Using Radiation in Natural Convection thermal model

[bblom1]

I am trying to model electronic enclosures from a system level but have some detail at the microchip level as well. I have been running my models without radiation and have been getting decent convergence and have not had any problems. I am now wondering if I should include radiation in my models as well. I made sure that I included a emissivity value for each of my materials that I am using in my model. I noticed that the default fluid air in CF2013 is e=1 is this correct? I am wondering if the emissivity value for air should be lower? I have run a radiation model and compared it to one where I did not use radiation and I am get a temperature delta of about 30deg C and I am wondering if that is unrealistic. I have a model that uses a tabled mounted domain where the walls are adiabatic and the top surface is set to P=0 and a film coeffeicient. I am wondering if I am missing something in my BCs for my radiation model? I was reading in the manual that there is another value called transmissivity, does this need to be included to run a radiation model? I want to make sure my setups are correct to verify I am getting the correct results.

Thanks!

[nmuoio · Accepted solution]

The radiation is calculated only between solids.  Air and all other fluids are not included in the radiation calculations.  The value of emissivity assigned to your air material is used at boundaries that do not include a solid, like external boundaries.

It appears from your description that your model includes an electronics enclosure that is surrounded by an external air volume.  The emissivity of the air is applied to the boundaries of the external volume.  So an emissivity = 1 is probably too high.  This assumes that your enclosure is surrounded by a black body, and all radiation incident to the external boundaries is absorbed by the external boundaries.

The value of emissivity assigned to the external air volume should be appropriate for your specific environment.  For example, if your enclosure is in a room with green painted walls and ceiling, then the emissivity assigned to the air should be the same as the emissivity of the green paint.

The transmissivity determines the amount of radiation that travels through a transparent solid.  If none of your solids are transparent, then transmissivity = 0.

[Royce_adsk · Accepted solution]

Nathan is right on the ball!  From your description I would lower the emissivity value of the air to probably 0.3 and assign a fixed temperature to the 4 walls (same temperature as your film coefficient) of your larger air domain, which is typically refered to as a 'bucket'.

-Royce

[bblom1]

Thanks so much this will help a lot! The current emissivity values I have been using for the air in the bucket compared to a radiation off model have shown a delta in temperatures of around 30 degrees C. Would a 0.3 value for the emissivity value of air inside the enclosure be appropriate? I am also comparing using a bucket approach versus applying film coefficients to the outside walls of the enclosure and not modeling the bucket. I am trying to compare both approaches to see if I can get relatively the same results with and with out radiation.

[Royce_adsk]

It is not advised to have film coefficients on the walls during a radiation analysis and if you mean 'not as a bucket', I would assume that you mean having 5 surfaces with a 0 pressure conditions.  This is also not advise for 2 reasons.  Having that many 0 pressure conditions on a domain of this shape and result expectation can sometime lead to divergences AND radiation is only trasmitted to walls.   Since a 0 pressure is an opening it will not partake in your radiation analysis which will give you some strange temperature results.

In terms of emissivity values, taking a value of 0.3 will probably result in higher temperature than what you have now so it is typically a good go to value to be conservative.  This value can be tricky to define because it is your environment emissivity and will depend on where the electronics will be used.

[bblom1]

I am looking at modeling an electronics enclosure, electronics box, and modeling it in a bucket or attaching film coefficients to the walls of the enclosure instead of modeling the external air volume. I am looking at sealed enclosures and vented enclosures and looking at how the temperature differs between modeling the external air volume or not. I am also interested in looking at how the software handles detailed vents in a bucket vs a FAR and how it handles using a FAR to represent vents without modeling the external air volume. In each case there is an internal air volume but it differes whether it is in contact with the external air volume or not. I have been modeling my enclosures using both approaches, bucket and film coefficients, and I am not looking at the effects of adding radiation to them to acheive greater accuracy.

Thanks!

[smataXBKXP]

Hello, I am trying to verify if my setup is correct. I have a natural convection scenerio with an electronics enclosure and i wanted to include radiation in my simulation. I have tried to read up on everything i could about proper setup and what is going on in the solver when including radiation in the calculation. While my results seem to be correct and better than the simulation excluding radiation, the plum looks strange and this is the only indicator I have right now to make me think something seems wrong. 

Here is my setup: 

1. Air modeled around enclosure, set to variable, custom air material with emissivity value set to 0.3. This air material is located in 2 areas. Air outside system and air inside system.

2. P=0; set at inlet on bottom and outlet at top.

3. T=70C set at inlet

4. gravity is set in the correct direction

5. Advection 5 is applied

6. Mixing lengths is applied

7. I have emissivity values set on each material and have clicked on radiation in the solver. 

Any insight would be helped. Perhaps I have missed something.