Meshing 4

[PipingEquipment]

I was wondering how to establish the correct meshing size in FEA. Where should a smaller mesh size be used and how much smaller should it be? Is it always best to use as small of a mesh as possible to get the most accurate results? Any tips or good books/articles on meshing would be appreciated also.

 

[GregLocock]

Th creect mesh size is dependent, among other things, on what you are looking for. For example if you are interested in modal analysis then your element lengths might be 10 times bigger than if you were looking at fatigue analysis. There are many papers, books, and threads on this subject, I expect google will find some.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[marty007]

Agree with GregLocock,

Smaller mesh sizes won't always solve your problem. If you're looking at fatigue in a corner for example, making your mesh smaller and smaller will just force your calculated stress higher and higher. Instead, post-processing may be suitable. Don't always rely on making your mesh smaller to solve your problem, it's all about knowing what problem you're interested in solving, and refining your mesh to suit.

Cheers,
Marty

 

[PipingEquipment]

I'm sorry, should have been more specific, I'm wondering about a static load case stress analysis.

 

[MartinShane]

As Greg and Marty indicated, smaller is not the automatic answer, if for no other reason than the fact that your computer does not have unlimited resources and you have to draw the line somewhere. What you may want to look into is Convergence Analysis - Basically using finer meshes until the results from multiple runs converge on a finite value - There is sure to be an online source that can direct you on how to do this within your specific FEA software. Some (most?) programs will provide tools or functions to do this automatically if you choose, but doing some manually to see whats going on in the background can't hurt. While looking into this you'll probably pick up rules of thumb about how and where to refine the mesh locally in your model in areas that may require a finer mesh in order to get acceptable results. First thing that comes to mind in that regard is areas with high curvature (ie, a filet) will typically require a finer mesh than a simple flat plate.

 

[Worldtraveller]

Your question is rather fundamental, and I would recommend brushing up on the theory/application of FEA in general to answer that. Any first year textbook about FEM would answer the question, especially for the straightforward case of static strength analysis. Keep in mind that, in addition to the possibility of finer meshing around details like a fillet, hole, or sharp corner, many of these things can be adequately accounted for post modeling with hand calculations and some conservative assumptions. FEA is not the answer to everything.

It's also not always the geometry alone that might determine a mesh. Any rapidly changing aspect of your model (like loads) might also require a localized increase mesh density to capture the effects of load/stress gradients.

I found a good (free) online intro to FEA here.

 

[rb1957]

research "mesh convergence".

consider too, FEA goes places that hand calcs couldn't, but things built with "good" hand calcs haven't fallen down, ie hand calcs are not bad (just as all FEA aren't good). this is leading to "why do you want a perfect stress from an imperfect analysis tool ?"

linear FEA is a very useful tool, but it's answers (like hand calcs) need to be used with experience. I don't think you should consider it'll give you answer to be right with any size mesh.

something to consider is how the structure is responding to the loads ... an aircraft fuselage can be reasonably well modelled with quite large elements.

Quando Omni Flunkus Moritati