radiator design

[jlangholzj]

Originally posted by Phil1988:
Sorry. i mistaken something^^ the statement with the area of twice as large wasn't right. I was a bit jazzed^^..

So but if i mount the fan not direct on the core. i have a big nozzle. All the airflow has to pass trough the fan. If its turning or not. That must produce lots of pressure loss? Do you have any thermal image where the fan isn't mounted directly on the cores? Any experience?

Regards Phil

Unfortunately life got in the way (aka capstone project at school) so my testing has been tabled until semester break. You are correct in that you'll effectively have a nozzle but I haven't had a chance yet to measure any effects on the pressure drop. What I CAN say, is that it provides an increased avg core CFM. We taped off the core in 1" squares and measured the apparent air speed with an anemometer. Core mounting really hurts this. From my understanding the shroud doesn't need to be very far away either (again, not had time to test it quite yet).

Its actually quite painfull, I almost have everything put together to test but I just plain don't have the time

[Phil1988]

I have the problem of deciding whats the best at the moment.

My cooler has the dimensions of 300 *200 mm.

I decidet to take a fan with an diameter of 180mm to make the pressure loss as a minimum. But a 180mm fan would not be neccecarry. So its overdimensioned.

regards Phil

[Jish in China]

Has anyone thought about the value of K, heat transfer coefficient? or the range? need ur help~!thank u in advance!

[Harry Clegg]

I have recently tested a VW Polo radiator, with core dimensions of 480*320*33 and found it to have a coefficient K in the region of 21.

heres the data I used:

Water Inlet = 58.9, Water outlet 51.2, (Giving Delta Tw of 7.7Kelvin) delta T air was 4 K. Mass flow rate of water 8.8Kg per minute. Mass flow rate of air 1.01Kg/second.

Hope this helps!

[jlangholzj]

Originally posted by Jish in China:
Has anyone thought about the value of K, heat transfer coefficient? or the range? need ur help~!thank u in advance!

The actual overall value of your heat transfer coef. "k" of the system persay depends on a lot of different things. Like the Reynolds number through the core, your fin geometry as well as a few other things. The actual heat transfer coef. used IN THE CALCULATIONS should be readily available through any heat transfer textbook.

[BigBoss]

Can anyone shed light on how they decided to apply a Factor of Safety to their sizing/design? What are some common values/ranges? Have Design Judges inquired why your FOS was too high?

[Warpspeed]

Direct mounting the fan onto the core is very bad.

A very significant part of the core gets no forced air help at all, and the small area that does get all the forced air exhibits a higher air pressure drop.

If you have the space to do it, fit a decent air box right over the core, on the down stream side of the core, and fit your extraction fan into that.
Something like this maybe...

The advantage is that you get a very even airflow over the entire core surface area, and higher mass airflow through the fan, because it has less differential pressure to work against.

If the fan blows onto the core, even with an airbox, the flow will still be somewhat uneven unless the air box is enormous.

[BigBoss]

Hi All,

I'm trying to reason through a radiator sizing script, based on BeaverGuy's prescriptions. Say I know the average heat rejection that I would like to dissipate, as well as all the temps, the mass flow rates of the coolant and air, heat capacity.

1.) Ch & Cc; find Cmax & Cmin of the 2; find Cr
2.) Find Qmax
3.) Find effectiveness; Qneeded/Qmax
4.) Find NTU based on Cr & e
5.) Guess a U
6.) NTU = UA/Cmin ; re-arrange so Aguessed = NTU*Cmin/Uguessed

Qoes U & A combination reject enough heat?

1b.) calculate Cc & Ch based on guessed A
2b.) calculate Cr, NTU, e
3b.) If we know Cmin from 1b.), we can find the new Qmax
4b.) From all the above, we can find the new, actual Q rejected
5b.) Check: Does new Q act ~ target Q act? If so, A & U is correct; otherwise iterate.

[BeaverGuy]

Originally posted by BigBoss:
Hi All,

I'm trying to reason through a radiator sizing script, based on BeaverGuy's prescriptions. Say I know the average heat rejection that I would like to dissipate, as well as all the temps, the mass flow rates of the coolant and air, heat capacity.

1.) Ch & Cc; find Cmax & Cmin of the 2; find Cr
2.) Find Qmax
3.) Find effectiveness; Qneeded/Qmax
4.) Find NTU based on Cr & e
5.) Guess a U
6.) NTU = UA/Cmin ; re-arrange so Aguessed = NTU*Cmin/Uguessed

Qoes U & A combination reject enough heat?

1b.) calculate Cc & Ch based on guessed A
2b.) calculate Cr, NTU, e
3b.) If we know Cmin from 1b.), we can find the new Qmax
4b.) From all the above, we can find the new, actual Q rejected
5b.) Check: Does new Q act ~ target Q act? If so, A & U is correct; otherwise iterate.

Wow, people are still referring to what I did a 8 or 9 years ago.

From what I recall, that is the basic method that my sizing script used. The initial guess at U was programmed in because there is a relatively small range of U for an automotive radiator. You will also need a few tube and fin parameters so that you can calculate the air side heat transfer parameters because fins aren't 100% efficient and to calculate the size of the core.

Later on I started using only an evaluation script. I abandoned calculating the core size. I picked tube and fin configurations and core size then just calculated Qact and ensured it was sufficient for the heat load. I did this for two reasons, the code was much shorter thus easier to check, I discovered a number of errors when the code was about a year and a half old, and solved much faster though that was a only a difference of a few seconds versus 1- 2 minutes.

The sizing script had a pair of nested iteration loops. The outer loop Calculated Qact on Uguess and Aguess and compared Qact to Qneed. The inner loop calculated Aguess based on Uguess, Qneed, and the tube and fin geometry. This made it somewhat slow to converge and also unstable in some situations.

[BigBoss]

BeaverGuy,

Hmm, interesting. My script utilized only 1 while loop that encompasses the entirety of the b portion. Although, I am having trouble converging. My U guess would start would say, 200 W/msqrd*K, and the program would ideally change the U based on what Part B would tell me (as well as the A combo).

As for converging, yup. It's sticky as hell. I would change the U initial from 200 to 25, and my Q rejected would change by a very, very small percentage until the program will exhaust itself of a feasible U value. Must be something I'm missing...