Going without radiator fans?

[BHJohnson]

Context: I've got a watercooling loop with an EVGA 1080 FTW and a Ryzen 5 2600, both overclocked but only to the stable clock I could get when they were one air. I haven't retried to see if I can get more on water. My case is a Thermaltake View 71 with a fan in every slot. The radiator is a top mounted 140x420mm from EK.

 

My goal is to reduce the noise signature of the computer. It's not bad now, but the radiator fans spinning up are super loud. This makes sense, since they have the most restriction and the turbulent air flow off the blade tips is hitting the metal fins that are at an angle to the original path of air. I've done some research on fan shrouds to distance the fans from heatsinks and fins, and they seem fairly effective at slowing the air down and evening the flow out to get a more even spread of air across the heat transfer surface with less noise.

 

So I logically took this one step further and said "What's the biggest shroud I can use" which started with measuring clearances in the case but ended in "The case". My plan then is:

 

Seal all the openings that air could vent besides the radiator. This will be done with weather stripping, flex tape, and silicon gasket maker. The PSU will have gaskets on the suction (bottom) and back. All the fans will have rubber gaskets or be sealed with silicon. All of the fans will then be pointed to push in, and they are all pressure optimized to some degree, because I only used Riing fans from Thermaltake as that's what shipped with the case, and they chose to use pressure optimized fans for case fans because reasons. Remove the radiator fans to lessen the airflow restriction as much as possible. Possibly either mock up a mount for the radiator out of bar stock or sheet steel because the radiator mount has a lot of material in the way of air flow. This should then cause all air to flow from the 6 remaining fans to blow through the radiator, theoretically providing enough air flow to cool the system.

 

The issue I can see happening on this setup is if there's too much pressure drop across the radiator and the fans just can't move enough air to cool the air cooled components on the mobo and drives. (This fear was partially confirmed when I did some testing of random airflow configurations and when all 9 fans on the case were pointed in, there were major heat soak issues from not being able to replace case air fast enough through the fairly large vent holes. 3 of these fans were pushing through the radiator into the case, so I think that was part of it, but it's a data point.) It has 4 HDD's in RAID 5, so they make some heat. The drive cages are setup to get air flow on their side from the front fans and the bottom drive at least will get air flow from the bottom fan. The rear fan is also an intake, so that will supply some air flow to the VRM's. And the two space heaters (CPU/GPU) of the case are both on water with an exhaust radiator so ambient case temperature should be low.

 

Has anybody tried this before? I'm working on a proof of concept build right now just to see if it's completely off the wagon trail. I'm using way more flex tape and gasket maker than I will on the final product just to do a quick and dirty operational check. It'd be cool if other people had tried this and had any lessons learned to share, or have some other guesses of possible issues to be overcome.

[brandishwar]

Turbulence is going to be your primary issue here, as you've already discovered. Your expectation of the air direction after sealing off every other possible avenue of escape is unlikely to mirror reality. Having exhaust fans is about directing the airflow in a particular direction.

 

This would work as expected if convection could evacuate the air at the same rate it was flowing into the system. If air is flowing in faster than it convection can take it out, you're going to have turbulence circulating the air in the system with little of it actually leaving until it has nowhere else to go. This is why it's important to have exhaust fans directing the airflow where you need it to go, as well as provide enough static pressure to push it away from the system so it doesn't get circulated back in via the intake fans.

 

In your instance, you're not going to get away with no fans on the radiators or pointed toward the radiators. That's the only way to break the turbulence and exhaust the heat in a standard computer chassis.

 

The only other way to ensure little turbulence would be a fan configuration similar to what's in a rack chassis, in which you have intake fans blowing a channel of air across everything, with it all going out straight through the opposite of the intakes. Since all the air is flowing just in one direction, any turbulence that forms is quickly broken and doesn't result in heat soak unless the fans are weak. In a typical desktop chassis, the intakes are at the front (and bottom), exhaust at the back and top. So even with just front-only intakes, you shouldn't get heat soak provided the fans have enough airflow to replace the internal air fast enough. Exhaust fans can further help prevent the heat soak by giving the air direction to flow and help with replacing the internal air.

 

With what you've described, you're creating a lot of turbulence inside the case, leaving the air and heat inside the chassis with almost nowhere to go. Again you're not going to be able to get away without exhaust fans, since you need to give the air direction to flow, especially when you have intakes flowing in from all but one side.

[BHJohnson]

I obviously didn't think of that. There won't be anywhere for the air to go besides through the radiator, so volume of air past the fans will equal volume of air past the radiator. The issue will be, as you said, areas of high turbulence where air can get stuck and just spin in place, causing whatever is in that location to get heat soaked over time. VRM's and memory are the two that worry me the most on that, as they are the only things that make appreciable heat that don't have direct air flow over them. Memory is easy enough to fix, since I've got an hold G. Skill memory fan/cooler I can hook up to keep air circulating over them. VRM's are right in front of the rear fan, but without air getting pulled up over them, the upper set might have issues. I could try to find a full cover water block for the VRM, but that's probably a bit much overkill. I'll do some air flow tests to make sure the VRM is getting the air it needs, and that there's no dead spots I'm not OK with.

 

I also think properly balancing fan speeds can mitigate a lot of that. I don't have the knowledge in fluid dynamics to predict ahead of time what they all need to be, but I just ordered a 6 channel fan controller with rheostats for each channel so I can tweak in the fans. With all fans running at 50% under a CPU and GPU stress test simultaneously, I got 71C on the CPU and 40C on the GPU after an hour. So I can dial the fans fairly low to clean up the air flow, and if vortexes occur where I don't want them I can raise or lower a fan speed on one side to get rid of it. I think the rear fan will have to be about as slow as it can be for this to work, and turning it off/blanking the hole might be the best bet, assuming the VRM's get enough cooling in that configuration.

 

I may also come up with a creative way to solve this when I get there. Maybe a way to directly cool the VRM's other than a full cover block.

 

Edit: Or a cowl/duct to redirect the rear fan air up to keep it from interacting with the air from the front fans. Or the front fans might have enough juice to cause both streams to turn up when combined with the bottom fans. I'm not sure, and I'm gonna have to do a lot of testing on this, then figure out what the problem is and fix it.

[Neo-revo]

11 hours ago, BHJohnson said:

Has anybody tried this before? I'm working on a proof of concept build right now just to see if it's completely off the wagon trail. I'm using way more flex tape and gasket maker than I will on the final product just to do a quick and dirty operational check. It'd be cool if other people had tried this and had any lessons learned to share, or have some other guesses of possible issues to be overcome

mostly my thought is even if you try to force ventilation by sealing every thing you can either win or lose.   not really in between.  either it will work,  or after entering the air current doesn't have the same pressure and  you get sub par  cooling,    there is nothing stopping air from back flowing esp if you set them on a curve and they slow down.

 

Certainly if you entire goal is to just use your case fans to 'move' air through the case only.  The idea sounds good on the out side till you start to think about all the metal seams, rivets.  Even if you tape/caulk/cover every thing there is nothing stopping the air from coming out the case door edge.  unless you put a gasket and screwed it in  with thumb nubs, or a more permanent method. 

 

this would be easier IMO to purpose build it out of wood or whatever you can 'work' with.  (and easier to seal the air out)

 

its a good concept. but  there will also still be dust and shit to consider (my intake fans get a lot dustier than my rad push fans)  and that would effect the performance in teh end (or make you clean your computer a lot, not sure)

[Neo-revo]

48 minutes ago, BHJohnson said:

Edit: Or a cowl/duct to redirect the rear fan air up to keep it from interacting with the air from the front fans. Or the front fans might have enough juice to cause both streams to turn up when combined with the bottom fans. I'm not sure, and I'm gonna have to do a lot of testing on this, then figure out what the problem is and fix it.

yea make you own case/box and try it. it will either work ok, or not really work.

 

its the fins on the rad that cause's the resistance.  sealing up everything else doesn't change this fact. 

The fluid moves differently through my system when i run one or the other, and they are set up like a push/pull in series right after each other,   when i run intake only its sucks well, but when i turn it off and turn on the 'push' pump it exits more violently.  together they are like a damn fire hose when i fill up my system at first. 

POINT BEING,  fans working together would be almost equivalent  but takes more physics to explain since you cannot get better performance like running pumps in series if you just stack all your fans up  air can be pressurized, water not so much.

[BHJohnson]

Dust isn't really a concern of mine. The case fans will be running regardless, so the same filter cleaning will be required.

 

As for rivets and seams and such offering an alternate path for air, I thought of that. But I'm not going to go that all out until I test it and see if the radiator actually offers that much back pressure. By allowing the air to expand to the volume of the case, the air slows down raising pressure. Basically the fan blades hit the air and throw it, giving it kinetic energy. That kinetic energy then changes to pressure when it slows, similar to how a centrifugal pump or turbocharger works. So by slowing down the air and using the whole radiator for flow vice leaving the air fast and forcing them through the relatively small areas on the radiator where air comes off the radiator fans, it should be a relatively low pressure drop. This more efficient laminar flow across a wider surface area of the radiator will, I think, be more efficient, as evidence by lower energy loss to noise.

 

And multiple fans in series can raise flow, if there is sufficient pressure. Putting too many fans in series raises the air speed, lowering the air pressure, so the air "tears", effectively causing a stall on the blades, which is synonymous to cavitation in a water or other liquid pump. Fans are just centrifugal pumps designed for low pressures, so they follow similar pump laws. Putting them in series raises the pressure they can "pump" against, while putting them in parallel raises the flow rate.

 

Oh, while we're talking about the pumping characteristics of fans, neat thing I just consciously realized from reading on this is the shape of air coming off a fan. It doesn't come off the fan evenly in any sense of the word. It comes off in a fairly concentrated cone going out at a 45 degree angle.