OC on Fan Stacking

[lolito_fr]

Yes please, for the spreadsheet.

Frictional heating is going to be a whole lot lower than predicted...
5psi instead of 17psi @3gpm
(Original BI graph was a worst case scenario)

[BillA]

ok, first I'll add the flow curves for 1/2" barbs - while I sort through Les' offerings
Ta

nope, got to finish all with 3/8 ID connections first

[HammerSandwich]

A-ha - a radiator thickness test with varying airflow! Yes! Thank you, Bill. I don't suppose you have any airflow data available?

[BillA]

only the fan specs

Mfgr Model A RPM static CFM noise, dB(A)
Papst 4412 FGL 0.105 1600 55.3 26
Delta WFB1212L 0.14 1750 0.104 61.8 30.2
Delta WFB1212M 0.33 2100 0.134 72.4 34
Delta WFB1212H 0.45 2470 0.189 86.5 37.6
Delta
Delta AFB1212VHE 0.9 3200 0.420 130 48
Delta AFB1212SHE 1.6 3700 0.571 151.8 53

[Les]

Bill
Thanks for Excel.
Only now noticed that you use LMT for water temps in C/W calcs.
Always assumed was water(in), should probably paid closer attention.
Should keep me befuddled until Friday(Pub) - especially since cannot copy/paste(or at least failing using Excel 97) any manipulated data.

[Cathar]

Would be particularly interested in the Panaflo L1A's as part of Bill's radiator/fan test.

Bill, are you actually measuring the dBA at all? Or just relying on manufacturer specs?

[lolito_fr]

Only now noticed that you use LMT for water temps in C/W calcs

Had a feeling that was the case...
Looks like some of the 'predictions' will need reworking a tad!!

[BillA]

difficult to use the word rely in the context of fan specs
just observing the factory numbers

L1As ?
here's the deal, I am only going to use 4 fans; low, med, high, very high
the high and very high are easy, VHE and SHE
med is the 30dB(A) fan we ship, the Delta WFB1212M
-> low should be a fan that Swiftech would wish to ship with a 'low noise kit'
I selected the Papst 4412 FGL
???
I am very open to suggestion
Note: the low noise fan should be so at 12V, and even less so at 7V

[Cathar]

What is a little "puzzling" about the BIP vs BIX curves is that either radiator appears to be exhibiting a peak at around the 1.5-2.0gpm mark.

What is puzzling is when we consider that the BIP is a single-row core, and the BIX is a dual row core. Both are two-pass cores.

The water velocity through the cores is going to be twice as high through the BIP as the BIX, so why do they exhibit roughly the same curve peaks if we are to assume that the improved performance is coming from increased convectional efficiency between the water and the tube walls?

Methinks that something else is at play here, and its effect just happens to be larger than convectional gains, but that's just a gut feeling. I can offer no suggestions as to what might really be going on.

Bill, have you re-assessed any of the Thermochills rads under your new test setup? Your old test data seemed to imply that they continued to improve at least up until the 3gpm mark.

[BillA]

the ThermoChill 120.1 is the 'same' as the BIX,
but the BIX II is quite flat also with the Papst, virtually no tail off

something else indeed, lol

you have thoughts on a good low noise 12V/7V fan ?
at least worth testing ? (cannot use on specs, must be tested)

[Les]

Yes, the results seem to be a little enigmatic.
For a first look-see chose what appeared to be most problematic area - a low fannage curve.

At equilibrium
Radiator heat in = Radiator Heat Out
Wwi + dPwQw = Wa + Wwo
TwiCwQw + dPwQw = Wa + TwoCwQw,
Wa = (Twi-Two)CwQw + dPwQw



The imaginary + or - 0.02c error and/or the inclusion of frictional heat (dPwQw) in the heat dissipated seem to help a liitle..
An 0.02c offset error (represented by top and bottom of error bars), on its own , is no better.

[Cathar]

No really good fan suggestions unless you can source "Made in Japan" Panaflo L1A's, as opposed to the noisier "Made in China" ones.

Fans rated by the manufacturer as being below 30dBA all seem to be a case of "suck it and see". Have yet to witness any reliable fan specification information for said fans from anyone but the big guys (Papst, Panaflo, Delta, etc).

Am wondering if the low-profile tank size has anything to do with it? Venturi effect, or something else, preventing even distribution of flow into the tube inlets for the second pass? Effect is "masked" in the larger radiators due to increased cooling on the first pass?

The effects being observed do seem to imply that the water flow is not being distributed evenly as the flow rate is increased. Looking at real heater-core tanks vs the tanks which are fitted to the BI rads, and one can see a marked difference in the height of the tanks. Larger high-flow heater-cores and radiators do have quite involved end tank designs. Has something been overlooked at the design end in the quest for a low-profile solution by the "made for computers" core makers?

[pHaestus]

Why not use the EHEs for "very high"?

[HammerSandwich]

Originally Posted by unregistered

you have thoughts on a good low noise 12V/7V fan ?
at least worth testing ? (cannot use on specs, must be tested)

The Nexus 120 would be a good match for the DDC with dBA in the low 20s. Independent free-air testing by MikeC at SPCR confirms official specs. I'd really like to see how it does on a rad. Also suspect Swiftech could work a deal with the OEM, Yate Loon.

[Incoherent]

Bill
I think there is a measurement error.
It is not necessarily an offset, or at least the offset could be proportional to T.
I note "*RTD calibration in hardware". What kind of RTDs are they? PT100, Ni100, Ni120, CU10 etc? they have different temperature coefficients, even within the same type (i.e. PT100s can have different coefficients depending on the standard adherred to). Could it be that you are using an alpha value for a different type? This would defy calibration of the RTDs against a "known calibration of one", or comparing at zero load, they would read the same at the same temperature, the error coming when they have a difference...
(Edit. Some misleadings statements here. The "known calibration" would work, if it is itself correct. The effect is a compression/dilation otherwise)

... is pHaestus using RTDs???


Using a PT100 RTD with a NI110 setting for example might give this error within the small delta T we are dealing with.

Sorry if this is way off the mark, clutching at straws really, I just do not like these curves.

[lolito_fr]

Thx for Excel.
Following up the frictional heating idea, now evidently not the major culprit!



Correct me if I'm wrong, "Adj dP" (column R) is connector dP ? raw or PSI?
Just curious...

[Les]

Originally Posted by Incoherent

Bill
I think there is a measurement error.
It is not necessarily an offset, or at least the offset could be proportional to T.
I note "*RTD calibration in hardware". What kind of RTDs are they? PT100, Ni100, Ni120, CU10 etc? they have different temperature coefficients, even within the same type (i.e. PT100s can have different coefficients depending on the standard adherred to). Could it be that you are using an alpha value for a different type? This would defy calibration of the RTDs against a "known calibration of one", or comparing at zero load, they would read the same at the same temperature, the error coming when they have a difference...
(Edit. Some misleadings statements here. The "known calibration" would work, if it is itself correct. The effect is a compression/dilation otherwise)

Using a PT100 RTD with a NI110 setting for example might give this error within the small delta T we are dealing with.

Sorry if this is way off the mark, clutching at straws really, I just do not like these curves.

Maybe something like this:



Using these RTDs and switching gives :

[Incoherent]

Originally Posted by Les

Maybe something like this:

If I am following you correctly Les, yes, thats essentially what I am driving at.
The curves for the RTD responses, ideal vs actual, meet somewhere but their roc is different. We are looking at small errors here in the context of overall accuracy...

Or something like this.

Bills calibration should take care of this but I am curious about the "RTD Calibration in Hardware" statement. I am not a fan of using the alpha temp coefficient for Resistive sensors, RTDs are nowhere near as bad as NTC thermistors, being actually incredibly linear, but there are residual errors if the hardware is using it is as the calibration method. Way prefer the 3rd order fit (Steinhart-Hart for thermistors) especially over a large range. This is not the error in this case though (If it is in fact an error)
At 35C the Cp should perhaps be 4154J/l ish, but this has a very small effect on the power calc.

[BillA]

"*correction in hardware" is my notation for a goofy calibration scheme
I have a ref Pt392 caled in 10° increments for which I have a curve, so I understand actual;
in a bath I then 'cal' the working sensor at its actual use temp (or midpoint) - a trim resistor for each RTD input; connected to the separately caled inst that will read it

I need to re-cal over a curve for each one
bleh, this is a lot of work

[aaronspink]

Originally Posted by unregistered

med is the 30dB(A) fan we ship, the Delta WFB1212M

Isn't the 1212M a 34db fan? Surely you ment the 1212L.

-> low should be a fan that Swiftech would wish to ship with a 'low noise kit'
I selected the Papst 4412 FGL
???
I am very open to suggestion
Note: the low noise fan should be so at 12V, and even less so at 7V

The 4412 or maybe the nexus 120 seem to be popular and available world wide, so either would see a good choice. The Nexus is a little quieter but may be harder for swiftech to source.

Aaron Spink
speaking for myself inc.

[Les]

God Bill, this is tricky. Glad I didn't open this can.
Yes, am truly befuddled
Anyone fancy checking some sums whilst I am in the pub tomorrow.

(1) Heat Balance Air
Heat In = Heat Out
Wai + Wa + dPaQa = Wao
TaiCaQa +Wa + dPaQa = TaoCaQa
TaoCaQa - TaiCaQa = Wa + dPaQa
Tao= Tai + Wa/CaQa + dPa/Ca

(2)
LMTD=(Twi - Two) - (Tao - Tai)
---- ln ((Twi - Tao)/(Two - Tai))

Then substituting from (1) :
LMTD=(Twi - Two) - (Wa/CaQa + dPa/Ca)
.... ln((Twi - (Tai + Wa/CaQa + dPa/Ca))/(Two - Tai)

(3) R= F (LMTD)/Wa ,
where F is correction factor (obtainable from Wolverine Fig 4.14 (for two pass) - I think).
Have values for all except the air-flow(Qa) and air's pressure drop(dPa).

[BillA]

as
yes, the WFB1212M is 34 dB(A) - our 'kit' fan

[Les]

Think Bill's published data as presented in ThermoChill assessment are/is more than adequate.
Can be used for system calculations taking pump heat at zero flow and adjusting for temperature drop across the radiator(((0.5 *1/'CwQw ((theoretical imprecise but error is negligible(checked with data))).
They/it characterise(s) the radiator.

Edit. Is data a collective noun ?

[BillA]

collective ? no, plural; singular datum

[Les]

Gaggle of geese.
data = gaggle of datums

Feel more comfortable using data as singular