Air flow vs static pressure for pushing and pulling

Control: management of fans, temp/rpm monitoring via soft/hardware

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
Shai-tan
Posts: 24
Joined: Thu Apr 22, 2010 6:49 am
Location: NZ

Air flow vs static pressure for pushing and pulling

Post by Shai-tan » Thu Apr 22, 2010 7:03 am

Basically I would like to know if static pressure is better for pushing or pulling and is air flow better for pushing or pulling.

I have a Noctua cpu cooler and wondering if I should use a static pressure fan to push or pull and the other end use a decent air flow fan to push or pull. Or should I just use 2 static pressure fans for both ends of the cooler?

I hope I haven't lost you but I can't seem to find an answer to this anywhere.

b_rubenstein
Posts: 273
Joined: Tue Aug 04, 2009 7:03 am
Location: Brooklyn, NY

Post by b_rubenstein » Thu Apr 22, 2010 7:41 am

For some reason people are under the impression that this is all black and white, and it isn't. No application of fans in computers are in free air; there is always an impedance, it just ranges in magnitude. In the case of fans, there is always some degree of leakage between the fan blades and the frame. For a given PRM, the greater the amount of leakage the less static pressure it can develop. It makes no difference in whether the fan is pushing or pulling.

CPU coolers are at the high end of impedance, so fans that can develop a higher static pressure are better suited for this application.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: Air flow vs static pressure for pushing and pulling

Post by ces » Thu Apr 22, 2010 7:46 am

Shai-tan wrote:static pressure
I'm not certain what you mean by static pressure.

But what I have learned from trial and error is that if you want to cool something, pushing air at it generally is most effective. If you are pulling air away from it much of the air current gets wasted pulling air away from other directions. That is why some fan manufacturers try to design fans with very directional air streams - so that even when you are pushing, less of the air stream gets wasted on things are you not trying to cool.

You get more air current / dB from two fans at 600 rpm each moving 20 DFM than one fan at 1200 rpm moving 40 cfm.

So I would use two slow fans in push pull rather than one stronger fan in push.

BUT DISREGARD ALL THE ABOVE! Because the only thing that counts is testing. Logic and bench marks are only good for making intelligent decisions as to what you are going to test. Air cooling has too many variables and is too unpredictable, to rely on anything other than doing your own testing on your own system.

But the Noctua comes with two good fans. Just don't forget to run them on low. Some people run them on high, and then complain they make too much noise. When you run them on low, they are different fans.

datapappan
Posts: 199
Joined: Wed Nov 10, 2004 1:44 pm
Location: Sweden

Post by datapappan » Thu Apr 22, 2010 11:22 am

Once again one must point to how fans and air flow works.

Air flow:
pushing air through a case has to overcome the pressure drop caused by impedance. Basically pressure drop is related to air flow by the square (double flow=four times pressure)

Fan:
Fans have more or less teh reverse relation, at 0 flow it gives the most pressure, and as flow increases, pressure decreases, up to a maximum (the cfm figure you can find in specs, actually at free flow in mid air)

So, for any given setup these to curves meet at the operating point.

With two fans working in parallell, they share the flow, but still has to put in the pressure. This is done by lowering rpm until the operating point is att the same pressure, at half the flow.

Again, two fans in series (push-pull) shares the pressure, but has to give the same flow. And again, lowering rpm gives a new operating point.

It's not obvious which setup gives the lowest noise, so I agree with the above - trial and error gives the best answer. It also comes into account where the fans are placed in the case, and where the case is situated in relation to the user. (A really silent PC will sit in another room...)

/ d
[size=75][CPU: 2.4 GHz P4 Northwood w. AC Freezer 4 (fanless-Bluefront PSU mod inspired) / MB: ABIT VT7 (Via PT880) / GPU: ATI Radeon X800XL w Zalman ZM80D-HP / HDD: Seagate Barracuda IV SATA 80GB / CD: Samsung CD-DVD combo / Seahawk ALU Case / PSU: FSP 350-THN (fan mod-Everflow connected to CPU fan header)[/size]

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Post by ces » Thu Apr 22, 2010 11:31 am

datapappan wrote:Once again one must point to how fans and air flow works.

Air flow:
pushing air through a case has to overcome the pressure drop caused by impedance. Basically pressure drop is related to air flow by the square (double flow=four times pressure)

Fan:
Fans have more or less teh reverse relation, at 0 flow it gives the most pressure, and as flow increases, pressure decreases, up to a maximum (the cfm figure you can find in specs, actually at free flow in mid air)

So, for any given setup these to curves meet at the operating point.

With two fans working in parallell, they share the flow, but still has to put in the pressure. This is done by lowering rpm until the operating point is att the same pressure, at half the flow.

Again, two fans in series (push-pull) shares the pressure, but has to give the same flow. And again, lowering rpm gives a new operating point.

It's not obvious which setup gives the lowest noise, so I agree with the above - trial and error gives the best answer. It also comes into account where the fans are placed in the case, and where the case is situated in relation to the user. (A really silent PC will sit in another room...)

/ d
I think you have a lot of core fundamental information in there condensed into very few words. But I think it might be too condensed for most readers. It took me a few reads to understand what you were saying. If you are so inclined, it might help others if you rewrote it with more words and some examples - to get this fundamental info into their heads.

It is my personal belief that very few people in these forums understand the above information and if they did, the discussions here relating to fans and cooling would have much more clarity.

~El~Jefe~
Friend of SPCR
Posts: 2887
Joined: Mon Feb 28, 2005 4:21 pm
Location: New York City zzzz
Contact:

Post by ~El~Jefe~ » Thu Apr 22, 2010 12:42 pm

the inside of a case has nothing to do with anything else designed or tested for air flow.

There is no design to a case at all. The whole layout is terrible for air flow. The worst place component acounts for 70% of the heat, the video card. It is upside down and horizontal and has no outlet. It is in the midle of the case and causes a hard stop to air flow. So, it really never mattered to me this inquiry as much as it matters how to cool the video card. Then, after that, how not to heat up the PSU. I do this by forcing more air into the case than trying to exhaust. This also causes turbulence which removes heat away from the tiny components that get super hot like Mosfets. Again, the mosfets arent in the best places and also are not normally focused on for cooling more than a simple flat heatsink at best. This is where you COULD save on wattage used if you cooled these things down to a much cooler level.

So, while this above info is very interesting for general fan use or for say cooling one single component, it is hard to use it in a computer case. The presence of harddrive cages and huge cables and 90 degree turns makes flow of air a bit lower on the totem pole. The idea of air pressure I have found is the key here, and that is simply? done by putting 2 120mm fans in the front and one slow one in the back. I might be making some physics errors but physics has no way of cooling a case unless you want to do a dissertation worth of calculations and observations.

ascl
Posts: 279
Joined: Tue May 05, 2009 1:15 am
Location: Sydney, Australia

Post by ascl » Thu Apr 22, 2010 1:25 pm

b_rubenstein wrote:It makes no difference in whether the fan is pushing or pulling.
This is not really correct. From here: http://martin.skinneelabs.com/Radiator- ... eview.html
Image

We can see that with this specific fan and radiator, push vs pull does definitely make a difference, and shrouds also make a difference. Martin concluded that:
Push Vs Pull - This depends on fan speed/power. The high speed fans at 2000RPM with a 38mm fan thickness provided the best performance in a push condition. The slow speed fans with 1350RPM with a 25mm fan thickness provided the best performance in a pull condition. I would estimate that performance line is likely to cross in the 1500-1700RPM range where they are equal. So.... slow speed = pull, high speed = push, medium speed = it doesn't really matter.
Of course, this is with these specific fans (Yate Loons), and on a specific radiator. Changing either of these will likely change the results.... so my post is basically pointless, what everyone else said is right "test test test" :)

b_rubenstein
Posts: 273
Joined: Tue Aug 04, 2009 7:03 am
Location: Brooklyn, NY

Post by b_rubenstein » Thu Apr 22, 2010 2:24 pm

OK, it makes more than no difference, but very little: 2.2% with only a fan running at 1350 RPM. As can be observed from his other graphs, and the quote you include, the effect is highly dependent on the fan's RPM.

Once you get into the RPM range of fans in quiet PC's (1000 RPM, or less) there may not be any inherently better configuration.

ascl
Posts: 279
Joined: Tue May 05, 2009 1:15 am
Location: Sydney, Australia

Post by ascl » Thu Apr 22, 2010 2:29 pm

Yeah its minor at best. I just thought it was interesting... as I said, my post wasn't really useful as such... just a curiosity :)

Shai-tan
Posts: 24
Joined: Thu Apr 22, 2010 6:49 am
Location: NZ

Post by Shai-tan » Thu Apr 22, 2010 2:39 pm

Thanks guys. This is the sort of answers I was looking for and I realize my specific case wasn't explained better (it was 3am in the morning here)

Basically I was wondering if there was nothing impeding the fan pointed to the rear exhaust fan should I use a fan with better Air flow vs Less Static pressure on the back of the CPU cooler.

The fans I'm using here are Noctua NF-S12B and Noctua NF-P12.
NF-S12B @ 900rpm
Airflow (m3/h) = 75.8
Static Pressure (mm H2O) = 0.76

Noctua NF-P12 @ 900rpm
Airflow (m3/h) = 63.4
Static Pressure (mm H2O) = 1.21

#1 at the front fan I'm using a NF-P12 because the static pressure is better suited for the impedance between the front and the CPU.
#2 at the CPU cooler I'm using another NF-P12 because the static pressure pushing through the Noctua NH-U12P SE2 would be better.
#3 on the back side of the CPU cooler (facing the rear of the case) I'm using an NF-S12B because there is nothing impeding the fan and the airflow goes straight to the rear fan
#4 Rear exhaust fan is a NF-P12 currently because I still haven't cut out the rear fan cover in the case and the pressure would be better suited for that.

I need to know whether I'd be better using another NF-P12 on #3 or sticking with the better air flow of the NF-S12B

b_rubenstein
Posts: 273
Joined: Tue Aug 04, 2009 7:03 am
Location: Brooklyn, NY

Post by b_rubenstein » Thu Apr 22, 2010 5:49 pm

If there is a cooler attached to the fan, regardless of the configuration, use a high static pressure one.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Post by ces » Thu Apr 22, 2010 6:10 pm

ascl wrote: This is not really correct. From here: http://martin.skinneelabs.com/Radiator- ... eview.html
That is all very interesting. I wen to that link and it gets a bit more complicated.

I did notice something. I have been told that you should always have at least one fan blade of space in front of a fan. Those numbers are consistent with that. They are using a 38mm fan. But the fan blades may be 30 or 32mm deep.

By the way my findings that pushing air is more effective than pulling it is consistent with the data on that site. They are testing fans that either have a shroud or are right next to the radiator. My tests were with fans some distance away that had to rely on air current to reach what they were cooling.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Post by ces » Thu Apr 22, 2010 6:17 pm

Shai-tan wrote:I need to know whether I'd be better using another NF-P12 on #3 or sticking with the better air flow of the NF-S12B
What you have done is all logical. It's close to what I probably would have done, though maybe I would have used two NF-P12s on the cooler and two NF-S12B fans on the case.

But logic only gets you so far. Now you have to experiment. No one can fully predict what you will find. I bet you will pick up a few degrees of improvement though. And don't forget to experiment with whether the cooler works best in a north south configuration or an east west one.

ascl
Posts: 279
Joined: Tue May 05, 2009 1:15 am
Location: Sydney, Australia

Post by ascl » Thu Apr 22, 2010 6:22 pm

Shrouds are great. They help reduce the noise of the fan on the radiator and they also boost the performance.... the only downfall is the extra space they take up.

The 'size of a single blade' rule is great! Where did you hear that? It does seem to match Martin's findings (and also, conveniently, about the same of a fan frame with the motor pulled out).

One thing I haven't seen, although I could probably figure out with the right numbers, is the benefit of using push/pull vs a single fan against the noise. What I mean is, if your aim is to have a constant noise level, are you going to get better performance from a single fan or from 2 fans running slower. Its going to depend on the fan I guess (extra noise source = +3db if I remember correctly vs 2 fans providing approximately 25% better cooling).

ascl
Posts: 279
Joined: Tue May 05, 2009 1:15 am
Location: Sydney, Australia

Post by ascl » Thu Apr 22, 2010 6:24 pm

ces wrote:
Shai-tan wrote: But logic only gets you so far. Now you have to experiment. No one can fully predict what you will find. I bet you will pick up a few degrees of improvement though. And don't forget to experiment with whether the cooler works best in a north south configuration or an east west one.

I cannot agree with or emphasis this enough. Fluid dynamics is extremely complicated and difficult to predict in something like a computer case. Sometimes obvious or logical changes don't have the expected outcome.... for example, I have had a case where increasing the fan speed of one of the intake fans increased the CPU temps.... testing is very important!

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Post by ces » Thu Apr 22, 2010 6:32 pm

ascl wrote:One thing I haven't seen, although I could probably figure out with the right numbers, is the benefit of using push/pull vs a single fan against the noise. What I mean is, if your aim is to have a constant noise level, are you going to get better performance from a single fan or from 2 fans running slower.
I am pretty darn certain that you are better off with say two 600 rpm fans pushing 20 cfm each than one 1200 rpm fan pushing 40 cfm. This is based on the logarithmic nature of sound. I have been wrong before, but I am pretty certain on this.

If you are good with numbers, you could probably use SPCR's data from the 1200 rpm slipstream to confirm this with some calculations. See:
http://www.silentpcreview.com/article83 ... .html#SS-M

Just eyeballing SPCR's data right now, it seems to me they confirm my proposition.

Shai-tan
Posts: 24
Joined: Thu Apr 22, 2010 6:49 am
Location: NZ

Post by Shai-tan » Thu Apr 22, 2010 6:43 pm

ces wrote:What you have done is all logical. It's close to what I probably would have done, though maybe I would have used two NF-P12s on the cooler and two NF-S12B fans on the case.

But logic only gets you so far. Now you have to experiment. No one can fully predict what you will find. I bet you will pick up a few degrees of improvement though. And don't forget to experiment with whether the cooler works best in a north south configuration or an east west one.
Thanks Ces and ascl

Have done some testing and the new front to back CPU cooler bracket (which Noctua has sent me out for free!) works better because it retains the same air flow of my Sonata III's front fan (on the rear side of the HDD bay which sucks). I'm going to have to upgrade the case soon I think... probably to a P183 and CP-850 PSU

Now running all fans at 900rpm and the Rear fan is now a NF-S12B after I cut the case cover out. The CPU cooler is now NF-P12s all around. I'm still retaining my NF-P12 at the front because the Accerlero 1 Rev 2 is restricting airflow. So at the moment everything is running around 30-35C under no load and its quiet!

If I bought some 1200 rpm slipstream's how do you think I should use them in my setup? Should I use them for case fans or use them for the CPU cooler or should I use them for everything?

I would love to see an SPCR "Noctua NF-S12B vs 1200 rpm slipstream" because it seems the Noctua NF-S12B hasn't been reviewed yet

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Post by ces » Thu Apr 22, 2010 7:06 pm

Shai-tan wrote:new front to back CPU cooler bracket
What exactly is the new front to back CPU cooler bracket?

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Post by ces » Thu Apr 22, 2010 7:08 pm

Shai-tan wrote:If I bought some 1200 rpm slipstream's how do you think I should use them in my setup? Should I use them for case fans or use them for the CPU cooler or should I use them for everything?
I would use them as case fans. But experiment with them as CPU fans.

ascl
Posts: 279
Joined: Tue May 05, 2009 1:15 am
Location: Sydney, Australia

Post by ascl » Thu Apr 22, 2010 7:13 pm

ces wrote: I am pretty darn certain that you are better off with say two 600 rpm fans pushing 20 cfm each than one 1200 rpm fan pushing 40 cfm. This is based on the logarithmic nature of sound. I have been wrong before, but I am pretty certain on this.
You are probably right, but your numbers are off. Adding an extra fan doesn't double the CFM, it maybe increases it by %25 (based on Martin's numbers). SPCR's figures aren't useful in this case, because we are talking airflow through a radiator.

At risk of getting shunned from this forum, I'm going to use figures from here:
http://www.madshrimps.be/?action=getart ... rticID=936

Not because I think they are particularly accurate, but because its one of the few "air flow through rad" type tests I can find (I thought I saw one with GT's, but can't find it now). I am also going to make the rude assumption that twice the air flow equals twice the cooling... but I am almost certain this isn't accurate either. But whatever, for the sake of the thought experiment... lets see what we get!

This is the graph from the Yate Loon high speed fan:
Image

Based on that, running the fan at a speed that produces 30 CFM produces about 68 db of noise. But we should be able to run 2 fans at a speed that produces 24 CFM (each, 24 CFM * 1.25 = 30 CFM), which should produce... about 65 + 3 db. Doh.

I suspect we need much more accurate figures, or someone that understands this stuff better to figure it out.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Post by ces » Thu Apr 22, 2010 7:25 pm

ascl wrote: You are probably right, but your numbers are off. Adding an extra fan doesn't double the CFM, it maybe increases it by %25 (based on Martin's numbers). SPCR's figures aren't useful in this case, because we are talking airflow through a radiator.
You need data that is rpm against CFM, madshrimps doesn't provide that info.

If you look at the SPCR free air data for the 1200rpm Slipstream is close the the fabricated example I made up. If they are both pushing into the same case, they each generate some impedance for the other, but not that much.

Subject to messy real world details, basically:
1. CFM adds up linearly.
2. Sound adds up logarithmically, and
3. Sound increases probably close to the square of the speed of the fan blades, at least at the speeds we are working with.

Tephras
Posts: 1140
Joined: Tue Sep 07, 2004 11:03 am
Location: Europe

Post by Tephras » Thu Apr 22, 2010 8:18 pm

ascl wrote:One thing I haven't seen, although I could probably figure out with the right numbers, is the benefit of using push/pull vs a single fan against the noise. What I mean is, if your aim is to have a constant noise level, are you going to get better performance from a single fan or from 2 fans running slower. Its going to depend on the fan I guess (extra noise source = +3db if I remember correctly vs 2 fans providing approximately 25% better cooling).
A comparison of 1 fan vs 2 on the Scythe Mugen-2: Cooling Results - Two Fan Configuration.
And this forum sticky might be of interest.

ascl
Posts: 279
Joined: Tue May 05, 2009 1:15 am
Location: Sydney, Australia

Post by ascl » Thu Apr 22, 2010 8:24 pm

Thanks for the links, I knew I'd read something here on 2 vs 1 fan! The forum sticky is talking about free air however, which is quite different to against a radiator.

The Mungen review however... seems to answer exactly my question:
2 fans
5V
11 dBA
38°C


vs

1 fan
13 dBA
38°C


ie same noise, more cooling!

Awesome, I guess this means I am ordering more fans... ! hehe.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Post by ces » Thu Apr 22, 2010 8:36 pm

ascl wrote:The Mungen review however... seems to answer exactly my question:
2 fans
5V
11 dBA
38°C

vs

1 fan
13 dBA
38°C
Isn't it less noise with the same temp.

ascl
Posts: 279
Joined: Tue May 05, 2009 1:15 am
Location: Sydney, Australia

Post by ascl » Thu Apr 22, 2010 8:38 pm

Hah, you are right... same thing tho, either way, it works out quieter with same cooling, or better cooling with same noise :) Either way is win!

Shai-tan
Posts: 24
Joined: Thu Apr 22, 2010 6:49 am
Location: NZ

Post by Shai-tan » Thu Apr 22, 2010 11:04 pm

ces wrote:
Shai-tan wrote:new front to back CPU cooler bracket
What exactly is the new front to back CPU cooler bracket?
This:
http://www.noctua.at/main.php?show=news ... =54&lng=en

I'm running an AM socket and the fans can only face up and down with the Noctua NH-U12P standard AM2,+,3 configuration.

Post Reply