obtain media stress of a panel

[lo85]

Hey, guys.
I wanted to ask a question for those who are more experienced than me in the matter,
How do we switch from the individual values of results for stresses in individual shells to an average value to be used in a verification?
example:
I have a meshato panel with 100 shell281 elements, so I have 100 mx values (moment around the x axis) defined with etable, how would you pass to a single representative average value of the whole panel that will be used as an action to check?
I hope you can help me, I can't find any solution about it.
Thank you.

 

[Onda]

take the maximum value

 

[lo85]

So where do you see the values? in nodal values, or in etables? because the maximum value is always referred to a single element and never to the total and the maximum is a value however ace if seen so.
That's what I want to understand.
which reference system used for shell etables? global xyz with mx,my,mz as local oquelli moments of each shell m11,m22,m12 where axes are 1e2?

 

[Onda]

If the material is homogeneous isotropic I would say that the global or local reference system is indifferent.
If instead the material is for example a laminate (so not isotropic) you have to go to use the moments in the single shell and compare them with the local break characteristics.
basic use is the maximum because for a resistance test the important is not the average value but the relationship between the resistance of the material and the maximum stress, therefore precisely the maximum value obtained from the shell.

in relation to the question whether to use the nodal values or those to the element, depends on whether the system uses a form of local media, which is calculated to the element and not to the node.
If the mesh is good, I would say to use the values to the element, which should however differ very little from the values to the node.

 

[lo85]

but then the mx,my,fx,fy... if you refer to shells or knots, they must be made general, i.e. if you want to make a meter verification, I should take that moment or force value and make it hypothetically agent on all the panel meter I want to verify?
That is, for example a mx on a shell element of 0.1 meters, should be multiplied by 10 so you have a moment on a meter?

 

[Onda]

if I remember well, the results are expressed per unit of length. so to find the voltage you need to divide by the express shell resistance module per unit length.
in practice the unit of measurement of moments is [N*m/m] and that of the resistance module is [m^3/m] in order to transcend from the length of the element. check all this on the manual.
However, the etable should already give you the results in the form of tensions and not moments, removing you from the impetus to go to calculate the resistance modules.

 

[lo85]

Since I thought I would take a direct axial forces and moments instead of tensions, I thought I would do this:

I take on a panel a group of shell elements,and average their value.(if I need an effort at the base of the panel I will take a group of shell elements that are near the base, and the same if I want the cut to the base) only that in the first case I will take n11 or n22 depending on the orientation of the shells and in the case of the cut value n12 (they are etables for the shell281 for example)
after that, since the stress values on the shell elements are linear, multiply for the base length of the panel if I want the average axial effort, while for the height if I want average cutting value on the panel,
I have noticed so, in the hypothesis of a set framed at the base, where therefore I have the reactions to the knots that the values that I obtain are tending to those of the sums of the reactions (preprocessor-->list result-->nodal reaction :fz in the case of axial effort, and fx or fy in the case of cutting), this could give me reason of comparison since the reactions to the nodes at the base however are correct values with which I can I can be. If that makes sense, depending on where they select the elements I would have an average value for that panel area for stresses.
Do you think I'm going to be able to figure it out?

 

[IlNanni]

I take advantage of asking a question.
exists a size/thickness ratio which indicates the limit beyond which it is better to do an analysis with shell geometry to have + reliable results?

Thank you.

 

[lo85]

I don't know sincerely. . .