BillA: re Last graph.
A chart and formulae of head in mH20 vs flow in LPM would save me some time. :-D
Seriously though, the formulae should not have a constant (the "c" part). Surely at zero flow the rig exhibits zero resistance?

Quite like Series4 and Series5.
Explanation of experimental observations?
Good cancellation of Velocity and Conduction errors* with superimposition of Dissipation curve(0.3 to 0.45° Tins difference) for DUT ???

* 4-198 "Temperature and Heat Transfer Measurements" Robert J. Moffat
“Heat and Mass Transfer” Kreith, F.; Boehm, R.F.; et. al.
Boca Raton: CRC Press LLC, 1999

Pub today - "compos mentis" problems later - Friday canceled - crutches unstable due to rain and high winds.


Edit1: Pleased to see the "CoolingWorks CoolRad-12T" reasonably fits the "Typical(?) 1PSI dP @ 7LPM Water" description(120x120mm Rads).
Details of the CoolRad-12T's core would, for modeling purposes, be interesting

Edit2: Attached Tentative Model

Edit3: Refined Attachment: Changed description "Turbulent Water" to "Transitional Water(Re=2300)"

Edit4: Added "Laminar Water @ 0.5 LPM"

LHG there are a number of things which mean that it can be any order.

Im fairly sure it should perform according to q^2 only. A constatn term could happen best guess is that its so small in this case that it may not exist, but a constant pressure drop can happen. There maybe be a cubic term coming in there due to mixing effects.

As it so happens the big book has a chapter on this but i dont have time to look at it this weekend

here you go LHG, understand that I'm curve fitting with Excel - do not read more into it than that
(sure the end sum has it all together, and different tests can characterize some aspects better than others)

bobo, forgot to attach the photo

Les
Series 4 and 5 are the same crosses + tubing, but swapped end for end (pressure taps reversed but not sensors)
the nominal ID went from 0.53" to 0.92" but as the RTD is protruding in from the branch the area calc is not worth the time, but for sure the velocity is greatly reduced
anyone care to comment on 'h' wrt the probe ?

this whole exercise is bothersome as I'm sure all this is well understood by those better versed in metrology
i.e. we 'measure' by associating a signal with a source, never mind that the sensor is potted within a sheath all elements of which have their own temp gradients
- so the first adjustment is to cal in the medium to be measured
(I cal in flowing water, does this mean that my air temps are off ?)
- now I find a flow (velocity related ?) temp correction ?
surely all this is known, apparently little different than pressure drop corrections
guess I need that book Les, seems experimentally oriented

in any case, using the Series 4 temp corr eq yields the comparison shown

<note to self> buy clear PVC primer</note>

:p

glad you could get something out of it
tip appreciated, but I just do it

Bothersome? I find it a bloody nightmare and I am not the tester
Have been busy postponing addressing.
Only posting to show have not "done a runner"
Not yet formulated any translucent thoughts,
Have niggles, though about the initial and dominating coolant Temp correction being +, this appears contrary to Robert J. Moffat expectations: "The Prandtl numbers of most liquids are greater than 1; hence, the recovery factor a is greater than 1 and
probes tend to read higher than the stagnation temperature in high-speed liquid flows. With thermistors
and RTDs in liquids, the self-heating effect and the velocity error both tend to cause high readings".
Similarly feel the that "air temps are off " maybe correct:
"In gases, where the Prandtl number is less than 1, the two effects are of opposite sign and may partly cancel
each other."
However need to do some reading


Busy refining and cluttering Model shown in Post127
From this:
Think the parameters from your data(SD,(C/W)in,and Eff(Fan +Rad)) can be designated a subscript m:
"Tm: The mixed mean fluid temperature. Used for internal flows where the intent of the calculation
is to describe the changes in mixed mean fluid temperature (e.g., heat exchangers)."
.........."For a given data set, the value of the heat transfer coefficient will depend on the reference temperature
chosen, and h should be subscripted to inform later users which reference was used: e.g., h¥, hm, or adiabatic"

With measured Heat Dissipation ~ 150 to 300w, I think the difference between " mixed mean fluid temperature" and "Adiabatic" parameters is 1.44%(2.16/150) and 1.25%((2.16+1.59)/300)
Have been correcting my Model(adiabatic I think) for the meddlesome dP*Q to obtain a " mixed mean fluid temperature" Model
Have,also, been correcting for "Fan Heating of Air"(2.16w,if present).
Attached .
Not suggesting should read anything into agreement

Edit Corrected Tube Spacing miscalculation in Model

LOL I can see that.
its the thought I dont have nearly enough scotch tape that gives me pause :p

I see the dual vs quad fan positions you mentioned
are you employing SS thermowells in the water lines?

IC
yes, the same as for the air temps
the intent is to use the longer time constant to smooth the readings (steady state avg)

as Les has noted, the cal procedure needs to be developed - what exists is terribly crude

Les
it is going to be frighteningly difficult for me to make a ref flowing air temp source to cal the air probes

not sure if self heating is an issue when caled in the medium with the inst, I see the effect every time I switch from front to back (wait 10 sec. to start recording)
and likewise any 'surface effect' as I cal in flowing conditions
??
most of this can be obviated with a tailored cal procedure I think

nice graph
the increasing dP should limit the efficiency, be interesting to put more pump to it

Or pray for error cancellation

prayer ?
is that the first step, or the last ?
I am not going to be a happy camper waiting on the second coming

some ramblings that may be informative
in the (incorrect - bah) determination of Cv, I had occasion to observe the decrease in temps
it would be interesting to verify the Cv (Kv in Europe), then see the shift from 15 to 35°C
I'll work out a procedure and try again (gonna have to cal the crosses @ 60°F, the spec)
it may be futile as Cv factors are for turbulent flow

Have I got this right?
In steady state.
An apparent temperature drop is observed across "6in of tube +bits" when coolant inlet is at the same temperature as the Environmental Chamber
This indicates an apparent heat dissipation from coolant to "environmental chamber"
Are we not applying a correction which implies heat absorption by coolant when coolant inlet is at the same temperature as the "Environmental Chamber"?
Or have I got it arse-about--tit?
Right or wrong, above does not consider the Environmental Chamber's response to apparent or real dissipation. This should be included in logic.

Tested CoolRad-12T with zero fan and zero temperature differential?


May edit as brain warms-up and discover I am writing crap


Edit
Thinking more in terms of psychokinesis than "waiting on the second coming".
This has to be the first step. Commonly used as the last step but this is pk moving the observer and not the DUT - cheating.

"Have I got this right?
In steady state.
An apparent temperature drop is observed across "6in of tube +bits" when coolant inlet is at the same temperature as the Environmental Chamber
This indicates an apparent heat dissipation from coolant to "environmental chamber"
Are we not applying a correction which implies heat absorption by coolant when coolant inlet is at the same temperature as the "Environmental Chamber"?"

yup
and note that the metrology considerations drop out wrt coolant temps
the offset in air temps due to caling in flowing water has me perplexed though (but will impact all calcs equally)
needs resolution re the procedure though

"Tested CoolRad-12T with zero fan and zero temperature differential?"
no, just the crosses; that whole series will be rerun in several hours

bobo will like this,
for 2+ days have been looking for a source of dissipation variation at high flow rates, not understanding that it was the same source as the hysteresis offset
- if the system is started at fulll flow some air is entrained that will not clear until the flow is below 1.5 gpm (using a March M3 pump), a coolant density variation

now that I (think I) have learned how to start the test bench, I wil recal the crosses and do the rad at 0 dT also

Perhaps not the 0.02c?

correct, that is an instrument/sensor 'offset' (there is no white and green)
I was referring to the effects of flow per se

Was suggesting in Post137 that the "Flow correction" should lower SD and Eff not raise them.
We are showing SD@10c > SD@10c w/o corr , I think this is wrong.
Should be as shown in attachment.