Multiple radiator exhaust vs intake investigation

[jamesbritt268]

Hello all -


I have been running my loop for some time but I looked around for any kind of consensus on case airflow with multiple radiators and top rad intake vs exhaust, with 3 radiator setup (however 2 radiators or any above 1 should also be somewhat similar).


There are a lot of claims and assumptions in various internet tech forums -- about certain things regarding this type of scenario:

+Balanced airflow is best school of thought (top rad exhaust)
+All intake is best

But underlying this is what drove me to test this myself, the other factors people had assumptions about regarding these two options --

+Pushing all warm intake air in case is bad (exhausting internal with all cold air intake for rads)
+Fans would be working against each other and not be as efficient (too much positive pressure)
+Various claims that, total air, heat exchange saturation would not be 100 percent, so that the exhausting rad would still benefit from air although some of the warm air would be being used to try and cool a top external exhausting rad

My testing setup:
-------------------
Single Loop
Ryzen 2700x OC PBO 1.5volts max 4.3ghz max, 4.2ghz all core boost
2080ti seahawk x - OC to 2140-60 mhz 110% power limit + 400 memory
Asus Crosshair 6 forumla
Phanteks Enthoo primo
alphacool ut 60 420 top rad
alphacool ut 60 280 bottom rad
hwlabs gts 240 rad front
dual d5 pumps (xspc bay rez)
EK waterblock for 2080ti seahawk
XSPC raystorm pro cpu block



Fans -- corsair ML 140 pro LED on one side, rosewill hyperborea 140 on opposite side for push pull rads (bottom and top) corsair ML 120 pro LED on front rad push only


Testing methodology (may be changed)
-------------------------------------------

Temp sensors placed at bottom and front rad air outflow into case (exhausting hot air into case for all intake radiator test)
Temp of water measured with sensor in rad drain port
(flow meter has not been installed yet)
Ambient temp sensor on desk with humidity measurement also
CPU socket
CPU temp
Motherboard temp
Chipset temp
CPU VRM temp

All intakes have dust filters left on, and case is closed - may test with side panel off as well.



Stress test will be 30 min of AIDA 64 CPU/RAM/GPU stress test

 

[jamesbritt268]

My goal here is to provide some measured data besides just maximum temperatures of components and actually see what is the optimal situation -

+How much cooling capacity can be affected and how much temp rise from air already passed through one radiator?
+Does having all fans as intake reduce efficiency / is balanced airflow better?
+What role does varying fan speed play (does slow or faster affect the outcome one way or another)?
+Does hot exhaust into case from rads all intake actually pose a problem?
+Which setup does have the lowest temperatures and for what components?

I'm sure I will also come to other potential valuable data points.

Now, some people will be quick to say I already knew one vs the other, but after a lot of searching on various tech sites forums, and personal experience, I decided to finally put something together once and for all for myself, and maybe for anyone else who ever came across this decision. Yes there are videos by some tech sites show single radiator testing.


I ran an initial test once I put the sensors in place one a few millimeters above each front and bottom rad exhausting into case --

You're gonna have to forgive the horrible image, because AI suite and HW monitor do not like each other --




The room initial ambient was 23.8 for my first test for my test run (ya a test of a test i know ). Humidity 46%.

The max ambient achieved after 38 minutes of Aida 64 was 24.7 -- this room is a small office room, about 10x12 or so -- I'll get exact numbers once i actually finish full real testing.

Therefore if we take max water temp -max ambient, we have our max delta T at full fan speed for all fans -- all radiators are intake (top bottom front) three fans exhaust 140mm each -- 1 back and 2 side

Case was on piece of glass, on top of carpet (more on why this may or may not be relevant later)
This is a slightly older picture but it is fully on the glass for this test - just wanted to give a general visualization.



Water temp is T sensor 1, bottom rad exhaust(internal exhausting hot air in) is water out temp, front rad exhaust is water in temp (internal exhausting hot air in).

Initial ambient started at 23.8 and rose to max of 24.7



We see that the max delta T is 32c - 24.7c = 7.3c

Max CPU temp is 78c
Max CPU socket is 48c
Max GPU was 41c
Motherboard max was 32c
Chipset max was 50c
Max VRM temp was 50c
Max front rad exhaust air temp was 30 degrees (internal exhausting hot air in)
Max bottom rad exhaust was 33 (internal exhausting hot air in)
Max water temp was 32c

Now this test's results also I notice something interesting -- the UT 60 on the bottom either has perfect water temp to air temp saturation, or in the max temp higher(?), 1 degree higher then that of max water temperature in fact.
The front rad max temp was 2 degrees C below max water temp.

So one thing that invalidates the results here for me, is that my air pressure in this setup is so positive, it spins the fans when their off (back exhaust fan for example). This concerns me because the bottom rad although thicker and in push pull, and lower fin density, has better air to water temperature saturation. I am concerned that because the case is on the floor, the bottom rad may be recycling some of that positive airflow. from the front bottom as the intake may be restricted for it sitting on the floor.

However we can draw one valuable conclusion - that regardless of blowing the hot air into the case - the CPU, CPU socket, chipset, and VRM will all more then exceed the max exhaust temperature from the radiators in this scenario, therefore the theory and commentary some make about heat building up as if it is a closed system would not matter. Any component that exceeds max water temperature, will be cooled by exhaust air from radiator as it is at worst within 1 degree of water temperature. Basically these components will still be getting sufficient air temp difference for cooling.

I will update with my findings as I continue to iron out the scenarios with case location and fan speeds.

 

[Blaylock]

Thanks for taking the time to do this.

Subbed

 

[ehume]

UT?

In general, I advise people to remove the rear "grill." If your case is aluminum, that was usually done before it left the factory. If it is steel, the "grill" is nothing more than holes punched in the back wall. Get rid of that and you can usually do without the rear fan and the noise it makes.

With nothing to impede air as it leaves, you can then go to all-intake with your rads. This is good because air inside a case is usually warmer than outside a case. So drawing outside air to the rads should produce better cooling, and the air inside the case will find its own way out -- silently. At least, that's the theory. The nice thing about what you are doing is that you are reporting data.

 

[EarthDog]

Cool testing!

The thing is, each situation will be a bit different. Generally, cooler intake air over the rads equals better temps.... at the expense of warming up the internals of the case (a bit or significantly depending on the load, length of time, etc). Whether top intake versus front exhaust or front intake versus top exhaust will depend on the case and internals used.

Since most users aren't chasing after the couple degrees C difference this can make, its sort of a non-starter. Either way, I look forward to seeing your results!

+How much cooling capacity can be affected and how much temp rise from air already passed through one radiator?
+Does having all fans as intake reduce efficiency / is balanced airflow better?
+What role does varying fan speed play (does slow or faster affect the outcome one way or another)?
+Does hot exhaust into case from rads all intake actually pose a problem?
+Which setup does have the lowest temperatures and for what components?

1. Will vary by setup
2. Whatever the water loop is cooling should run cooler since it is using the coolest air.
3. A couple/few of C in most cases
4. Typically not.. but this will vary. If you are pushing the motherboard hard and have little airflow it could tip the scales being a few C warmer due to the warmed air being circulated.
5. If all rads are getting fresh intake air, then whatever is attached will have the coolest temps. If you run with airFLOW (front intake, top exhaust) for the rads, the comonents in the loop may be a couple C warmer, but the internals will be lower due to the lower temp air coming in.

 

[Blaylock]

UT?

Alphacool UT60 Radiator.

 

[pgdeaner]

Cool testing!

The thing is, each situation will be a bit different. Generally, cooler intake air over the rads equals better temps.... at the expense of warming up the internals of the case (a bit or significantly depending on the load, length of time, etc). Whether top intake versus front exhaust or front intake versus top exhaust will depend on the case and internals used.

Since most users aren't chasing after the couple degrees C difference this can make, its sort of a non-starter. Either way, I look forward to seeing your results!

1. Will vary by setup
2. Whatever the water loop is cooling should run cooler since it is using the coolest air.
3. A couple/few of C in most cases
4. Typically not.. but this will vary. If you are pushing the motherboard hard and have little airflow it could tip the scales being a few C warmer due to the warmed air being circulated.
5. If all rads are getting fresh intake air, then whatever is attached will have the coolest temps. If you run with airFLOW (front intake, top exhaust) for the rads, the comonents in the loop may be a couple C warmer, but the internals will be lower due to the lower temp air coming in.

OP, you are on the right track and that's a good analysis by EDog there. What I have also seen is people talking about warm air rising and such. This is the same lack of understanding that makes people think that loop order can have a huge impact on temps. It's basic fluid dynamics. Think about this. A single 2000RPM Noctua fan is rated at about 70CFM. That's seventy cubic feet of air per minute. Most people have several fans on their case so the entire volume of air in your system will be exchanged with outside air many times per minute depending on your case config. I have tested on every build I have done and I get the best results as EDog suggests with the radiators pulling in outside air and a fan at the back of the case blowing out. If you could do an airflow test in a chamber with colored gas, there would be pockets of stagnant air depending on your case design but that is why most cases have large honeycomb areas on the back for airflow these days. Where I have seen people struggle is where the system sits under a desk or somwhere it can't get fresh air or warm air gets pulled back into the system. I'll be interested ot see your results.

 

[Blaylock]

^ this

 

[DaPoets]

My tower sits under my desk and it's so big (Antec 1200) that I wouldn't dare keep it on my desk. Because of this I do have to pay attention to exhaust air flow and it does make a noticeable difference. Once you're on water though, loop order doesn't make much of a difference true, however many of us here look for the "best" or most efficient way to run our systems given our unique environments. For me loop order didn't matter, but exhaust order did.

 

[pgdeaner]

My tower sits under my desk and it's so big (Antec 1200) that I wouldn't dare keep it on my desk. Because of this I do have to pay attention to exhaust air flow and it does make a noticeable difference. Once you're on water though, loop order doesn't make much of a difference true, however many of us here look for the "best" or most efficient way to run our systems given our unique environments. For me loop order didn't matter, but exhaust order did.

If that beast is a tight fit under your desk I'd bet you had airflow challenges.

 

[jamesbritt268]

So after raising my case a few inches (on a dolly with open middle and casters) to make sure i had enough ground clearance to avoid any positive air recycling I ran another run fo AIDA 64 full speed fans and what I saw was very similar:




I ran AIDA 64 for a few hours as I was also testing putting FSB up for PBO on 2700x:

Heat load for CPU and GPU:






Water temp:



Front rad air temp exhausted into case:



Bottom Rad air temp exhausted into case:



Ambient of room low was 23.8c , high was 25.2c

CPU maximum core temp was 91.4c
GPU maximum temp was 41c

Fans are hooked up to 3 PWM hubs with like fans being controlled by MOBO on each. 2200 max speed for 120 ML pros, 1800 max speed for 140 ML pros, 1300 for hyperborea 140s.

Chipset max was 50
VRM max was 48c

What we can tell here is that with thicker rads, as bottom intake, that you can easily saturate the air to water temp. If the air is already the same temperature as the water, it would provide very little if any cooling, there fore making a top externally exhausting rad potentially irrelevant or useless in a similar scenario.

I will be interested to test out both fully now that I have the case off the ground to be able to prevent some air recycling that may have occurred.

When I switch the top rad to exhaust, the side fans will become intakes again, so we will have some additional cool air coming in besides the exhausting bottom and front rad's and this may help or change the outcomes. This would be more common I feel with most peoples case setups and mine in the past.

I may also run the tests and also run with a temp sensor on top rad and see if the air temp being pulled into top rad has changed at all or what it's value is during each test.

Stay tuned!!!

 

[TeslaHUN]

Thanks for the testing ,very useful . It must have been time consuming to measure all
I had a Primo case few years ago :

Its a special case with so many exhaust vent ,can accept 2x 140mm fan at the back. So basicly u dont really need any exhaust fan and still warm air will leave @ 8x pci slot cover +2 rear fangrill.

 

[Silver Surfer]

It seems the concern is for CPU temperature when actually no matter the flow argument we are still running motherboards that the voltage regulators around the CPU socket were designed to be cooled by a CPU cooling solution that cooled those as well. We remove the designed airflow cooler and replace it with a water block and do not supply the same airflow intended to cool around the socket. Which in the long run is responsible for at least 80% of water cooled overheating motherboard failures.

It really doesn't matter about what's left of case airflow when it comes to cooling if the only cooling concern is the CPU itself, what does matter is just how capable the cooling actually is, and what temperatures can the cooling actually handle. Fortunately the CPU protects itself no matter the circumstances it is running in, but unfortunately the motherboard does not, and it is one of the most overlooked items in 98% of water cooling setups. The more expensive overclocking motherboards have good heat sinks over the surrounding socket VRs some have their own cooling fans built in, some even have water cooling capability, so obviously that is necessary to even their design specifications. Unfortunately even some of those with good heat sinks are still designed to use the stock CPU coolers designed airflow to get the units cooled.

If you are running a motherboard that was 100% dependent on the stock air coolers air flow and you removed that cooling in lieu of a CPU water block, make direct cooling those surrounding socket VRs your next priority. I think many don't remember the earlier days when motherboard providers of overclocking motherboards actually provided independent clip on squirrel cage fans to cool the socket VRs if the stock air cooler was removed and replaced with a water block. Seeing as how some CPUs today don't even come with a stock boxed cooler it falls more on our shoulders to keep those VRs cool around the socket, with either dedicated air flow, or a case airflow design that takes cooling those elements into consideration.

To the OP, nice work it will be interesting the various flows affecting the VRs temperatures, unfortunately not every water cooler has that good of a motherboard.