Material affects on FEA stress results

[Reelman]

Can someone explain to me why changing the material of a mechanica model (only material changes from one run to next, same loads & contraints) changes the stress results? I am thinking stress calculations are based on load and geometry, and that material properties should not come into play in this calculation. What am I missing??

 

[ron_roberts]

First thing that comes to mind are a couple of the material specs, particularly Poissons's ratio and the modulus of elastisity. Hookes Law states that the stress is proportional to the strain and in its generalized form, the strain is a function of the modulus and Poisson's ratio. Is this the case with your materials? Are the modulus and Poisson's ratio different?

 

[Reelman]

yes, it is the modulus and poisson's that changes between the 2 materials. I have done some more research and came to the same conclusion. I guess I was thinking the stress values outputted came from a simple force/cross sectional area relation. This is clearly oversimplified, as I just read that FEA uses material properties (E andpoissons )to determine the stiffness matrix, from which the displacements are calculated.


I also just did a little experiment in Mechanica. Same part, same loading and constraints, same material, but E and poissons were changed from run to run. When the E only is changed, the displacement only is affected in the results plot. When the poissons only is changed, the displacement and stress results are affected. This implies that FEA does not use the E to calculate stress. Does this sound right?

 

[ron_roberts]

In FEAthe modulus of elastisity (E)is definitely used to calculate stress. First off, E brings the correct units (psi) into the picture and it relates stress to strain and vice versa. I would be careful about drawing conlcusions from random changes to related variables. As the saying goes "correlation does not imply causation". Based on your experiment described above I am guessing that you were not varying E enough to see any effect on the displacement, its there, try looking at the displacement in a single direction as opposed to the displacement magnitude.If I were you, I might dust off my mechanics of materials textbook and brush up on Hooke's Law and strain and material relationships.

 

[burnsp]

Changing the modulus E will not change the stress, only the displacement.Simplistically, stress=Mc/I, i.e. the stress is related to the applied force (or moment M) and part geometry (c/I), not the Modulus. The Modulus of the material determines how much that material will displace with an applied load, just as Reelman saw in his experiment. The Modulus E is the stiffness of the material, which is the linear slope on the stress-strain curve.


Hope this helps and clarifies.


Paul

 

[Reelman]

Thanks for the confirmation, burnsp. However, I am still stuck one one thing. If the E value is not used to calculate the stress, do you know the exact formula that the FEA software uses to calculate it?

 

[Reelman]

I think I got it now, after re-reading your post, burns. Since E is the slope of stress v. strain, then changing E only will cause the strain to change accordingly such that the stress does not change.

 

[burnsp]

Yes - think of it this way...two identical beams, one plastic, one steel with an applied load. The max stress is the same in both beams but the plastic beamwill typically deflect much more. Also, the plastic beam will likely fail before the steel beam, not because the stress is higher in the plastic beam, butbecause the plastic beam'stensile strength is much lower than the steel beam's.


Sorry, I don't know how Pro/E calculates the stress. As ron_roberts stated, it may be using the modulus to calculate the stress from themeasured strain. Load creates strain (not stress) and strain is a directindication of the stress in the part, based upon Hooke's Law (E=stress/strain).


Your experiment showed that changing E did not affect the stess but if you change the load, the deflection will increasand subsequently sowill the stress. Perhaps Pro/E calcualtes the stress from the measured strain using E.

 

[Luis Aguirre]

Here are my two cents.



Most Finite element analysis programs use energy methods to find a
solution. What this means is the finite element analysis first finds
displacements from the load applied. Displacements are the only results
that one needs to calculate the rest of the quantities the post
processor provides. What I mean with this is that displacements can be
diffirintiated to calculate strains. Strains then are use to calculate
stress by using the "E".



So the question now remains. Does material properties affect stress?
The answer for most cases should be NO. Stresses are not a function of
material properties unless the stresses are located close to the
constraints or the model is indeterminent. Other wise the stress should
be the same or maybe slightly different do to knumerical error (less
than 1%).





Luis