Thermal Analysis of Electrical box 1

[Paosy27]

I am a mechanical engineer (not the strongest thermo background) and have been tasked with solving a heat transfer problem. I have a junction box with electrical devices inside that produce 30W of heat. The box's surface area is 0.26m. The box is made out of steel. The ambient temperature is 35C. The electrical devices are near the bottom of the box and I want to find the surface temperature of the box.

Correct me if I am wrong, but I am right to neglect radiation in this instance. Also, would a resistance network be the best route to solve this problem since radiation is neglected?

Any help would be great!

Mike

 

[IRstuff]

"but I am right to neglect radiation in this instance"

That would depend on the area of the devices, but, in general, you ought not ignore radiation, particularly in a enclosed space. You've not given sufficient information about the size of the heat source, so the interior temperature might be extremely toasty, literally.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[DRWeig]

I'm an EE who gets jammed into thermo too. I use the free 2-D heat transfer program (THERM6) from LBNL It's quick and easy to learn. As long as you understand the basics of heat transfer, you'll be fine.

Now, your problem will be setting the boundary conditions. Will your box be next to a smelting furnace, or will it be outside at Vostok Station in Antarctica? You'll have to make some assumptions or else run the analysis many times. That's pretty easy to do, though.

It'll do your finite-element or resistance-network thing for you, and it accounts for internal radiation between members that have calculated or defined surfaces. It'll give you two planes of analysis, which should be plenty. If concerned, or if your box is assymetrical, spin it 90 degrees and solve it again, then mesh your results.

http://windows.lbl.gov/software/therm/therm.html

Get the architectural materials database with it. You can then easily model example walls to which your box might be mounted. The architectural package also has a better array of metals in it. Don't forget the manual.



Best to you,

Goober Dave

Haven't see the forum policies? Do so now: Forum Policies

 

[Paosy27]

DrWeig:
I will check out the software! Thanks for the tip. The boundary conditions are such that it will in a climate/room that will be roughly 35C and mounted on a wall.

IRstuff:
"in general, you ought not ignore radiation, particularly in a enclosed space."
Why is this? Wouldn't it be mainly convective inside of this box?

Thanks,

Mike

 

[IRstuff]

If the convection is low, then the temperature will be high, and an increased portion of the power will be radiated. A 4-in cube at 80C will exchange a net of 5W into the surrounding environment. Radiation should always be verified for significance.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[Paosy27]

Ok that makes sense.

So I can assume these electrical devices will be radiating power since the box is enclosed? I think I have enough information to get me going.

 

[DRWeig]

Yes, they'll be radiating. The walls of the box closest to the device will be warmest by a surprising amount, in spite of the metal's high conductivity.

Best to you,

Goober Dave

Haven't see the forum policies? Do so now: Forum Policies

 

[DRWeig]

One last thought -

Your box is a kind of small and inexpensive. I would build one up and measure the surface temperatures - either to verify my calculations or as a substitute for calculating. 35°C ambient shouldn't be too hard to produce. If it is, run your test at 25°C measure temperature rise over ambient. It'll be close to the same, as long as your 30W load stays constant.



Best to you,

Goober Dave

Haven't see the forum policies? Do so now: Forum Policies

 

[LiteYear]

DRWeig, I'm very interested in your use of THERM to model enclosures. It seems the OP's question is a common task in general engineering and I've never found a reliable way to solve it. In the past I've tended to use a combination of the online calculators that various enclosure manufacturers publish on their websites, or find the thermal conductivity of the material and plug it into the extremely simplified equation: H = -k*A*dT/dx. With a given wall thickness, surface area and heat source, I get a ball park estimate for delta T between outside and inside. Where possible, a quick experiment can firm up the data, but it still doesn't provide much for extrapolating to direct sun exposure scenarios, for example.

I've had a bit of a play with THERM and with the libraries you mention. One question I have, and perhaps it will reveal a greater misunderstanding, is how do you model the heat source? Or is the idea not to enclosure your model and instead model just one half and take for example, the left as outside and the right as inside?

Any further resources on applying therm to electronics enclosures that you know of?

 

[DRWeig]

Hi LiteYear,

I tend to use the half-n-half method you described for pieces that are fairly symmetrical, with single heat sources or at least grouped heat sources.

For stranger things, I'll put little convoluted dents in the side of the panel on which the heat sources will mount.

OK, I just typed a page of description of the two methods I use. I promptly deleted them after proofreading in preview mode. I need pictures! I'll dig out a couple projects and post them later (might be a couple days) with some annotation.

Best to you,

Goober Dave

Haven't see the forum policies? Do so now: Forum Policies

 

[chicopee]

For an electrical box, I would forgo the calculations because whatever you get for results, you will be lucky to get within 50% of the actual value. A better and faster solution is to get a pyrometer for surface temperatures. Such instrument is great to find out which electrical fixture, junction box, circuit breaker, motor, etc... is overheating.