Plate element models and stress concentrations

[dktoao]

I know that there is a stress concentration due to sharp corners in solid models, but does the same hold true for plate models? Also, because there is no way to add fillets to a plate model is there a way to correct for this concentration? My supervising engineer tells me that we can just scale the stresses by the extra thickness added to the plate by the fillet. (e.g. the fillet has a radius of 1 in and the plate is also 1 in thick, so we assume that it is adding 1.414 - 1 = 0.414 inches to the plate and scale down the stresses by a factor of 1.0/(1.0+.414) on the elements at this corner. I can see no reason why this is should be correct or representative of the real system. Anyone have any experience with this?

 

[deandm]

I am having a hard time picturing your part. Can you attach a sketch that is representative of it?

Digital Engineering and Manufacturing

http://www.deandm.com

 

[dktoao]

Here is a good example of what I am concerned with.

 

[SWComposites]

that's a shell model? looks like a solid model? Is it a hollow duct or something?

can you post a cross-section drawing in the location of interest?

first thought is that the mesh is way too coarse...

scaling the stresses to account for a fillet is wrong ...

more details needed. what is the purpose of the FEM?

 

[dktoao]

Yes, it is a hollow duct of sorts, more specifically a waveguide. With a more or less rectangular cross section.

I would show a longitudinal cross section but I think that my boss would not like that much due to the propriety of this part.

The purpose of the FEM is to ensure survival under random vibration loading.

All I am really worried about is whether my supervisor's assumption (scaling of the stresses) is right or not. And if not what the correct way to go about it is.

 

[deandm]

To just verify a theory, you may want to make a simpler model. It sounds to me like you want to verify the existance (or non-existance) of stress concentrations in your model.

Can you make a solid model with a filleted step in it and load it in tension? Then make a corresponding model with plates. One section with a different thickness than the other. Then compare the results. See if the ratio of stresses are what your cohort is suggesting.

Digital Engineering and Manufacturing

http://www.deandm.com

 

[rb1957]

i think you've modelled a waveguide as a four sided box, with plate elements of constant thickness. and want to correct your FEA for the internal corner radius (not modelled). the model supports the sides with the thickness of the plates. the extra thickness of the fillet rad would make the corner stiffer in bending and attract more moment. using the fillet thickness (some 40% higher than the nominal) is somewhat reasonable but is probably unconservative (slightly). i imagine that there is some pressure acting on the waveguide ?