Airflow: Amount of CFM in / through a case

Enclosures and acoustic damping to help quiet them.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Devonavar

Post Reply
spookmineer
Patron of SPCR
Posts: 749
Joined: Sat Nov 11, 2006 6:02 pm

Airflow: Amount of CFM in / through a case

Post by spookmineer » Sun Nov 26, 2006 4:49 pm

First off, this will ofcourse be highly dependent of what case and what components are used.

I'd like to know if there are general guidelines as for an optimal airflow, taking cooling and noise into consideration.

I've got an old Thermaltake Xaser II 6000 case with 5x 80 mm fans:
2 front intake
1 side intake
2 rear exhaust

They are standard sized 80x80x25mm and will do 2000 RPM. They are rated at 32 CFM at "a quiet 21 dB(A) noise output".
True, they are not that noisy, but still audible.

For intake, the total amounts to about 96 CFM, outtake will be about the same with a reasonable PSU (or slightly less maybe).

Is 100 CFM considered to be a normal amount of airflow through a case, could it be less (sacrificing somewhat in temperatures in hardware, gaining in silence) or could it be a bit more with the correct fans?
From what I've read in different reviews and forums, 21 dB(A) is not that bad for 32 CFM, this will most probably not be the real noise of these fans though.

If I were to replace these fans, should I be looking for more silent fans with the same CFM, or can I drop a bit in CFM and get even more silent ones?

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Sun Nov 26, 2006 5:36 pm

a) the specs are probably lies

b) even if they are not lies, they are measured in "free air", so will be nowehere near what you will actually get in your case

c) the only way to know is to try it and see what your temps are like

ciz28
Posts: 141
Joined: Fri Apr 11, 2003 12:17 am
Location: Redmond, WA

Post by ciz28 » Sun Nov 26, 2006 6:40 pm

I think sufficient airflow is incredibly dependent on case layout and clutter. I have almost no cable clutter in my NSK2400, and my (very roughly estimated) 60cfm of total airflow is plenty to keep everything cool. This number should probably go up to compensate for any cables or devices impeding airflow and possible inefficiencies in case layout or fan placement. As was previously stated, the only real rule here is to try out your idea to see where the actual temps end up at.

spookmineer
Patron of SPCR
Posts: 749
Joined: Sat Nov 11, 2006 6:02 pm

Post by spookmineer » Sun Nov 26, 2006 6:45 pm

a and b: agreed

c: there are no general guidelines as to the amount of CFM through a case should be?
Fro instance, there seems to be a consensus on the amount of RPM a heatsink fan should operate on (120 mm: 800 - 1400 for decent cooling and low noise).

If these guidelines should be present it would save me some money, I wouldn't like to buy 5 fans that turn out to be very quiet but don't provide enough airflow. If most people here would say "my system runs with an airflow of 80 CFM" I know what to expect and what to aim for.
Ofcourse, if I suspect my system would be a bit hotter or if I think my airflow is restricted, or if I want lower temps in exchange of a little bit more noise, I can adjust my search for fans.

If people would list their airflow in CFM it should show some optimum, provided enough people will.

I'll start :wink:
My airflow is about 100 CFM (corresponding to specs from manufacturer*)

[edit] ciz28 thnx [/edit]

ciz28
Posts: 141
Joined: Fri Apr 11, 2003 12:17 am
Location: Redmond, WA

Post by ciz28 » Sun Nov 26, 2006 6:54 pm

spookmineer wrote:If people would list their airflow in CFM it should show some optimum, provided enough people will.

I'll start :wink:
My airflow is about 100 CFM (corresponding to specs from manufacturer*)

[edit] ciz28 thnx [/edit]
I'll elaborate on my airflow situation in case it helps you... I have two 120mm fans from original Antec Sonatas. I don't know any specs on these, but I went out on a limb and assumed they'd be fairly close to the newer Tri-Cools they are shipping. They are both setup as intakes at 5V, so I estimate they are ~30cfm each, making a total of 60cfm in/out of the case.

Edited because I don't know how to proofread :(

Felger Carbon
Posts: 2049
Joined: Thu Dec 15, 2005 11:06 am
Location: Klamath Falls, OR

Here is the engineering answer to "how much CFM"

Post by Felger Carbon » Sun Nov 26, 2006 7:24 pm

Actual CFM = 1.738*Pw/dC where CFM = actual, at sea level
Pw = power in watts going into the computer box
dC = degrees Centigrade rise in exit air vs input air

In my computer case I get about half what you would expect from the exhaust fan in free air where CFM is concerned plus I'm at 4100 feet with 86.4% air density vs sea level.

Obviously, the greater air temperature rise you allow, the less cooling air you need. Contrarywise, if you want lower air temps, you'll need more CFM. Use the mobo integrated video and run cool or install a 200W video card and run really hot. There's lots of tradeoffs. But the formula in the first line of this post is correct.

I've posted the derivation before and I won't do it again. If interested, search "1.740" or "1.738" and my name as author.

Cerb
Posts: 391
Joined: Tue Apr 13, 2004 6:36 pm
Location: GA (US)

Ugh, long! The results of procrastination!

Post by Cerb » Sun Nov 26, 2006 8:31 pm

spookmineer wrote:First off, this will of course be highly dependent of what case and what components are used.

I'd like to know if there are general guidelines as for an optimal airflow, taking cooling and noise into consideration.
Yes!
http://www.silentpcreview.com/article75-page1.html
The first two pages are just that. Also peruse the case reviews, and see what is done with cables and fans, especially the P180, SLK3000, and SLK3700.

For example, take Ralf's cablegami mastery, here:
http://www.silentpcreview.com/article76-page2.html
You don't need to manage anything near that level of perfection to get great results. The key is making sure there's plenty of open space without cables in the middle. Tucking them away and making ugly zip-tie arrangements does a perfectly good, if not picture-perfect, job.
I've got an old Thermaltake Xaser II 6000 case with 5x 80 mm fans:
2 front intake
1 side intake
2 rear exhaust

They are standard sized 80x80x25mm and will do 2000 RPM. They are rated at 32 CFM at "a quiet 21 dB(A) noise output".
True, they are not that noisy, but still audible.
Anything can be 21dBA, but they certainly are not measuring at 1m. Get a Nexus, note that there are real measurements here, and be amazed at the difference. The last time I trusted Thermaltake it nearly fried a Duron (the old Orb coolers). Everything I've heard since hasn't been any better.
For intake, the total amounts to about 96 CFM, outtake will be about the same with a reasonable PSU (or slightly less maybe).

Is 100 CFM considered to be a normal amount of airflow through a case, could it be less (sacrificing somewhat in temperatures in hardware, gaining in silence) or could it be a bit more with the correct fans?
From what I've read in different reviews and forums, 21 dB(A) is not that bad for 32 CFM, this will most probably not be the real noise of these fans though.
Nowhere close, and you're not going to get 96CFM. That's in free air.
If I were to replace these fans, should I be looking for more silent fans with the same CFM, or can I drop a bit in CFM and get even more silent ones?
With otherwise good components for quiet cooling, get Nexus fans and be done with it. With a decent case layout, you might be amazed how easy it is to cool stuff. A cramped case with cable messes and drive cages in the way of a good air path will get you nothing but trouble.

OEMs often short-cut things by having just enough cables, have them made just long enough, no extra parts, etc.. For standard PC cases, though, it takes more work, because everything has to be made with flexibility in mind, and then we must bend it to our will (sometimes forcefully, with pliers).

In the standard configuration of your case, one intake fan will have tons of static pressure relative to the other with the config you're meant to use, vibration dampening is minimal, the 3.5" drives need to be hard-mounted to aluminum, not counting suspension options in the 5.25" bays. The PSU and rear intake are way too close together to use a PSU with a bottom fan very well (without ducting the CPU HSF to an exhaust fan, anyway). The fans will all be turbulently trying to move the same air around. Ducts help, as does using lower-speed fans.
c: there are no general guidelines as to the amount of CFM through a case should be?
Fro instance, there seems to be a consensus on the amount of RPM a heatsink fan should operate on (120 mm: 800 - 1400 for decent cooling and low noise).
No, because that's not anywhere near as much of a concern as a clean case layout, a sturdy case, dampening, low-power parts, etc.. RPM can be used better, but still not perfectly, because noise always goes up drastically with RPM, but even that's not set in stone.

If you want quieter, neaten up cables, replace the fans with Nexus ones, maybe get rid of the side fan, and mount the hard drives somewhere in the 5.25" space (or make a rig in the front-bottom and replace the cages). Getting it nice and quiet should certainly not be too hard, but it will not happen with any Tt fans in there. Their PSUs they don't make, if you've got one in it, and it could be fine. On a very bright side of things, the Dragon that your case is made from used to be one of the best cases for good cooling out there, and certainly can be made to do a decent job, quietly.

Come to think of it, you may as well try out two Nexus (or just normal opaque low-speed Yate-Loons if you can find them in 80mm) fans in the rear, no side fan, no front fans, then see if that does the job. It just might, saving you money, time, and precious dBAs. Just keep track of your temps and see how well it works. If it doesn't, you may be out $4-10 additional shipping (depending on store), but if it does, you've saved at least another $20.

You may also want to try ducting the CPU fan, if its fan points down onto the CPU, to the side intake. If looks aren't important, cardboard and duct tape should handle it well. If it's a tower-style heatsink, you could try ducting to one of the real exhaust holes.

RAFH
Posts: 129
Joined: Sat Sep 16, 2006 10:30 am
Location: Kahaluu, Hawaii
Contact:

Post by RAFH » Mon Nov 27, 2006 1:25 pm

My Antec SOLO has two 92mm intakes rates by Nexus at 27 CFM each with 19 dba and the 120mm exhaust is rated by Nexus at about 37 CFM and 23 dba.

The PSU is on a different air flow circuit and runs at a constant air flow and noise without influence from the rest of the rig. I installed a custom baffle to separate it.

You should realize its not the total of intake and exhaust. Just because you have, say like my rig, 54 in and 37 out doesn't mean you have a total flow of 91 CFM. The intakes reduce the resistance the exhaust sees or you can look at it they increase the intake pressure vs the outside for the exhaust fan. Alternately you could say the exhaust fan lowers the back pressure the intakes see. Either way, they are helping each other but certainly not at 100% rating even if that was good.

Also, as point out, those specs are iffy at best, unless you have certified independent testing laboratory test results to support them. And then there is the matter those are free air tests, which have very little to do with your actual situation.

As a result, I guesstimate I maybe have a flow through my system of about 30 CFM, maybe even less. But I also have a positive pressure and that keeps the dust infiltration down.

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Mon Nov 27, 2006 2:07 pm

You should realize its not the total of intake and exhaust. Just because you have, say like my rig, 54 in and 37 out doesn't mean you have a total flow of 91 CFM.
Yeah, I should have pointed this out before, when fans are placed in series the airflow is not additive (ie you can't simply add intake + exhaust to get the total CFM figure); in a free air situation two fans in series doubles static pressure (but CFM does not change), two fans in parallel doubles airflow (but static pressure does not change). Of course in a real case it depends on the impedance; in a high impedance case fans in series will increase airflow more than fans in parallel, but the opposite is true for a low impedance case.

See what a can of worms you opened up! :lol:

ciz28
Posts: 141
Joined: Fri Apr 11, 2003 12:17 am
Location: Redmond, WA

Post by ciz28 » Mon Nov 27, 2006 2:22 pm

jaganath wrote:Yeah, I should have pointed this out before, when fans are placed in series the airflow is not additive (ie you can't simply add intake + exhaust to get the total CFM figure); in a free air situation two fans in series doubles static pressure (but CFM does not change), two fans in parallel doubles airflow (but static pressure does not change). Of course in a real case it depends on the impedance; in a high impedance case fans in series will increase airflow more than fans in parallel, but the opposite is true for a low impedance case.

See what a can of worms you opened up! :lol:
I believe he already accounted for this by only considering his three intake fans of 3x32cfm for 96cfm, or ~100cfm, his stated figure. He also stated that his exhaust roughly matches his intake, assuming that his PSU fan is close enough to one of his other 80mm case fans. This would make me think he already considered the idea you mentioned?

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Mon Nov 27, 2006 2:32 pm

It's questionable whether the side intake counts as "in parallel" with the front two intakes, sure it will pressurise the case but it won't add to airflow, so a more realistic figure is 64CFM, and that's without accounting for case impedance (which could be significant, depending on case layout). Basically SPCR 101 says use as few fans as possible, spinning as slowly as possible. Trying to get more quantitative than that is really not helpful, as each situation is different (ambient temps, case heat load etc).

spookmineer
Patron of SPCR
Posts: 749
Joined: Sat Nov 11, 2006 6:02 pm

Post by spookmineer » Mon Nov 27, 2006 2:49 pm

This is more advice then I hoped for!

I know the airflow is not additive, and I only listed the specs of the manufacturer because that's all I have...
So according to Thermaltake, intake is (2x front and 1x side) * 32 CFM, the exhaust is (2x rear) * 32 CFM + whatever the PSU will move, but I doubt this will be 32 CFM.
Total airflow would be a bit less then 100 CFM (intake equals exhaust) but I didn't take into account all losses from restrictions etc.

It's just a starting point, the more I read here and in other places, we might as well just take a calculated guess as to what the specs really are and probably be more accurate then what is listed on the products... The good thing is, there are loads of people who have tried a lot of products and spent a lot of time to post their findings.

Before I started to think about this, I had little hope that there would be a simple lay-out / starting point of what a good airflow looks like, but that's an illusion (well there is a basic lay-out but that's not really helping much - and it's how the case is built so I have to work with that).
It would never occur to me to not use the front and side fans, and only use the rear instead.

The more I think about it, creating 3 ducts or chambers for the a. heatsink and motherboard, b. the PSU and c. the graphic card sounds more appealing but that would mean dividing the available fans over these chambers, and that would imbalance the needed airflow for the seperate parts (heatsink would need much more CFM then PSU, vid card maybe needs somewhere between these two).

I like opening cans of worms!
hm... I'm so sure anymore.

ciz28
Posts: 141
Joined: Fri Apr 11, 2003 12:17 am
Location: Redmond, WA

Post by ciz28 » Mon Nov 27, 2006 2:52 pm

jaganath wrote:It's questionable whether the side intake counts as "in parallel" with the front two intakes, sure it will pressurise the case but it won't add to airflow, so a more realistic figure is 64CFM, and that's without accounting for case impedance (which could be significant, depending on case layout).
I would think that side case fans would still count as being in parallel with the front intakes. The only way to really get a fan to be in a serial configuration would be to have the intake of one fan pull from the exhaust of another fan. Since all three case fans are pulling air from outside the case, I'd be inclined to consider them all running in parallel, thus all contributing (although not a full 32cfm) to the total airflow of the case.
jaganath wrote:Basically SPCR 101 says use as few fans as possible, spinning as slowly as possible. Trying to get more quantitative than that is really not helpful, as each situation is different (ambient temps, case heat load etc).
I completely agree with this advice. There's really not much more that can be said except to try it for yourself or compare to other peoples' very similar configurations.

ciz28
Posts: 141
Joined: Fri Apr 11, 2003 12:17 am
Location: Redmond, WA

Post by ciz28 » Mon Nov 27, 2006 2:56 pm

spookmineer wrote:The more I think about it, creating 3 ducts or chambers for the a. heatsink and motherboard, b. the PSU and c. the graphic card sounds more appealing but that would mean dividing the available fans over these chambers, and that would imbalance the needed airflow for the seperate parts (heatsink would need much more CFM then PSU, vid card maybe needs somewhere between these two).
I don't know that I would necessarily agree with this imbalance that you are proposing. My take on ducting is that you want to make sure hotter zones don't have a negative impact on cooler zones. Although you'd ideally like your CPU zone to run as cool as your PSU zone, you don't need to increase the airflow if you are willing to accept a higher temperature for this particular zone.

RAFH
Posts: 129
Joined: Sat Sep 16, 2006 10:30 am
Location: Kahaluu, Hawaii
Contact:

Post by RAFH » Tue Nov 28, 2006 11:40 am

As far as ducting goes, you would definitely not want to just put the same fan on each of three ducts serving three different purposes.

In my case, the PSU's fan is sufficient for itself and the opticals. The fans I did install all go to the main compartment as my case is just too small to start breaking up the flow beyond that, unless I was to isolate the graphics cards with a supply coming from the vents over the expansion slots and exhaust somewhere else (or vice versa to avoid bleed from the main exhaust to the intake over the expansion slots).

As to serial/parallel flows and trying to work it out mathematically, you might as well throw theory out the window. Its great for giving you a basic concept as to what to shoot for and how things ought to work, but in practical application, unless you have a very detailed cad model of your case with surface characteristics and all accurate specs on every fan and a very good thermodynamic flow modeling software, you are just as well off guessing.

The real alternative is to go at it empirically. Try different configurations and see what works best. Theory will get you close enough to avoid wasting huge amounts of time or endangering your components but real world testing is going to get you home. The problem is not many of us have accurate pressure and velocity gauges. You'd probably want to get yourself a clear side panel so you could watch 'smoke' plumes. It could easily get very expensive and time consuming to get that last 5%.

The best 'rules' I can offer are similar to SPCR's:
-Keep it simple. The fewer fans the better, unless you want to devote your life to monitoring fans.
-Keep it slow and soft General rule, slower big fans make less noise than small fast ones.
-Keep it clean. This is really an offshoot of the KISS above, the less clutter, the easier it is for the air to move and hopefully go where you want it to.
-Maintain a positive case pressure and use a filter. This is just a matter of Keep it clean and Keep it simple. If you have negative case pressure, there's no telling where all the little holes and gaps will direct air or what they will let in. If you have positive case pressure, you don't have to care, unless they are so many or so large they waste your air. In that case, a bit of duct tape or similar will fix that simply enough.

Post Reply