Dead zone impact on radiators

[extremis]

Which is more effective for cooling on a radiator?

- Yate Loon 88CFM (120x25)
- Panaflo 115CFM (120x38)

The Panaflo has a larger dead zone than the Yate Loon. However, the Panaflo has a higher CFM and static pressure than the Yate Loon. Is this tradeoff in CFM/static pressure versus dead zone worth it?

[KaptCrunch]

panaflo, but dead zone only counts on free blowing

[extremis]

Shrouds are made to deal with the dead zone caused by all fans. It would make sense that a radiator without a shroud would do better with less dead zone.

Why do you say it only matters with free blowing?

[Vapor]

If you look at the flow of water through a radiator, every part it flows through gets ample airflow, even with a deadspot.

I'm not convinced a dead spot itself limits cooling much at all, to be frank.

I'd bet shrouds work () more because:

1) of an actual increase in net airflow.
2) slightly evens out air distribution (horizontally--AWAY from center).


Here's some (theoretical) math:
Let's say a hub (and therefore deadspot) is 40mm in diameter and the actual airflow is 120mm across. If you have an even donut-pattern of airflow of 100CFM, I'd bet the majority of airflow is actually going over the center channels, probably ~70-75% of airflow will go over the middle 50% of the radiator. The outer quarters will thus only get a small portion of airflow (and therefore cooling). This is because top and bottom of the donut will go on the center channels, while the outer edges of the radiator will only get a part of the airflow each. What makes this even more troublesome is that airflow is not an even donut-pattern, it weakens at the edges (with most fans), thus giving even less cooling to the edges.

I can't find my calculator, and I sure don't want to the integration by hand, so this is all theoretical, but just imagining it in your head should help. (maybe I'll draw one and have Photoshop do the math for me )

[Cptn Vortex]

Depends on the RAD too. Some, like the Thermochill, have a greater distance from Fan to actual cooling fins, which is kinda a shroud in a way. If it was sitting right up against the cooling fins, it would be much less effective.

[Vapor]

According to Photoshop, 55.75% of the red pixels (of the donut) are in the center half.


Will try to feather (to simulate further distance from the rad) and remeasure....if that doesn't work, I guess I'll have to dither or something :-/

[extremis]

I have the X-Changer 240 and 360. The distance from fan to fin is about the same as the Thermochills. I was going to put the Panaflos on the 240 and the Yate Loons on the 360.

My other thread was asking the question of which blocks should go on which radiator.

I have 2 Panaflo fans and 4 yate loons. It makes sense for me to put the Panaflos on the 240 given my current limitations. However, I could always pick up a few more Yate Loons.

Vapor, I am very interested in learning more about the airflow physics you are trying to explain. I'm willing to bet that others are as well.

[Vapor]

And this is probably more what airflow looks like without a shroud (but with some minor spacing, like a PA or MCR)....unfortunately PSCS3 doesn't want to measure the density of red for me





EDIT: anyway, I think regardless the shroud's main advantages aren't from spreading the air out (though that probably helps ever-so-slightly)...but rather from an actual increase in airflow (fans don't like have something RIGHT against them) and from a decrease in turbulence noise.

[Ashraf]

So a shroud is worth getting? Or it depends on what fans I am using?

[Vapor]

Okay, used dissolve blending on the red portion above, then measured the red

(yes, the original image I've been working with has been 800x800 the whole time, I've resized to 500x500 for bandwidth/viewing sake...but resizing with a dissolve does not work so well )

52.8% of the red is in the center half, so airflow is basically even (this is without a shroud, just the small spacing already provided). Also note the corners probably don't actually have 0 airflow.



(And of course note this is not an exact science, at all, but just a series of approximations, but considering how near-50% they're landing, it's saying something, IMO)

[Vapor]

Okay....more words from me:

I think the reason shrouds worked so well in the past was because the high-end watercooling system involved using a heatercore. Heatercores are a) thick and densely finned and b) frameless. Fans had to be right up against the dense heatercore and airflow probably suffered a lot. Shrouds made sense...get the fan away from the heatercore and give it some room to 'work'. The deadspot probably wasn't the issue the whole time, though the shroud did fully even out the airflow (though it probably was already near-even in terms of left-right distribution [because waterflow is going in one dimension of two, airflow only needs to be distributed in the other dimension--horizontally]).

Newer rads have a frame providing a few millimeters of spacing, which is pretty important (it's especially better than NOTHING, and especially in pull I'd imagine), and they also have significantly lower fin density (and some are thinner), so actually moving the air isn't as hard either. Shrouds do still 'eliminate' the deadspot, but practically, I'm not sure how important that is compared to the other things it can do.

[extremis]

Thank you.

[billb]

Originally Posted by Vapor

Okay....more words from me:

I think the reason shrouds worked so well

Isn't this more like how your "donuts" work out with a shroud?

[Giannis86]

check this thread out
http://www.overclock.net/water-cooli...t-shrouds.html

in that testing using a shroud improved temps by 2-3 degrees..

i will be doing some testing myself with 6 scythe ultra kazes at different speeds, on push/pull configs, with/without shroud, and on 2 rads (120.3 and GTX360).

only problem is that i have exams coming up at uni

[C'DaleRider]

Originally Posted by Giannis86

check this thread out
http://www.overclock.net/water-cooli...t-shrouds.html

in that testing using a shroud improved temps by 2-3 degrees..

But realize that testing you linked is done on one of the most densely finned radiators out there......the GTX series.

The TC PA series, on the other hand, probably would exhibit nowhere near the effect of having a shroud over not having one. As is mentioned on TC's website:

....on the PA series... shrouded vs unshrouded exhibits minimal benefits, in the region of up to 2%.

So, up to a 2% difference with a shroud on a PA series radiator (and I tend to trust TC's own testing over most other testing out there....). That would mean at a full load temp of like 50C, a max 2% difference (notice TC said up to 2%, meaning you may not even see a whole 2% difference) would mean that there would be a 1C difference at best, less more likely.

I'm not surprised the GTX exhibited a larger difference given the fin density on their rads.......

[Giannis86]

i realised that but i am more than willing to give some of my time to test those..i already have the PA 120.3..now waiting to get the GTX from petras, along with some other blocks..it could also be that the 2% relates to the fan used as well.

anyway when the time comes, let the numbers do the testing

[Anemone]

There is a second contributor which you touched on, the fin density. Placing fans close to any resistance, rad or filter, lowers its CFM ability due to resistance. Placing some distance between the resistance and the fan itself lowers that effect. The most simple way to test this is to use the "old school" shroud of an empty fan housing and an active fan and then a filter over the far end and pulling air through. If you alternate putting the filter directly on the active fan vs putting it on the far end of the shroud you'll get a noticeable (wet finger test) better airflow with the shroud. ProCooling, back in the day, did some testing on this topic of shrouds and found that it could give a 1-2 degree improvement. Mind you, the most common setup was a simple pull setup. Around here we have a far greater percentage of push/pull configs and with two fans working in concert, I'm going to hedge a bet that you would find far less improvement by adding shrouds to the mix of a push/pull setup.

We should do a poll on what % of rads are around here, but from the water cooling rig pics, some % have black ice in rigs due to the smaller size, so it's not all PA's, but PA's do likely cover a huge % of XS.

$.02

[iboomalot]

might consider this

at 2000 rpm the tip of the 120mm fan is going to equal 112mm*3.14159(pi)*2000rpm = 703716.1mm/25.4/12 =2308 ft traveled

at 2000 rpm the near middle of the 120mm fan is going to equal 60mm*3.14159(pi)*2000rpm = 376990.8mm/25.4/12 =1236 ft traveled

Also the the size of the fin on the fan is smaller too.

Thus the dead spot/smaller hub on the YL will make very little difference due to location if the fins on the YL were larger on the outside then YES a big difference could be seen

[extremis]

Originally Posted by billb

Isn't this more like how your "donuts" work out with a shroud?

Does push impact the donut differently than pull?

To clarify, I will be using the traditional pull design.

[septim]

pull fan config, with the help of a few mm away from the fins (with gasket, no air leaks) = minimum dead spot or no dead spot at all, because air will be pulled from every where, even the edges or the center so long as is can get through the fins...

[Snyxxx]

I think this picture from eva2000 best shows the fan dead spot on a rad.

http://i4memory.com/reviewimages/coo...adBox_025.html

[gojirasan]

I am currently trying to fabricate a shroud for my PA120.3. Another issue that may or may not matter is noise. Is there a drop in noise the further you move the fan away from the fins? I have noticed that putting my hand on either side of my San Ace fans increases the noise. I am wondering if a shroud would allow a higher fan RPM for a given level of fan noise. Even an increase of 200 RPM might make a shroud worthwhile. The idea that a shroud might not make a difference seems counter-intuitive to me.

[F9292]

Too scientific for me. Basically when it comes down to it, The fan with the bigger dead spot is putting through a higher amount of air through your rad, dead spot or not! So it will be cooling the non-dead spot area of the rad a shat load better than the lower cfm fan! A higher cfm fan is always going to win, no matter how big the dead spot is! Just my 2c

[septim]

snyxx good picture, basically dust buildup on the outer non dead spot area, but after a while that area in his dead spot would probably also fill up, there is already some build up there in his picture...

[billb]

Originally Posted by extremis

Does push impact the donut differently than pull?

To clarify, I will be using the traditional pull design.

With a pull setup, the dead spot doesn't extend as far from the fan as with a push configuration.
The shroud with a pull configuration can be thinner, but why mess around with that? Anything over an inch, push or pull, should be fine.