Point of diminishing return?

[Boostage]

Im trying to find out at what point having the highest flowing possible fans produce no better results than lower cfm quieter fans. and if I will be reaching that threshold with my setup.

Loop will consist of a single dual 120 rad and a single quad 120 rad. cooling a cpu,NB and GPU.

out of the 8 fans in my illustration. would running 6 220cfm deltas perform any different than standard 55-90cfm fans?

http://i12.photobucket.com/albums/a249/Afrodziac/fans.jpg

 

[CharlieCS]

welcome to OCF .

First of , you not going to be able to stand 6 220 cfm deltas for more then 10 minutes .Radiators are not optimized for that kind of airflow and will scale poorly after some point , at which will depend on radiator maker and model . Black ice GTX radiators are optimized for high speed fans , most other rads are optimized for low/ medium speed fans . Now to say they wont gain few degrees is not true , they will but it will not be worth it. Also your setup is inefficient , you are taking hot air from the case and putting it trough the radiator not just that , but afterward you are taking the same air ,now even warmer ,trough another radiator .

They will perform better but it is a point of diminishing returns .

 

[Bobnova]

If you're using really high speed fans (listen to youtube clips of 'em) get a few automotive heater cores, their dense fin count doesn't really come into it's own until you start moving a lot of air. They're way, way cheaper then the high end rads.

 

[Boostage]

welcome to OCF .

First of , you not going to be able to stand 6 220 cfm deltas for more then 10 minutes .Radiators are not optimized for that kind of airflow and will scale poorly after some point , at which will depend on radiator maker and model . Black ice GTX radiators are optimized for high speed fans , most other rads are optimized for low/ medium speed fans . Now to say they wont gain few degrees is not true , they will but it will not be worth it. Also your setup is inefficient , you are taking hot air from the case and putting it trough the radiator not just that , but afterward you are taking the same air ,now even warmer ,trough another radiator .

They will perform better but it is a point of diminishing returns .

I knew someone would say this but on the contrary teh air coming out of the case is very cool because I have one sythe ulta kaze 133cfm fan blowing from the cpu outside. and it is very cool. case temps are in the 40's all the time.

and my rads are both swiftech MCR series

 

[ricflairw000]

I knew someone would say this but on the contrary teh air coming out of the case is very cool because I have one sythe ulta kaze 133cfm fan blowing from the cpu outside. and it is very cool. case temps are in the 40's all the time.

and my rads are both swiftech MCR series

my CPU and GPU are in the 40s

 

[thorilan]

the whole point of watercooling is to find that sweet spot for YOU where your hearing and your max stable OC meet. because every system is different we all play a balancing act through a little trial and error and best guess to find out where that is .
many newer rads are better designed for todays range of fans so you really have to start off with what kind of noise is acceptable to you to determine which fans you will use then choose your rads from those that can perform well using those chosen fans.
i have found doing it this way people are less likely to upgrade as often and are usually happier with their setups initially ( some like to push the envelope further)

you see a lot of people that want to start out with high pressure fans and they quickly change their mind when they see systems performing nearly as well with literally 1/10th the noise or less

 

[hokiealumnus]

I actually ran a small comparison with an MCR-320 and Ultra High Speed Panaflo fans (119CFM). After running 7v vs. 12v with the same ambient temperatures, it gained maybe 1-2C. No more, that's for sure. I was very surprised.

Then again, I had a very cool running dual-core E8400 and it may have been as low as the cooling system could get it, regardless of fan speed. You couldn't even feel heat coming off the rad. Things have changed. I've got an i7 860 now and you can definitely feel the heat coming off the rad. I'll try and re-run the experiment when I have the chance.

 

[ricflairw000]

I actually ran a small comparison with an MCR-320 and Ultra High Speed Panaflo fans (119CFM). After running 7v vs. 12v with the same ambient temperatures, it gained maybe 1-2C. No more, that's for sure. I was very surprised.

Then again, I had a very cool running dual-core E8400 and it may have been as low as the cooling system could get it, regardless of fan speed. You couldn't even feel heat coming off the rad. Things have changed. I've got an i7 860 now and you can definitely feel the heat coming off the rad. I'll try and re-run the experiment when I have the chance.

really!!!
i must be a wuss then because i think my ankle is going to get cooked when im doing prime95 or soemthing haha
and this is at 40*C coolant

im going to have to get more temp readings

 

[Otter]

Im trying to find out at what point having the highest flowing possible fans produce no better results than lower cfm quieter fans. and if I will be reaching that threshold with my setup.

Loop will consist of a single dual 120 rad and a single quad 120 rad. cooling a cpu,NB and GPU.

out of the 8 fans in my illustration. would running 6 220cfm deltas perform any different than standard 55-90cfm fans?

http://i12.photobucket.com/albums/a249/Afrodziac/fans.jpg

It depends on your heat load and your chips. Once the coolant gets close to ambient, more air isn't going to help. And there may be a wide temperature band where you get more or less the same OC.

Thorilan is right about starting with your noise tolerance. It doesn't really matter whether or not the CPU will clock 5% higher if the cooling system is so loud you turn the computer off.

Are you stacking the two radiators? If so, get the dual rad away from that quad so they can both draw fresh air. Odds are that the quad alone will perform as well or better than the two stacked together.
http://www.xtremesystems.org/forums/showthread.php?t=220874

 

[Boostage]

I actually ran a small comparison with an MCR-320 and Ultra High Speed Panaflo fans (119CFM). After running 7v vs. 12v with the same ambient temperatures, it gained maybe 1-2C. No more, that's for sure. I was very surprised.

Then again, I had a very cool running dual-core E8400 and it may have been as low as the cooling system could get it, regardless of fan speed. You couldn't even feel heat coming off the rad. Things have changed. I've got an i7 860 now and you can definitely feel the heat coming off the rad. I'll try and re-run the experiment when I have the chance.

ok so it pushes 119cfm @ 12v what does it push at 7v?

 

[ricflairw000]

ok so it pushes 119cfm @ 12v what does it push at 7v?

would cross mutliplying work ?

 

[hokiealumnus]

Good question Boostage. 119CFM is just a rating. They may put a little more or a little less; ratings aren't exact. Unfortunately (and to answer Ric), trying to extrapolate the CFM from the voltage, we would need a lot more data. I am nowhere near versed enough in fluid dynamics to know how to figure it out, but it it is not a linear relationship (such as 6v = 1/2 of 119CFM, or 59.5CFM).

 

[muddocktor]

hokie pretty much hit this one square on.

You would have to have some sophisticated equipment to even try to measure this and the mathematics of calculating it are way beyond me. But, you could get a rough idea of what it could be putting out by monitoring the rpm the fan is turning at 7v and then looking at lower speed models of the fan series. If the manufacturer hasn't changed airfoil design and pitch on the impeller, then checking the cfm rating of a lower speed fan that turns approximately the same rpm as the high speed fan at 7v, you can guess at the approximate cfm and noise rating of the 7v fan.

Here's a hypothetical example:

You are looking at a Panaflo 120x38mm Ultra High Speed FBA12G12U1BX fan, which is rated at 114.7 CFM @ 2750 RPM, 45.5 dBA. So you undervolt it to 7v and it ends up turning 1700-1800 rpm. You then find that Panaflo makes a lower speed version of the same model line fan (120x38mm Ultra Quiet FBA12G12L1BX), which is rated at 68.9 CFM @ 1700 RPM, 30 dBA. Assuming that the impeller design hasn't changed, then the undervolted fan should be psuhing around the lower fan's rated cfm and noise. Fan specs were pulled off of Sidewinder's site, BTW.

 

[hokiealumnus]

Heh, oops...my memory failed me. Please disregard all claims that my fans are 119CFM and pretend I said 114CFM. Sorry about that.

 

[muddocktor]

Heh, what's 5 cfm amongst friends.

 

[speed bump]

You can figure out your CFM that your fan moves using a simple Pitot tube and a manometer. Build a box to seal around the fan and mount the pitot tube downstream from the fan and get the pressure reading. After you have a pressure reading then throw that in Bernoulis (make sure your conversion factors are correct) and you can extract velocity. After that use the continuity equation to find your volumetric flow rate.

As far as picking out a sweet spot what I would do is figure the change in water temperature you want across your Radiator and then the mass flow rate you have going through your system. Throw that in the heat exchanger equation q=mdot*Cp*delta T where in this case Cp is for water. Then using a psychrometric table figure out your enthalpy values for air at the temperature change you want to affect (probably inlet water temp and ambient) put that in the equation q=mdot*delta enthalpy (you don't have a consistent Cp on this side). Set your q's= to each other and solve for mdot of the air. Taking that and the density of air at ambient (from the psych chart) you can figure out your mass flow rate.

If you want a better approximation that can be done using Differential equations to figure out the best possible temperature change you can get across your Heat exchanger but if I remember correctly its a numerical approximation.

I would personally reccomend big rad and lower noise fans unless you only run computer when your using it and like wearing closed ear head phones.

One last thing with the stock fans how you have shown, if you do that don't expect them to last long because the present a restriction for those deltas and will probably be over speed by them to the point where the bearings go.

Before I forget EC would be proud of your fan choices.

 

[baditude_df]

For me, in my recent testing here, I found that the best performance came from the 10v setting with these fans. Full 12v spec, actually resulted in slightly higher core temperatures that the 10v spec in some cases. In most cases however the 12v was rougly the same or slightly better than the 10v, but nothing to write home about.

 

[Otter]

ok so it pushes 119cfm @ 12v what does it push at 7v?

Roughly half that. You might find this article helpful.
http://www.overclock.net/water-cooling/503407-fan-testing-part-2-radiator-tests.html

 

[voigts]

I personally would try to see if there are posted C/W values for the specific radiators you are looking at using at Skinnee Labs. For instance, the case of the Black Ice GTX480, there can be quite a bit of difference in temps given low and ultra high airflows. I'm seeing about a 8.5c difference between Yate loons at 1350rpm vs Scythe fans at 2930rpm at a rather large 570w heat load.

However, I completely agree with Thor. Even 8c in that scenario probably isn't going to matter on what kind of OC you get, and you will find as most of us have that the noise simply gets unbearable in short order. It really is better to get fans that put out a noise level you can live comfortably with, and then design you rads/fans/setup around that. We really do tend to stress the temps thing too much sometimes.

 

[hokiealumnus]

Alternatively, you could get a good fan controller and the strongest fans you feel like getting. Just turn them down. These panaflos would drive me (and my family) insane if I didn't only turn them up when benching.

Consequently, in very unscientific testing (simple observation while overclock stressing), the reason the fans made no difference was the little E8400 had as much heat dissipated as it could put out. It most definitely makes a difference with this very hot i7 860. I'll do actual measurements eventually (time is scarce right now), but know that 12v vs 7v most assuredly makes a difference on the MCR320 with a chip that puts out ridiculous amounts of heat.