Heat transfer and fluid flow analysis using FEA vs FVM CFD 1

[Tunalover]

Guys:
I just need a sanity check on these assertions. Most FEA codes claiming prowess in heat transfer analysis of problems with convection (e.g. ANSYS Workbench) require you to assign convection coefficients (h) to all wetted surfaces. But to know what "h" values to use is almost always a WAG at best? And the convection coefficient h, especially over large surfaces, is not at all constant?

Don't CFD codes based on the Finite Volume Method (where convection coefficients are not even mentioned because the domain includes the fluid) do away with that nasty guesswork and error created by choosing what "h's" to use? Does it then make sense that such FEA codes should NOT be used for heat transfer problems with convection? It seems to me that these FEA programs if used to solve problems with convection will be giving guesses for results since the convection coefficient(s) are guesses from the get go? Then is it safe to say that these FEA codes should only be used for problems without convection?

Just wondering because I've always believed this stuff but never had an educated audience (like here) provide validation. TIA for your remarks.




Tunalover
Electro-Mechanical Product Development
UMD 1984
UCF 1993

 

[IRstuff]

Depends on what you are trying to do, doesn't it. CFD codes don't do conduction, which means that anything inside a surface isn't even being simulated.

So, which one provides a more complete solution, if that's what you need, vs. one that is only interested in the fluid domain?

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

 

[L_K]

CFD codes inside Ansys Workbench (CFX, Fluent) can also model solids (conduction).
Heat transfer coefficient between solid/fluid surfaces can be outputted and used
in subsequent finite element modelling.

Heat transfer coefficients used in FEA are empirical values, which in many
cases yield good enough, conservative results. A good practice could
be to do a sensitivity analysis (how changing h affects the results
you are interested in, for example max temp).

If FEA model with WAG h is good enough, is a lot dependent on
what you are trying to achieve.

 

[Tunalover]

IRStuff,
OK the CFD codes I've dealt with (IcePak, FloTherm, FloEFD) all target electronics applications so they actually include conduction. I didn't stop to think that maybe there are still CFD codes out there that do not. Can you provide examples?

L_K,
Regarding ANSYS Workbench, my experience with WB goes way back to Rev. 13. At that time these programs were not part of WB. This is news to me since my only recent dealings with CFD codes was with FloEFD and I haven't looked at WB since Rev. 13. However I cannot imagine when a WAG would suffice in any situation.

Thanks guys!


Tunalover
Electro-Mechanical Product Development
UMD 1984
UCF 1993

 

[L_K]

I should have written empirical correlation instead of WAG in my text.
Both CFX and Fluent are part of Workbench but require an additional license.

 

[Tunalover]

L_K
When I used Fluent in the 90's, it was a FVM program. Is it still? It would be odd to have a FVM program embedded in an FEA program like Workbench!

Tunalover
Electro-Mechanical Product Development
UMD 1984
UCF 1993

 

[L_K]

Ansys Workbench is the name for a suite of programs including structural analysis, fluid dynamics and electronics.

Ansys Mechanical (which is based on finite element formulation) is used for example structural and heat transfer analysis.

So Fluent is not embedded in Ansys Mechanical. Both being part of the Workbench suite enables easy coupling between
them (fluid-structure interaction).