mesh addiction modal analysis

[piersilvio]

hello to all, in wb 12 I launched the modal analysis of a system with elastic linear behavior. I have noticed that the ways of vibrating and the excited masses depend on mesh, but I do not understand why. in particular with some mesh seems that the system vibrates by exciting masses in x directions in the first way and in y in the second way.. then ovviamante the modal answers combine. in other mesh instead the modal response seems characterized by excited masses in the 1st wayfor 50% long x and 50 % long y. In short, think of a plate on 4 piedritti that for alune mesh presentmodi to vibrate long x for the 1st and long y for the second while for other mesh vibri along the diagonal! I honestly don't know how to answer it, but I'm urgent to solve it. Who can give me advice? :confused

 

[Mattia_v]

changing the mesh changes the rigidity of the structure therefore it is obvious that they change the ways of vibrating, you have to see how close the frequencies of the various ways are, if they are very close can happen that changing the mesh the ways you exchange position

 

[piersilvio]

hi, the first two ways are practically identical but perpendicular to each other and directed according to the axes x and y.. or at least must be! vary the mesh happens that they are perpendicular to each other but are oriented according to the bisettrice of the quadrants and therefore inclined 45° to the axes according to which they should be directed. I don't understand why the mesh changes the rigidity of the structure?!? In reality the matrix of rigidities changes, but in proportion the number of knots also changes. There should only be numerical differences in frequency values.. Where am I wrong?
thanks for your interest. .

 

[Mattia_v]

for the structure you can put a photo so we know what we are talking about:biggrin::biggrin: stiffness changes because for example if you have a thin structure and put only one element in the thickness risk of having problems of shear locking (if you do a circle in the internet you should find something), as general rule less knots you put more rigidity because you impose a way of deforming (according to the question of form of the element

 

[cacciatorino]

I think the problem you're dating is normal. You should gradually refine the mesh (magari with submesh in the most intricate corners) until you see that the solution no longer varies from time to time, that is, it tends towards an asymptote.

 

[piersilvio]

Of course, here is my situation: 1) mesh quite thick; 2) mesh not dense but sufficient for the size of the problem;

 

[piersilvio]

In fact I tried in some cases with refinition in the section of only ink, as the model being of the order of meters and solid type generates a large number of g.d.l and make things difficult to solve. thanks for the advice