Push vs. Pull Tested

[amuffin]

*NOTE* THIS IS A SCIENCE PROJECT I NEED YOU GUYS TO HELP ME EVALUATE IT!! (Please post the scores that you think I deserve with this Rubric: https://docs.google.com/viewer?a=v&pid=sites&srcid=Y3VzZHNjaG9vbHMub3JnfGNtc19zY2llbmNlX3Jvb21fOHxneDo2NmU4YTJjZjg1YmI1ZmQ2 )
(Currently transferring project, photos etc. over to forum post from google docs.)

PC Cooling: Push vs. Pull, Which Performs Better?

Abstract
I tested the performance differences between Push (Fan is pushing air through the heatsink) and Pull (Fan is pulling air through the heatsink). In order to get temperature results I ran stress tests 3 times and got the averages of each fan setup. I found out a fan pushing air through a heatsink will provide better cooling, than a fan pulling air. Not only can this be the deciding factor between heatsink A or heatsink B, this can be very helpful to first time users wondering where they should mount their fans.

Introduction
In every modern day computer, there is a fan and heatsink to take heat away from the processor. A heatsink is designed to absorb the heat generated from the processor and a fan is used to dissipate the absorbed heat from the heatsink by pulling/pushing the heat away. Without a heatsink, processors would burn out by overheating.

How a Heatsink Works

1) Processor generates heat.
2) Heatsink absorbs generated heat via heatpipes/plating.
3) Heat is brought up to heatsink fins (hot air rises).
4) Fan brings cool air across the heated fins, causing them to become cooler.
5) Absorbed heat in fins are dissipated.
6) Cool air molecules absorb the heated air molecules.
7) Heated air molecules are removed from the case via exhaust.

Materials and Testing Methodology
The system was ran in both overclocked and stock configurations for minimum and maximum temperature readings. Ambient temperature* in the room was a cool 20.5C which did not change between each fan configuration. The processor was overclocked from a stock of 3.3ghz to a speedy 4.5ghz and the voltage was raised from 1.1v to 1.34v. The fan used in this experiment was an Aerocool Shark 120mm Fan, which pulls around 85.5CFM (Cubic Feet Per Minute). The Shark was ran at a max speed of 1600RPM (Revolutions Per Minute) for consistent results and maximum performance.The fan was then mounted in both Push and Pull configurations on the heatsink and temperatures were taken with CPUID’s Hardware Monitor.

To stress the processor and take temperatures, I used the following programs:
Stress Testing**
10 Minute Prime95 Run (Used to check for a stable overclock by stressing processor)
http://www.mersenne.org/freesoft/
10 Passes of Intel Burn In Test (Maximum stress testing for all cores)
http://downloads.guru3d.com/IntelBurnTest-v2.3-download-2047.html
*NOTE* None of these will simulate real world temperatures, these are just used to indicate which fan configuration works BEST for YOU.
Temperature Reading
Hardware Monitor
http://www.cpuid.com/softwares/hwmonitor.html
*Ambient temperature is the “temperature of your surroundings” essentially, room temperature
**In order to generate consistent results, each test was ran 3 times and the average temperature was taken.

Test Hardware
CPU: Intel Core i7-2600k @4.5ghz 1.34v
Motherboard: Asrock Z68 Extreme 4 Gen 3
RAM: 16GB Patriot Sector 7 DDR3
Heatsink: Cooler Master Hyper 212 EVO/Aerocool Shark 120mm
Thermal Paste: Arctic Silver 5
Case: CM 690 II Advanced
Power Supply: Antec HCG 620W Power Supply
Graphics Card: Nvidia Geforce GTX 550ti

Aerocool is a reputable company when it comes to fans, their shark series of fans performs very well and is able to generate consistent results when mounted on the Cooler Master Hyper 212 EVO.

Aerocool Shark 120mm Fan Compared to Other Fans

Fan mounted on heatsink:

Thermal Paste applied:

Cooler Installed:

Fan Differentiations
Push

Pull

Notice how the fan on the heatsink in the “Pull” photo is flipped? This is what denotes the difference between the Push and Pull methods!




Test System Fan Setup

Fan Setup: 140mm Intake 2x 120mm Outtake 1x 140mm Outtake 2x 120mm Intake

Results
*these for some reason don't work, so click on them for results*
http://img526.imageshack.us/img526/8673/intelburnin.jpg
http://img443.imageshack.us/img443/5541/prime95c.jpg



Performance Differences:
Push offers 2C-4C difference in performance when compared to Pull. This is on average a 2.8-5.6% difference between the two. Even though the temperature difference doesn’t seem like a lot, it can be crucial to users who want the best possible temperatures.

As the data shows, when a fan pulls hot air from a heatsink, it will perform better than a heatsink pushing the air. Keep in mind these temperatures will never be reached during real world applications and uses such as gaming and common tasks like internet browsing. However, temperatures will be different between both fan setups regardless of the task being given. Remember, temperatures may be different for everybody, this is because ambient temperature is different in many parts of the world.

Discussion
As shown, temperatures between Push and Pull methods do not vary much! As said before, these do not simulate real world performance. They only show which fan configuration is best and works for you. You may ask yourself, why would someone want to use the Pull method when the Push method works better? Well, there are some cases where RAM Modules get in the way of the fan and block the fan from being installed. That is the only reason why one would use the Pull method instead of the Push method.


Acknowledgments
I would like to thank 4Ryan6 of Tomshardware.com for allowing me to use his CFM and airflow results of the Aerocool Shark 120mm fan. I would also like to thank Tomshardware.com and Cooler Master for the photos on airflow. The original Tomshardware.com article on PC Aircooling: http://www.tomshardware.com/reviews/cooling-airflow-heatsink,3053.html Without Tomshardware.com Forums, I would not have been able to come up with the idea for this project! I would also like to thank FinneousPJ and many other people for helping me with ideas and evaluating this project.

 

[kajabla]

hooray muffin
Your photos don't work, though.

 

[amuffin]

Because I haven't added them

 

[Motopsychojdn]

Cool project man, however this series of experiments only shows that xx fan performed better in xx setup on xx heatsink
for a truer indication of the old 'which is best' you would have to repeat the entire set with different fans and heatsinks,
Seems like a perfect opportunity to grab some freebies from Companies in the name of 'research'
Hehe, I like it man and look forward to your further adventures
Moto

 

[amuffin]

I don't know how to grab freebies though, plus I don't think any companies would send fans to a kid like me

 

[Motopsychojdn]

Lay out your brief, include some data like you already have, and Email it to fan/heatsink companies saying you intend to evaluate say 20 popular fans on the top 5/10 heatsinks and would they be interested in participating?
at least a couple of guys on here get stuff for reviews and evaluation purposes,
as well as the modders who get sponsor offers,
age is irrelevant, if you present the plan in a well thought out fashion and show your professional intentions, you may be surprised at the replies you get
Moto

 

[kajabla]

Yep. Watch your formatting. Making a professional first impression is important. I'd be happy to check your initial letter over if you like.

 

[horaciopz]

I want pictures ! pretty well done experiment I was wondering how "big" will be the difference between push/pull fans configurations...

I have a Hyper212 with its stock cooler in push configuration, I wanted to try how good or bad would be if I install a small and fast fan pulling air from the front (just under the DVD drive) blowing air through rams and directly to the heatsink, and move the Stock Fan in the pull configuration taken the air just to the back... Still I haven't enough time, I guess that I will be doing that this weekend !

Keep it up, it would be great if you get some kind of help and check fans/heatsink configuration for our knowing ;D !

 

[FinneousPJ]

Are you looking for comments before you present this at school or did you just want to post the results?

 

[HEXiT]

you will get even better temps if you use pull but block the sides of the heatsync so your pulling air all the way through the fins rather than getting high pressure at the edges and low in the middle. at least in theory it should be more efficient.

 

[spentshells]

I want pictures ! pretty well done experiment I was wondering how "big" will be the difference between push/pull fans configurations...

I have a Hyper212 with its stock cooler in push configuration, I wanted to try how good or bad would be if I install a small and fast fan pulling air from the front (just under the DVD drive) blowing air through rams and directly to the heatsink, and move the Stock Fan in the pull configuration taken the air just to the back... Still I haven't enough time, I guess that I will be doing that this weekend !

Keep it up, it would be great if you get some kind of help and check fans/heatsink configuration for our knowing ;D !

Push/Pull did very little for me but 1-1.5 C.....I have a 120 exhaust 4" from the back of the Heatsink

Waiting for pics

 

[amuffin]

I want some comments before I give my teacher the project on monday! If you guys can evaluate me too.....that would be awesome! I'll fix up the photos soon ok?

 

[13thmonkey]

Look forward to seeing the pics and graphs.

Am curious about hexits thoughts on blocking the sides, i tried this with push and blocked sides were warmer, but I was surprised by the result. I also wanted to test a plenum chamber between the fan and the fins to get rid of the dead zone in the middle.

Definitely ask for samples, it might be the start of something interesting (I imagine it's how toms originally started back in the day), at the very least it'll be interesting and a differentiator when it comes to CV time.

 

[yumri]

ok in your starting you forgot to include if that was 1 16GB DDR3 stick or 2x8GB or 4x4GB to get a total of 16GB of DDR3 you also forget the speeds in MHz and CAS latency as these also factor into it
you forget to include what kind of cooling that was used on the Graphcis Card which i suggest against as using the intergrated graphics processor will give off less heat inside of the case then any cooler on any decaited graphics Card
links to all the parts should be included and sited in your list of sources at the bottom of your report
the liquid colling system and the liquid used was not listed and you should also list its the chemical makeup of it and compare how it changes if at all when the temperatures raise and fall
i did nto see a spot in which the results were posted if they are posted make it more clear where they are
the Pull seciton was blank please fix that before turning it in
Like the people above adding in pictures will add interest into your project

 

[amuffin]

RAM Sticks, Speeds don't matter.
Cooling on the graphics card does not matter.

As for photos and actual editing, I am working on that now...

 

[amuffin]

Lay out your brief, include some data like you already have, and Email it to fan/heatsink companies saying you intend to evaluate say 20 popular fans on the top 5/10 heatsinks and would they be interested in participating?
at least a couple of guys on here get stuff for reviews and evaluation purposes,
as well as the modders who get sponsor offers,
age is irrelevant, if you present the plan in a well thought out fashion and show your professional intentions, you may be surprised at the replies you get
Moto

Going to email some companies. Which ones should I email? I need a suggestion list...BTW can some of you guys help me write one?

List:
GELID
Noctua
Phobya
Cooler Master
Antec
Scythe

BTW Where do I find each companies email

 

[FinneousPJ]

This is a high school project right? I think you should get a B.

Some things you could add/change:

A bit more in-depth look into a heatsink. How does it actually work, why is there a fan. To really impress your teacher, look up some basic concepts of thermal physics and how they relate to this engineering problem.

It would have been cool if instead of revealing your results in the first chapter you had some discussion as to what you think is going to happen and why. Then you present the results, and again it would be great if you had some analysis as to why and what is going on.

Use less exclamation!!!

With these changes it would pass a university course I'm sure, good job anyway.

 

[13thmonkey]

Your pull image has the heatsink rotated, but the fan is still pushing through the heasink, you need to turn the fan round on the image too.

 

[FinneousPJ]

Oh yeah, what you claim is Pull is actually Push - and what you claim is Push is pushing in the wrong direction. Flip the fan to achieve Pull.

 

[amuffin]

Are you sure? I thought Pull was the fan in the front of the heatsink pulling air (intake) But push was on the back? :/

 

[amuffin]

This is a high school project right? I think you should get a B.

Some things you could add/change:

A bit more in-depth look into a heatsink. How does it actually work, why is there a fan. To really impress your teacher, look up some basic concepts of thermal physics and how they relate to this engineering problem.

It would have been cool if instead of revealing your results in the first chapter you had some discussion as to what you think is going to happen and why. Then you present the results, and again it would be great if you had some analysis as to why and what is going on.

Use less exclamation!!!

With these changes it would pass a university course I'm sure, good job anyway.

Middle school project! Alright, I need to add those things.

 

[DJDeCiBeL]

Are you sure? I thought Pull was the fan in the front of the heatsink pulling air (intake) But push was on the back? :/

You got confused there. Yeah, the fan is taking in cool air, but it's also "pushing" cool air through the fins. Pull would be the other way, pulling cool air through the fins.

 

[amuffin]

Changing stuff now...

 

[13thmonkey]

the pull is definitely wrong, (I can't tell if you are still editing). The fan direction on the pull model should from the heat sink to the case. On push the fan should be from the case to the heat sink.

Also your model of 'how a heat sink works' is missing some key elements, the 'fan dissipated generated heat' for instance is not really telling the story, 'fan generates air flow', 'heat is transferred from fins to air', the heat rises element is not important, as thats only airflow, in fact in 90% of PC cases it rises away from the heatsink. Perhaps putting the transfer mechanism on each element would be useful, convection, conduction etc.

 

[amuffin]

did some moar editing.