OC on Fan Stacking
 

http://www.overclockers.com/articles1155/

One thing i dont quite understand is exaclty what joe means by fan stacking? does he mean 2 fans on top of each other? or push pull? or somethign else?

slater..

.....
.....

Joe seems to have a completely different opinion on this from my own. My opinion is that the performance difference is typically minimal and the noise increase is HUGE. I would never do this on my systems.

:) i emailed

talking about push pull
 

He's talking about push pull

got me baffled
here http://www.overclockers.com/articles750/ we can see Joe's wind tunnel
I cannot figure out how Joe is assessing the DUT flow rate with its own fans ?
(please no Laing second air pump to neutralize . . . )

someone more observant than I ?

now this could be done if the dissipation were measured for both and the DUT fan flow rate back calculated,
but Joe did not do this

According to the manual , the aux blower (yes, Laing second air pump :D ) is adjusted so that the static pressure is zero. This gives the DUT operating point.

...was that the question?

He Mounted the radiator to the chamber with one fan on the outside(blowing) and one on the inside(sucking) with the air drawn through the chamber(opposite of the diagram he has posted) the chamber can operate in either flow direction. His measurments air the actual recorded flow values.

I guess that leaves me to put 2 fans on and measure the actual change in dissipation

this is how I would prefer that Joe present the data (and test accordingly)

http://www.thermal-management-testin...20by%20fan.gif

the system C/W predictors are gonna cry over these curves
(what a difference an environmental chamber makes, and bumping the air temp resolution by using RTDs)

hmm... those curves look odd: is there really a drop in heat dissipated over 1.5gpm??

A Swiftech kit for the first to guess the identity of the rad, and the reason why the disp. drops off with increasing flow rates...
hehehe :)

Interesting that the sweet point appears to move depending on the fannage!?

Out of curiosity, tests done with or without shroud?

Edit: Just an idea.

http://img19.exs.cx/img19/4117/f7toriginalbihydpwr.gif

:shrug:

Not a stupid one.
I think this potential energy is converted to heat and lowers the Coolant dT (used to calculate Heat Dissipated).
Think should be added to FlowxCpxdT to give the Heat Dissipated.

BTW Your sums appear to be correct.

Edit: Changed "kinetic energy" to the correct(I hope) "potential energy"

this is going to be a hijack of JoeC's thread

here or a new thread ?

I don't know Bill. Seems as if this is as good as any way to use this thread. We already answered the original question and have moved on to more interesting matters. :D

Interesting graph, Bill, thanks. I wouldn't have expected the friction->heat contribution to have become measureable even by 2gpm. It's either that or measurement error.

I would like to suspect measurement error, but I'm running out of ideas as to the 'source' of the discrepancy - if any ??

a brief outline of the methodology:
tested in an environmental chamber using RTDs, recirculating chiller, and mag flow meter; all cal'ed as appropriate
rads mounted horizontally w/baffle plate to reduce re-circulation
cycle run from max flow to min, readings at equilibrium
pressure and temps for coolant at insulated crosses at 2" from connections, lines insulated from before cross to DUT

my question: what could be causing a reduction in the temp measured at the rad output ?
- yes, frictional heating is a candidate

but I'm not too sure that there is an error involved, looking at the data there is an apparent shift of the max diss to higher flow with more air flow as well
and note the difference in the decreased diss at max flow, were the airflow increased more the dissipation curve would likely go flat

-> but if there is no error here, then there was an error in the ThermoChill 'setup' - what would cause the higher flow rate data to indicate a 'greater' dissipation
(I re-ran the 120.1 and the apparent dissipation tailed off as with this data

lolito, Les
yes, there are a number of 'energy' considerations - but are they so relevant to the bench testing evaluation of a rad ?
for example the head loss is very much a function of the connector installed, which has no bearing per se on the dissipation characteristics of the rad
- am I missing something ?

In reply to the original question, but not in relation to water cooling. Many of the Shuttle range of computer users now run with push-pull fan setups on the ICE cooler (heatpipe technology). Two low noise fans like the SilenX, running at reduced rpm gives far better cooling than a single fan at higher rpm's and noise levels.
Radiators are obviously more restrictive to airflow though, so the benefits are probably a lot less, and the restriction will certainly add more noise.

Is it possible to run the tests "the other way 'round" with the liquid side cooler than the air side?
Would this shed some light and maybe prove/disprove the frictional heating possibility?

slick idea, not sure off the top of my head what - but something will be learned
sure, will do later today after this series

Water In - out +0.06 degrees?

http://w1.863.telia.com/~u86303493/C...ad_fans_os.jpg

Me and my offset errors. Deja vu.

Might fit.

Edit. Are you using RTD's for water temp or just the air? Thermocouples?

Yes, if you are measuring them and you are including any "heat gain from dP across fittings +rad"( "frictional heat" with numbers)

RTDs (4/wire, 0.01°C res) for water and air

what is the 'source' of the +0.06° ?
if this is a systemic error the correction should be applied to everything
(but the BIX curves with BIG air are pretty flat ??)

Have added frictional heating as per above:

http://img53.exs.cx/img53/5294/q3araddissbyfanfudge.gif

This is as 'nice' as the curves will look with my primitive rad P/Q model (0.13.Q^1.85 in lpm & mH20)

agree with Les,
also like Incoherents curves more than mine lol.

w/o frictional heating, eh ?

http://www.thermal-management-testin...BIX%20diss.gif

http://www.thermal-management-testin...0BIX%20C-W.gif

http://www.thermal-management-testin...0rads%20hl.gif

am rather unhappy to be (considering) adding the fitting(s) into the rad characterization
Les, lolito, Inchoerent - I'll send the spreadsheet if you want it

Please do, this does not feel right.

Yes please.

Ramblings:
The quantity many are interested in is (Two -Tai)
This gives the WBin temp.
For the simple case (using MTD not LMTD, and ignoring air's frictional heating) get* :
(Two -Tai)/Wa = R + 1/2QaCa - 1/2QwCw + dPw/2CwWa
and
(Twi -Tai)/Wa = R + 1/2QaCa + 1/2QwCw + dPw/2CwWa



* Using
Wa= dTmtd/R , dTmtd = 0.5(Twi + Two) - 0.5(Tao + Tai), Twi -Two = (Wa - QwdPw)/QwCw and Tao -Tai = Wa/QaCa ,
rearranging and substituting.
Where,
Wa= Heat dissipated into Air., R = (Convective + Resistive) Resistance of Radiator.
T= Temperature, Q= Flow-rate , dP = Pressure drop , C= Specific Heat.
Subscript "a"=air , Subscript "w"= water, Subscript "i"= in , Subscript "o"= out .

Edit 1: Deleted some erroneous manipulation. Will re-edit when corrected.
Edit 2: Will add further manipulation in post where used. Corrected typo.