Stress near constraint 1

[ahad29]

I have a beam with circular cross-section. The beam has a taper at one end. The taper-end fits into a fixture with threads and keys to restrain it from translating and rotating. In my finite element model I have fixed the taper end in all DOF. The load is applied at the other end transverse to the axis of the shaft. But the problem is that I am getting very high stresses right on the curve where my constraint ends. I compared it with simple hand calculation and it is way off. What is the right way to transition form a "total constraint" to no constraint. And how do you deal with this problem.

ahad

 

[johnhors]

Ahad

Your FE model is a cantilever where one end is free and the other end fully fixed. A cantilever is a mathematical concept, i.e. it does not exist in the real world, everything has "give" in it, nothing can be completely restrained. An infinitely stiff object is a mathematical singularity that your FE analysis is trying to model, hence you see very high and meaningless stresses at the restraint.

What you can do is mimic the "give" by using springs to earth instead of rigid supports. But, the difficulty there is knowing what spring stiffnesses to use, you could try some simple calculation based on the supporting structure size or experiment and "eyeball" your deflections.

Alternatively, and usually the best solution is to apply a balancing set of forces and moments at your "fixed" end, and use minimal supports. This will totally eliminate your nonsense stresses and give you a very clean solution.

 

[ahad29]

Johnhors,
Thanks for your input. I tried to prove that the stress i was getting at that location is not realistic by moving the constrtaint a little away from the current locatin and then observed the stress at that location. Now I don't see the sudden change in the stress gradient but at the same time I realize that the problem is not fully solved yet. I need to research a little about what you said i.e. "..apply a balancing set of forces and moments.." My loading condition has 3 forces and 3 moments each in along an axis of the global co-ord system.

ahad

 

[corus]

johnhors is really referring to measurements in comparison to FE results where the measured results may be less. In comparison to a hand calculation, which ahad has done, there should be agreement. I suspect that the circular cross section may be giving you a stress concentration effect at the restraint where the circular section is prevented from deforming into an ellipsoid. Try plotting the results up to the high stress and see if your hand calculation agrees with the FE if you take out the stress concentration effect. If not then your hand calculation is probably in error.

corus