search

BriefFiniteElementNet / BriefFiniteElement.Net

BriefFiniteElementDotNET (BFE.NET) is a library for linear-static Finite Element Method (FEM) analysis of solids and structures in .NET
GNU Lesser General Public License v3.0
154 stars 57 forks source link

Too much error in TriangleElement Stress Validation #84

Open epsi1on opened 3 years ago

epsi1on commented 3 years ago

This is a comparison between BFE output and Abaqus output for a model which only contains triangle element. Take a look for more info:

https://github.com/BriefFiniteElementNet/BriefFiniteElement.Net/blob/master/BriefFiniteElementNet.Validation/Data/FlatShell%20-%20triangle/IBeamTorsion/Intro.md

To understand why there is such error, we must first know how abaqus handles calculations for this element and unfortunately i do not know what is the element used in this Abaqus model. Do you guys know which element is used in this Abaqus model? I think it is STRI3 element.

I'll try to find the abaqus input file for this validation case, then will continue finding the problem.

Thanks...

epsi1on commented 3 years ago

Hi @rubsy92,

I did some changes in BriefFiniteElementNet.Validation project. Validation of clamped plate is now inside this directory: BriefFiniteElement.Net/BriefFiniteElementNet.Validation/Case_02/

I'm doing some edits to make it report to a HTML file instead of csv file but I need element stress at integration points which are not inside the CSV file you included. Can you please also add the rpt file to that folder? Thanks...

Also if not hard, please add following items to folder (like this one)

Thanks

rubsy92 commented 3 years ago

Hi @epsi1on,

The aforementioned link (https://github.com/BriefFiniteElementNet/BriefFiniteElement.Net/blob/master/BriefFiniteElementNet.Validation/Data/FlatShell%20-%20triangle/IBeamTorsion/Intro.md) is not working but I know the I beam example.

Regarding the required element type: you need to use the STRI3-element of Abaqus if you want to compare the results with BFE. The default S3(R) element of Abaqus can be used for both thin and thick shells and I think Abaqus does some optimization in the background. Furthermore the (R) version uses reduced integration; which is definitely not the case for the triangular element in BFE.

Regarding the I beam results: I performed that simulation some months ago and as far as I can remember I got consistent results with Abaqus (STRI3 element). Comparing the results must be done manually since the element numbering is not the same and this process is very time consuming… I’ll look into it later on.

Regarding the clamped plate: • Do you want all stress components or just the von Mises? I think I deleted the rpt file, but I’ll look into it (the data is on my other PC). • What do you mean with input .odt file? The input file of Abaqus has an .inp extension and the output file is an .ODB file. • I provided an image of the model in the original Excel file, but you requested a .csv file; which removed the image. I’ll look in my backups.

epsi1on commented 3 years ago

Regarding the I beam results: I performed that simulation some months ago and as far as I can remember I got consistent results with Abaqus (STRI3 element). Comparing the results must be done manually since the element numbering is not the same and this process is very time consuming… I’ll look into it later on.

Maybe no need for all nodes. I have seen some validations in thesis PDF which only compares displacement values of 3 nodes and elements, not all nodes. As you said this is time consuming to reorder nodes.

• Do you want all stress components or just the von Mises? I think I deleted the rpt file, but I’ll look into it (the data is on my other PC).

Preferably both, or at least stress components.

• What do you mean with input .odt file? The input file of Abaqus has an .inp extension and the output file is an .ODB file.

Sorry my mistake.

• I provided an image of the model in the original Excel file, but you requested a .csv file; which removed the image. I’ll look in my backups.

Thanks, I did recover image file from .xlsx file and added to folder. No need to recover from backups.

Just for your info, as we are going to release nuget package, I need to automate the execution of validations and take a look at result, this is why i asked you the stress components. This is related project : Project#4. Also suggestions are highly welcomed. Thanks

rubsy92 commented 3 years ago

Hi @epsi1on,

I was validating the I-beam and the values seem to be okay; however we need to agree on the probe location for the stresses. The "envelope" function I wrote earlier doesn't always seem to provide the same values as Abaqus. However, the stresses for the bottom/top look good. This thus goes hand in hand with the shell face orientation. I think we need to discuss how we define the top or the bottom of a shell and visualize it? Do we have the normal of the face available?

Maybe I can already do an export of the stresses for all components (top/bottom/envelope) separately?

epsi1on commented 3 years ago

Fortunately I've just found the ABAQUS model files from my friend... Do you know which files i should place inside the validation folder? And which file is temporary and can be deleted?

image

rubsy92 commented 3 years ago

Hi, basically you need the following files: • .cae • .inp • .odb • .msg • .dat

So how would you like to compare the stresses? Do we check top/bottom separately?

epsi1on commented 3 years ago

• .cae • .inp • .odb • .msg • .dat

OK, included these file.

So how would you like to compare the stresses? Do we check top/bottom separately?

what you do suggest? i think comparison of a top and button stresses of a few elements (like 2 or 3) is enough

rubsy92 commented 3 years ago

Hi,

I created a pull request with a comparison for top/bottom for the flat plate. Do note that there were some issues with my branch and as such I think I might have deleted the following files:

Could you check this?

Also could you add the CAE/ODB file of the I beam? Then I can do the same comparison for that example.

Thanks!

epsi1on commented 3 years ago

Hi,

Hi,

I created a pull request with a comparison for top/bottom for the flat plate. Do note that there were some issues with my branch and as such I think I might have deleted the following files:

* Case 02\Abaqus comparison\Elements.png

* Case 02\Abaqus comparison\Nodes.png

* Case 02\Abaqus comparison\Validation_Elements.png

* Case 03\kibnrzi2.bmp

Could you check this?

Files are already there. That is OK.

Also could you add the CAE/ODB file of the I beam? Then I can do the same comparison for that example.

Added files to BriefFiniteElement.Net/BriefFiniteElementNet.Validation/Case_01/files/

Thanks