problems with beam normals 1

[machstudent]

Hi there,
I am working on a paper for my university degree.
therefore I have to build a Model in Abaqus using truss and shell elements.
As an exercise I aam working on the Crane model in the Abaqus tutorial.
Now I am experiencing difficulties with the beam normals. In the tutorial it is requested first to define the beam section orientation by using the n1-vector orientation. After that, there is a short chapter about the beam normals which I do not understand:

"In this model you will have a modeling error if you provide data that only define the orientation of the approximate -vector. Unless overridden, the averaging of beam normals (see “Beam element curvature,” Section 6.1.3) causes Abaqus to use incorrect geometry for the cargo crane model. To see this, you can use the Visualization module to display the beam section axes and beam tangent vectors (see “Postprocessing,” Section 6.4.2). Without any further modification to the beam normal directions, the normals in the crane model would appear to be correct in the Visualization module; yet, they would be, in fact, slightly incorrect."

Here comes my question: If I do define the n1-vektor to define the beam orientation, why can there be still a problem with the beam normals? the tutorial only says, that this problem can be solved by manually editing the keywords with inscrutable data.
Can someone please help with this? I am getting more and more desperate, as the abaqus manual does not help me.
Thanks!

 

[xerf]

I suspect the issue mentioned in the paragraph you cited is the default normal averaging at nodes. I think Abaqus tries to (automatically) establish the normal at nodes by using the orientation of beam elements specified by the user and in certain cases it does some averaging....

 

[machstudent]

Thank you xerf. But I still don't get it.

The steps are as follows:

- defining the "beam section orientation" (which is done by defining the n1-vector)

- generating sets in the assembly which contain certain beams

- editing the normals of the beams (again?) in the keyword editor

So the nodes' normals aren't touched...?

I've attached the tutorial to this post. You don't have to read the hole thing, but it might help to understand the issue reading paragraphs "Beam normals" and "Using the Keywords Editor and defining the job". "Creating an assembly" explains how to generate the sets used in "Using the Keywords Editor and defining the job".

Thanks a lot!

 

[machstudent]

Sorry forgot the attachment...

 

[xerf]

I might be wrong but I think your confusion relates to the following:

Abaqus consists of 3 main -standalone- programs:
Abaqus/CAE - pre/post processor (for creating models - more like a CAD software with meshing capabilities and for visualizing results.)
Abaqus/Standard - implicit analysis code
Abaqus/Explicit - explicit analysis code

When you submit a job , Abaqus/CAE will create and input file (.inp) and call either Standard or Explicit (basically depending on the type of Steps in your model.)

Input file is a text file with keywords. You do not need /CAE to generate the input file, you can write it by hands. See the Abaqus Keywords Manual for reference.

Initially, this was the only way to create FE models. Then /CAE came along. /CAE will simply generate the .inp for you. However, /CAE does not have support for all keywords. In this case, you can use the Keywords editor to add or modify the keywords not directly supported by /CAE.

I suspect the intention of this tutorial problem was to illustrate how the Keyword editor can be used.

The input file can be used independently of /CAE. You can submit the job to one of the solver product from the command prompt using: abaqus -job <job_name> -input <input_file_name>.

This will run input_file_name.inp and create job_name.odb which you can open in /CAE to visualize results.

Best.