How to model "fictitious" air convection using the finite element code ANSYS Mechanical

[GuillaumeT]

Good afternoon,

I am currently working on a project involving a "box with holes" and cylindrical packages inside. The aim of the study is to study the heat transfer within these cylindrical packages using ANSYS Mechanical (only).

The box is submitted to radiation as well as to convection. The box has holes on its sides which allow thermal convection to develop inside the box and then to heat the cylindrical packages.

The problem is that I do not know how to model the convection phenomenon using a finite element software.
My idea for this time is to model the "air" as a solid material (using solid elements filling the space in the box) with a fictitious equivalent conduction value taking into account the convection phenomenon (And maybe Radiation phenomenon).

In addition, I would like then to add the radiation effect inside the box, (using the cavities radiation tool of ANSYS Mechanical) but... As you can guess, that's impossible to use if I fill the whole space with a "Solid Air" material. (This second point is less important).

Do you guys have already dealt with such a question ? Should you have any idea about how to deal with that, I would be glad to receive your proposition.

I enclosed to this message an illustration in order to give you an idea of the problem.

I thank you very much by advance for the help.

Guillaume T

Pink - Packages (They are more than two in the reality - The box is totally filled)
Blue - The box (Their is a lid on the box which cover the packages - I removed it for visibility)

 

[IRstuff]

Sounds like you are using a hammer to install a screw...

A conduction model is woefully inadequate for any realistic modeling of convection. Convection is driven by buoyancy difference of gas at different temperatures, which is not possible to correctly model with conduction. If you really need to do convection, you probably should be using a CFD tool.

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

 

[FEA way]

You can use ANSYS Fluent for that. See this example:

https://www.youtube.com/watch?v=9tKLL7P5_t4

Apart from CFD it's a common approach to model air domain with 3D elements. Most FEA programs have special heat transfer elements that can be used to discretize fluid domain in non-CFD thermal analyses. For example in Abaqus they are called convection/diffusion elements.

Finally, you could try applying convective heat flux and radiation boundaru conditions directly to these cylinders but it would be a large simplification.

 

[GuillaumeT]

Dear IRstuff, FEA way,

Thank you very much for your fast replies. Obviously, I know that it would be far efficient using a CFD calculation software to solve this question considering the convection problem which is deeply linked with buoyancy and Navier-Stokes equations... Unfortunately, I am forced to use ANSYS Mechanical to solve this problem and I have to deal with what I have to use.

I will give a look to these special heat transfer elements you talked me about. I think they can maybe help me a lot.
I have already thought about applying a calculated convective heat flux and radiation boundary condition directly to these cylinders and I am currently launching calculations with this hypothesis but as you said, that's a pretty huge simplification, that's why, I am currently looking for some other ways

Thank you both very much for your reply.

Yours faithfully,

Guillaume

 

[IRstuff]

But, your conduction model is similarly a simplification, but worse, in some ways, than the alternative simplifications.

Have you considered something like OpenFoam?

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

 

[GuillaumeT]

Dear IRstuff,

I agree with you... The model I want to set up is not a very nice model...

To put things on their context, I work as a young graduated engineer on a fire safety project. The final aim is to study the temperature elevation on the pink box resulting from a fire agression. In this way, several softwares have been validated at the beginning of the project such as ANSYS (For FEA), FDS (For CFD Fire Simulation, ...). Unfortunately OpenFOAM does not belong to them....

Using OpenFOAM could have been a solution, unfortunately it requires from us to do a qualification file which is difficult regarding the time affected to this project as well as the financial aspect...

However, regarding the stake of the safety aspect, We are working with a "upper approach" - which mean that everything is overvalued (Fire Profile, thermal agression, and finally our hypothesis). Therefore, I think that although these hypothesis are not very correct, they can be sufficient to validate the first phase of this study.

I totally agree with you on everything you are saying, about the hypothesis, the irrelevance of using FEA to model convection... but I have currently no choice, and I base everything on the fact that my hypothesis are overvalued :/

I thank you very much for your help

Guillaume THIRIET