My goal is to determine and/or estimate the inner air temperature of a thin-walled, six-sided, empty, aluminum box with dimensions of 12 inch x 12 inch x 18 inch (304.8 mm x 304.8 mm x 457.2 mm). The box will be situated in a hot desert environment with one side facing the sun for solar heating, and the remaining five sides insulated to minimize heat transfer to the atmosphere. It will essentially be a solar oven.
I imported a STEP file of the box into PrePoMax to conduct a thermal analysis and made what I thought were the required inputs that are listed below. Meshing and everything else went smoothly, but the results are obviously wrong, which reflect a maximum internal temperature of 37.8 degC (100 degF), which is the ambient temperature. I changed a few loads for troubleshooting, and the results are all the same. What am I doing wrong?
Reference Point
Created in the middle of the box
Material
Aluminum
Density (2,770 kg/m^3)
Thermal conductivity (177 W/m*degC)
Specific heat (875 J/kg*degC)
Sections
Solid_Section-1
Initial Conditions
All six sides selected with 37.8 degC (100 degF), which is the ambient temperature
BCs
Tried leaving these alone, and tried inputting 37.8 degC as a troubleshooting step, where the results are all the same
Loads
Radiation (inputs are the same for all six sides)
Surface Flux for front side facing the sun (Side 1) = 950 W/m^2
Surface Flux for remaining five sides = 0 W/m^2 (insulated)
Convective Film for all six sides = 2,096.3583 W/m^2*degC (intended for internal convection; tried 0 W/m^2*degC as a troubleshooting step with the same results)
Then you should model it as a box with aluminum walls and air volume inside (apart from CFD, stationary air can be approximately modeled as a solid in FEA).
Reference points are only used by rigid body constraints, so not needed here.
Make sure that you apply the initial temperature to the whole part (not just outer surfaces) - I assume that’s the intent.
This explicitly enforces that temperature on all boundaries during the whole analysis so not the way to go here.
How did you define it ?
The default condition for a boundary with no prescribed temperature or external flux is q=0 (perfect insulation).
Thank you for the quick and thorough reply. I will research and adjust accordingly. In regard to BCs, what do you recommend? Should I just leave these alone?
Look for articles discussing CubeSat analyses, they use the same concepts, just mostly in vacuum.
The only BC type in thermal analysis is prescribed temperature. So it should be used only when you know the temperature on some boundaries and want to enforce this value there during the whole analysis. Otherwise, use heat fluxes.
I reviewed my inputs based on your feedback and example file and made changes. Here’s a summary:
Created a “box” of air, called the “Air-Cavity”, to reside inside the aluminum box, called the “Hot-Box” via FreeCAD, and imported the resulting STEP file into PrePoMax, so the model now has two parts: the Hot-Box and Air-Cavity
Applied a thermal conductivity of 177 W/m-degC to the aluminum Hot-Box, and 0.0338 W/m-degC to the Air-Cavity via Materials
Initial condition temperature of 37.8 degC applied to both the Hot-Box and Air-Cavity
No BCs applied
Loads: Radiation for Hot-Box only
Cavity radiation = Yes
Selection = Hot-Box selected as an entire part
Sink temperature = 37.8 degC (100 degF)
Emissivity = 0.6 (based on the example file you provided)
Loads: Surface flux
Side facing sun = 950 W/m^2
All other sides = 0 W/m^2
Loads: Convective Film: All sides’ film coefficients = 2 W/m^2*degC (the reason for my original high value was based on a thoughtless calculation error on my part)
The results make a little more sense (max temperature in the center of the Hot-Box side facing sun = 82.3 degC or 180 degF), but the internal Air-Cavity appears to be unaffected, where the resulting temperature is at a uniform 37.8 degF. It’s as if the air wasn’t there, and it’s this internal air temperature estimate that I’m after.
I found no specific material part-selection provision in the Materials inputs, other than my assumption that the name of the material should match the name of the part. Is this correct?
In regard to Sections, I incorporated two sections, one for the Hot-Box and one for the Air-Cavity. The analysis results don’t seem to distinguish between the two sections. Is this normal? Is having two sections incorrect?
Do you see anything else I’m doing wrong, or something that I am not doing?
For internal surfaces ? Cavity radiation means surface to surface radiation so that only the surfaces facing each other can exchange heat this way. For external surfaces, you should use radiation with ambient.
You have to make sure it’s connected with the box. Either by compounding both parts (like in my model) or using tie constraints. Only this or thermal contact can allow it to exchange heat if the meshes are disconnected.
The names can be arbitrary. Materials are assigned via section definitions. So you define material (e.g. Material-1) first, then create a solid section (e.g. Solid section-1) with this material selected and assign it to any parts if you want. Of course, proper naming helps to understand the model.
You probably have the same material selected in both sections. Change it in one section (e.g. for the air volume if it has the box material assigned).
Your recommendations worked! Thank you!! I now have reasonable results. (Ballpark results in my case are OK.) I may tweak the CAD model a bit to determine feature-change effects.
My Hot-Box has a couple of holes in it (to accommodate attached “plumbing”) such that I can see the Air-Cavity effects/results. (I had to adjust the Constraint-Tie settings a little to prevent analysis failures.) For your model, how do you look at the inner core temperatures in the results? I ran an analysis on your model as is, I see the external results, but I don’t know how to see the internal results, assuming, in the first place, that the inside is empty (air-filled).
It seems to me that it should be higher. A hot box that is perfectly insulated on five sides, with thin walls and which is a very good conductor of heat (with almost no temperature gradient across the walls), seems like a one-dimensional problem . Five perfectly insulated sides mean that heat neither enters nor leaves the box because the external surface of the insulator is at the same temperature as the heat sink. All convective and radiative equilibrium should be achieved through the front face, shouldn’t it? In the steady state, it is irrelevant whether there is air, water, vacuum or even a solid block of aluminium inside. If I remove those five faces from the problem to get 146°C.
I don’t understand why it is said Convective Film: “All sides” film coefficients = 2 W/m²*°C’. There is something contradictory or physically incompatible in those boundary conditions, isn’t it?
In the Results tab, you can hide parts. Here, there are two - the inner air volume and the outer box. You can also use view cuts to look inside the model.
But yeah - it’s important to make sure that the BCs and loads are applied correctly and to the proper faces.
If only one face is exposed to solar radiation, then that face should have a proper heat flux while the other faces are only exposed to atmosphere. Even if they have some insulating layers, you can model them and apply convective film to their outer faces. Otherwise, with 5 perfectly insulated faces and one loaded face, there will be no actual exit for the heat.
As I’ve mentioned, you may want to check the CubeSat models. They often have one side exposed to sun. Of course, the other sides use only radiation to ambient. Plus, there can be some cavity radiation inside.
And consider running a transient analysis too. This can help you see how the model heats up.
Thanks for the feedback. Your conclusions are correct, but my description was admittedly incomplete. The Hot Box has two holes in it to accommodate plumbing, so the air has inlet and exit paths. The convective film coefficients are applied to the six inside surfaces to account for any free and forced convection mechanisms.
I plan to conduct an analysis run with a completely closed box minus the holes and convection film coefficients, so I’ll be checking for the higher temperatures as you described.
Keep in mind that free convection will generally have a lower coefficient. And that you can only model it approximately without CFD. This whole approach with air modeled as a solid is just a simplification assuming no flow phenomena.
Also, even without the holes, if some walls have insulation layers that are exposed to air on the outside then there should be some convection there too. It’s like with double glazed windows also having an air gap inside:
Yes, I agree. But for what I am doing, ballpark estimates are OK for the time being. Any CFD work would come later for design optimization ($). Thank you for your help and insight.
You could use OpenFOAM for that - FreeCAD has add-on workbench for this solver and PrePoMax can import results (non-uniformly distributed pressure and with some modifications also temperatures) from external CFD simulations. CalculiX can be even coupled with OpenFOAM via preCICE for FSI or CHT.
But I agree that CFD can often be skipped, especially for the initial design evaluation. However, you have to keep in mind the inaccuracies due to highly approximate HTCs (maybe you could estimate them more accurately based on the literature because the tables provided above have very wide ranges).
