Laminar flow PC cooling

[PabloDons]

So this is very obviously inspired by the 2019 Mac Pro with the cheese-grater design. I was thinking about the functionality of something like that. How would a case built with laminar airflow in mind perform?

 

So to get this out of the way: Turbulent airflow picks up heat better, but is more noisy. I don't know the details (because aerodynamics is a hell of a drug), but I think it's pretty safe to consider this factual.

 

Now looking at the Mac Pro, there is a huge amount of surface area the air is in contact with such that essentially no volume is wasted. I'm thinking this should be possible to design for a standard PC build as well. You just make a large, wide tube and line it with plates along the airflow direction. You might stick heat pipes through it, but that would disturb the airflow and cause noise. Then you literally just stick one to the CPU and another to the GPU and put fans on either side of the tubes and have their RPM synced. The challenge with this design however is cooling literally anything else. What about your motherboard and components? Also the PSU doesn't usually have any custom cooling support, so noise from that would render this kind of project pointless.

 

I don't know though. I don't know components very well, but I think I have a good enough grasp of the physics that I could have a productive discussion about this. Do you have any suggestions?

[emosun]

While fun to think about , it's a terrible idea. lol

 

Your biggest issue is trying to get quietness VIA laminar flow. Turbulent flow can actually be more quiet when surface area isn't a problem. Meaning a giant normal heat sink with a giant fan spinning slowly can be very quiet and easy to make and engineer vs trying to design a heat pipe that works without perpendicular fins.

You'll also run into the "server room" problem.... which is that a 100% silent pc can be achieved with any normal computer sitting in a different room and having some extension cables go into the room the user is in.

I'm actually struggling to think of an application where laminar flow for sound levels would be.... applicable. High end computer can be quieted by far easier means , and low end setups can be 100% passive. and when a loud machine is unavoidable , it can simply be moved to another location

 

[PabloDons]

I do think there are realistic use cases. I mean the Mac pro is a great example. It's aesthetically pleasing. It's designed to sit still and look pretty. That's reason enough to build something like this imo (assuming of course it's doable to make it pretty). 

 

Also with your example, you'd still need intake and exhaust. Depending on the performance of a laminar cooling solution, it might still have a chance of being quieter.

 

Either way, easy is not what I'm looking for anyway. If it's doable, I wanna do it. Call me old fashioned.

[Demonic Donut]

I've never had noise from airflow be an issue. Its always the fans themselves that are loud. The air being "chopped" and creating pressure waves. Higher RPMs are obviously worse. Both due to pitch and volume. Silent wings fans and Noctuas are designed for quietness and efficiency and are very quiet, especially at low speeds.

 

But a big thing about air noise (really the biggest other than straight fan/motor noise) is simply from air velocity. That's why the cheese grater works. Nice big holes that allow the air to move slowly, and therefore queitly.

 

While the idea of laminar flow is neat, you generally don't hear air "whooshing" inside the case, you hear it through the holes in the case, or you hear fan noise. The best way to get a quiet PC is to have lots of big fans spinning nice and slow with as little restriction as possible at entry and exit. That and sound insulation. If case size isn't an issue just get acoustic insulation and you're golden.

 

Simple box shape with (3) 140mm fans on each end. Normal components and a tower cooler with a fan like normal inside. Now extend 3 inches past the fans and box it in, leaving one side open for flow at a 90 degree from the fan flow. Line with acoustic insulation. Done. Low restriction, lots of large fans so they can move slow and still move a good amount of air and insulation to keep it all quiet.

 

 

[PabloDons]

10 hours ago, Demonic Donut said:

Its always the fans themselves that are loud. The air being "chopped" and creating pressure waves.

I've heard something similar. You know how it's often recommended to set up fans to blow into a heat sink rather than blow out of it? Apparently this also has to do with turbulence. Airflow into a low pressure zone is laminar and therefore not very efficient at picking up heat. I suspect this is also where the noise from fans come. Quiet fans are likely designed to be less turbulent for this reason, but I can imagine that for both efficient AND quiet fans, there is a compromise.

 

Now you are probably right about the relation between air velocity and noise, but how can you be sure it applies to laminar airflow? Either way this is part of the idea as well. 2 pairs of large fans blowing through a large tube inside a case with intake on the entire front and exhaust on the entire back. Depending on this relationship, I could maximise the size of the tubes to blow more air at slower speed. If I can get my hands on fans solely designed to minimize turbulence, it could get extremely quiet.

 

Now sound isolation works I suppose, but I'd call that a hack, not a solution.

[Overly_Analytical]

You pose an interesting question, but it shouldn't be a question of Laminar vs Turbulent flow. Since you understand the physics, let's actually think about this proposal from a heat transfer perspective. The concept of heat transfer as it pertains to "Laminar Flow" vs "Turbulent Flow" really boils down to something called the Nusselt number, which in simple terms is the ratio of the amount of heat removed  from a surface via convection versus conduction. Now, this only applies to the fluid; this isn't the conduction of heat like you would see from a CPU to a heatsink, but rather the spread of heat through the fluid itself. The idea is that when heat is being transferred away from a surface by a fluid (think of this as a single, bouncing particle) it can either bounce off the surface once and go away or many times and stay close to the surface. For conduction imagine the particle bouncing off the hot surface, colliding with another particle, transferring that energy (this is the conduction of heat through the fluid), and getting bounced back towards the surface. For convection, the particle hits the surface, picks up some energy, then goes flying away without transferring any to the other particles. In a flow with a low Nusselt Number, a relatively large percentage of heat is being removed by conduction, which means that particles need to have a pretty high chance of bumping into each other and transferring that energy. Those properties are determined primarily by the nature of the fluid (Water, Air, Nitrogen, H2, Mercury, etc). Conversely, a flow with high Nusselt number will remove much more heat via convection, which is largely governed by the flow dynamics (this is described using something called the Reynolds number, which is essentially a measure of how turbulent the flow is). Unfortunately for your proposal, air just isn't well suited to conducting heat in at low Nusselt numbers, since the density is so low that those particles won't be as effective at conducting heat between themselves. Air is overwhelmingly more effective to remove the heat with a higher Nusselt number, where a gas particle can pick up the heat and get blown away through convection.

 

So, let's apply this principle to your concept with the goal of cooling a computer using only laminar flow. In order to remove enough heat to cool a computer, you would need to slow down your flow enough that your Nusselt number with Laminar flow would be larger than a Nusselt number with Turbulent flow, while maintaining the same air flow. This would mean giant fans and heat exchangers to get enough surface area to transfer this heat (law of diminishing returns is a killer), as well as devices to correct your flow since fans don't naturally produce laminar flows. Pretty soon your system no longer fits in your Mac Pro form factor, unless you install a processor that is low powered enough to stay within the the thermal envelope you have set. At that point your Mac Pro is going to be pretty limited in performance while under load, and you still won't know how much audible noise your system will produce (although it would probably be relatively quiet with the larger fans). I personally think the ambient-cooled system that was reviewed a while back is pretty aesthetically pleasing, delivers silent operation, and would behave in largely the way you are describing (flow between the cooling fins will probably be close to laminar since the only thing driving flow is the buoyant forces from the warm air, but you would have to measure it).

Laminar flow is cool and all, but it sounds like you have created a solution looking for a problem.

[PabloDons]

5 hours ago, Overly_Analytical said:

Laminar flow is cool and all, but it sounds like you have created a solution looking for a problem.

Wow. Thanks for the reply, I learned a ton. Basically air is bad at conducting heat, so when a large percentage of the heat transfer is through conduction instead of convection, you get a pretty low total heat transfer.

 

Also if there are no fans that can create laminar flow without turbulence, it breaks the whole idea really, because that's the source of noise I'm trying to eliminate. Correcting the flow wouldn't serve any beneficial purpose. Might as well make the tubes with noctua fans, which wouldn't be much better than a regular setup in terms of noise.

 

By the way, for curiosity, where would you place noctua fans on that graph? Also how would you graph fan loudness (of the linear kind) with Reynolds number (assuming they are related)?