General Mechanical

General Mechanical

mesh convergence study

    • m.caragiuli
      Subscriber
      Hello,nI know it would be better to run a mesh convergence study to assess the accuracy of the results. I have always done it by decreasing the elements' size in the region of higher stress by 1.3 until stress was asymptotic. However, I have noticed that Ansys allows inserting the tool convergence under a solution result for instance the stress. But how does it work? What is the allowable change? If I leave 20% as default I obtain in the tabular value a change of about 38% and stress increases of about 0.10 MPa, that is a high value for my simulation. How can I set the size reduction by 1.3 factor?nMoreover, I know that the structural error tool represents the region of high stress due to mesh inaccuracy, but what does it mean? Is it energy? nThank you for your courtesy.n
    • Govindan Nagappan
      Ansys Employee
      Array nIn details of solution, you can specify Max refinement loopnWith these settings, program will compute the first solution. nThen refine the mesh and compute the second solutionnIf the change in result is less than 20%(allowable change), the solver stops. It will not refine the mesh again.nIf the change in results is more than 20%, it will refine the mesh again and solve again. Solution stops after reaching max refinement loopn========================nYou can scope the result(stress) to region of interest and then insert convergence to the result object. This way only the scoped region is refined. nIf there is stress singularity, and if you refine the region with stress singularity, this will result in high stresses and lead to divergence. ==========================nError plots are used to identify regions where large energy changes occur between adjacent elements. The actual energy value in the legend is of little significance on its own. nExample: highlighted region where the contour colors change quickly from red to blue could benefit from mesh refinemenetn17.6. POST1 - Error Approximation Technique (ansys.com)nn
    • m.caragiuli
      Subscriber
      Thank you, just a couple of questions.nHow can I decide the best setting for the allowable change? Since the stress should be asymptotic I think 20% of change is too much so maybe 1-5% it would be better, is it?nYou said that if there is a singularity stress increases also in case of mesh refinment, in that case do you suggest to improve the mesh with a local sizing and then run again the simulation?nThanks!nn
Viewing 2 reply threads
  • You must be logged in to reply to this topic.