Possible Rad Testing Procedure

MaxxxRacer · 03-12-2005, 07:56 AM

Before i say anything on this matter, bill please dont rip my throat out for my following suggestions..

Since getting exact numbers for rad testing that are generic and are useful and mathematicaly aplicaable to a wide variety of setups seems just about darn near impossible i thought of coming up with a stanardized test setup in which a number of rads could be tested on, and compared to eachother on this setup. the numbers would not be aplicable to other systems (obvously) but it would help to give us a judge of how a rad will perform compared to others in respect to pressure drop (no exact numbers on pressure drop, but relative restriction), heat disipation and various flow rates and heat disipation at various fan speeds and types.

The test setup would be a cpu block mounted on a cpu with the thermal diode mod or a die simulator, along with the block inlet temp. Flow meter, rad inlet temperature, rad outlet temperature, a powerful pump to test as a various flow rates, and the other odds and ends in a normal loop.

With this you could get the the affect of the rad on cpu temperatures, the amount of heat the rad is removing from the water at a given flow rate, and the flow rate. Obviously room temp would need to be kept constant for all of the tests, but with this data you could get usable performance numbers for rads.

So now that I have given my idea for rad testing, feel free to tell me im a moron and that this will not work. (bait)

BillA · 03-12-2005, 11:08 AM

any test is better than no test
there is a general lack of understanding about testing rads because so few have done any
good / reproducible results require good equipment used with exactitude
none of which is defined above

I now consider all my previous work primitive - relative to the present data 'quality' I can now maintain
but I live in a different place, eh ?
(I'm charged with designing these things.)

jaydee · 03-12-2005, 12:16 PM

You missing one (probably many more) important thing, CFM. The airflow through a rad should be thought of as water flow through a cooling system. It needs to be measured at various air flow rates to generate a curve.

I think you misunderstand how a rad works. The flowrate of the coolant dosn't dictate the rads performance. You can run 1GPM or 50GPM and the rad will still disipate roughly the same amount of heat. The flowrate of the liquid is not what you want to base your tests on.

In my uneducated opinion how to get anything resembling usefull data you need an airflow chamber that you can measure and adjust CFM through the rad while running liquid at a known temp, viscosity, and thermal properties (distilled water probably the easiest).

You can't find the efficency of a rad just by flow rate of the liquid. Your test methods above (probably mine to) would not give anything resembling useable results IMO.

Rad testing is much harder than water block. Stick with getting block testing right first.

Les · 03-12-2005, 12:43 PM

Quote:

Originally Posted by MaxxxRacer

.......
The test setup would be a cpu block mounted on a cpu with the thermal diode mod or a die simulator, along with the block inlet temp. Flow meter, rad inlet temperature, rad outlet temperature, a powerful pump to test as a various flow rates, and the other odds and ends in a normal loop.

With this you could get the the affect of the rad on cpu temperatures, the amount of heat the rad is removing from the water at a given flow rate, and the flow rate. Obviously room temp would need to be kept constant for all of the tests, but with this data you could get usable performance numbers for rads
.....
(bait)

"with this data you could get usable performance numbers for rads"

Doubt it.
See my rough analysis of Phaestus's data in this thread

MaxxxRacer · 03-12-2005, 03:45 PM

BAH!

jaydee, i know that the air is the most important factor in all of this. Obviously I have NO way of measuring the actual air flow so i was going to use a variety of commonly used fans including panaflows, sayno denkis, papst, and anything else that ppl commonly use. I would run them at 12, 7 and 5 volts each. These numbers will not be able to turned into the equivlanet of a pq curve for a pump, but atleast it will show the performance at different fan speeds and and types. I think that knowing this information is more useful to the user than actually haivng cfm ratings. I believe this becuaes who in the hell knows how much cfm is going through their rad. 100 bucks says bill is the only one who has the slighest idea. The rest of us dont have anythign even close to measuring it. and that most surely includes the end user of this stuff who hasnt the slighest idea. Giving the end user something that is a bit more tangible and easy to compare is imo a better way of handling this. most surely its not as scientific but getting scientific requies an investment that non of us can even come clost to afford, much less wouldnt buy if we had the money.

Les, looking at that data, my predictions of what would happen with the rad air temp in to out are totaly wrong.. it seems the higher the actual airflow the lest dT for the air temp. which does make sense now thinking about it. each molecule is heated less, there are just and X number of more molecules to take the heat up.

Les, reading through that thread im throughly disheartened. When i initially looked at the data i thought something useful was there. but once you brought up those CW numbers and the other graph, it shows that the data has gone screwy lewy, and there is no definite answer for this.

But all of that aside, I am still going to eventually try my meothod of testing stated above. As soon as I get a review unit rad other than my 2-199 weapon modded HC, I will do a test run and see what becomes of it. It will be a month or more before I can go any testing, but when i do i will most surely post all of the data that i collected. maybe les can do those lovely graphs he always does.

Bill do you have any suggestions to make the test better? Other than a 100 thousand dollar wind tunnel with thermal control? If that is your only suggestions, i wonder if JPL would let me use their wind tunnel for a few short hours

jaydee · 03-12-2005, 04:27 PM

Quote:

Originally Posted by MaxxxRacer

BAH!

jaydee, i know that the air is the most important factor in all of this. Obviously I have NO way of measuring the actual air flow so i was going to use a variety of commonly used fans including panaflows, sayno denkis, papst, and anything else that ppl commonly use. I would run them at 12, 7 and 5 volts each. These numbers will not be able to turned into the equivlanet of a pq curve for a pump, but atleast it will show the performance at different fan speeds and and types. I think that knowing this information is more useful to the user than actually haivng cfm ratings. I believe this becuaes who in the hell knows how much cfm is going through their rad. 100 bucks says bill is the only one who has the slighest idea. The rest of us dont have anythign even close to measuring it. and that most surely includes the end user of this stuff who hasnt the slighest idea. Giving the end user something that is a bit more tangible and easy to compare is imo a better way of handling this. most surely its not as scientific but getting scientific requies an investment that non of us can even come clost to afford, much less wouldnt buy if we had the money.

I am not against you in trying this BTW. I just belive if your goingto do it then at least learn from everyone's mistakes and do it better otherwise what's the point?

Bill said "any test is better than no test". I disagree. Bad tests are worst than no test IMO. No test are not misleading, bad tests can be if the errors are not known.

You say most do not know the CFM through their rad and use that as an argument. Now think about flow rate through a water block. Do you also not need a flow measurment because most people don't know the flow rate in their systems?

If you can't afford the equipment to do a job then do you do the job without the equipment needed and expect good results? I know I would be fired from my job if I showed up without my tools...

Anyway you havn't stated how you are going to record the temps? You need an expensive thermometer for such measurments as your temps changes will likely be below .5C in many cases.

Accuracy below .5C is not cheap. Also your flow meter will need to be within 1% accuracy or better.

MaxxxRacer · 03-12-2005, 04:43 PM

oh i will be getting a fluke 51 52 52 or 54. donation from pharmacutical company. so accuracy is not a problem. the fluke will have nhit or nist (cant remember which) certification so its garunteed accurate. hopefully i get the 54 so i can do delta T's and data logging esier.. if i get the 51 or 53... well delta t's will not be possible.

as for the cpu temp it will be recorded by the maxim chip via the cpu socket mod and the smbus.

I knew you would make the argument about most users dont know the flow that they have. but with ph's tests its alot easier to get a guestimate of your flow as his setup is about what would be equal in restriction to most loops with a gpu block on them. (his flow meter and temp sensors should be equal to or more restriction than a gpu block).

Dont get me wrong, i would love to do flow tests with actual meters and temp controled air. that would be awesome. but we cant have everything that we want in life. I will do the tests as stated, and if I dont come out with some results that are actually usable i will post my findings here, and let everyone know that they are horriblly innacurate and try to see where i went wrong.

jaydee · 03-12-2005, 04:54 PM

Good luck.

MaxxxRacer · 03-12-2005, 04:56 PM

oh come on.. no more suggestions. Im always interested in a way to improve testing.. that is as long as its remotely feasable...

jaydee · 03-13-2005, 12:45 AM

Quote:

Originally Posted by MaxxxRacer

oh come on.. no more suggestions. Im always interested in a way to improve testing.. that is as long as its remotely feasable...

heh, well not to sure what to add. One thing that might pose a probelm is the Fluke 52II and 54II only have 2 TC's. The 51 and 53 only have one. The 53 and 54 are data loggers but that won't do any good because you need 3 temp measurments.

I am also thinking you're making it harder than you have to to get the same data. Instead of measuring air temps you can just measure the water inlet and water outlet of the radiator.

Say the water in is 26C and the water out is 25C with a Panalfow 120MM at 5 volts. Then you raise the voltage to 12 and the water in is 25C and water out is 23.5C....

At 7V you get 1C of cooling.
At 12V you get 1.5C of cooling.

If you have a decent flow meter you can add in a GPM to make it a little more usefull.

I can kinda see how that might be usefull or interesting if nothing else.

MaxxxRacer · 03-13-2005, 04:42 AM

well i really only need two temperature measurements. see in my loop the rad outlet is about 5 inches of tubing to the wb inlet. so they are essentially the same thing. so i would put one temp measurment right at the rad inlet and one right inbetween the rad outlet and the wb ilnet.. works well that way.

As for air temp, thats not something that I will really include as a valid measurement as Im using the radio shack gauge for that.. lol.. I will just try to keep it at 25C (my room doesnt like to get cool)

i will also get another temp probe to get air temp with the fluke.. just unplug one of the water temp probes for a sec and grab the air temperature to see how far off the radio shack gauge is off so i have a point of reference for testing.

freeloadingbum · 03-13-2005, 11:57 AM

Quote:

Originally Posted by jaydee116

I am also thinking you're making it harder than you have to to get the same data. Instead of measuring air temps you can just measure the water inlet and water outlet of the radiator.

Say the water in is 26C and the water out is 25C with a Panalfow 120MM at 5 volts. Then you raise the voltage to 12 and the water in is 25C and water out is 23.5C....

At 7V you get 1C of cooling.
At 12V you get 1.5C of cooling.

If you have a decent flow meter you can add in a GPM to make it a little more usefull.

I can kinda see how that might be usefull or interesting if nothing else.

The delta across the rad doesn't change with changing airflow in a closed loop, only with changing waterflow. The delta represents the load so with 100 watts at 1gpm the delta will be something like .37C ( I don't remember the actual number) and at .5gpm it will be double at .74C regardless of airflow. The airflow will change the input temp though so you would see something like this

fan@7volts water in 26.37C water out 26C
fant@12volts water in 24.37 water out 24C
assuming 1gpm and 100watt load

Remember that Bill uses an open loop test with a constant input temp so he can use the rad delta to measure performance

Personally, I don't trust Bill's (or Joe C's) method of using rad delta as I feel the measurement is too easily distorted by temp gradients in the waterflow which I think explains pH's fluctuating rad delta as well.

Such a gradient is problably most evident when you look at the results of the bi extreme rad with low airflow. The front row will remove a higher percentage of heat than the rear row creating two distinct water temps with only a split second to converge before reaching the output sensor.

The method used to keep the input water temp constant may cause a temp gradient as well, compounding the problem.

In a close loop test, a temp gradient will be too small to have any effect on the final result. The main problem is you would need to know how the pump heat changes at different flow rates in order to generate a proper c/w measurement.

MaxxxRacer · 03-13-2005, 04:32 PM

well i will be restricting the pump not controlling voltage. so heat dump will be constnat. but your statement on the delta being the same is one that intrigues me. I will have to see what happens to my test setup.

As to the water not having time to combine i dont think thats a huge factor but it could play a role... with that said, i will situation the rad outlet temp closer to the cpu block. it should let the flow even out a bit more in the tube before i take a reading.

freeloadingbum · 03-13-2005, 05:25 PM

Quote:

Originally Posted by MaxxxRacer

well i will be restricting the pump not controlling voltage. so heat dump will be constnat. but your statement on the delta being the same is one that intrigues me. I will have to see what happens to my test setup.

The pump draws more current at higher flow rates regardless of the voltage being constant. It's one of the characteristics of how a pump works.

MaxxxRacer · 03-13-2005, 06:25 PM

hmm... need to get a high grade ohms meter for this. but afaik the impeller spins at roughl y the same speed despite flow. which would mean the same same power draw.

and also, the impeller isnt directly connected to the motor anyway. hence mag drive. so they are independt of eachother. the impeller may spin at a different speed than the motor is running at.

Incoherent · 03-18-2005, 11:23 AM

There might be a rather simple way to get a C/W number for a radiator.
The maths behind it is rather complex but try this.
You need to know:
1. Water temp
2. Water volume in litres
3. Water heat capacity in j/l°C

Run system with no load except the pump itself and the fan running and run until the water temperature has stabilised. Record temperature.
Switch on load and log temperature and time. Run system until the temperature has stabilised. I mean really stabilised.
Find the time t (in seconds) at which the temperature was half way towards the final loaded temperature.
t/(natural log of 2) = tau
tau/(Cp*volume)= C/W of radiator

This is the UI for some rather complex maths, which is possibly (or even probably) wrong but I think it works. Half lives, timeconstants, charging phenomena which I have attempted to hack away at.
I believe it works without knowing the watts or air temperature, they are intrinsically bound to tau. Also it does not matter what your starting point vs end temperature is.

T=(1-e^(-t/R.Cp.v))*(R.Cp.v.(Q/v.Cp)+Tamb-Tinit)+Tinit

T=(1-e^(-t/tau))*(Tend-Tinit)+Tinit

lolito_fr · 03-18-2005, 11:42 AM

Sweet idea!
You would probably want a large mass of water (big res) to help negate errors due to (unknown) thermal capacity of WB/rad/tubes/etc...

BillA · 03-18-2005, 12:06 PM

will give it a go
time to equilibrium will depend on the nature of the heat source;
i.e. a ttv with ~20% secondary losses will come to equilibrium in 15 min or so (if hot, with a change in load; cold ~45 min but that is due to the chamber mass) - but here W is by no means well quantified
- but with a well insulated heat die the time to equilibrium varies between 1.5 and 3 hrs for a hot change in load, 6hrs from a cold start - and W is known with 'good' precision

if the system tc is known then many shortcuts are possible
in any case an environmental chamber will be found essential, as will the ACCURATE characterization of the fan inlet air temp (not the same as ambient by any stretch of one's imagination)

I'm thinking that if one already has the equipment to test directly, the proposed methodology may not compare too favorably
- no way to judge really, got to try it out
jeez, another project

Incoherent · 03-19-2005, 04:40 PM

Quote:

Originally Posted by unregistered

...in any case an environmental chamber will be found essential, as will the ACCURATE characterization of the fan inlet air temp (not the same as ambient by any stretch of one's imagination)

Indeed a well controlled air temperature is essential.
However I believe this method does not need air inlet temperature or ambient - except to check the method result and it is essential that they are constant.

I have made some measurements and calculations:

The lack of air temperature control is visible here as a deviation from the ideal curve. This will be throwing the reading but I am now pretty certain this is a valid method.
Important to note that this is the System heat dissipation C/W, not just the Radiator.

BillA · 03-19-2005, 05:25 PM

ah, post #16 had me going the other direction
I'll be testing systems again shortly so I'll collect some data

pardon the simple question, but since the system tc is not known a priori, what is the significance of then defining 1/2 T rise ?
clearly I've missed something: is T in (T/ln2) actually 1/2 T rise ?

not difficult to test: put in a chamber and keep the air and coolant equal, compare the heat input and T

EDIT: lol, whee that would be a test

MaxxxRacer · 03-19-2005, 09:05 PM

well so much for my simple rad testing procedure.. lol... though this one doesnt seem to hard to be able to do as long as u take care of the math.

Incoherent · 03-20-2005, 02:48 AM

t, T, τ

Lower case t = time (to 1/2 Temp rise) = half life
Upper case T = temperature
Greek letter tau τ = time constant (t/ln(2)) = mean life

The mean life equals the time constant.
These problems (heating/cooling, charging/discharging, radioactive decay) all follow the equation e^-t/τ. τ (tau) is the time constant, in our case: thermal resistance*volume*heat capacity. An electrical equivalent is τ=R*C, C, capacitance I am equating to volumetric heat capacity and resistance R to C/W. Can't find a decent link right now, later.

Les · 03-20-2005, 04:40 AM

Quote:

Originally Posted by Incoherent

...........Important to note that this is the System heat dissipation C/W, not just the Radiator.

Very true.
However, as noted by lolita fr , does this not require some account to be taken of Heat capacity of the system components(+fittings)

Still a bit brain dead and hope you excuse repetition of lolita's comment.
[Recovering from Strong Flow's defeat in the Gold Cup(6th, perhaps the drying ground was against , Nicholls reports "He is a bit shouldery")
Very little difference between predicting cooling and racing behavior].

lolito_fr · 03-20-2005, 04:54 AM

lolita in my dreams as well
haha

Incoherent · 03-20-2005, 07:35 AM

True, the other components certainly need to be accounted for . Not quite sure how to approach it, might just be to sum the Cp's. I am not sure they are particularly significant especially with a decent volume.
Perhaps best to take it as a relative number for comparing radiators in the same system, the secondary heat capacities are then just a constant. Annoying, I prefer absolutes.

03-12-2005, 03:45 PM	#5
MaxxxRacer Cooling Savant Join Date: Sep 2004 Location: Los Angeles, Ca Posts: 310	BAH! jaydee, i know that the air is the most important factor in all of this. Obviously I have NO way of measuring the actual air flow so i was going to use a variety of commonly used fans including panaflows, sayno denkis, papst, and anything else that ppl commonly use. I would run them at 12, 7 and 5 volts each. These numbers will not be able to turned into the equivlanet of a pq curve for a pump, but atleast it will show the performance at different fan speeds and and types. I think that knowing this information is more useful to the user than actually haivng cfm ratings. I believe this becuaes who in the hell knows how much cfm is going through their rad. 100 bucks says bill is the only one who has the slighest idea. The rest of us dont have anythign even close to measuring it. and that most surely includes the end user of this stuff who hasnt the slighest idea. Giving the end user something that is a bit more tangible and easy to compare is imo a better way of handling this. most surely its not as scientific but getting scientific requies an investment that non of us can even come clost to afford, much less wouldnt buy if we had the money. Les, looking at that data, my predictions of what would happen with the rad air temp in to out are totaly wrong.. it seems the higher the actual airflow the lest dT for the air temp. which does make sense now thinking about it. each molecule is heated less, there are just and X number of more molecules to take the heat up. Les, reading through that thread im throughly disheartened. When i initially looked at the data i thought something useful was there. but once you brought up those CW numbers and the other graph, it shows that the data has gone screwy lewy, and there is no definite answer for this. But all of that aside, I am still going to eventually try my meothod of testing stated above. As soon as I get a review unit rad other than my 2-199 weapon modded HC, I will do a test run and see what becomes of it. It will be a month or more before I can go any testing, but when i do i will most surely post all of the data that i collected. maybe les can do those lovely graphs he always does. Bill do you have any suggestions to make the test better? Other than a 100 thousand dollar wind tunnel with thermal control? If that is your only suggestions, i wonder if JPL would let me use their wind tunnel for a few short hours __________________ Excuse me, I believe you have my stapler.

03-12-2005, 04:43 PM	#7
MaxxxRacer Cooling Savant Join Date: Sep 2004 Location: Los Angeles, Ca Posts: 310	oh i will be getting a fluke 51 52 52 or 54. donation from pharmacutical company. so accuracy is not a problem. the fluke will have nhit or nist (cant remember which) certification so its garunteed accurate. hopefully i get the 54 so i can do delta T's and data logging esier.. if i get the 51 or 53... well delta t's will not be possible. as for the cpu temp it will be recorded by the maxim chip via the cpu socket mod and the smbus. I knew you would make the argument about most users dont know the flow that they have. but with ph's tests its alot easier to get a guestimate of your flow as his setup is about what would be equal in restriction to most loops with a gpu block on them. (his flow meter and temp sensors should be equal to or more restriction than a gpu block). Dont get me wrong, i would love to do flow tests with actual meters and temp controled air. that would be awesome. but we cant have everything that we want in life. I will do the tests as stated, and if I dont come out with some results that are actually usable i will post my findings here, and let everyone know that they are horriblly innacurate and try to see where i went wrong. __________________ Excuse me, I believe you have my stapler.

03-12-2005, 04:56 PM	#9
MaxxxRacer Cooling Savant Join Date: Sep 2004 Location: Los Angeles, Ca Posts: 310	oh come on.. no more suggestions. Im always interested in a way to improve testing.. that is as long as its remotely feasable... __________________ Excuse me, I believe you have my stapler.

03-13-2005, 04:42 AM	#11
MaxxxRacer Cooling Savant Join Date: Sep 2004 Location: Los Angeles, Ca Posts: 310	well i really only need two temperature measurements. see in my loop the rad outlet is about 5 inches of tubing to the wb inlet. so they are essentially the same thing. so i would put one temp measurment right at the rad inlet and one right inbetween the rad outlet and the wb ilnet.. works well that way. As for air temp, thats not something that I will really include as a valid measurement as Im using the radio shack gauge for that.. lol.. I will just try to keep it at 25C (my room doesnt like to get cool) i will also get another temp probe to get air temp with the fluke.. just unplug one of the water temp probes for a sec and grab the air temperature to see how far off the radio shack gauge is off so i have a point of reference for testing. __________________ Excuse me, I believe you have my stapler.

03-13-2005, 04:32 PM	#13
MaxxxRacer Cooling Savant Join Date: Sep 2004 Location: Los Angeles, Ca Posts: 310	well i will be restricting the pump not controlling voltage. so heat dump will be constnat. but your statement on the delta being the same is one that intrigues me. I will have to see what happens to my test setup. As to the water not having time to combine i dont think thats a huge factor but it could play a role... with that said, i will situation the rad outlet temp closer to the cpu block. it should let the flow even out a bit more in the tube before i take a reading. __________________ Excuse me, I believe you have my stapler.

03-12-2005, 07:56 AM	#1
MaxxxRacer Cooling Savant Join Date: Sep 2004 Location: Los Angeles, Ca Posts: 310	Possible Rad Testing Procedure Before i say anything on this matter, bill please dont rip my throat out for my following suggestions.. Since getting exact numbers for rad testing that are generic and are useful and mathematicaly aplicaable to a wide variety of setups seems just about darn near impossible i thought of coming up with a stanardized test setup in which a number of rads could be tested on, and compared to eachother on this setup. the numbers would not be aplicable to other systems (obvously) but it would help to give us a judge of how a rad will perform compared to others in respect to pressure drop (no exact numbers on pressure drop, but relative restriction), heat disipation and various flow rates and heat disipation at various fan speeds and types. The test setup would be a cpu block mounted on a cpu with the thermal diode mod or a die simulator, along with the block inlet temp. Flow meter, rad inlet temperature, rad outlet temperature, a powerful pump to test as a various flow rates, and the other odds and ends in a normal loop. With this you could get the the affect of the rad on cpu temperatures, the amount of heat the rad is removing from the water at a given flow rate, and the flow rate. Obviously room temp would need to be kept constant for all of the tests, but with this data you could get usable performance numbers for rads. So now that I have given my idea for rad testing, feel free to tell me im a moron and that this will not work. (bait) __________________ Excuse me, I believe you have my stapler.

03-12-2005, 11:08 AM	#2
BillA CoolingWorks Tech Guy Formerly "Unregistered" Join Date: Dec 2000 Location: Posts: 2,371.493,106 Posts: 4,440	any test is better than no test there is a general lack of understanding about testing rads because so few have done any good / reproducible results require good equipment used with exactitude none of which is defined above I now consider all my previous work primitive - relative to the present data 'quality' I can now maintain but I live in a different place, eh ? (I'm charged with designing these things.)

03-12-2005, 12:16 PM	#3
jaydee Put up or Shut Up Join Date: Dec 2001 Location: Spokane WA Posts: 6,506	You missing one (probably many more) important thing, CFM. The airflow through a rad should be thought of as water flow through a cooling system. It needs to be measured at various air flow rates to generate a curve. I think you misunderstand how a rad works. The flowrate of the coolant dosn't dictate the rads performance. You can run 1GPM or 50GPM and the rad will still disipate roughly the same amount of heat. The flowrate of the liquid is not what you want to base your tests on. In my uneducated opinion how to get anything resembling usefull data you need an airflow chamber that you can measure and adjust CFM through the rad while running liquid at a known temp, viscosity, and thermal properties (distilled water probably the easiest). You can't find the efficency of a rad just by flow rate of the liquid. Your test methods above (probably mine to) would not give anything resembling useable results IMO. Rad testing is much harder than water block. Stick with getting block testing right first.

03-12-2005, 04:54 PM	#8
jaydee Put up or Shut Up Join Date: Dec 2001 Location: Spokane WA Posts: 6,506	Good luck.

03-13-2005, 06:25 PM	#15
MaxxxRacer Cooling Savant Join Date: Sep 2004 Location: Los Angeles, Ca Posts: 310	hmm... need to get a high grade ohms meter for this. but afaik the impeller spins at roughl y the same speed despite flow. which would mean the same same power draw. and also, the impeller isnt directly connected to the motor anyway. hence mag drive. so they are independt of eachother. the impeller may spin at a different speed than the motor is running at. __________________ Excuse me, I believe you have my stapler.

03-18-2005, 11:23 AM	#16
Incoherent Cooling Savant Join Date: Sep 2003 Location: Vallentuna, Sweden Posts: 410	There might be a rather simple way to get a C/W number for a radiator. The maths behind it is rather complex but try this. You need to know: 1. Water temp 2. Water volume in litres 3. Water heat capacity in j/l°C Run system with no load except the pump itself and the fan running and run until the water temperature has stabilised. Record temperature. Switch on load and log temperature and time. Run system until the temperature has stabilised. I mean really stabilised. Find the time t (in seconds) at which the temperature was half way towards the final loaded temperature. t/(natural log of 2) = tau tau/(Cpvolume)= C/W of radiator This is the UI for some rather complex maths, which is possibly (or even probably) wrong but I think it works. Half lives, timeconstants, charging phenomena which I have attempted to hack away at. I believe it works without knowing the watts or air temperature, they are intrinsically bound to tau. Also it does not matter what your starting point vs end temperature is. T=(1-e^(-t/R.Cp.v))(R.Cp.v.(Q/v.Cp)+Tamb-Tinit)+Tinit T=(1-e^(-t/tau))(Tend-Tinit)+Tinit Last edited by Incoherent; 03-18-2005 at 11:37 AM.*

03-18-2005, 11:42 AM	#17
lolito_fr Cooling Savant Join Date: Dec 2003 Location: France Posts: 291	Sweet idea! You would probably want a large mass of water (big res) to help negate errors due to (unknown) thermal capacity of WB/rad/tubes/etc...

03-18-2005, 12:06 PM	#18
BillA CoolingWorks Tech Guy Formerly "Unregistered" Join Date: Dec 2000 Location: Posts: 2,371.493,106 Posts: 4,440	will give it a go time to equilibrium will depend on the nature of the heat source; i.e. a ttv with ~20% secondary losses will come to equilibrium in 15 min or so (if hot, with a change in load; cold ~45 min but that is due to the chamber mass) - but here W is by no means well quantified - but with a well insulated heat die the time to equilibrium varies between 1.5 and 3 hrs for a hot change in load, 6hrs from a cold start - and W is known with 'good' precision if the system tc is known then many shortcuts are possible in any case an environmental chamber will be found essential, as will the ACCURATE characterization of the fan inlet air temp (not the same as ambient by any stretch of one's imagination) I'm thinking that if one already has the equipment to test directly, the proposed methodology may not compare too favorably - no way to judge really, got to try it out jeez, another project

03-19-2005, 05:25 PM	#20
BillA CoolingWorks Tech Guy Formerly "Unregistered" Join Date: Dec 2000 Location: Posts: 2,371.493,106 Posts: 4,440	ah, post #16 had me going the other direction I'll be testing systems again shortly so I'll collect some data pardon the simple question, but since the system tc is not known a priori, what is the significance of then defining 1/2 T rise ? clearly I've missed something: is T in (T/ln2) actually 1/2 T rise ? not difficult to test: put in a chamber and keep the air and coolant equal, compare the heat input and T EDIT: lol, whee that would be a test

03-19-2005, 09:05 PM	#21
MaxxxRacer Cooling Savant Join Date: Sep 2004 Location: Los Angeles, Ca Posts: 310	well so much for my simple rad testing procedure.. lol... though this one doesnt seem to hard to be able to do as long as u take care of the math. __________________ Excuse me, I believe you have my stapler.

03-20-2005, 02:48 AM	#22
Incoherent Cooling Savant Join Date: Sep 2003 Location: Vallentuna, Sweden Posts: 410	t, T, τ Lower case t = time (to 1/2 Temp rise) = half life Upper case T = temperature Greek letter tau τ = time constant (t/ln(2)) = mean life The mean life equals the time constant. These problems (heating/cooling, charging/discharging, radioactive decay) all follow the equation e^-t/τ. τ (tau) is the time constant, in our case: thermal resistancevolumeheat capacity. An electrical equivalent is τ=RC, C, capacitance I am equating to volumetric heat capacity and resistance R to C/W. Can't find a decent link right now, later. Last edited by Incoherent; 03-20-2005 at 02:56 AM.*

03-20-2005, 04:54 AM	#24
lolito_fr Cooling Savant Join Date: Dec 2003 Location: France Posts: 291	lolita in my dreams as well haha

03-20-2005, 07:35 AM	#25
Incoherent Cooling Savant Join Date: Sep 2003 Location: Vallentuna, Sweden Posts: 410	True, the other components certainly need to be accounted for . Not quite sure how to approach it, might just be to sum the Cp's. I am not sure they are particularly significant especially with a decent volume. Perhaps best to take it as a relative number for comparing radiators in the same system, the secondary heat capacities are then just a constant. Annoying, I prefer absolutes.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)