were measuring temperature down the plate). We also measured the inlet temperature. So, we could measure Re, the Prandtl number was known, therfore, we could solve for the Nusselt number.

The Nusselt number can also be found by Nu = h*L/k

h - convective coefficient
L - length
k - thermal conductivity

So, from above, we solved Nu based on our Re, and we then we used the second equation to get our h, since we measured k and the L was the same.

[SoulsCollective]

Sorry for the revival, but did quad tests ever end up happening? And what's happening with availability of this stuff?

[skinnee]

I just received the next two mixes to test. I am setting up a second bench so I can run these tests while I rerun all of the CPU block tests. Trust me this additional wait is a good thing.