Limit on SCF

[izax1]

Hi all

I have a kind of philosophical question. Is there an upper limit to a SCF (Stress Concentration Factor)? If you have a perfectly 90 deg corner without any radius (fillet), would the local stress in the corner be infinite?

I have not been able to calculate one with FE analysis. The finer I make the mesh in the corner, the higher, but also more local, the stress.

This is (usually) not a real life problem because: 1. It is not possible to manufacture (machine) a sharp 90 deg corner and 2. your material will be ductile and redistribute the high local peak.

But I think I have encountered such a case with composite material. The process is capable of producing a (almost) sharp corner and the composite material is brittle and will not redistribute the stresses. This design has been proposed due to easiness of manufacturing, (rotational winding of GF) and I have cautioned this design.

I would really welcome your opinion on this matter, before I tell the manufacturing people that they have to find another (and more expensive) method.

Thanks in advance.

 

[SAIL3]

Usually a crack will form before the stress level reaches any meaningful level....depending on the type of loading the ductility and redistribution of the stress may not get an opportunity to occur....ductulity and redistribution is a very useful capabilty of material, but it will not occur if the the design is such that another failure mechanism kicks in before the stress reaches that level....that is one of the reasons that 90deg corners should be avoided....

 

[SLTA]

If I remember from Materials class in college - stress cracking at 90 deg is exactly why all windows in airplanes have rounded corners. It also happens in concrete, which clearly can't get exactly 90 deg.

 

[izax1]

Thank you for your answers, but I didn't feel I got an answer to my question.

I will try to be more precise.
My question is: If you have a perfectly 90 deg corner, will your local (very local) stresses be infnite?

 

[patswfc]

In FE analysis yes, the finer you make your mesh the higher the reportred stress, going to infinity, due to assumptions of FE analysis.
In reality no, as the stress is redistributed.

 

[acrmnsm]

In any analysis FE or hand, the calculation creates a singularity at the 90 corner and is infinite...
I find that to avoid convergence problems I always describe corners with a min of 15 degree angle between one face another, or to put it another way always have a min of 6 elements around any radius.

This link gives a reasonable explanation as to why a 90 degree bend makes a singularity.

http://andreweib.wordpress.com/2010/12/14/stress-singularities/

 

[izax1]

Thank you very much for your answers, patswfc and acrmsnm.

The background for rising this question is that a nearly perfect 90deg corner and almost no redistribution would be the case with composite materials. So I'm left with the Stress Concentration. I think we will have to find another manufaturing process, that does not leave us with a 90 deg corner.