we did fea in Solidworks as well as in Ansys, kept the setting same for both the software but the results are varying very much, the location of high stress is coming in the same area. how much variation in result can be expected?
the static study was done with centrifugal loading. All the material's data was inputted the same.
HI Jim,
same geometry, also boundary conditions are kept same. i have attached the mesh details.
IMHO your aspect ratio is way too high - could cause a singularity and give a distorted/false reading.
For SWX I prefer aspect <100; < 3 to be as close to 100 as possible; > 10 to be as low as possible; Jacobian to be as close to 0 as possible. Since you have so few nodes you could refine your mesh in areas of concern (3M is not that many).
What is the comparison of your output between the two packages? If deflection is w/in 10% of each other and there is only discrepancy between stress values then check as I note above.
absolute max principal stress is differing by 20%. deflection values I need to check. what are the ways of keeping aspect ratio within 100?
Anup B. wrote:
absolute max principal stress is differing by 20%. deflection values I need to check. what are the ways of keeping aspect ratio within 100?
Usually local mesh refinement or simplify the geometry to something more IRL.
BTW 20% difference means you are not doing something right or the measurements are not the same, see above.
local mesh refinement has been done in high-stress areas.
<3 is 98.7% and >10 is 0.04%.
0% distorted elements.
Anup B. wrote:
<3 is 98.7% and >10 is 0.04%.
0% distorted elements.
Yes, I saw that, however your highest distorted element was > 100 which indicates a problem area. Find that and it may explain your results.
As Dave Laban suggests, as the mesh is refined it becomes more uniform especially in the areas of greatest distortion (but not always) and the values from one element to the next don't change drastically. This is indicated (usually) by the stress 'leveling' out. Now if it shoots to infinity then you usually have a geometric anomaly (think a shrinking fillet to a sharp corner) which produces a stress-riser. I doubt that is your case.
yes, I will run the simulation by further refining at the distortion areas and will check the results.
have you compared your fea results with Ansys? if yes can you share your experience?
Anup B. wrote:
have you compared your fea results with Ansys? if yes can you share your experience?
If your FEA does not agree across platforms then you are doing your FEA wrong and you need to see which one is closest to hand calculations.
Have you done a mesh sensitivity study to make sure the model is as refined as it needs to be?
mesh sensitivity study, I suppose you mean mesh refinement. Yeah, I have done mesh refinement in high-stress areas.
or is there a way for doing the mesh sensitive study?
It's about getting to a point where as the mesh is refined further, the results don't change. If you've levelled out those curves then the results in each software should correlate well.
For what it's worth, I would regularly get good alignment between SW and Ansys when I was regularly using both.
Attach file here.
Are you certain you are comparing apples-to-apples? I am not certain but many FEA programs provide element averaged values vs. node averaged. If you have distorted elements this can make a huge difference in the values reported.