Cooling Tube diameter choice and a little theory

What size tubes do all of you use? It seems that the major concensus is that 1/2" tubes supposedly give you the maximum performance.

Flow velocity is determined by flowrate and cross-sectional area of the tube by the following relationship:

Q = A x U

Where Q is the flowrate (the output of your pump ideally and is constant but can be changed due to flow resistance), A is the cross-sectional area (related to the inner diameter of your tubes), and U is the fluid velocity.

So this says that smaller tubes yield higher velocities (which is sorta good in a way, since you get some nozzling effect when it dumps the water into your waterblock), and flowrates aren't impacted much by size of tube (assuming your flow resistance is minimal). But water is a viscous fluid so tube diameter will determine how much of the water is outside the viscous boundary layer, and the more of it that's outside, the more of it will flow at the velocity U calculated above. So we'll see some flow resistance, but we need to calculate Reynold's numbers, kinetic viscosities and yucky stuff like that to figure that out. Also adding things like Water Wetter change the viscosity so that'll also have an impact.

How much of an impact in cooling do you guys see between 1/4", 3/8" and 1/2" ID tubes?

Maybe I'll go do that this weekend if there's no info on this, but I need to find tubes first. The experiment would be connect the same lengths of tube up to a pump (about the amount you would use in a water cooling setup), and run the pump and see how long it takes to fill a bucket up to a certain volume, and the flow rate would be given simply by that volume/the time it took.

 

This is what i dont get about these new 3 barbed blocks:

There's LESS time for the water to absorb the heat from the block cause it spends fraction of the time in it compared to that of the old maze's. Yes, you get direct die cooling (effectively), but its in and out quicker than you can shake an electron at it (which is uber fast believe me).

Im not an engineer or a physical chemist, im only going by logic :/

I think more water = better, but youve got to use it effectively. You need to find the heat conduction speed of water for starters.

3/8 to 1/2 = something like a 44% more volume of water. Someone worked it out on OCS ages ago.

 

I think those triple barb blocks actually slow down the water flow after running past the core. So you get high velocities when it the water hits the core (higher velocites = better cooling), and the slow down right allows the water to absorb heat from the block (not the core). It also plays with pressure to get that nozzle effect from the input, since in order to get an effective nozzle in a closed system like that, you need a dramatic reduction in pressure from the nozzle (effectively 0 guage pressure), and the best way to do that is simply to increase the pipe diameter (or in this case use 2 pipes to get effectively double the pipe diameter).

Doesn't really change the AMOUNT of water that flows across your processor, it just changes at what speed that water flows past. A smal amount of water zipping by at a high speed will make better use of the thermal properties of water than a large body of water moving by slowly (I mean it makes sense that, since you're constantly replenishing the top of the processor with cold water instead of letting it sit there and get warm you get better cooling).

-Lemur 6

 

Ah now i see.

 

Different types of waterblock work differently, as I'm sure you can imagine.

The White Water, was probably the first block which was designed from the ground up for optimum performance.

Water spending a long time in the block is not beneficial if the time is not used efficiently. This was the problem with maze style designs. Granted the water spent a long time "in" the block, but most of this was too far from the die to be of any real use, because the temperature difference required to get thermal energy from the die to the outer reaches of the block is too large.

The fact that the white water exhibited a significantly lower thermal resistance than the maze 3, despite the water being in contact with cooling surfaces for considerably less time is a good indication of this. By concentrating a large surface area above the block, which ALSO has a thin barrier layer of slow/stationary water at the surface, (the thickness of which is determined by the interaction of water with the block, more turbulence, thinner barrier layer) optimum performance is achieved.

Turbulence was achieved with jet impingement which virtually abolished the barrier layer at the cooling fins in the white water. The cascade is better than the WW since it uses dual impingement. Ie there is impingement on two types of surface above the core rather than just the one in the white water.

The reason the dual outlet was used was to allow for an even flow of water through the cooling fins, while allowing the water to get out of the block as easily as possible. Once the water has flowed over the fins, it is of no more use, and the best bet is to get it out of the block as quickly as possible.

Hope this helps

8-ball

 

Let me reiterate, dual outlets does NOT increase the amount of water that exists the block and does NOT increase the speed of the water. The AMOUNT of water stays the same, if 10 mL of water enter the block then 10 mL has to exit the block, no more, no less, simple mass conservation. The speed SLOWS DOWN, also because of mass conservation. The flowrate equaition I gave above: Q = A*U, tells you this. Q is constant, so if A increases U has to decrease, and vice versa.

Flow resistance also is unchanged between single outlet and dual outlet, since both of them sooner or later merge into a single hose via a Y split. HOWEVER, if you did not merge the two and had a reservoir with 2 inputs and connected the waterblock's hoses directly into the res without using the splitter you'd get less flow resistance and more flow (how much might be negligible though since the rest of the system is also responsible for flow resistance).

-Lemur 6

 

I think I know this.

Anyone could tell you that it is not possible to have more water leaving a block than entering it.

Let me explain EXACTLY why the white water had a dual outlet.

Firstly, a central input was used for obvious reasons. The coldest water is applied to the point where it is most important, ie directly above the core.

The reasons behind the dual output are less obvious.

Essentially, it would require a more complicated design of the base to allow for even flow in both directions through the channels while maintaining a single output. This also reduces the pressure drop associated with water leaving the block, which reduces the overall pressure drop associated with water travelling through the block, since the total pressure drop is the sum of the pd for water entering, flowing through the fins, and leaving the block. Reduce the pressure drop associated with the water entering the block through the jot nozzle, simply by removing the nozzle, and performance suffers, since jet impingement is the key design principle behind the white water. Reduce the pressure drop of the water flowing through the fins by increasing the fin spacing, and the surface area available for cooling is reduced and performance suffers. So neither of those is preferable. However, reduce the pressure drop associated with water leaving the block, and performance doesn't suffer, but the overall pressure drop of the block is reduced. If the overall pd is reduced, then using the same pump in the same loop, the flow rate will actually INCREASE. Increased flow rate, without adding additional heat by using a larger pump will increase the heat transfer efficiency of both the block and the radiator, though the latter will be affected less.

Both theoretical simulations using computational fluid dynamics software and practical experiments showed that the white water performed better with a dual output setup as opposed to a single output, by as much as 3 degrees. (AMD system I think)

The key point is that it allowed for symmetrical flow through the block to either side of the input.

This case is not the same with the cascade however, since each cup has it's own jet nozzle, and water which has already passed through one of the cups has little effect on water passing into the cups. In this case, two outlets was shown to have little benefit to cooling efficiency when compared with the single outlet.

Finally, your point about flow resistance not changing since the water eventually flows back into the same single 1/2"ID tubing is flawed. Flow resistance is not limited by the most restrictive component, but is a sum of the resistance, at a given flow rate of all of the components, so reducing the flow rate of any part of a watercooling loop, provided doing so does not directly affect the operation of a heat exchanger, (like removing the fins from a water block) will reduce the flow resistance of the whole loop and improve performance. The trick is to know which actions will reduce flow resistance, since some which may at first seem to obviously reduce flow resistance may actually do the opposite.

Sorry for ranting

8-ball

 

8-ball, you lost me... Maybe it's your terminology, or maybe it's your view point, but I don't get it... Maybe it's the assumptions you're making and not stating...

Maybe it's my theory clashing with your real world knowledge. Ah well, there's always testing...

So, returning to the original point, which tube diameter is better in your eyes: 1/4", 3/8" or 1/2"?

-Lemur 6

 

I think Pug would have something to say about that. I tend to work with theory as well.

For high flow blocks such as danger den, swiftech, anything by cathar, dTek etc, particularly when used with a low pressure drop radiator such as a heater core/black ice, then 1/2" is clearly the preferred option.

As for the typical german low flow designs, there appears to be little benefit to running large diameter tubing, however, this is something I plan to test in the near future.

Hope that helps.

8-ball

 

Agreed.
This is why I wonder why it's fitted with the same size barbs throughout - Anyone can visualise that an effective application (in a closed loop system) would be to reduce the diameter of the two outlets to balance out the larger single diameter of the inlet.
The only way I see the two outlets' pd negation coming into effect is if both tubes meet in an open reservoir.

[OT]Out of interest, can I ask if you know if this was found to be the case in a tower based setup or just on a desktop based simulator?
The reason I ask is slightly unrelated but I keep meaning to see if top inlet has an advantage over bottom inlet (or even whether side to side is preferable).
I'll save my reasonings on this for another thread though.[/OT]

and

See my first point. Thus -

_____________

I should have the remainder of most of the parts I need for testing various tube diameters by the end of next week, hopefully.
Need to find some Tygon/Clearflex in the bigger sizes yet and get a few large barbs in but I have plans to test 5.5mm, 6mm, 8mm, 10mm & 13mmID (American translation - really small up to 1/2"ID)
Just to prove a (personal) point (to myself), I have OCWC Atlantis blocks running fine on 6mm ID on a continuous tube rad here (have been for months now) so both camps appear to be able to suffer a little give and take. Last time I checked, these were apparently "High Flow" blocks...

Hey, it all starts with theory. I just find it more useful when people label it as such.
*/puts away his model M*

Damn, there just so little to disagree with here.
I'd add the parts in blue as my contribution though.


As an aside and something that doesn't often get a mention, I think one thing that may set the Cuplex Evo and the upcoming spate of microchannel blocks apart for use as so-called "Low Flow" blocks are their ability to harness increased internal surface area without the inherent restriction being compounded by a high flow rate.

[OT] Water Wetter users can actually suffer with high flow rate systems too, as the coating it is supposed to leave on your blocks gets washed away by high (turbulent) flow and deposited in tubes (where the flow is more laminar and friction coefficient higher) hence the whitening many people see. (Hey, this is a theory thread, right? )[/OT]

 

To be honest, i still stand by my theory of a Copper (hydroxide) solution in distilled water with brass barbs. Im sure thatll work better than glycol based stuff.

 

Didn't I mention that already? I do think the block would benefit from having two outputs of the same diameter as the input, but both outlets would have to empty into a res.

I think I know what the problem is here. This whole time I was assuming the waterblock has the least amount of cross-sectional area for fluid flow in the system (or atleast smaller than the input and output). I was assuming the water block acted as a nozzle, and that the 2 outlets was a way of creating the needed pressure drop for an effective nozzle (so there's an increase in flow velocity in the block as well and not while the water is just entering the block).

So in high flow designs, the internal volume of the block is such that the flow velocity would slow down (assuming one input and one output, and when using smaller ID tubes), is that correct?

But regardless, if flow volume isn't affected by tube diameter, then there would be no difference between using large ID tubes and small ID tubes, correct?

I thought inches were British units I'm sorta biased toward metric myself

-Lemur 6

 

No, we're pretty much metric in the UK, but there are still a few areas where imperial measurements are still used. Get's really confusing.

Also, the pressure drop argument I mentioned is really minor in comparison to achieving a symmetrical flow either side of the central inlet. This only really applies to white water style blocks, though there may be others.

8-ball