I seem to have a stress singularity in these fillet portions of a nozzle to plate intersection which propagates into the stiffener. However, I’m not sure how else I can eliminate this singularity as I have tried using a fillet to remove the sudden geometry change . If the stress scale is adjusted for the yield strength of the material at 260MPa, the area in red is much larger.
I am also using local region refinement in those areas, the global mesh is left fairly course.
I am ultimately trying to understand what the max (converged) stress is e.g if it’s below or above yield, and whether the sections need to increase in thickness to accommodate or some other design changes e.g additional stiffener need to be included.
Any advice on what to do in this instance would be greatly appreciated.
I just wanted to check on you that the definition of the remote force is set up correctly.
Right now the force is acting on the point highlighted here.
Also since in comparison to the mesh sice. the thickness of the plates is pretty thin. Maybe refining the mesh or using 2nd order elements can help with that.
I wanted to add moments as well as forces to each of the nozzle faces but the only option to do that was using the “remote force” I didn’t pick an area so I thought it would just apply the moments at the face I selected. Do you think I should it apply it differently?
Hi Dan_RT.
If you want to use it like that change the reference point to the center of the face, to ensure that the force is applied at the right position.
Thanks for your tips, they worked great for first order mesh but I was curious to see how much the results changed if I used second order mesh. So when I tried it with the second order my results no longer converged and kept increasing with each finer local mesh.
Do you have any tips what I can do achieve convergence? e.g I tried adding a small fillet to the area where the singularity is but it hasn’t helped.
glad that I could help since my suggestion solved your problem I have marked it as a solution to your topic.
On which simulation run do you have the issue with the second-order elements?
Just to update you, I was struggling yesterday with convergence still so I was playing around with the numerics control and I have changed the solver the MUMPS, I believe this may have solved my non-convergence issues but I’m not sure why? I’ll report back if I have any further issues, but do you think that’s a reasonable fix, just seems odd to me that simple change like that would be the fix?
“If the deformable option is used and the number of nodes of the assigned entities is large (>1000), it is advised to use either the MUMPS or PETSC solver instead of Multfront since the performance of Multfront is not optimal for this kind of equations.”
therefore I think that changing to the mumps solver is the right solution.
