I am new to thermal management.

I want to lift about 250 W of heat from hot end of thermoelectric cooler/peltier cooler (TEC).

The temperature of hot side of TEC is 30°C.

I want to use some kind of heat exchanger/chiller to accomplish this task.

I am using equation

$$ P = h S (T_s-T_f) $$ where

$P$ = heat to be removed = 250 W

$h$ = heat transfer coefficient

$S$ = area of contact between hot side of TEC & fluid (air/water) used for convective heat transfer

$T_s$ = temperature of hot end

$T_f$ = temperature of fluid (air/water)

I do not know how to calculate $h$. If I calculate that, it will give me temperature of fluid.

How do I compute $h$?

And is my approach correct?

  • $\begingroup$ I might be asking a silly question, but the hot end is 30 C? I think it would take a large heatsink to dissipate 250 watts from such a small temperature difference. I haven't used peltier, etc. coolers myself but I would expect a regular semiconductor device to be a lot hotter. I'd then size a heatsink based on the difference between the desired device temperature and the surrounding air. $\endgroup$ – Andy Oct 6 '16 at 11:03
  • $\begingroup$ You've left out the flow rate of your coolant, for one thing. Are you intending for passive flow, or using a fan/pump? There's a lot more design info required. $\endgroup$ – Carl Witthoft Oct 6 '16 at 13:02
  • $\begingroup$ @CarlWitthoft: I am using water as coolant; and a pump. Let us assume flow rate of water is 4 liters/minute. $\endgroup$ – atom Oct 10 '16 at 3:57

This is a convective heat transfer problem you're asking about. Calculating the coefficient isn't normally possible.

What you can do is read up on stuff like fin geometry on heat sinks, and CFD. The long-term solution here is to derive an equation for your system, something like P = f(X) where X is (for example) the length of a fin. Then, perform a CFD simulation for X = 1 to get P for that particular geometry. Then you can scale your fin until you get the right value for P.

It's get a bit more complicated because you probably shouldn't use dimensioned parameters for something like this. Instead, consider something like the Reynold's number of the system.

Some more reading to help you get started:

| improve this answer | |

Not the answer you're looking for? Browse other questions tagged or ask your own question.