Electrical Enclosure Heating Calculations

[Jieve]

Hello,

We are deploying field equipment in standard (rectangular) electrical enclosures and are trying to get an understanding of heating/cooling requirements/modalities with the aim of keeping the components inside the enclosure at a max of 95F. The enclosure manufacturer gives the equation Qdot = UA(Tenc – Tamb) for heat conduction/convection out of a sealed encosure, with specified effective A values for different mounting types, and U values for different enclosure materials. What we’ve found in lab testing is that these values don’t seem to match what we’re seeing (the enclosure internals are much hotter than predicted), so I’m hoping to get some feedback here on what might be going on (discrepancy between calculations and real life).

To test, we’ve mounted a resistive PTC enclosure heater inside the enclosure near the bottom, and we can monitor heater voltage and current (voltage is controlled by a dimmer switch). Inside, in the upper ¼ of the enclosure, we have a temperature transducer. Externally we have a thermocouple set to measure ambient temperature. We set the heater to different wattage values, allow everything to come to steady state, then record the Qdot (I x V), A (enclosure surface area), Tenc, and Tamb and calculate U. To simulate the wall mount condition, we mount a rubber pad behind the rear enclosure wall, effectively insulating it.

What we find is that U is not constant with constant A; in fact, U seems to follow the reverse trend of what we’d expect to see (the more insulated the enclosure, the higher the U value). Also, the higher the heater wattage, the higher the U value. I was under the impression that U should be a constant. We’re not really understanding why we're getting these results.

Our hypothesis at the moment is that the temperature gradient in the enclosure air and surfaces is affecting the effective enclosure surface area in the calculation. The lower half of the enclosure walls, along with the bottom of the enclosure, are more-or-less at ambient temperature, so don’t really seem to be conducting much if any heat. On the other hand, the upper 1/3rd of the enclosure surface seems to be quite warm, with the top surface likely conducting a majority of the heat. Our presumption is that as we increase the wattage, the more heat would be conducted out over a larger surface area as the enclosure surface temperature gradient from top to bottom becomes larger. However, this of course doesn’t really lend to easy calculation of the internal temperature based on a constant A, U, Q dissipated, which is ultimately what we are looking for.
One thought was for us to use a fan internally to circulate and equalize air inside the enclosure. However, this won’t be the case in the field, so we’re interested in the region of higher temperature in the enclosure.

So my questions are:
1) Is the above equation with overall heat transfer coefficient valid in this case?
2) Do we need to make adjustments in surface area in the calculation based on temp gradient to make this work? I.e. would it maybe make sense to only use the area from the upper half of the enclosure, or upper 1/3rd, etc.?
3) Is our understanding of U correct, that this should be a constant?
4) Are there any suggested modifications we can make to this experiment/calculations that would make them more accurate or useful for being able to predict internal enclosure temps based on the other variables shown?

We can modify experiments using additional temperature sensors, or measuring surface temps as opposed to air temps if needed, but ultimately we’re looking for a model to be able to make accurate predictions based on ambient air temp and enclosure air temp, so not sure how helpful measuring surface temps would be.

Thanks for any input.

 

[itsmoked]

1) Any equation puked-up by enclosure makers is just that: Puke. Disregard. Those are historic BS that fit only theoretically non-realistic boxes put forth by enclosure makers who are asked about temperatures in enclosures.

2) That would make a whole lot more sense than the usual stuff handed out by the clueless enclosure people.

It's as you've discovered, no exchange in the bottom 2/3s of any enclosure. You need a math whiz or a sophisticated simulation program that will take into account the fact that the front and sides do ~80% of the dissipation (owing to vertical convection) and the top does considerably less, (due to it only able to contribute horizontal convection) and that the back, against a surface usually contributing nothing.

The program must integrate over the side surfaces since they are in gradients and then take into account the top and how it will dissipate while surrounded with three-side significant convection surrounding it.

Not for the faint-of-heart.

3) No! U is in no way even close to a constant. Not only that but it changes with every degree of ambient and in-bient. If there are any drafts or any non-convective air movement externally then U is even further removed from expected.

4) Short of developing a model as I discussed above and then living with fudge-factors intelligently applied to those results due to individual site differences, I predict the results would be unreliable.

I typically take enclosures that will be filled with sensitive equipment to be thermal systems requiring thermal designs to support them. That's not a quick application of an equation and away you go. It's modifying the enclosure and the even the enclosure surroundings with detailed installation instructions, based on the heat load inside the enclosure, the annual ambient extremes, and any potential solar input.

In most cases with enclosures inside you can ventilate them. This removes all the issues discussed. If they must remain sealed then you have lots to deal with.

If the sun can ever hit an enclosure it usually will cook anything short of maybe a motor-starter, who's overloads will be wrong often due to temp excursions. Require a sun-shade.

Check this thread:
thread248-340558



Keith Cress
kcress -

http://www.flaminsystems.com

 

[Jieve]

Hi Keith,

this was literally probably one of the more helpful responses I've gotten to an engineering question Thank you for this!

I read the prior thread from 2013, such great information. It looks like in your early days, you went through everything we're going through now in terms of testing and modelling dissipation of heat loads. Glad to see at least that our hypothesis of the effects of the temperature gradient and surface areas corresponded with your experience as well.

I'm interested in the solution you proposed to homogenize the enclosure air; sealing the bottom and sides of the panel, and using forced circulation to circulate air around the panel. Unfortunately the enclosures we build are not that large (used for data acquisition equipment and not motor drives, etc.) , so maybe 2ft H x 2ft W, so the fans would need to be small... also the enclosures are often in pretty harsh environments that require NEMA 4X ratings/to be completely sealed so we can't always ventilate. What size fans were you using to accomplish that circulation? Once you were able to get a more constant air temperature inside, did you find that the exterior surface temps of all sides of the enclosure were more consistent and you had more constant heat transfer across all surfaces?

Short of full FEA modelling, for locations without direct sun exposure (usually the case for us) as a conservative calculation, I'm wondering if we can use a combination of natural convection equations (in still air) for the sides using the upper 1/3rd of the enclosure, then natural convection of the top surface to get somewhere in the ballpark. I'd probably use average temps for the sides (lots of assumptions here) as opposed to integrating (or assume some type of temp distribution and use that for integration), of course this is off the top of my head at the moment so not sure it's feasible... but hopefully this would give us the "intelligent fudge-factor" values we're looking for to make useful design decisions through some sort of usable calculations.

 

[IRstuff]

If you can't ventilate, then you need to basically make your enclosure into some sort of heat exchanger; one could imagine putting fins on both sides of a wall and circulating interior air against the inside set of fins which then transfer heat to the exterior fins.

Note that FEA is simply a highly discretized application of more involved equations; the middle ground would be some sort of lumped model with, say, 10 to 20 nodes or so

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[itsmoked]

Hey Jieve;

Your situation is better than I imagined, which was all sorts of panel sizes and power dissipation variations.

We were specifically focused on an industrial PC based design and so we had a lot of flexibility in how the panel was laid out.
We used standard PC fans. When mounting them to a panel we found they moved a lot more air blowing into the narrow space behind the panel than sucking from that space. These days I'd 3D print a plenum off the panel to get a better more acceptable space for the fan to work in. Doing that you might find suction to be better AND you may be able to pressurize the rear plenum anywhere near the bottom optimized to miss stuff you mount on the panel.

When we sealed the sides and bottom we found the sides front and top of the enclosure to be uniformly warm. That is the measure of your circulation system. That would allow a chance for an equation to better describe the thermal situation.

Using some form of standoff system can get you a lot more heat transfer due to stack effect behind the enclosure where the air is now being rapidly circulated behind the panel. A standard Hoffman/Saginaw type back panel has bent lips at the top and bottom. That's useful at the bottom for sealing but a major PITA at the top where you have to cut it off - a major job. Now days, since laser cutting is so cheap and fast, I'd custom design an aluminum panel including all the hardware mounting and fan holes and have them laser cut by the bushel. Then just add some form of gasketing needed to constrain the airflow. BTW do not use any of the various gasket materials that are like polyurethane or whatever dissolves in about 2 years into powder flakes or you'll likely regret it.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[3DDave]

It would take only a small fan to make a difference, but making a channel to constrain airflow over fins on the inside to force the air to blow over fins will improve that a lot. The limit is on the outside as there would be no fan there, but it's also not trapping heated air.

To the surprise of colleagues on a similar situation the exterior temp will go up. They thought the box would be cooled better but since heat was more effectively removed from the heat generating components inside the box, well, that heat had to go somewhere. For a constrained heat sink I'd go with a small blower than axial fan.

 

[chicopee]

If the equation given to you is properly used, it will account for convection inside and ouside the box and for conduction thru the walls of the box but that equation does not account for the increase in energy for the components within the box and for the walls of the box. My suggestion is to increase the size of the box and allow air flow within the box. Perhaps you may have underestimated the amount of heat input within the box.

 

[Jieve]

Thanks all for the responses.

@itsmoked: Currently we are only testing two enclosure sizes, that may change for the future but we foresee these two sizes currently. Funny, we actually do have an industrial PC mounted inside also. The problem is space for the fans, we'd definitely need to go an enclosure size up if we include this type of solution, which we're hoping not to do (the size increase part). One positive is that we're using Rittal cabinets; the backplates don't have the lips that the hoffman panels have. Also a great idea about the custom panel. Did you ever have issues with fan reliability? Did you put your fan solution in all of your panels from that point forward? Were the fans running constantly or did you put them on a thermostat and only circulate the air when above a certain temperature threshhold?

@chiopee: At steady-state, we're assuming there is no more energy increase for the interior components and walls. Also we're measuring the heat input by measuring the voltage and current input of the resistance heater in the enclosure, so heat input measurement should be pretty accurate.

@3DDave: I like the 3D printed manifold and blower idea ... will think about this some more, it would certainly save space over the axial fan.

 

[itsmoked]

Rittal enclosures? Plastic?

When I was doing that work there were NO industrial PCs, that was what we were doing. Coming up with something that would house a PC in an outdoor enclosure. We eventually included a 1000nit large daylight visible LCD on it. We eventually cut a rectangle out of the back of the enclosure leaving about 1" of perimeter to the back of the enclosure. We then had a custom heatsink cast that was exactly the size of the enclosure 30" x 20". It was 1/2" thick with 1" vertical fins. We ran the ducted forced air behind the panel. We monitored the fan and the internal temperature for OTemp conditions. We found we needed quality fans with ball bearings. We ran the fans above a certain temp.

I think you need to get more creative in working a fan into your panel. I find it hard to imagine a panel that you can't find a couple of square inches to duct a fan up to. The fan could even be 90° to the panel just with an organic swoop to a mounting foot surrounding a port in the panel for easy screw down.

Example fan mount examples I'm referring to.

Something like the red one should get you onto the panel somewhere. You can also use the mini blowers instead of fans.


Keith Cress
kcress -

http://www.flaminsystems.com

 

[3DDave]

An example of closed-box thermal management and the usefulness of air ducting:

https://www.youtube.com/watch?v=R6qnePPCdHg

While it does have an external cooling system, if the air isn't flowing in the box it doesn't do much good.

 

[Jieve]

The rittal enclosures are Wallmount, powdercoated steel.

Great ideas, thanks. Watched that video as well, definitely informative. Will think about a way to design a duct setup and update on the results.

Thanks everyone!