Sign in with
Sign up | Sign in
Your question

Would air from 2 fans negatively interact & reduce flow?

Last response: in Overclocking
Share
December 10, 2006 11:51:42 AM

Now, correct me if I'm wrong (and I know you will) butt... and it's a big butt...

I'm designing a killer system. The sucka is gonna generate enough heat to get to be #1 on Al Gore's hitlist. Since the case is in the next room and I wouldn't hear it if was running a rocket engine, I wanna take both the sidecovers off the huge tower and plunk a monster three-foot wide industrial fan next to it. Crank the fan up and watch Dorothy fly through the room looking for Toto.

The question is, should I take out all the fans that are currently in the system? From what (little) I understand about fluid dynamics, it would seem to me that the smaller fans would negatively interact with the flow created by the monster fan and actually provide less airflow to the components.

Let's not discuss the PSU since that's a box and it's rather sheltered from the massive airflow. Let's just concentrate on stuff like the HSF and GPU.

This should be interesting... 8)
December 10, 2006 3:19:15 PM

You are somewhat right, but the problem is that if you remove those fans youll be removing concentrated airflow to those parts, and CPU and GPU need a fan of their own... feel free to remove every other case fan tough because youll kinda have enough airflow...
December 10, 2006 3:31:48 PM

Quote:
You are somewhat right, but the problem is that if you remove those fans youll be removing concentrated airflow to those parts, and CPU and GPU need a fan of their own... feel free to remove every other case fan tough because youll kinda have enough airflow...


Fair dinkum, but let's say that I go with some monster air CPU cooler like the Scythe Infinity I could definitely run that in passive mode. Sticking a 120mm on that would be overkill when it's four inches away from a three foot fan, right?
Related resources
December 10, 2006 4:11:54 PM

Why don't you buy an air cooler an stick that next to your case.
December 10, 2006 4:33:37 PM

Quote:
Why don't you buy an air cooler an stick that next to your case.


I'd love to but the room the case is in has no windows. That would be great if I could make it work, but I can't.
December 10, 2006 4:45:40 PM

Hello,

I would suggest that you might be disappointed with the results coming from pushing massive quantities of air in from the side of your computer. As the flow approaches the flat plate of the motherboard, it will divert and flow out the sides/top. There may not actually be very good flow rates close to the surface of the motherboard, and most importantly the cpu, gpu, and ram. On my computer, all of these components are passively cooled, and I can get away with this because I have a <100 watt CPU and very good case ventilation.

The current trend with computer air-cooling is to design the case so that air flows from front to back in a sort of "wind-tunnel". I'd suggest you look up how the BTX form factor has been designed for thermal considerations. In many cases, modern cases achieve better cooling with the side panel on rather than off. If you really want your industrial-sized fan to provide effective cooling, I would recommend fitting it with a custom funnel to push air into the bottom half of the front of your case, and leave the side panel on so that the air is forced to flow out through the hot parts of the computer. In this case, you should make sure to remove the intake fan by the hard drives (replaced by the huge fan and the exhaust fans at the rear. I'd leave the PSU alone, but you might find the forced flow through the fans causes annoying vibrations. Leave the cpu and gpu coolers as they are designed, so if they came with fans, leave those on for strong local airflow.

And finally, if you are planning to stick with the side-injection airflow, make sure you don't have a front air intake fan fighting the flow. Your huge fan will be the only air intake for the system.
December 10, 2006 4:46:56 PM

One drawback to your cooling solution is that dust, (and if you have pets) pet hair will get thrown into the case. With my old 486 system I cleaned it out like every other year because there was only the power supply fan, with my current system with 3 fans (rear, processor, powersupply) im cleaning it out at least 3 times a year.
December 10, 2006 5:01:25 PM

If I go with the "closed case" option then I could use a "nautilus squirrel cage" type of fan that is going to have an even higher CFM rating than the 3' wide industrial fan. The windspeed inside that case would be inconceivable. I wouldn't be surprised to see 30mph. As long as any vibes it sets up generate noise but don't damage the components I don't care as I couldn't hear anything that's going on in that room. I also don't have to worry too much about dust as that little room is sealed, I have no pets and there is no carpet, etc. Well, you guys got me thinking... and that is a dangerous thing! :lol: 
December 10, 2006 6:31:38 PM

These guys have valid points.

I was just wondering if you would not mind to add water colling with that assembly then you might have a better cooling than just air power. You still need to redirect air flow though.

Best airflow design is if you have the same amount of air coming in as it and out. Like HOTFOOT said check the BTX design.

Proper location of components is important for optimize airflow.
December 10, 2006 6:36:38 PM

Quote:
These guys have valid points.

I was just wondering if you would not mind to add water colling with that assembly then you might have a better cooling than just air power. You still need to redirect air flow though.

Best airflow design is if you have the same amount of air coming in as it and out. Like HOTFOOT said check the BTX design.

Proper location of components is important for optimize airflow.


I've used WC in quite a few builds, but that was mostly in the bad old days when we were using aquarium pumps. The build I'm considering now is a dual quadcore with 5 internal HDs and top-end DX10 cards and I just don't wanna run plumbing to all this stuff. What I've found in some WC builds is that the plumbing is set up properly for the CPU and GPU and since the guy is getting good temps on the readout he doesn't concern himself with the air cooling on the rest of the equipment in the case, usually ending up with searing hot HDs, RAM, Northbridge, etc. The setup I have with the case in another room really begs for a killer air system and I'm pretty well stuck on having it that way.
Anonymous
a b K Overclocking
December 10, 2006 10:05:57 PM

I think that one of your best bet would be to get a good case like this one:
Tj09 and stick super high CFM fan in there. With 4 high CFM 120mm fans I wouldn't worry about cooling!

Neat thing is the dedicated fan for video card, One 120mm on top of each quad core, dedicated fan for the HD. I am sure it would cool it properly

Imgaging sticking 3 of Those as exhaust!


BTW I think this case is set to have 25mm fans but You can probably make it hold 38mm no problem!
December 11, 2006 6:40:08 AM

@labbbby

Those are definitely killer fans. The main reason I really wanted to go with an external single fan is to plug directly into the mains and bypass the PSU that is already gonna be feeding two quadcores and DX10 cards, etc. PSUs that can handle that job no problem start costing $500 and up and that's just too much damn money for something that is little more than a fancya$$ transformer. :lol: 

@JMecc

That looks great! Kinda like a wind tunnel for your components. Great airflow too. It would be fun to put that into an acrylic case with a few LEDs!!!
December 11, 2006 7:08:26 AM

I would have to advise against the one monster fan; we're trying to cool specific hot-spots here, not blow a gale through a prairie.
I'd recommend simply a good air cooler (such as the Tuniq Tower), with some >200 CFM Delta fans for cooling.
As Rexter said, however, I'd have to side with some extreme water cooling. With no noise limits, you could do some crazy things; industrial pumps, quad radiators with Delta's... but there's always the price issue.
December 11, 2006 7:24:36 AM

Bad idea for several reasons.

First some principles of forced air convection:

1. Efficiency of forced air convection is proportional to the velocity of the air, to a point of diminishing returns.

2. Forced air will follow the path of least resistance.

3. Metal fins, pins, ribs etc typical of heatsinks provide significant resistance.

4. Velocity of air produced by a fan is the greatest at the fan blade and diminishes gratly with distance unless restricted by a constant cross section duct.

Some of the many reasons why a large 12" fan will not cool as well as several small fans attached to the heatsinks:

1. A large CFM, large diameter fan will likely produce lower air velocity than a small diameter high RPM fan.

2. Large fan will be far away from hetsinks producing significantly lower air velocity than a small fan attached directly to the heatsink

3. Large fan, if not ducted via venturi such that all of the air volume is forced into the heatsink fins; air will travel path of least resitance, over and around the hetsink, not through the fins, due to increased resistance in the heatsink.

You are much better off with using typical cooling heatsinks + fans and replacing the fans with same diameter high RPM units. Adding the big fan can help feed the small fans with fresh cool air but cannot be a substitute for spot cooling where necessary.
December 11, 2006 7:44:45 AM

Thanks for the info, Cirtes and Grim! I'm glad I prefaced this thread with "From what (little) I understand about fluid dynamics" :D 

Ok, so can we have a consensus on the following:

a) Huge airflow through a case is not of paramount importance.
b) Individual heatsink fans do not negatively interact with huge airflow.
c) You're better off very efficiently cooling the known hotspots than trying to flatten Saskatchewan.

Now I also have questions about ambient temps. It seems to me (from a recent thread) that someone in a 90F room is not going to get the low component temps with WC than if he were in a 68F room. After all, no matter how efficient the rad, it's gonna be quite difficult to get the water much below ambient air temp (unless TEC or other cooling is applied.)

Now the room where I have my case is a small, windowless room which is not air conditioned. In the summer, ambient temps in that room can easily exceed 100F, and that's without a system running and adding even more degrees to the mix. The room where I sit is wonderfully AC'd but I can't duct that AC to the little room where the case lives.

Would the hot (!?!) setup be to get a monster triple, quad or whatever rad and set it up in my room, with the hoses running through the small hole in the wall where the cables now pass? The prob is that I emphatically do not want fans of any kind in my room, even the "silent" ones make an unacceptable amount of noise. My room is used for a lot of video viewing and the last thing I need in here is BUZZZZZZZZZZZZZZZZZZZZZZZZzzzzzzzzzzzzzzzzzzzzz.

If I wanted to go with a truly passive, massive rad for the WC system, any suggestions?
December 11, 2006 9:25:40 AM



Saw the picks of the kind of fan you want and if the proper methods are not taken you will have real serious vibration this is my pick of the build, great care has been taken to produce a reduction of vibration. The fan cooling the rad is mounted on a stick of wood with rubber hosing under the stick of wood and same as the back one but it was a bit tricky to get the vibration away from the case. I found a method that allows the fan to slightly move, this reducing the vibration of the case.
December 11, 2006 10:00:30 AM

Nice build! Cosmetically it resembles the electrical panel in the basement of my friend's 1920s farmhouse, :lol:  but since function is the key element, it's really great!

Now I don't think that even in this setup you're reaching that kind of windspeed but wouldn't there be some sort of low-end inertial cavitation near the fanblades of the expeller fan? That's basically what I meant by 2 fans negatively interacting. Or would the effect be so minor as to be irrelevant?
December 11, 2006 10:41:37 AM

Not in this build, one blowing in the other blowing out, the holes in the back fan reduce the back splach some what more. I can say this, even with the psu out of the case the air blowing out is quite warm. You should see when the back fan is on speed 3 wind tunnel for sure and the case temp go's from 22 to 18 while system is oc'ed that is with the psu not in the case.
December 11, 2006 11:16:03 AM

Yes, I realize that the airflow in the case is from left of the photo to right. Both fans are pointing towards the right. What I meant by inertial cavitation is the effect that you get when you have two fans blowing in the same direction in a line. Let's call the fan on the left of the photo Fan A and the other Fan B. I understand that if Fan B is generating, say, 500 CFM by itself, if you turn on Fan A and have it generate 500 CFM of its own, blowing on the back of Fan B, the airflow will interfere with the exit volume of Fan B, bringing it from 500 CFM to, say, 400 CFM. Now I don't know whether the difference would be from 500 to 499 or from 500 to 300, but that's my understanding anyway. I very well may be dead wrong.
December 11, 2006 12:13:03 PM

Great link. Thanks!

"In both series and parallel operation, particularly with multiple fans (5, 6, 7, etc.), certain areas of the combined performance curve will be unstable and should be avoided. This instability is unpredictable and is a function of the fan and motor construction and the operating point."

Is that the "cavitation" effect I was discussing?

"It is also important to consider fan placement in the enclosure. Pressurizing the enclosure is the preferred method, since incoming air can be readily filtered. In addition, a pressurized enclosure will prevent dust entering through cracks or crevices. The fan is also handling cooler, denser air, and it will therefore have a slightly higher pressure capability (this may be a very slight advantage for low heat dissipating systems). An important feature of a pressurized system is that the fan life and reliability are increased due to the fan ambient temperature being lower. The disadvantage of pressurization is that heat generated by the fan is dissipated into the enclosure."

Now what extent of pressurization are we talking about? Is this the conventional pressurization of a fan blowing in more air than is easily let out, or actual pressurization of a noticeable amount which would require compressed air or ???

Regardless it was a very informative article!
Anonymous
a b K Overclocking
December 11, 2006 2:06:59 PM

Hehe nice thread you ended up making =), informative!

I would say the wind tunnel is really the best bet, but that pressurization thing is cool!

One more advice, I would suggest a Dual PSU configuration for this type of rig. One powerful one to feed the CPU/GPUS and nothing else and a ~300W or even lower for the remaining peripheral and fans.

Or two moderate one, using one for the CPU's and the other for GPU's... There's a few great DUAL PSU case out there, the upcoming Antec P190 might end up pretty good.

Anyway just another thread for you to explore, I am sure you can get by with ~300$ or so of PSU!

Good luck with your project =)
December 11, 2006 3:00:24 PM

Quote:
Hehe nice thread you ended up making =), informative!

I would say the wind tunnel is really the best bet, but that pressurization thing is cool!

One more advice, I would suggest a Dual PSU configuration for this type of rig. One powerful one to feed the CPU/GPUS and nothing else and a ~300W or even lower for the remaining peripheral and fans.

Or two moderate one, using one for the CPU's and the other for GPU's... There's a few great DUAL PSU case out there, the upcoming Antec P190 might end up pretty good.

Anyway just another thread for you to explore, I am sure you can get by with ~300$ or so of PSU!

Good luck with your project =)


Thanks, labbbby! I've got this tower case that stands close to three feet high and I can hide Snoop Dogg's entire stash in there. I could put 4 PSUs and still have room for my surfboard! :lol: 

Dual PSUs sound like fun since most of the really heavy duty PSUs I've seen are north of 5 Cnotes. However, that Tagan 1.1KW has caught my eye. Just over 3 bills. Hell, if my system needs more than 1100W I couldn't afford the electric bill anyway! :D 
December 11, 2006 3:04:39 PM

Maybe I should have scaled the fan to still show some on the vid card... You still need individual component fans to force air through the fins like Cirtes said, but a large airflow through the case like I showed would keep the case internal temperature to almost room temperature rather than 10C more and therefore would help cooling (and having a cool-looking unique case, which you have to admit is part of the point).

Quote:
4. Velocity of air produced by a fan is the greatest at the fan blade and diminishes gratly with distance unless restricted by a constant cross section duct.

Uh - what I have IS a duct, and the components REDUCE the x-section, ACCELERATING the flow at those places. Although the components are not aerodynamically shaped, the large fan will still provide much more CFM than 120mm fans trying the same thing in normal cases.

In most cases, even expensive ones, the vid card(s) have an almost static air supply - they suck back in much hotter air than room temperature. The heat below the card stays there (feel below your vid card - the heat sink is not cool like your processor's heatsink). The top of the vid card (and the back of the mobo at certain points) get hot and are never cooled (possibly causing the iPod charging killing mobos issue). In my design, these spots do get cooled and the large airflow mostly decouples the cooling of the individual components (i.e. vid card being hot doen't provide a hotter intake for the cpu & power supply).

So, you DO still need to cool the hot spots with local fans. BUT the huge front fan keeps the ambient temp lower, cooling the medium spots that unpredictably get hot and lowers the intake temperature of the local coolers. One large fan probably costs less than two 120mm's and will blow way, way more and to places you never thought to cool. It is also easy to deflect the flow to where you want it with angled pieces of sheet metal.

Jo
December 11, 2006 3:07:20 PM

I would have to agree with the idea of massive watercooling with the idea that you could provide your massive airflow to the heat exchanger much more efficiently than you could to your PC components. You could duct the airflow from a very large fan to the radiator and get a huge amount of effective cooling - as opposed to simply blowing a gale across the motherboard.

Also - you may want to consider the room you are going to house this pc in. If the room is not being actively cooled - such as a small closet - then the ambient temperature of the room will rapidly rise - reducing the amount of actual cooling from any solution you provide. If the room is already being cooled - then you should be in good shape - otherwise consider this as part of your solution.

Additionally - you may simply want to put this together as a bench solution - if all of the hard drives and the MB are on a raised platform - then ambient air cooling will be very effective. Also your water cooling (if performed) will be easier to impliment. Heck - you could even put in two water cooling solutions - one for the CPU and one for the vid and memory - each seperate. If this is the case - you might want to look into making this room reletively clean - to minimize dust.

Good luck.


Good luck -
December 11, 2006 3:09:25 PM

Humm, I think if I had both fans blowing towards each other that would be what you are explaining negitive air pressure not good at all.
I have a fan cover with a great filter that has to be cleaned 1 a month, interesting thing about the pressure thingy is when I take the top off the case with the fans going the pressure sucks the lid down. I have a 120mm fan to help the pressure excape



December 11, 2006 3:11:24 PM

Very interesting info, JMecc. Maybe you can enlighten me on something that has puzzled me since day one.

Since most systems these days are towers and the standard orientation is with the slots down what rocket scientist came up with the standard that all the electronics are on the bottom of the cards? Did they ever hear of heat rising? Why in the name of Jimmy Crack Corn don't they just build the cards with the goodies on top so that the fan can vent up instead of down??? :?
December 11, 2006 3:33:11 PM

Quote:
Since most systems these days are towers and the standard orientation is with the slots down what rocket scientist came up with the standard that all the electronics are on the bottom of the cards? Did they ever hear of heat rising?


I'm not sure why they do that - I even have a passively-cooled vid card with a heat sink pointing down which holds the hot air pocket in the same place. It makes no sense. Some cases hold the mobo upside down and in that case, the video card fans are facing up, which is the right way I would say. I'm not sure who designs heat sinks for electronic companies anyway. I'm not sure that Electrical & Computer engineers even take heat transfer; from what I remember it was only a mechanical and chemical engineering thing. Although the electronics companies may have a heat dissipation team, it is likely that these are not people directly educated in heat transfer. This is ok if they are well-educated in the workplace (engineers are good learners), but I would imagine that heat dissipation would not be where the company puts the most emphasis. If you work for a group that the company doesn't care about as much as the other groups, your performance will likely suffer.

Jo
December 11, 2006 3:38:19 PM

Quote:

Also - you may want to consider the room you are going to house this pc in. If the room is not being actively cooled - such as a small closet - then the ambient temperature of the room will rapidly rise - reducing the amount of actual cooling from any solution you provide. If the room is already being cooled - then you should be in good shape - otherwise consider this as part of your solution.

Additionally - you may simply want to put this together as a bench solution - if all of the hard drives and the MB are on a raised platform - then ambient air cooling will be very effective. Also your water cooling (if performed) will be easier to impliment. Heck - you could even put in two water cooling solutions - one for the CPU and one for the vid and memory - each seperate. If this is the case - you might want to look into making this room reletively clean - to minimize dust.


My setup is as follows:

Main room. Air conditioned. Monitor, keyboard, mouse, etc. in this room. There is a one inch hole knocked in the wall behind the desk which routes the various cables to the:

Small room. Not Air Conditioned. No windows. This is where the PC case sits with all the goodies humming away. No problem in winter as this room stays at normal indoor ambient temp. Big problem in summer. Temps will exceed 40C there on a regular basis, and that's without any electronics increasing the temp even further. Small room is not air conditionable, not ductable, not anythingable. It is what it is. Fortunately it's spotlessly clean and has no carpets or fibers in it to generate a lot of dust.

The MB now sits in a humongous Tower Case almost three feet high. The only fans in it are: CPU, GPU and PSU. It once had two 120mm blowing in and one 120mm blowing out but I found that in normal ambient temps I could drop my CPU temps by 4C by just taking off both sides of the case and forgetting the case fans.

So my thought was to WC the rig and run two hoses through the current hole (ok, so maybe I have to ream out the hole a little) to an absolute killer monster rad set in the Main Room. The rad could be off of a 1974 500 cubic inch V8 Cadillac Eldorado. I don't care. Got lotsa room. The important thing is that it have absolutely no fans on the rad. Why? The main room is used as a media room and complete and absolute silence is of paramount importance.

Therefore, by keeping all the noisy bits in the Small Room and only running a 100% passive rad in the Main Room, I could have my cake and eat it too. So the $1 million question is... is there such a passive rad in existence that would do this job on a killer double quadcore rig with a smokin' DX10 card, etc. etc. etc.? :D 

Quote:
Humm, I think if I had both fans blowing towards each other that would be what you are explaining negitive air pressure not good at all.
I have a fan cover with a great filter that has to be cleaned 1 a month, interesting thing about the pressure thingy is when I take the top off the case with the fans going the pressure sucks the lid down. I have a 120mm fan to help the pressure excape


Actually, this pic will show what I mean. Excuse my crappy photoshopping...



The area behind Fan B, according to what I believe are the principles of inertial cavitation is actually moving less air with Fan A on than Fan A off. What I'm trying to determine is whether this is right or it's just my hallucination! :D 
December 11, 2006 4:03:27 PM

Actually the air in the pic is not how mine is blowing, reverse the order, the air is sucking through the rad and blowing the air to the back, the back fan is sucking the air out. The most important parts of the MB, CPU, Ram, I really notice and see the difference with the transformers, coils, voltage regulators are cooled this way.

If you look at the blades and the shaft you see the fan blade was taken off the shaft and turned around. I spent lots of time, wasted money on cases. I created a perfect air, water cooling system that produces real difference and for 1.5 years this has work perfectly.


<--------------------------------------------------------<
crappy drawing skills
December 11, 2006 4:05:50 PM

If the 2 fans are both pushing air in the same direction, they should be helping each other. How do you know the airflow is higher with only one on (which is A and which is B anyway)? Are they both sort of the same (run at the same speed, same CFM)? Is the case enclosed (you put a lid on the top)?

About this cavitation thing, see http://en.wikipedia.org/wiki/Cavitation - they explain a lot about cavitation. Effectively it happens in water - the back of the fan blade moving creates a high-pressure zone on one side of the blade and a low pressure zone on the other side (like a van has high pressure on its windshield and low pressure on its rear). The low pressure on the back of the blade can get so low that the water boils at its cold temperature (water boils easier at low pressures). This makes tiny bubbles which collpase on themselves in microseconds, hitting the impeller so hard locally as to make a pit in the steel. After much of this goes on, it wrecks your fan. This is not a process happening in your case.

Jo
December 11, 2006 8:42:01 PM

Cap'n

You are asking for a solution that is not very practical and are imposing restrictions that make it near impossible to implement.

Both Intel and AMD release design guides for motherboards, CPUs and chassis/system design. In the design of any computer component the first thing that is stated is the target environment temperature, which tends to be in the 20 C range.

Then you add the standardized industry practice which is used in most deigns:

Interior chassis temperature should be no more than 15C higher than ambient temperature.

Most manufacturers produce two designs; one for Server markets and a second for Consumer markets. For Server markets, the component target operating temperature is usually 50C and Consumer design goal is 35C.

You have stated a design goal of operating your equipment in a room that often reaches 37C even without the proposed 1000W gear dissipating heat in a small unvented environment.

The only viable method you present is passive WC, which can be done if you do install a large number of heat exchangers in a controled temperature location but it will not solve the problem, only help with reducing it. Even if you WC the CPUs, GPUs, CHipset, HDD's etc there are all sorts of voltage regulators, your 70% efficient power supply, hundreds of capacitors etc working beyond their design specification for prolonged periods of time that are not cooled by the WC.

Don't get me wrong, your system will run for a while, but you are essentially going to de-rate the MTBF of the system well below the original design goals.

This will make your system more likely to fail in the first year of operation by a significant amount. Most designs target ~ 1% failure rate in the first year when the system is run in a room with 20C temperature. Probability of failure doubles for every 5C above designed so your 40C room will likely cause the system to have a 16% chance of failing in the first year of operation. If the room with all the extra heat from the gear goes to 50C the failure rate probability in one years time goes to 64%.

These are not good odds for a multi-thousand $$ investement.

I would seriously consider changing the requirements such that you are able to cool the small room. You can get a floor standing AC unit relatively cheap, provided you have a means of installing an air duct that leads out of the room.
December 11, 2006 8:51:43 PM

Quote:
Actually the air in the pic is not how mine is blowing, reverse the order, the air is sucking throw the rad and blowing the air to the back, the back fan is sucking the air out. The most important parts of the MB, CPU, Ram, I really notice and see the difference with the transformers, coils, voltage regulators are cooled this way.

If you look at the blades and the shaft you see the fan blade was taken off the shaft and turned around. I spent lots of time, wasted money on cases. I created a perfect air, water cooling system that produces real difference and for 1.5 years this has work perfectly.


<--------------------------------------------------------<
crappy drawing skills
Bump to make sure you know the air flow
December 12, 2006 12:32:41 PM

Our beloved ATX cases are indeed upside down. Research BTX - particularly with respect to component orientation i.t.o. cooling - and then wonder why ATX is still the favourite...
December 12, 2006 3:01:52 PM

Hey Capt,

I think the idea of running the pipes through the wall to a fanless radiator in your AC'd room is a great idea. I've heard good reviews of the Zalman Reserator 2, but I couldn't tell you if a single rad would handle your whole system. You might have to run two loops, and this would start to cost you mucho $$ and you'd have extra holes in your wall.

As for your ongoing airflow cavitation question, I haven't seen anyone give this question an answer yet, so I'll attempt it. For my credentials, I have taken every aerodynamics course offered at my Alma Mater, including grad-level. However, I have very little direct knowledge of turbomachinery, so someone else may have a more qualified answer for you.

Firstly, the concept of "cavitation" does not exist for airflow problems. Cavitation occurs for underwater propellers when the localised pressure drops too low and vapor bubbles form. This causes large amounts of vibration and noise, and certainly is a problem for submarines looking to avoid sonar detection.

For fans operating in air, there is no cavitation. There can, however, be flow separation, which when extensive enough, results in stalling. Stalling occurs when the angle of attack of the lifting surface is too high, and is also a function of turbulence. With respect to having two fans working in tandem (in line, blowing in same direction), the first fan will increase the turbulence on the second, and so increase the tendency towards separation. However, the extra velocity of the air will reduce the effective angle of attack at the fan blade, and so reduce the tendency towards separation. The latter trend is by far the dominant trend in most cases, so you don't need to worry about separation.

Fans, however, are most effective when they maintain a substantial angle of attack just below the stalling point. For most tandem fan operations, contra-rotating fans are employed so that the spin induced on the flow by the first fan will increase the angle of attack at the second. In this way, it is possible for the fans to work together synergistically. I would recommend using fans spinning in opposing directions for your setup.

There have been a few aircraft built where there are two propellers working in tandem, and these have in general been very successful. Consider the close-coupled contra-rotating fans on the Russian bomber and transport aircraft, or some of the private aviation twin-boom pusher/puller designs.

Once you have your computer set up like a wind-tunnel, you need to make sure that your heat sink fins are lined up parallel with the flow. The heat transfer will be most effective if you do duct all the flow through these heat sinks, but even in an open environment some of the flow will go through the heat sinks, and not all around it like some other posters have suggested.

You can think of the airflow as electricity going through a circuit, with two resistors in parallel. One resistor is large (your heatsink), and the other is small (the free path). If you apply a voltage, there will be current through the large resistor. If you want to increase the flow through the heatsink, you can do one of many things. You can increase the fan capacity (higher voltage), increase the resistance of the small resistor (restriction of airflow or ducting for non-heat sink air paths) or decrease the resistance of the heat sink path. You are already set to do the first option. The second option is hard to accomplish, and has the disadvantage of depriving the rest of your mobo from effective cooling. The third option is a good idea. I would recommend either using your liquid cooling idea for the hot-spots, or use a passive heatpipe-tower of fins cooler like the ThermalTake Sonic Tower to cool your CPUs. With significant airflow, these towers will certainly keep your CPUs nice and cool.
December 12, 2006 6:51:32 PM

Quote:
Hey Capt,
You can think of the airflow as electricity going through a circuit, with two resistors in parallel. One resistor is large (your heatsink), and the other is small (the free path). If you apply a voltage, there will be current through the large resistor. If you want to increase the flow through the heatsink, you can do one of many things. You can increase the fan capacity (higher voltage), increase the resistance of the small resistor (restriction of airflow or ducting for non-heat sink air paths) or decrease the resistance of the heat sink path.


Oh I just love it when folks go to the Physics store and use circuit modeling.

Let me use a different model that is a touch more accurate.

The whole case is a large block of copper. The heatsink is a much smaller block of iron embedded in the copper block. You apply a voltage on two sides of the copper. Does any current go through th iron? Not really.

The physics behind obstacles in an airflow are as follows:

An obstacle creates increased pressure on the leading edge and reduced pressure on the sides. The air must overcome the increased pressure by increasing the velocity as it goes around the obstacle since volumetric flow remains constant. This is great if you are trying to build a wing on an airplane. With modern heatsinks you have a large number of fins that are close to each other. This generates a cumulative pressure increase on the leading edge of the heatsink. If the density of fins is high enough, the air will simply flow around the heatsink as if were a single solid obstacle. This is highly likely since the volume of the case is orders of magnitude greater than that of the heatsink and can accomodate with ease the miniscule increase in velocity needed to completely bypass the high pressure point.

BTW, I have designed ~ dozen enclosures for PCs and telecom equipment. Each one was tested with smoke trails to validate exactly what I described above.

This is exactly why heatsinks on CPUs with fans directly attached have really tightly packed super thin fins but the hearsinks that are passively cooled tend to have well spaced out fins, like the chipset coolers.
December 12, 2006 7:55:15 PM

I have to disagree with your statement that your analogy is any more functionally accurate than mine. You've taken a 1-D simplification of a problem and added extra dimensions to it to make it more accurate. This may be true, but airflow isn't electrical current to start with, so there's not much point in getting too elaborate with the dimensional accuracy. My analogy was for simplification.

Also, I must offer a counter-point to the idea that you must have a fan on the heat sink for it to be effective: my own computer. I have a case with one 120mm intake fan and on 120mm exhaust fan, plus the fans in the PSU. My CPU heat sink is the TT SonicTower, with no fan attached. I run my 95-Watt processor at 100% for three days straight and my temps don't go over 55 Celsius. This is in a 21 Celsius room.

I agree with you that most of the air will go around the heat sink if the free path is available. However, there is, at least in my case, still sufficient flow through my heat sink to to a good job. Even if the heat sink were a solid block, there would still be cooling through the flow boundary layer around the cooler (convective heat transfer). Consider that you replace this solid surface with one that is covered in fins, and the boundary layer heat transfer is increased. The boundary layer that is developed over the surfaces of the heat sink is in no way represented by either of our electrical model simplifications, but the point of a simplification is so that people who haven't studied aerodynamics can get the picture quickly.
December 12, 2006 8:27:38 PM

I think the industry term for our discussin is "violent agreement". Yes, the analogy is just one of those things I cannot resist when the Physics store gets involved. Remember the Physics store where you can buy frictionless pulleys and string that does not stretch? It was a fun diversion.

The TT sonic was designed as a passive cooling device. You are deploying it in an environment that meets the original design criteria. The OP will be running his solution in a room whose temperature regularly exceeds 40C so I doubt your good experiences with a low airflow convection sollution will necessarily be applicable to their dillema.

In any event, this is not a right or wrong thing. If our experiences help a fellow out, it is all good. I have had my work scrutinised and approved by Telcordia, UL and NEBS all of which are regulatory enforcement bodies in the US that deal with topics like this. You seem to have comparable credentials and I hope the OP can get some good information from both.
December 12, 2006 8:42:56 PM

Quote:
I think the industry term for our discussin is "violent agreement".

I'm going to have to use this phrase at least twice over Christmas with the in-laws. Hilarious. :lol: 
December 12, 2006 10:52:57 PM

So I have to ask, what causes my lid to be pulled down, when I lift the top to look in- side the case, I can hear and feel the suction. Like I mean, if I smoke a cig, the ashes would be sucked into the case. WHY?
The cooling is awsome.
December 12, 2006 10:53:40 PM

One question that some people have hit on. But I don't think it was answered is... Is it possible to somehow vent the "closed" room where the PC will be located. Like placing a couple of vents into a crawl space or attic? Because that will help vent some ofthe excess heat out and away from the system, and it will allow for possibly some cooler air to be brought into the space. This can be done with 2 vents and a couple 120MM fans (with industrial dust filters of course). And that would alleviate some of the trapped heat issue.

-ouch1
December 12, 2006 11:40:32 PM

Quote:
So I have to ask, what causes my lid to be pulled down, when I lift the top to look in- side the case, I can hear and feel the suction. Like I mean, if I smoke a cig, the ashes would be sucked into the case. WHY?
The cooling is awsome.


Oh, my favorite Physics experiment! Take two standard sheets of A4 paper. Hold them with both hands at the corners of the short edge in the following manner: One sheet between the thumb and index finger, second sheet between the index finger and middle finger. THis should make the two sheets of paper more or less hang in parallel planes. Now, jently blow in the ~ 1 cm gap that is defined by the sheets of paper.

DId they blow apart or did they stick together?

Your computer case is essentially a tunnel with two fans at each end. When you flow air through it by increasing the pressure at one side of the case and reducing the pressure at the other end you are forcing a column of air in the center of the case to flow rapidly from one end to the other. Now, the air that is near the walls of the tunnel will move slower than the air in the center of the flow which in turn causes a pressure difference between the tunnel wall and the center of the case. The center of the flow is at a lower pressure than the case wall and you observe a vacuum effect.

This marvel of science is how a carburator works.

This is all explained by Bernoulli's Law and effect:

For horizontal fluid flow, an increase in the velocity of flow will result in a decrease in the static pressure.
December 13, 2006 2:41:54 AM

Quote: "Now, the air that is near the walls of the tunnel will move slower than the air in the center of the flow which in turn causes a pressure difference between the tunnel wall and the center of the case."

Yes, the flow near the walls is slower (developing boundary layer), but there isn't a significant static pressure difference between the flow at the wall and the flow in the centre. This all comes from potential flow theory. Take a simple example: a straight tube with fans at either end. If you draw the streamlines of the flow, they are roughly parallel to the axis of the tube (here we neglect fan-induced spin and a slight choking effect caused by boundary layer development). The static pressure is constant for all points on a line perpendicular to the streamlines (these are called lines of potential). This means that the static pressure near the wall of the tube is equal to the static pressure at the centre. The static pressure, in fact, only changes along the length of the tube.

In order to get the suction effect you're noting, you would need to have a venturi-type restriction (a waist shape to the cross-section of the tube). Alternatively, blockages within the box may be causing significant reduction in cross-section and therefore you would also get a venturi effect. Otherwise, if you have a lot of suction, you might just have stronger total exhaust than intake on your system.
December 13, 2006 6:39:49 AM

Try it. Get a short length of 3" plastic sewer pipe. Put a 3" computer fan on each end in a push-pull configuration, drill a hole mid point in the tube.

Remember that in this example there is a fan at both ends, hence there is no net pressure difference along the length of the pipe. The only thing creating static pressure is the velocity gradient.

For reference look up Poiseuille's Law regarding the velocity gradient. Then go back to Bernoulli's Law to tie in the drop in static pressure across the tube diameter.

Now this is all models and mathematics. Ouch1 may very well have a venturi by virtue of heatsinks, VGA card and other components in the box causing a reduction in cross section.
December 13, 2006 9:13:07 AM

Sorry I've been AWOL for a while. Duty beckoned and I had to heed the call to keep the world safe for "my brand" of democracy against godless communism! :) 

@ gomerpile

Sorry for screwing up the direction of airflow. I can now see that the fans are pointing backwards. The question I have is what is in the rad. Is it a conventional water rad or is it a compressor like in a fridge? Are you getting any condensation off that rad? Any chance of water droplets being torn off and tossed onto the electronics? Regardless, it looks really nice!


@ Private Messenger

Thanks for the great information and diagram. If I'm not overly mistaken, I think that gomerpile's system uses this "reverse water cool" by sucking air through the rad in his box. I'm not sure though what kind of rad it is, but I've asked him and I'm sure we'll find out soon.

@ Cirtes

Yeah, the problem is not an easy one. When temp in the Small Room reaches 40C, I can pretty well assure you that the inside of a closed case could be 55C. That is damn hot no matter how you look at it. Thanks for the probabilities of failure analysis. Much appreciated. All those percentages are far too high considering that by the time I'm said and done, there's gonna be about US$10K of goodies in that tower case.

Ducting in a near impossibility:

UP: I can get to the outside by smashing a hole through 12" of concrete reinforced with rebar the size of your arm. Next...

DOWN: I'd have to run a duct through three walls and two floors and get to an opening to the outside that is below grade. Not exactly optimal. Next...

EAST: I'd have to duct through 75' of someone else's property. Next...

NORTH: I'd have to duct through four walls and then 60' of someone else's property. Next...

SOUTH: Two walls and 40' to open air. Next...

WEST: I can go through the Media Room. About 25' to the outside wall. That's still a helluva long way to run an AC duct to a window or split AC. Could be done if I had a gun to my head and is definitely the least unlikely of the options.

Now the Small Room could "fairly easily" be ducted into a very large three storey broad stairwell by just knocking a hole in one wall that separates them. Problem is that the stairwell is just as unAC'd as the Small Room. But at least you wouldn't have a tiny, airtight space trying to dissipate some serious heat.

So what that leaves is the following hypothesis:

What if we open a hole from Small Room to Stairwell. At least now we have a 4,000+ cu. ft. air volume to play with instead of 350 cu. ft.

Now let's enclose the case in "some form" of refrigeration. Keep the whole damn case cool. Didn't Asetek have something like that? I've just checked their website and only see a few killer CPU coolers but not whole case coolers. Didn't they once have a VapoChill that would basically AC the entire interior of your case? It was a garden-variety small compressor like one of those that you could buy in a cheap AC at Walmart for $80 and they branded it real nice and put an MSRP of like $700 on it. Nice biz if you can sell 'em. :) 

@ Ouch1

You've proposed an interesting theory. Thank you for that, Ouchy! So the bottom line is, since I couldn't give a rodent rectum about how much heat I'm pumping into the Stairwell, what if I:

1) Open up a good sized hole between Small Room and Stairwell.

2) Set up a standard window AC in the small room. (Can't just make a hole in the wall and set up a window AC in the wall hanging out into the stairwell or I'll catch hell. But if there's just a hole in the wall with a vent on it, nobody will care.)

3) Duct the "cold side" vents directly into the case.

4) Let the "hot side" just vent into the Small Room and then just let air circulation warm up the Stairwell (trust me, nobody will notice).

I can certainly see ambient temps in the Small Room climbing up into the 50s C but hey, I don't care since the inside of the case "should" be around 15C!

Ok, so what have I missed here? Seems too simple! :) 

@ Mugz

It's not like the personal PC business is worth tens of billions of dollars a year (or likely much more) that little things like setting up the electronics upside down wouldn't have been rectified. This is totally nuts. Unfortunately BTX was not picked up by the general market that seems to have a ravenous appetite for midtowers. Just looking at a midtower design, I can't really see any major problem in designing one that sits the MB on the other side and lets the vid cards cooling face up! It really makes you wonder what these guys who design cases are thinking of!

@ HotFoot

Running a couple of Resarators could work, but then again, you'd run into the valid point Cirtes makes which is that the rest of the circuitry in the setup that I'm stuck with could be running into air ambient temps that could conceivably spike close to 60C. I like the idea of using a 1974 Cadillac Eldorado rad. Would make a fascinating conversation piece, especially if it was chromed! :) 

Thank you very much for clarifying my stupidity on cavitation. Your definition of flow separation is what my dull mind was trying to describe. The contrarotating fans is a really interesting idea! I can't comment on the difference of opinion you and he have on your circuitry analogy since I'm just a dumb backcountry Po' Boy and was too busy checking out the training bras filling up in high school physics class to pay much attention to anything else. Thanks once again for clearly and concisely setting me straight. I appreciate that! And I hope that you and Cirtes resolve your "violent agreement" :) 

@ JMecc

Sorry that I was using cavitation improperly. I was definitely using the wrong nomenclature. HotFoot set me straight.

As for Bernoulli's Law, now that I do understand. You should always buy Bernoulli Extra Virgin Olive Oil. Not only because of it's rich green colour and wonderful aroma, but also because there are so few Virgins going around that you should never miss an opportunity to get one!

Now Poiseuille's Law I'm familiar with. I've lived by it, especially when I was in France. "Always get all the Poiseuille you can!" :twisted:
December 13, 2006 10:43:06 AM

Quote:
Very interesting info, JMecc. Maybe you can enlighten me on something that has puzzled me since day one.

Since most systems these days are towers and the standard orientation is with the slots down what rocket scientist came up with the standard that all the electronics are on the bottom of the cards? Did they ever hear of heat rising? Why in the name of Jimmy Crack Corn don't they just build the cards with the goodies on top so that the fan can vent up instead of down??? :?


You would do well to stop pretending you know this. Cooling what you propose is extremely easy and yet this silly thread about ways to cause problems.

It is irrelevant whether heat rises, because that is insignificant in an actively cooled chassis. Further, hot parts are sunk with through-board vias to larger copper areas on the back of the card, that is one of the major heatsink paths even if a chip has a 'sink on it, and because fiberglass and copper are less tolerant to heat than silicon wearing a heatsink, is.

As for the prior poster who claimed their passive video card has a problem, they obviously shouldn't be building systems with parts they don't know how to implement, because a passively cooled card that runs hot enough for it to matter, requires a chassis airflow design to accomodate it. Merely flipping a card over won't help enough and will interfere with proper cooling via passive (chassis) flow.

Given a bad chassis, the remaining option is to leave the adjacent slot empty and pull that slot cover bracket off, letting case negative pressurization pull air past the card- and it will be cooler than if it were flipped over, because air travels along the face of the card instead of straight up to exhaust points.

Believe it or not, much research was put into the standards, they DO work fine. The problem is when individual parts manufacturers try to reduce cost with inferior cooling, or make bad judgements in design or component real-estate allocation.

The key to good cooling is focusing on the specific high heat areas, not trying to brute-force more air into the chassis which is unnecessarily loud, more expensive, higher dust accumulation, often more power used, more chances for mishaps from people, pets, stray wires, bugs, etc.

If you're not trying for a really quiet system, there is no reason a 40C room would cause 15C temp rise in the case.

Ducting this system across the house - LOL! Entirely unnecessary but it's a project alright!

If you have a really bad case and need extra cooling for SLI'd video cards, put a side panel fan in. Beyond that, and having a front drive rack fan so the side intake doesn't rob the drive rack of airflow, there should be nothing more necessary. If parts are still overheating, gale force winds won't help as much as a suitable heatsink would, because the first couple cm of metal the heat travels are by far the most critical. The airflow itself has quite diminishing returns if the 'sinks are adequate.
December 13, 2006 11:12:23 AM

Quote:
You would do well to stop pretending you know this. Cooling what you propose is extremely easy and yet this silly thread about ways to cause problems.


I can agree with a sizeable amount of what you've stated, but I can't understand these two sentences. I can interpret them in two ways. One is an honest expression of your opinion and the other merits a :p  :p  :p  :p  :p  :p  :p  :p  :p  :p  :p  :p 

:lol: 
December 13, 2006 12:16:08 PM

My point in bringing up potential flow theory was to highlight the point that Bernoulli's equation can only be applied along streamlines (parallel with the flow), not across them. It is just mistaken physics that you, Cirtes, keep trying to apply Bernoulli's equation for two locations that aren't on the same streamline: near the wall and at the centre of the flow.

If you're trying to say that faster-moving air within the case has a lower static pressure than that on the outside of the case because the of the flow within the case, then I can't argue with you. I just want to get this very fine detail ironed out, because as you say, people like to go to the Physics store and start talking about air flow.... and something about a frictionless pulley and string that doesn't stretch.

In any case, it seems we've agreed on everything except for the details in how to explain what's going on. I'm not interested in running a full Navier-Stokes solution for this particular problem. If we're both so interested in an aerodynamics question, we should probably get our own thread.
December 13, 2006 2:11:19 PM

Quote:
It is irrelevant whether heat rises, because that is insignificant in an actively cooled chassis.


What? Do you really think the WHOLE chassis is getting good airflow? Cases are built with a chassis fan ABOVE the vid card and only quasi-static air below, especially with PCI devices. Heat rising does matter to passively-cooled devices, as well as having the case air motion. Video cards would do better flipped upside down IMO. The cooling they have now IS adequate for the cards, but would be BETTER upside down which is important for OCers.

Jo
!