CPU Cooling Push Push, Thermodynamics & Heat Transfer

[alansdaman]

Do you guys ever try push / push. Some people say they "fight" each other but that doesn't ring true to me. If the mass flowrate increases, the temps will be lower (in Q=mc delta t, if m increases, delta t decreases). The air will blow out the sides of the cooler and flow in thru the front and back. This might create some compensating factors (such as portions of the heat exchanger not receiving airflow) but those exist in the dead spot of the fan in Push or Push Pull anyway. The only other thing that would make me think push pull could be more efficient and effective than push push (or pull pull) is if the fluid resistance of the smaller sides is far below that of the front and back. I would think this would only become a factor with extremely high cfm thru the cooler.

So I guess my question is:

Has anyone actually tested this? Writing it off before trying it is kind of silly. Those compensating factors might be more of an impact, but without using computational fluid dynamics software (like what I use at work), testing is the only real way to know.

I think that especially at lower RPMs the push push would excel. So for those looking to do hyper efficient and quiet computing it might work well.


Also, in a cooler where the open area of the sides and the front are the same (a square if looking down on the cooler) I think push push would excel further as the compensating factor of higher fluid resistance (with the smaller sides) would be gone.

You could also place a cooler on the front and side (so they make an "L" shape around the heat exchanger) which should eliminate the dead spots in the center of the fan (if you look at a fan, there is not any flow thru the center where the fan motor and rotor rest and no fins lie). If you could duct the exhaust from the cooler out of the case (or duct the fan exhaust in a pull pull which might be easier) then your inlets would be lower, your delta would be lower, and your cpu temp should be lower for any given load and fan speed.

I know I'm talking a lot of thermodynamics and heat transfer / fluid flow, but most of those concepts scale absolutely (I just wonder if the degree to which effectiveness increases are measurable). I would love to know if someone has given these ideas a try and would love to help coach them thru what I am thinking. I am planning to build a rig in the future and I want to totally geek out on it, but right now I'm saving for a down payment on a condo so it's somewhere on the horizon.

FYI I am thinking about the asus p8p67, i5-2500k, coolermax 212 plus, using the case from my Gateway LX6810-01, 8 gb of ripjaws, a corsair builder 650w psu and some type of hd 6870 with my existing raid 150 vraptors and and 600 gb storage drive and using a ducting system to create hot aisle / cold aisle in my case.

 

[alansdaman]

PS. I still think it might be more effective to install one fan with more area for the prime mover and a higher speed than using two.

 

[rubix_1011]

This might work on an open-sided air cooler, but running push/push or pull/pull on a watercooling radiator will result in a much different ending. The radiator is essentially a sealed box that the fans bolt to. Push/Push or Pull/Pull of equivalent fans would net you zero airflow in either scenario.

 

[alansdaman]

Yes of course, a flowpath must exist.

 

[4Ryan6]

This might work on an open-sided air cooler, but running push/push or pull/pull on a watercooling radiator will result in a much different ending. The radiator is essentially a sealed box that the fans bolt to. Push/Push or Pull/Pull of equivalent fans would net you zero airflow in either scenario.

I totally agree with everything except, "This might work on an open-sided air cooler", but my post is not directed at you rubix it's directed at the OP.

The same thing will happen to an air cooler, if you have 2 fans pushing against each other they will effectively cause a stall in air flow, simply because none of the heat pipe designs are setup for that type airflow scenario, they are all flow through designs.

Even if some air can escape out of the sides, the heat sink is not designed to take advantage of it, the aluminum fins laid out through the center are not just there to hold the heat pipes together they have a cooling transfer function to perform, transferring heat from the heat pipes to the cooling fins to the air flowing through the heat sink.

Some of the transfer fins have differing designs including winglets, dimples, special bends, anything to take as full advantage as possible to the air flowing over their surfaces, principle being to transfer the heat to the air and get it out of the fin area, not keep some of it trapped, or slow, or stall the air flow.

If you take into consideration of the dead air space created by the fan motor body itself blocking airflow, having 2 fan motor bodies back to back with dead air sandwiched between them, would effectively create not only a fixed dead air space but a heat zone between them, because they produce heat themselves.

From all of my experience using air coolers especially the heat pipe flow through designs, they all have their own peculiarities some work best with a certain airflow and increasing it doesn't change anything and some seriously benefit with increased airflow, the same thing happens with water cooling radiators.

Some cool better with the fan mounted on the backside with an old fan used as a shroud to allow full airflow through the cooling fins unobstructed by the fan motor body, they all have one solid need, unobstructed airflow.

If you are using air to cool, then the air needs an entrance and exit, so it can effectively remove the heat it collects, if you block either, the cooling efficiency seriously suffers.

We've already had users here complaining about their heat sinks not cooling properly and finally discovering they were accidentally testing the OPs theory because they had mounted the fans wrong, corrected the problem got the fans working together and the proper cooling, the heat sink was designed to do was attained.

Case air flow is very important for maximum cooling, to attain the best cooling your internal airflow needs to be as turbulent free as possible, so you don't end up with pockets of trapped air caused by fans working against the airflow, when it's all working together to exhaust the heat, the entire computer runs cooler.

From a clearly common sense point of view the concept of push/push or pull/pull with heat sinks not designed to take advantage of that principle, is a waste of time even testing, but to be fair to the OP, why don't you do the testing you're asking everyone else to do, and post your findings here, if your theory pans out I'll follow up and test it myself!

 

[rubix_1011]

I agree- if you are going to use 2 fans, you really need to do a push/pull setup...otherwise you are simply crippling the performance of using 2 fans to begin with.

There are a couple of things here I would like to emphasize-

why don't you do the testing you're asking everyone else to do, and post your findings here

Agree.

I realize this is a forum and meant for discussion, but let's try to keep all the 'what if's' to a minimum...try and see...give us your results. (I wish I could get people that keep starting the 'I want to put my PC inside of a refrigerator/freezer- think of the great temps! What do you think??' threads...most of us already know the answer- it's been tested before and is posted in many stickies...otherwise, give it a shot for yourself. (another gripe...people not reading stickies...)

We've already had users here complaining about their heat sinks not cooling properly and finally discovering they were accidentally testing the OPs theory because they had mounted the fans wrong, corrected the problem got the fans working together and the proper cooling, the heat sink was designed to do was attained.

Agree.

Incorrect or questionable information being posted where users might stumble across it and simply try it out. I am a firm believer that you are responsible for anything you decide to do with your rig, but let's try to minimize questionable information as much as possible until there are some firm results to either support or oppose it.

 

[4Ryan6]

^ Totally Agree!

 

[alansdaman]

I plan to try it. I feel like you guys are lecturing me. If someone has tried it before, why would I not want to benefit from their experience.

What misinformation am I spreading? If by misinformation, you mean things written in textbooks, then you got me good.

To 4Ryan6's points, the crappy airflow in the case that would result is something that worries me. I was thinking if I tried this I would use shrouds to route the hot exhaust. And it is a good point that the coolers are designed to operate in a certain way and one would like to think that way was most optimal. But my counter to that is that these are mass produced items, and looking at the metal on the 212 +, its hard to imagine that they really did anything herculian to make flow direction as important (such as vanes, dimples, channels etc) although I will be able to examine it more closely when I pick it up from the cleaners tomorrow morning. Features like that are expensive to produce and I doubt I will get them with a 20 dollar cooler. In that same vein, I am sure most coolers will have their own peculiarities, so this might be a solution for some and awful for others, or it might be something that works at low rpm but is terrible at high rpm. That's why I thought I would share the device I plan to use. The face of the cooler has an area of 18444 sq mm, and the sides have an area of 8109. So less than half. It should be interesting.


I plan to test the crap out of my theory at some point. I am ordering a PSU today, and I have my coolers and paste already. I have a board picked out (but I'm waiting to order from Newegg until I see if an ebay auction pans out). And the first chance I get near a MicroCenter I'll get my processor.

I am really anxious to see how well I can manage temps with low fan speeds. Does anyone know of or have any experience with fan speed control apps other than the bios management or the automatic fan speed control?

My experience with VFD software is on a much greater scale (10000 sq ft data halls) but I think dead spots wouldn't be a huge issue as the turbulence inside the cooler will create little mini cyclones. Although at the same time, those types of effects are undesirable, so I am really torn with what I think will happen. I have definitely been surprised before by the degree to which funky airflow (and how much air temperature differentials can move) can positively and negatively impact cooler effectiveness (but as a rule, effective removal of hot air from the environment ALWAYS kicks butt).

Also, does anyone have experience undervolting with Asus P8P67 boards?

 

[4Ryan6]

I plan to test the crap out of my theory at some point.

Great when you have some actual facts PM me with what you have and exactly how you arrived at the results and listing the hardware you used for testing, after I duplicate your testing if your claims are legitimate, I'll reopen this thread for you to post your results.