assigning properties to shell elements and offsets

[kontiki]

Bye to all,

are a beginner in the use of patran and in general of fem analysis.
as from title, I write here to ask if someone can clarify these two (for now) doubts:


1) the first is on the assignment of properties to the elements, and more generally on the links between: geometry - elements - properties: reading the manual, tutorials, google, etc... I was unable to understand if the properties (for example, for a beam element, the section, the offset, the orientation, the material, etc.) can only be assigned to a geometry or if it is possible also to assign them directly to elements.
I try to explain: I have a line (geometria) that represents a beam: I create a set of properties and attach them to them and then create the mesh: at that point the elements that are created (and which should be associated with geometry), if for example use 3d display, have section and offset set in the properties.
Now, if I try to copy (translating according to patran), mirror, etc... only the elements (and not the geomentary), the new elements that represent a beam identical to the first, do not have (or are not shown with) the same section of the first but are line traits (they are / modimensional appearance). I wondered if:
a- the new elements actually have properties and is it only a fact of refresh/redraw? But I tried to give the relative commands but I continue to see the original elements with the correct section, while the new ones always appear undimensional;
b- is it possible to assign the properties directly to the elements? It would seem to me the most sensible thing (even because I think we can also create directly "manually" individual elements where necessary for meshare more accurately) but in the form for the selection of ownership, if I click on an element, this is not selected, while the selection only works on geometries...in fact I have seen that there is also the selection option of regions: trying it I noticed that such mode selects any type of object in the graphic window, but does not add anything in the form (in fact I have not yet been able to understand what this mode is for);
c- if I want to reproduce identical parts of the structure I have to translate, mirror, etc. also geometries and then either repeat the same steps (application of the properties + mesh) or associate to these any translated elements, mirrored, etc and then apply the properties to new geometries? ....

Ultimately, what would be the best (faster, more effective) method to use if parts of the structure that are repeated identically are present in the model to be prepared?

2) the second macrodubbio is on how to use (and if to use) shell offsets: if for example model a beam a beam (to "l" for example) with surfaces that will then become shell elements, it is necessary, so that the analysis is correct, that the elements, once displayed in 3d, and then with the thickness assigned in the properties, reproduce "faithfully" the geometry of the beam section? because if for example I create the beam with two surfaces arranged at 90 degrees each other that are touched along a edge (the edge at the bottom at the left of the "l") and then check the properties (the thickness in particular) and the mesho, when the view in 3d, since the thickness is symmetrically assigned to the surface, the edge of the "l" appears on the other side from a triangle. so if the program used such geometry is as if the section I am using for the analysis was different from what I intend to study (addition without a portion of matter in a part of the edge).
if instead, for example, I create the beam to "l" extruding a modimensional curve to l, obtaining a single surface, when I apply the properties and visualize the elements in 3d the section appears tapered to the edge: that is, the thickness is symmetrically given perpendicular to the starting curve, so where the curve used to create the surface has a angle of 90 degrees, there the thickness is given regarding the angle bisettrice, then the section is tapered near the edge.

Then I asked myself: whereby using shell elements for beam modeling, the aspects highlighted are important and therefore different surfaces should be used with relative offsets to obtain a three-dimensional geometry of the faithful geometry that you want to study or is not necessary and therefore the offset, as in the case of beam elements, is to be used only to align/disalign the loads with/from the center of the sections?

Thank you!
Faithful

 

[kontiki]

bhè...the first point I solved it....in fact I saw that you can select the elements and apply directly to these the desired properties, after selection.... of the selector...!!!!...I saw that patran allows deafault to select only the geometries when you want to apply properties.
on the edge of the property form, however, appears a series of choices to decide on which elements the graphic selection can operate. choosing to select elements and not geometries you can apply to the first properties....

 

[Onda]

hi federico, I see to answer your point n.2 in a very short way
When you do an analysis, it's important to go through a series of approximations. What you're noticing is one of these.
It is up to you to assess how much this approximation may or may not affect the result. As you say, where you put the offset the beams are shaped with missing areas and overlapping areas. but this is a feature of approximation of solids with shell elements.
You could just take a cad software, and calculate the real inertia, and then the inertia of your representation. In this way you value how much error you are introducing. normally, for thin thick shell, the low error. If the thickness was remarkable and the required high precision, you would better shape solid. Otherwise you find the axis that minimizes the error and uses an offset on this plane, which is not said to be the average plan, often it is not.
It happened to me, for example, that I had to model a thick composite tube. being in composite I modeled them in shell but to avoid running into the error you showed I moved the average line of representation so that the inertia of the missing material over the shell line was compensated by the inertia of the additional material under the shell line.

Keep in mind that depending on the analysis you can or may not make these variations.
to solve a static you can use any offset, but for a buckling the previous versions of nastran required the model to be made on the average line, otherwise the results would not have been valid.