Is there a difference heat exchange wise with these designs?

[CheapMeat]

I'm sorry for the very basic and simplistic drawing. But say all components are exactly the same, same load, same ambient temp. Is there a difference in the efficiency or cooling ability of one over the other? Is it better to forcefully blow air unto the components? Or forcefully suck air away from the components?

Image/Design:

http://imgur.com/a/4frqgyr

I'm designing custom rack chassis's for my homelab builds and want to do something different. I plan to have a mid-plane fan setup but wanted to simplify the question.

I have some pretty good fans, even one of the only truly working contra-rotating stacked fans (140mm Silverstone variants), so was thinking the high pressure forcing air onto the heat sources would be better.

But I like the idea of something with all IO and some hotswap bays on the front instead (with custom bezel) and only exhaust and power in the rear for hot air extraction in my setup. Does higher pressured or faster air blasted on a component work better? Or is the "sucking" away work just as well? Not sure if this makes sense as a statement (not an engineer ).

 

[MrPotatoHead]

It's always better to expel the hot air instead of trying to cool it, so I'd say the second picture would provide the best cooling as you have cool air being pulled over the hot components and hot air being pushed out, though an even better approach is the push pull system that most cases are setup to provide, so unless this is a specific usage scenario then a well balanced push and pull fan setup would be recommended

 

[TheMadDutchDude]

They’re the same airflow path, look at the arrows more carefully.

I’d personally opt for the one on the left. It’ll cool surrounding components with little heat pick up vs. dumping the hot air from the main source into the chassis. You’ll need some very potent airflow and static pressure for either scenario, though.

It’ll be loud... so keep that in mind.

 

[MrPotatoHead]

They’re the same airflow path, look at the arrows more carefully.

I’d personally opt for the one on the left. It’ll cool surrounding components with little heat pick up vs. dumping the hot air from the main source into the chassis. You’ll need some very potent airflow and static pressure for either scenario, though.

It’ll be loud... so keep that in mind.

Look at the case fans..

Ideal scenario would be the second pic but with the fans at the top as in the first one, if that's possible

 

[TheMadDutchDude]

One is blowing air through the chassis, the other is sucking through the chassis. Like I said, look at the airflow arrow...

 

[CheapMeat]

Here's the plan (with a basic drawing) for something a bit more unique, although I've seen something similar done before. On the rear I could have two PSU's (redundancy or extra SATA cabling power for way more drives (15+ 2.5"), maybe use SFX PSU's for better space optimization; using Phantek's Power Combo) or just use one PSU and lengthen the fan area. Mostly trying to make something workable but interesting for myself (just an enthusiast). But if I do this "IO in front" idea, I was a bit worried if "sucking away" heat would be worse off. I have an inclination that "sucking away" would be better off for a crowded PCIe slot area (using all 7 slots on boards).

Since the IO shield and often PCI devices & their covers don't have a lot of holes, I would supplement some of the intake holes by adding perforations on the side front too.

 

[MrGRiMv25]

What about a straighter airflow across the entire setup with either the one or two PSU's in the middle instead of the sides, that might draw the air through more efficiently?

Depending what way around the PSU/'s are they'll take the middle airflow in through their fan and expel it too (in theory? er...)

Witness my paint skills in all their glory.


Could also reverse the airflow so it's flowing vertically in relation to the pic above.

 

[Vya Domus]

The hot air has the tendency to rise to the top of your case, your airflow shouldn't go against this natural flow.

 

[r9]

I'm sorry for the very basic and simplistic drawing. But say all components are exactly the same, same load, same ambient temp. Is there a difference in the efficiency or cooling ability of one over the other? Is it better to forcefully blow air unto the components? Or forcefully suck air away from the components?

Image/Design:

http://imgur.com/a/4frqgyr

I'm designing custom rack chassis's for my homelab builds and want to do something different. I plan to have a mid-plane fan setup but wanted to simplify the question.

I have some pretty good fans, even one of the only truly working contra-rotating stacked fans (140mm Silverstone variants), so was thinking the high pressure forcing air onto the heat sources would be better.

But I like the idea of something with all IO and some hotswap bays on the front instead (with custom bezel) and only exhaust and power in the rear for hot air extraction in my setup. Does higher pressured or faster air blasted on a component work better? Or is the "sucking" away work just as well? Not sure if this makes sense as a statement (not an engineer ).

I would pick the first option, because the heat would not enter the case at all.

 

[Athlonite]

If you talking about a desktop case then either scenario is crap in that configuration the one on the left would be better if you turned the fans over so they expel the air faster same with the one on the right as it would then suck cool air into the case much like the Silverstone raven series of cases do 3 x 180mm fan on the bottom blowing straight up and hot air exhausting out the top your also not fighting thermodynamics as heat naturally wants to rise so give it a helping hand not a hindering one

 

[FordGT90Concept]

Image/Design:

http://imgur.com/a/4frqgyr

If the "Heat Source" is a radiator of some kind, then the one on the left will be better because you need pressure to force air through the fins otherwise the air will take the route of least resistance which is to go around it.

View attachment 129545

As noted above, you're going to want your fans just before choke points to force air through the choke point. It's difficult to tell where the choke points are in this diagram.

 

[Vario]

The hot air has the tendency to rise to the top of your case, your airflow shouldn't go against this natural flow.

An additional reason is having the impeller facing upwards is more stressful on consumer fans that are FDB or sleeve or cheaper ball bearing designs. These fans will start rattling and eventually fail.
I run a top intake because I have a unique case design, the fan I use is a Yate Loon from a power supply that is dual ball bearing and hardy enough to operate in that position.

As far as thermal direction, while it will naturally rise when there are no fans, once you add fans to the case, it goes where your fans push it.

I would want to run the left image with the fan at the top. It positively pressurizes the case which helps prevent dust build up, and if the only exit ventilation point is on the bottom, the heat will exit easily through the bottom as the exhaust air is expelled. You could also put a fan at both the top and the bottom. It may not be necessary to run both an intake and an exhaust fan if the only exit ventilation holes are at the bottom and all wiring and other obstacles are tucked away so the pressurized air does not generate turbulence around these obstructions. You want to run an exit ventilation that is very minimally restrictive, such as a wire grill rather than a perforated steel mesh, because you want air to pass through the case with minimal pressurization of the case, its about moving a high volume of air in and then out.

 

[lilhasselhoffer]

The left hand design is preferred. The issue is there's a bit more to this. Let me explain.

1) Hot air is less dense than cold air. In modelling terms, PV=nRT. The n and R are fluid based, so PV=cT. Pressure is relatively speaking constant at atmospheric, so T increases mean a V increase, and V increase without mass means less density. Nerd speak free, air expands as heated and the same push moves less actual air.

2) Heat transfer, in this case is largely conduction. This means energy transfer rates are dependent upon the largest possible delta in air and component temperatures. Basically you want the hottest component nearest the cool air intake, so more energy can be pulled from it.

3) Pushing cold air over the components is preferred to pulling hot. Read the CFM ratings on fans. The RPM rating on the fans is generally given. This means that a constant RPM fan will move less actual air in a pull configuration than a push just due to density. Conversely, the same fan will draw less energy in pull configuration.

4) Don't add mesh or holes to the sides. Air experiencing a bunch of holes will rapidly become incoherent, changing from laminar(ish) to turbulent flow. The problem here is that laminar flow is largely conductive heat transfer (the best) while turbulent flow is convective (far less efficient). The trade-off here is less coherent flow for greater fouling (dust collection) and lower noise levels. The problem cited is not to minimize noise, and it definitely isn't a situation where greater fouling is ideal.



The above is why, of the two solutions, the fan pushing cold air in is better. The argument that pulling the hot air out is easier can be had, but most fans are RPM limited rather than power draw limited (PWM). You will have to concern yourself with fouling, and faster air hitting obstacles will drop debris, but a yearly cleaning isn't exactly a hardship.

The question, at the end of the day, is will it make a huge difference? I'd hazard to say either would be at most a couple of degrees different at absolute most (with my money being sub 1 C), which doesn't seem to be an immediate concern if you're only purchasing a single large fan to cool the entire system. Heck, most people do push-pull to make sure you get consistent flow. Another $20-$30 (for a 120mm fan at least) would give you a second fan, and if that's a financial impossibility then there are far deeper issues with the build. If you've got one huge expensive fan, then this is all a joke. A huge static pressure and CFM functionally void concerns about placement because fricative losses are such a small percentage of the total applied flow.


Edit:

For some pretty pictures, check out the following:

Computational Fluid Dynamics | Dojo Research & Consulting

e use COMSOL Multiphysics as our CFD platform. Besides traditional CFD simulation, we have capabilities in heat transfer, microfluidics, two phase flow and phase transitions.

www.dojoresearch.com

Note that the energy output (TDP equivalent) is 1 watt, the velocity is 0.05 m/s, and the temperature delta is 1 K (=1 C measured from absolute zero). Bump the output energy to your desired TDP, the delta in temperature would then be a function of v (note that v increases as density decreases because the fluid expands) and the input temperature. Your response is either to run a pull at a greater RPM than the push to get the same volume of air, or get better performance pushing with a greater power draw from the fan. Again, the power draw difference, because air is so very low density, would be miniscule. The resulting delta in actual steady state temperatures would be similarly miniscule.

 

[CheapMeat]

Sorry I just realized how the orientation of the image can be a bit confusing. The design is meant to be horizontal inside of a rack like this:

I'm reading through the replies, thanks guys. Thinking more about this all now.