Return to

Physics discussion related to Gamers Nexus' AIO video

I was watching Gamers Nexus’ AIO video and one of the PCs I built has the AIO installed in a definitely sub optimal way. This led me to thinking about how to configure the AIO in an appropriate way.

So for the next days I thought about what the video said and how it applies to my situation. I came to the conclusion that one (very) specific portion of the video just didn’t made sense (to me). I therefore wrote an email to GN regarding this portion and how I had trouble understanding it/how it might be incorrect. They didn’t seem to be in the mood of arguing (which, to be fair, was also stated clearly in the reply to my second (and last) email)..

However, as I already have my arguments in written form I decided to post them here.
I would really like to know if my arguments hold up or what I am missing/misunderstanding:

For this whole post I am going to concentrate only on one specific radiator position/installation:
An AIO in the case “front mounted and with the tubes down”.

The specific part begins at time stamp 19:44 (

Finally when front mounted and with the tubes down this air
bubble issue is no longer a problem.
Ideally the pump is still below the top of the loop, because
the other tank should now be at the top of the loop.
If that’s the case we don’t need to worry about air accumulating in the pump,
because it won’t,

What I am searching for is the explanation that is missing here (well, technically there is an explanation: “because it won’t”).

It continues with:

and we also don’t need to worry about it getting pulled through the barbs,
because that’s not where the air is anymore it’s now at the top tank
that is not connected to them.
There’s one huge reminder for this orientation though:
do not mount the pump in such a way that it is on the top of the loop
with the barbs at the bottom.
Because then it hasn’t fixed anything.
If you intentionally position your radiator lower on the front panel
[…] rather than higher it is possible that the pump ends up at the top of the loop
at which point we’re dealing with all the same concerns
as showed earlier when the pump was located at the top.

This is the part were I do not understand how they got to this conclusion.
I believe that his reasoning is: “Air always finds the very top of the loop”.
After spending a considerable amount of time thinking about it I came to an opposite conclusion:
That having the pump higher than the top of the radiator is perfectly fine.

To explain my reasoning I give the following scenario:


Air in a closed loop is NOT able to move to the highest point in the loop!
This is not an assumption; also see: the Wikipedia page “Air lock”: :

The gas, being less dense than the liquid, rises to ANY high points.

The air trapped inside the loop is not going to find the highest point in the loop.
All that matters is: there is a low section (completely filled with water) and multiple higher sections where air can be trapped.
The important detail is that the relative maximum height of the higher sections does not matter: air gets trapped regardless.

When looking at an AIO in the case “front mounted and with the tubes down”: there is a low section, where the tubes of the pump enter the radiator, filled with water. And there are two higher sections the top of the radiator and the pump, in BOTH of which the air is going to be trapped.
The point I am trying to make now is that the exact height of the radiator or the position of the pump does not matter!
The physics (of trapped air) is exactly the same for:

  • The pump is higher than the top of the radiator
  • The top of the radiator is above the pump


Based on the ideas of the scenario the following contradiction contained in the video develops:
They recommend that the pump can be higher than the (water filled) tubes, but do not recommend installing the pump higher than the top of the radiator.
However, if we look at the scenario and we know that installing the pump higher than the tubes is (at all) OK, then the implication is: “the pump can be mounted even higher than the top of the radiator”.

Again in different words:
Looking at the scenario, it does NOT make sense to say:
Air is ONLY going to be trapped in the radiator if the pump is mounted lower than the top of the radiator.
<=> Air is not going to be trapped in the pump if the pump is mounted lower than the top of the radiator.
<=> How it is described in the video

In fact experience/observation with “Air Locks” tells us: air is always going to be trapped in all highest sections (therefore in the radiator AND the pump).

So the question now is: “How does the pump get rid of trapped air?”

My hypothesis is: “The pump intake is ~100% water -> Air bubbles residing at the pump are getting swept away!”

In their video they seem to be arguing similarly:

[…] we also don’t need to worry about it getting pulled through the barbs,
because that’s not where the air is anymore it’s now at the top tank
that is not connected to them.

The Wikipedia article about “Air lock” states:

Flushing the system with high flow or pressures can help move the gas away
from the highest point […]

Based on the given information I would argue that the pumps should be able to getting the air bubbles swept away easily.

A relevant next experiment could be: “How strong (or how much time) does the pump need to be to get rid of trapped air?”

Regarding the outcome of this experiment: I would claim the mounting height is going to be negligible in all reasonable cases. (If somebody is curious about why I think this is the case; please ask. This post is already pretty long :wink: )

So, again, the conclusion I decided on is:
Installing the pump higher than the top of the radiator is perfectly fine. Again for: An AIO in the case “front mounted and with the tubes down”

I would really like to know if my arguments hold up or what knowledge I am missing/misunderstanding

If there is a place for the air to collect elsewhere in the loop, then yes it’s probably fine. In an AIO with no reservoir the small radiator tank is all you’ve got for that. It’s a pain in the neck for the pump not to be the lowest point in the loop anyway as it makes filling and priming the loop difficult. In a pre-filled AIO you don’t have to worry about that, but they’re not filled 100% full from the factory, and permeation is a thing, so you’ve still got to account for some amount of air in the loop and where it will settle once everything has reached equilibrium.

1 Like

So, just to synthesize, you’re arguing that:

  1. Because the pipes are at the bottom of the rad, we have an airlock
  2. Because of the airlock we can say that the bubbles are trapped, but we can’t say where they’re trapped
  3. If the pump is moving fast enough, we can assume the air is pushed out of the pump-side, through the air lock, and into the rad?

I think that’s plausible, but it still doesn’t change the fact that the top mounted radiator is still ideal (but that’s not what you’re arguing)

Yet my AIO keeps making bubbling noises and so on… No matter it has always been top mounted.

well, ideal is a relative term…

1 Like

Basically, yes :smile: .

I agree, but that would also be pretty much true for any front mounted radiator installation. (One exception may be pump completely beneath the radiator.)

Is it a possibility though that even with the air in the tank end of the rad, for sure, and then placing the pump higher that:

The “pipes” in the rad, that is the very small metal ones that are attached to the fins which pass a small amount of liquid through, are small enough AND the pump forcing water up through one side of the rad causes a splash in the tank end that will create small bubbles which get drawn into the very small pipes in the rad. Because the pipes in the rad are so small there is no room for the water to pass by the bubble and let them rise so continues to push it down the small pipe and then through the tube to the pump.

The pump will as you say push the air back which is fine, but it will result in the millisecond of noise where the pump has less resistance and spins up creating a whine as it passes the bubble and also a momentary spike in temperature as the bubble passes through the much smaller fins means it is split up and take up relatively more space.

Not really a big problem, but less ideal in that now you have momentary noise every now and then and a not 100% perfprmant cold plate.

Though upon further thought this would happen even if the pump was below the top of the rad anyway if it happens at all.

With the liquid flowing it is unlikely for air to accumulate in the pump because the flow will just push it straight out, I work with pumps and once they get going, even if they had a lot of air in them, they sort themselves out. The problem would be air accumulating in the pump when it is not flowing, causing it to run dry when it is started.

In a sealed loop, especially if it’s stationary I wouldn’t think this would be a problem, in the sense that the air (what little there is) won’t be moving around by itself, and so long as it works in whatever orientation you have it in it should continue to work.

Still, having the air in the top of the radiator rather than trapped in the hose somewhere is a good thing. Air locks will restrict flow as they create a high pressure point and in water cooling flow rate is key. I’ve had situations where air locks have completely stopped flow but it would be unlikely in a water cooling loop.

1 Like

I say none of this matters if you tap the pump. If there’s still air in the pump it’s because you didn’t tap hard enough. /s but not really

1 Like

I think that what GN said makes intuitively sense. The point they made is to place the fittings on the rad at the bottom and make sure that the pump is not higher in the loop compared to the top of the rad.
Like if you ever had the chance to play with fluids as a kid it surely works as a way to make sure the pump is always fed with fluid as best as possible.

The link to the airlock seems to prove what GN said, in my opinion. You can consider the radiator as a collection of very small tubes, like the wikipedia example, but, in an AIO, there is a constant flow of that’s going to overcome the ability of the small air bubbles in the fins to stay trapped and they will collect in the end tank.

In the end the more the pump is placed lower than the end tank of the radiator, the better it will work and flow for sure.

If ya’ll got big spanking air cooler we won’t be here discussing but instead we’d be cracking a cold one and make bad IT jokes on sysadmins (it’s just a joke)

1 Like

I am not sure I understand what you mean by “end tank” (do you mean the radiator or the pump?)

Well, I disagree; the reason is that the energy needed for the (same) water flow is independent of the height: the pump is connected with an inflow and outflow tube. Both tubes are going to be filled by water. So now, if you increase the height of the pump you will need more energy to lift the water up. However, you will gain the same amount of energy when the water leaves the pump and flows back down.

Both tubes are going to be filled with water because

  1. Inactive pump: Air wont find the way (because of the Air Lock)
  2. Active pump: Air won’t have a chance at accumulating at the pump as most (read: all) of the inflow is going to be water.

The only case when the tubes may contain a significant amount of air would be directly after the installation or after a significant amount of time. (Until you turn on the pump…)

(Please correct me if I am wrong; but that is my current understanding)

Yup, my personal personal computers are all air cooled. However, some people mainly care about looks; and to them an AIO looks pretty cool :upside_down_face:

1 Like

I mean the other side of the radiator where there aren’t any fittings.

I’m going to nitpick for the lols and say “actually you can’t gain all the energy back because there will be some lost in unrecoverable energy like heat due to the friction between the impeller and the motor and the water, the block and the tubes”

Jokes aside I didn’t mean to say that air will accumulate in the tubes. What I was trying to say is that air has nowhere to go beside the end tank if you install an AIO the “GN way”.
If you take a bottle, fill it up with some air at the top and put a tube on the bottom of the bottle you’ll see that as soon as you lower the top of the tube below the air gap in the bottle the water will start flowing out of the tube.
This is my mental image that explains what GN said, if it makes sense.

Last blur: my systems are all air cooled too, but I’ve been considering watercooling if I swap the case for my ITX build.

Hmmm, I don’t think you can really use the conclusion of this experiment for a closed loop AIO.

I am proposing a more accurate mental picture:
You can use the same bottle, but the tube is going to be arranged differently: the end of the tube is always on the same height where the tube is connected with the bottle (i.e. the inflow/outflow pipes of an AIO are on the same height).

For as long as the water level is above the connected tube the following holds true:
After the water started flowing, it is going to continue flowing without any additional energy.
Note that the middle of the tube (i.e. where the pump is going to be) can be as high as it wants to be or make as many loops etc.; it does not matter.

Related Wikipedia article to this phenomenon:

It was already linked in the “Air lock” page… (That’s not how I found it though…)

If the pump is the highest point in the system and loses its priming then you won’t have any flow. Plain and simple.

The other problem you could run into is cavitation, which at worst makes things noisy, and at worst destroys your pump impeller.

1 Like

First up: I don’t really understand what “loosing priming” means. (I guess it describes the situation when the suction of the pump is not enough to get to the fluid?)
Based on the information I found by searching for “pump priming” it sounds like it is mainly possible/a problem if the seal fails/air can leave and enter the loop. I don’t know if that is possible in an AIO. My assumption is that an AIO is sealed shut for air.

If you think “priming” is possible in a sealed AIO please elaborate why “loosing the priming” would be a problem exclusive to “the pump is higher than the radiator” (i.e. “the pump is the highest point in the system”).
Or: How would the pump gain priming when it is not the highest point (but still higher than the bottom of the radiator)?

I could see an argument being made about “if the pump is too high it fails to gain priming”. However, I would reject the idea that the highest possible position is dependent on the height of the radiator.

(I only argue about the case: “front mounted and with the tubes down” with the pump being higher than the bottom of the radiator)

Yes, it is. That’s what permeation is. Priming is just the process of “feeding” the pump with enough water such that it starts working as you would expect. A D5 pump run dry will not pump air the same way it pumps water. In fact, if you run it completely dry it’ll kill its self almost immediately.

This is incorrect. All AIOs will have air in the system and over time will lose liquid volume.

The design of the pumps in these AIOs will not effectively move air. They need at least some fluid to work, even if that fluid has air in it.

If the pump is the highest point in the system or just a high point alone makes no difference if there is enough air trapped at the high point where the pump resides.

I think you have a decent grasp on the hydronics at play here but you’re missing the details in how these pumps are designed and what they can and cannot do.

1 Like

thats not strictly true but a wet rotor pump does use the fluid as its bearing lubricant.

Its not an immediate death though. They’re tougher than you think.

Probably true. I haven’t really checked, and it’s only happened to me once or twice when I’ve done something boneheaded and turned a pump on when I shouldn’t have. I just haven’t been that interested trusting the machine to a pump after it makes that awful screeching noise. :confused:

1 Like

I 100% agree with this conclusion.

1 Like