Does loop order matter? (pressure)

No, bends increase water speed. Friction is irrelevant, since it's a force present in the whole loop. The water goes round the corner, but it doesn't stay in the same orientation, if you understand this. It stays more to one side, and spins a little. This is due to centrifugal force. You know when you spin a bucket of water around your head on a rope and the water stays in the bucket? Well that's because the centrifugal force is greater than the forces around it, namely gravity. The effect is similar to that of a point bar in a stream. At least, this is my understanding. Also, you pointed out that the water speed would increase when going through a confined point, this is part of the same principle. Forcing it to take a different path increases the speed of the water since the flow of water still coming from behind remains the same and so the water must speed up because of the increased pressure. And most water blocks are narrower than the tubing, so the speed would increase here as well.

It is possible to do water cooling without a pump, but since the water will be moving significantly slower the temperatures will be reduced due the water only being propelled by convection, which means that it has to get hot before it can move on. With a pump though, the differences in speeds don't matter much because the flow of water coming from behind a single point in the water will always be the same.

At least, that's my understanding of the principle, i don't know if it's entirely true but it makes sense in my head. I do know however that water speed increases while going around a bend, so everybody's concerns that bends and constrictions in the loop will reduce water flow are meaningless. I also know however that after a certain point water speed is irrelevant to the temperatures, and if the water moved too quickly i assume the temperature would rise since it wouldn't get enough time to transfer hear energy from the CPU or other components to the radiators.
 
It wouldn't slow back down again as much as you might think, you have to imagine you ARE a bit of water. You've just gone though the bend and sped up. The only forces making you move are:

A) Gravity
B) The water behind you

The water behind you has just gone though the bend also, so that's pushing you forward at that speed.

If you're having trouble understanding the principle then watch this:

http://www.bbc.co.uk/programmes/b00lz9fp

And find the bit where they're making a water powered jet pack. They explain their design and why in it they used bends to increase the speed of the water leaving it.
 
I fear you are very confused Superewza.

Firstly friction is never irrelevent. Ever.
Secondly that is centripetal force, which is a psuedo force anyway and nothing like friction.
Thirdly even if you chose a bend with a smaller cross section than the rest of the pipe, and so succeed in water velocity increasing, mass flow rate through any cross section is going to be constant because mass is conserved and the speed after the bend is going to be the same as before.

The reason this is a stupid way to design bends is that it exaggerates the effects of friction so you lose even more kinetic energy to the bend, and as velocity is a function of kinetic energy you then lose flow rate. One bend slows down the entire loop by a little bit.


Water will not flow through your loop without a force. It just won't, that's friction again. You can supply this using a pump, or through care with convection systems, and it doesn't change that more resistance in the loop means more force is needed to maintain a given flow rate.

The reason you are convinced of this is that your reasoning is poor and you are unwilling to accept different viewpoints than your own. Bends and restrictions in the loop are not meaningless, you can at best make the case that if the pump head is sufficient then you can ignore a large amount of restriction. Which is kind of what this thread is about, how to overcome a large peripheral resistance.

Your final point of if the water moves too fast it wont be able to take heat away from compenents is also complete rubbish. There is always water in the block, it is always cooler than the block, so heat always flows into the water.

A large part of your argument relies on breaking the principle of conservation of mass, and the rest on 'there is no friction' yet you have faith that you are correct regardless.

gah
 
Last edited:
The speed will increase marginally, but only due to how the tube reshapes when you bend it. It'll be slightly thinner on the bend than a straight. Like a syringe, when you push an equal amount of water from a thicker area through a thinner tube, the velocity increases and pressure decreases. Simple hydraulics. Even so, while it'd speed up slightly through the thinner area, it would also slow down again to "fill" the pipe when it widens again, so there is no net speed increase.

Even if the width of the pipe were exactly the same on a bend as it were on a straight, there would still be no net speed increase. The outside water would travel faster, as the distance is greater, and the inside water would travel slower, assuming the water is always "together" all the way throughout the pipe. If every molecule of water travelled at the same speed round each part of the bend, you would be left with a pocket with no water in, and since that is eliminated when you flood the loop, the water MUST travel faster at the outside edge, and slower on the inside, which levels out when the bend becomes straight again.
 
Last edited:
I'm not confused at all, i'm just explaining the laws of physics. I was talking to Jokester about this a while ago, ask him. Did you even watch the clip i told you to?

It is very possible to create a convective water cooling loop, but definitely not with a TDP any higher than 45W, and that would be pushing it a lot, for the reasons i explained earlier.

Although as i said before also, the speed of the water is pretty much irrelevant after a certain point. To slow and there's just going to be hot water around the CPU, and not cold water to take the heat away. To fast and the water won't be able to transfer the heat from the CPU to the Rad. The whole water speed balances out across the whole loop, depending on the pump.
 
Last edited:
In the video he's using the bends to create thrust (in the same manner that JonJ678 describes a few posts above - force on the pipe walls), it's not to do with velocity.

And water flowrate (not velocity) is largely irrelevant when it comes to cooling due to the twice as fast, twice as many times in the radiator.
 
What do you mean by "the twice as fast, twice as many times in the radiator. "?

For a given water temperature, at anything above a minimum flow rate through the radiator i'd expect heat dissipated to be near constant. The minimum flow rate being the point where diffusion is no longer the predominant means of heat transfer through the water. However below this threshold I'd expect it to suffer, heat transfer through near static liquid is a lot less efficient than through a mixed one.

I was fairly certain that flow rate didn't matter as long as it is enough. This is supported by turning the pump off. However from zero flow rate up to some value (which I think is around 1gph for most loops) the system performs better, anything above this shows negligible improvements until excess heat dumped by the pumps starts to lower performance.

Similarly, at low flow rate the ek block I have in mind underperforms because it's impingement mechanism cannot function, although I believe this to be a seperate issue to the above it was what started this thread.

This is not me agreeing with you Superwza. I'm disregarding your arguments until I see a response to either my own, D D Danneh, fornowagain or MikeTimbers posts which make the flaws in your theory very clear. At best you just don't know what the terms you're using are, despite fornowagain underlining mass flow rate for you. You still haven't said if you actually have a water cooled computer, I'm starting to suspect that this is educated trolling on your part.
 
Last edited:
Which is what i've been saying all along. Water speed is a precarious balance, but people who say that bends are bad for the loop are wrong. Sure, they're not exactly that good for it but they have no negative effect. It's impossible to have zero flow rate (when the computer is switched on) because of convection.
 
Superewza said:
Which is what i've been saying all along. Water speed is a precarious balance, but people who say that bends are bad for the loop are wrong. Sure, they're not exactly that good for it but they have no negative effect. It's impossible to have zero flow rate (when the computer is switched on) because of convection.
Click to expand...

But if the bends reduce the flow rate to a point where the water is in the radiator so long that the water has pretty much reached an equilibrium and cannot cool any more, the cooling capacity of the radiator would be wasted, and the overall temperature would increase. However, that would have to be incredibly slow for that to happen, and bends would almost never cause it to go that slow. So yes, you are right that bends make no realistic difference, but there is a minute impact.
 
JonJ678 said:
What do you mean by "the twice as fast, twice as many times in the radiator. "?
Click to expand...
Say you're flowing 1l/s going through a rad that just happens to be 1l in capacity that is cooling for 1C across it. If you now double the flowrate to 2l/s it spends half as much time in the radiator per pass, cooling for only 0.5C but actually passes through twice in the same time.

It's a bit more complicated as you will really only reduce the temp differential across the whole loop, rather than actually making the water colder on the second pass, so instead of there being a 1C delta in the first case there might only be a 0.5C delta in the second case, meaning you have a slightly better cooling performance.
 
Jokester isn't agreeing with you, he's saying that flow rate doesn't make a significant difference.

Water velocity is not the same thing as flow rate, and bends definitely do reduce flow rate. As does tubing, blocks, reservoirs and basically everything that isn't a pump.

Thanks D D Danneh, that would be the problem with excessively low flow rate. i agree that it would take a hell of a lot of 90 degree bends to drop it that low, but what about an ek supreme, five chipset blocks and three radiators? Becomes a bit harder to guess :(

Got you with regards time spent in the radiator, I agree that doubling flow rate is hardly going to double heat extracted. However I remain certain that the flow rate has to be sufficient for the water in the reservoir to be mixing, otherwise it'll set up an internal gradient with hot water in the middle and cold beside the copper/brass surface. Flow rate needs to be sufficient to prevent the stagnent layer of water next to the inner surface of the pipes being too thick or performance will suffer
 
Last edited:
I know that water velocity is not the same thing as flow rate, i've been saying that in my posts for a while now. But they are both interdependant. If the water speed increases then surely the amount of water moving in a certain amount of time will increase also.

Also, Jon... who has 5 chipsets to cool on one loop?
 
Superewza said:
I know that water velocity is not the same thing as flow rate, i've been saying that in my posts for a while now. But they are both interdependant. If the water speed increases then surely the amount of water moving in a certain amount of time will increase also.

Also, Jon... who has 5 chipsets to cool on one loop?
Click to expand...

Someone with a server stack, all watercooled in parallel by a pond pump? :p

5y6c84.png
 
Superewza said:
If the water speed increases then surely the amount of water moving in a certain amount of time will increase also.
Click to expand...
Velocity is a function of the flowrate and the cross sectional area of the thing it's flowing through, but increasing the velocity doesn't increase the flowrate. Indeed, an increase in the velocity at one point in the loop indicates that you've got a restriction at that point, to increase the flowrate you want to remove that restriction (and inversely reducing the velocity in that section, but increasing it elsewhere).
 
Only if pressure is constant Superewza, and it isn't.

Jon has that many things to cool, though it can be argued that I've lost track of why I started overclocking in the first place when the watercooling costs more than the hardware.

Mosfets x2, northbridge/southbridge, ram, 8800gt, cpu. Considering the wisdom of water cooling the psu as well. This is why I've developed an acute interest in water pressure :)
 
You must log in or register to reply here.
Back
Top Bottom