Why is pull-pull better than push-pull?

[cpl1234]

Been reading around and a lot of people cofirmed that pull-pull config is better than a push pull one.

So I've done this:



However, I have no idea why it works, can't seem to find any information about it either.

You see I don't like doing things without knowing how it works so can someone explain it to me please?

[alexwai]

Originally Posted by cpl1234

Been reading around and a lot of people cofirmed that pull-pull config is better than a push pull one.

So I've done this:



However, I have no idea why it works, can't seem to find any information about it either.

You see I don't like doing things without knowing how it works so can someone explain it to me please?

What I heard is completely different from the point you mentioned.

BTW. Do I know you?

[cpl1234]

Originally Posted by alexwai

What I heard is completely different from the point you mentioned.

BTW. Do I know you?

Yes, you do know me

A few people confrimed pull-pull is better but they didn't explain.

From all the threads I read/post, I gathered it's something to do with fans stalling each other in push-pull, better pressure with pull-pull and something to do with dead spot problem when pushing air through the radiator.

[Varsis]

A few people confrimed pull-pull is better but they didn't explain.(obviously because they got no proof to back it up)

From all the threads I read/post, I gathered it's something to do with fans stalling each other in push-pull,(how would 2 fans blowing the same way stall eachother?, think about that for a minute) better pressure with pull-pull (possible but doubtful) and something to do with dead spot problem when pushing air through the radiator.(dead spots do occur, hence shrouds. push or pull they happen, a push and pull setup helps much more with thick heater cores or getting higher airflow throw rads like bix etc.
pull pull in my opinion are 2 fans trying to do the same thing with one spinnen its wheels for no reason

[cpl1234]

Originally Posted by Varsis

A few people confrimed pull-pull is better but they didn't explain.(obviously because they got no proof to back it up)

From all the threads I read/post, I gathered it's something to do with fans stalling each other in push-pull,(how would 2 fans blowing the same way stall eachother?, think about that for a minute) better pressure with pull-pull (possible but doubtful) and something to do with dead spot problem when pushing air through the radiator.(dead spots do occur, hence shrouds. push or pull they happen, a push and pull setup helps much more with thick heater cores or getting higher airflow throw rads like bix etc.
pull pull in my opinion are 2 fans trying to do the same thing with one spinnen its wheels for no reason

Don't know about the explaination... that's why I'm asking...

They all said better temperature with pull-pull but all unsure why it works

[msv]

Try moving one or two of Your fans to the pushing side of the rad (lots of work just for checking out, but if You´re *really* curious nothing beats it) and check out how it affects Your CPU temp.
Add a shroud to the pushing fan to decrease the effect of dead spot.
Good hunting.
I could throw myself into a discussion about how more fans will help eachother overcoming the flow restriction through the rad, but since I don´t know jack shit about it I´ll spare You my BS.
regards
Mikael S.

[cpl1234]

The black fan in the middle is gutted acting as a buffer between the two fans... So I only have 2 fans

[pHaestus]

That middle piece isn't doing what you'd hope.

[AngryAlpaca]

A gutted fan is bad. It seems to be fairly restrictive. I think the idea behind pull/pull is to shroud them so they are in parallel, not series, but I could be wrong.

[Meethoss]

Try push/pull, push/push and pull/pull and then tell us what temps (well, the differences) ya get That will be more useful!

[Fyber]

Hmmm, it makes sense.
It would explain why Delta makes dual blade fans (basically 2 fans in one)

I think it would increase pressure, I'm not sure though. Ask one of the more experienced ones on this forum. I wouldn't say I dont know what I'm talking about, just I'm not as experienced as some others.

[JoeKamel]

Originally Posted by pHaestus

That middle piece isn't doing what you'd hope.

Why is that? Not trying to start an argument, just looking for clarification, because I'm on cpl's train of thought and figured that you want to make sure that the airflow is directed, and you want a buffer of sorts between to two fans (more for low noise than extra efficiency).

[redleader]

I'd think that putting one fan on each side is better simply because the radiator is in all possible configurations, but only acts to reduce turbulence when its between the two fans. So when you put it in the middle you get a "free" buffer.

However I have nothing to back up my view.

[Anemone]

So Ph what works better? Pull pull or push pull or is it too close to much worry which you choose?

My 'guess' would be that you don't need the open item between the two fans, that putting two in a pull pull scenario increases pressure, much like multiple compressing fans in a turbojet engine, they don't need a separation between to do that job, but there would be a drop in efficiency, ie the fans won't double the cfm througput.

My additional guess is that push pull would work about as well, IF using a shroud on the input fan. But many users who do push pull don't put a shroud on the front in many of the design screenshots I've seen.

So to continue this guess (PH your input could be very useful after this), the rad, shroud, pullfan, pullfan design and the pushfan, shroud, rad, shroud, pullfan designs should come very close to each other in efficiency but the thicker the radiator, thinking that would decrease the airflow through it, the more likely the pull pull will work better because the actual pressure increase from two fans working right next to each other would be better to overcome the thick rad pressure drop than push pull. The further benefit is the need for one less shroud in the lineup decreasing size.

All guesses, but the scientists are free to confirm or refute

[bobkoure]

Originally Posted by pHaestus

That middle piece isn't doing what you'd hope.

Hmmm...
My only data point is with two undervolted (5V) panaflo L1A fans stacked pulling through a be-cooling core (6x6) in a coolingworks shroud. Adding a gutted 25mm fan as a spacer reduced noise a bit and dropped reported-by-CPU-diode temps by ~.5C (sometimes same as before sometimes 1C lower).
To the extent that fan blades are wings and use the coanda effect, you need to not obstruct the top (lifting) surfaces. It's possible that one set of blades (or more likely, one set of struts) was interfering enough that a spacer helped. Might be something else going on, too.

With all that said - what is bad about using a hollowed out fan frame as a spacer?

[krazy]

The basic explanation for a pull configuration being more effective than a push configuration is that when a fan is pushing, more air flows through certain parts of the radiator (mostly in a doughnut-shaped ring corresponding to the blade area of the fan). When the fan is pulling, it does a better job of creating an even partial vacuum, which pulls air through all areas of the heatercore much more evenly. The mroe surface area of the heatercore you take advantage of, the more heat you can transfer.

[nexxo]

Overclex.com did a comparative test on Push, Pull, and Push-Pull configurations (unfortunately not a Pull-Pull), with a BIX. They found that a Push-Pull has negligable advantage over Pull, and that Pull worked better than Push. But we're talking, like, one degree C or so difference between various setups, so I figure efforts to optimise cooling are better invested somewhere else (e.g. ditch the double fan, concentrate on getting good access to cold outside air).

[bobkoure]

Originally Posted by nexxo

... so I figure efforts to optimize cooling are better invested somewhere else ...

Well, once you've done the other things and if one of the things you're trying to optimize for is quiet/silence (below ambient noise level) then you try things like stacking fans on the pull side. You might get a better noise / static-air-pressure balance than if you simply raised voltage to the fan. It happened to work in my case - and might not for you (and IMHO isn't worth bothering with if you either don't care about quiet or your idea of quiet is 30dB or louder).
I've since ordered a couple of surplus cage fans and will checking out the same noise / static-air-pressure trade-off with these.
I suspect it'll be better than stacking axial fans (it is based on the fan/blower curves I've been looking at - but I may be overlooking something important)
Bob

[pHaestus]

Your hollowed out fan case is a venturi tube

You'd be better off to use long screws and duct tape or similar in between fans.

[bobkoure]

Hmmm... I only have two data points
I started out with a simple "wrap of tape around screws" and was concerned with vortexes coming off sharp edges and so "hollowed out" a fan frame, radiusing everything.
As measured incredibly crudely by looking at the temp shown on my CPU diode - I saw no difference. I'd expected to see a drop, but would certainly have noticed a rise.
And... I'd also been concerned about the inside lip of the coolingworks shroud that I used (inner hole is 105mm or so - standard 120mm fan blade swept area diameter). The panaflos I used faired out in an oblong way to 130mm across the corners (less across the flats, of course - no "meat" there). So I carved the shroud to match. Again, no difference in CPU diode temps.
I suspect that the fact that I'm using these fans way undervolted makes a lot of this not matter - or my measurement technique is so crude that I'm not seeing the difference that's actually there. (The diode temp on this CPU does go up and down in lockstep with my room temp so I think I'd see a delta - but maybe not - hardly near your standards...)
Anyway, the next time I take this setup out, a squirrel cage is going in (whole new set of issues to learn about)

[JoeKamel]

Originally Posted by pHaestus

Your hollowed out fan case is a venturi tube

You'd be better off to use long screws and duct tape or similar in between fans.

Why is the hollowed out fan a restriction though? From the description of a Venture Tube, a VT is a place in the path of the flow of a liquid where there is a constriction or other type of decrease in diameter of available flow. A hollowed out fan, isn't going to provide anything in tha respect as it is going to have the same diameter opening (80mm to another 80mm, 120mm to another 120 mm) with nothing in its path because the an itself has been hollowed out leaving nothieng but a hollow ring.

[cpl1234]

Originally Posted by JoeKamel

Why is the hollowed out fan a restriction though? From the description of a Venture Tube, a VT is a place in the path of the flow of a liquid where there is a constriction or other type of decrease in diameter of available flow. A hollowed out fan, isn't going to provide anything in tha respect as it is going to have the same diameter opening (80mm to another 80mm, 120mm to another 120 mm) with nothing in its path because the an itself has been hollowed out leaving nothieng but a hollow ring.

That's exactly what I thought originally.

I also came across this :

There's not a simple answer and we do not have 'one size fits all' situation here I am affraid.

Basically it all depends on your hardware d00de. If you happen to be a happy owner of high performance small and very restrictive radiator (some car heater cores) your best bet is going to be push/pull config. It holds for low pressure fans (25mm generally, oh I am talking 120m fans here since smaller are jst waste of time in performance cooling)
with high pressure fans (35mm thick and up) you may get better results with all pull config. All the above assumes well designed and sealed shroud in action.
The theory behind it is the same as with pumps in WC loop.

Right, lets put things in order here, shall we?

1. Case One - restrictive rad and low pressure fans.

What we need here is more pressure to be able to push air through in respectale volumes. there are two ways of doing so, you either stack fans and have pull-only/push-only situation or you go for a fan on each side (push-pull). I found through experiments (aoerodynamics is too complicated to entertain here and I cannot say I have it mastered sufficiently to write about it) that stacking two fans is optimal and stacking four fans produces no air flow at all . Stacking two fans gives better results than push pull config due to almost double pressure produced by such an assy. Another possible config is mounting two 25mm fans in pull only config side by side with tapering shroud. With restrictive rads it does not work since all you gonna get is air blown out instead of being sucked (push config) and not much coming out in pull config. I works wonders with low restriction rad since CFMs almost double but ressure stays the same (with low-res rads we do not need pressure)
Let's proceede to next case

2. Case Two - Restrictive rad and high pressure fans (35+mm thick fans rated 130CFM and more)

The best config here is for two fans side by side in pull config mounted on a tapering shroud. We do not need increase in pressure, we already got it but more CFM would be nice so we mount fans in parallel. Pull config and shroud ensures no dead spots as gav2000uk very righly observed.
Reasoning behind is that with high pressure fans can push close to rated max CFMs (which number describes CFM with 0PSI back pressure and conversly max back pressure is obtained with 0CFM - same as with pumps and water) so increase in presure wouldn't yield almost any increase in performance (stacking/push-pll config)

3. Case Three - Low restriction Rad and low pressure fans

What is worth remebering is that the less restrictive the rad is the poorer performer it going to be assuming no limits on fan power. In other words low-res rads have a point of diminishing returns where increase air flow does not produce any more cooling. With PC water cooling such conditions are not attainable .
This case offers very quiet radiator subsystem (low pressure fans are, and shoudl be low noise designs) but needs to be of decent size to offer needed performance (at least 240x240 if not 240x360mm of finned area)

4. Case Four - Low restriction Rad and high pressure fans
same as case four but fans are usually ran at 7volts to limit noise levels but are able to be set to fuul steam for benching/gaming periods


The choice of a system config. is dependant on ones brief. Performance?, low noise? size?, for each combination of requirements there is the best combination of hardware. There isn't anything like The Best config for everybody

I hope it helps, if I slipped anywhere feel free to correct me and if anything is mud-clear ask

As comestic is quite a important part of my system so I think I'll pass with the carboard and duct tape idea.... However, I might go as far as cutting aluminum and a bit of silicone sealant.

I'm the pround owner of 2X 130CFM Y-S tech fan so I think a tapering shroud is what I should be making in my next upgrade.

[bobkoure]

I'd love so see some radiator air pressure resistance curves.
The only online one I've been able to find is from BillA. Radiator Heat Dissipation Testing (pg4) - scroll down to the chart labeled "Static Backpressure vs. Air Flow". This is now very outdated in the sense that there are only two radiators listed that you can still get:
- The "big momma" radiator is a 6x6x2 heater core (with restrictive barbs, but we don't care about that for air resistance).
- I think the "original" Black Ice is the same as the current Black Ice Pro (only the BIP now comes with 1/2" beaded inlet/outlet).

I'd love to see more info about other radiators/cores if someone had the equipment and felt so motivated (read: "doesn't have a life" )

Did cpl1234 have some data elsewhere? IMHO, the comments are all common-sensical, but I'd love to see which radiators/fans he's basing that on.

Oh - then there's the whole issue that axial fans, thick bodied or not, are really not appropriate for pushing/pulling air through a heater core - unless maybe those $100 for-Toyota cores that are 1" thick or so have dramatically less air resistance. I'm guessing that they're aluminum anyway (aluminum block engines and radiators, why introduce copper/brass even if you do have anti-corrosion additives in your antifreeze?). The engines probably do have bronze in 'em - but probably not where water touches.

[nexxo]

Well, once you've done the other things and if one of the things you're trying to optimize for is quiet/silence (below ambient noise level) then you try things like stacking fans on the pull side. You might get a better noise / static-air-pressure balance than if you simply raised voltage to the fan.

True... true... In terms of combatting noise, it could be a worthwhile additional investment. In any case, if it works, it works!

[HAL-9000]

Its pretty basic why pull pull works better though I hadn't thought about it 'till I started reading this thread.

Basically, so much air has so much cooling capacity. More air the better, and we fool ourselves into thinking more CFM=more air. But really a 50CFM fan at sea level will beat a 60CFM fan in Mexico City, because there's just less air per cubic foot up there around 7000 feet.

The pull-pull config probably moves a smaller volume of air (no data on that, just a thought experiment) than the push-pull, but at a higher presser per CF. Basically as far as cooling goes, the more air molecules blowing over my rad the better. If I have a higher pressure, that means I have more of those molecules in a given volume, so my radiative system can do more work. Each turn of the second stage fan gets to scoop more molecules of air than if it was just sitting there open-faced.

Its kinda like a supercharger intercooler, makes the air denser before it goes into it, hence there is more oxygen to burn in the fixed volume of the engine.