Increasing Stress with finer Mesh

- December 20, 2021 at 6:35 am
  
  thomaschi
  Subscriber
  
  Hello!
  I am Real Ansys Beginner and I am wondering why Stresses increase a little with finer Mesh. ist that normal? In the first steps the Stress sinks.
- December 20, 2021 at 6:42 am
  
  thomaschi
  Subscriber
- December 20, 2021 at 7:29 am
  
  Rameez_ul_Haq
  Subscriber
  
  ,so you might be conducting a mesh refinement study of a specific region (or just an element/node) in your model, to observe how the stresses would change with decreasing mesh element size. As you have already observed in your plot, the stresses increase and then kinda become stable 12267 MPa, which indicates that stress is converging in that location. The stresses that you get after decreasing the mesh element size (or in other words, increasing the number of mesh elements) are more accurate than the stresses that you get with higher mesh element size. The stresses not necessarily always increase while conducting a mesh convergence study; they might decrease as well at that location. But what we know is that the stresses do become more and more accurate with mesh refinement.
  If you have had a singularity in that region and you were doing a mesh convergence study for that, then you wouldn't had seen a converged stress after increasing the number of mesh elements in that location. The stresses would keep on diverging with increase in mesh elements, which is a clear cut indication that the stresses are not at all correct there and are a result of an FEA artifact, i.e. a singularity, which will not occur in real world.
- December 21, 2021 at 5:40 am
  
  thomaschi
  Subscriber
  
  Thank you! I did know, that increasing Stresses shows singularities. And that convergence is a good sign for the Results. I just was unsure because I have only red about konvergence and my results are not really mathematically converging, they are swinging about one worth in the last steps. But I think you mean that the last changes of the Result are so small that we can interpret it as converging...?
  
  Thank you very much for your answer above. It helped!
- December 21, 2021 at 7:15 am
  
  Rameez_ul_Haq
  Subscriber
  
  ,yes exactly. I mean it would never converge to a single value unless the mesh size becomes infinitely small. And since the max value of the stress is too high, you can accept it to be converged if the next stresses are changing within, for example, 1%, or even some else percentage, whichever you consider as a limit.
- December 21, 2021 at 7:19 am
  
  thomaschi
  Subscriber
  
  For the deformations i got stranger Results but its converging too. Can you tell me if it is an acceptable Result?
- December 21, 2021 at 7:25 am
  
  Rameez_ul_Haq
  Subscriber
  
  ,doing mesh convergence study would make both, the displacements as well as stresses become more and more accurate. If there is a singularity and you are conducting mesh convergence study on that singularity region, then you should know that the displacements would still converge to a certain value (within a specified range, 1% for example) but the stresses would diverge. So it doesn't matter if that is a singularity or not, the displacements always have to converge and become more and more closer to the accurate values.
  In the graph, you see that the min is 0.6821 and max is 0.68222, right? Do you know how much that has a difference? Only 0.0175%. Thats nothing. Just take any value from the graph of displacement against number of elements, and assume it is not changing at all since the difference is almost negligble.
- December 21, 2021 at 10:12 am
  
  thomaschi
  Subscriber
  
  Thank you again!