Topic: Joint / Element / Loading generation with loops

Dear Sir / Madame,

I am currently modelling an experimental wind turbine tower structure. The structural form is a truncated cone (horizontal section circle ring, vertical section trapezoid). I would like to perform a parametric study with multiple geometries. As the wind loads are varying in every surface point as a function of height and azimuth angle as well as the temperature loads as a function of the azimuth angle the modelling with multiple geometries with the GiD-Atena workflow would be very time consuming. However the structure can be very easily defined through a dozen geometrical parameters. I thought that maybe a direct definition of the numerical model in a parametric batch input file would be more suitable for the task.

My question is if there is a possibility for the generation of the Joints, Elements and Loading with the help of Loops including predefined parameters and the Evaluate command. I have read the Input File Format and Troubleshooting Documentations but I havent found anything regarding Loops. Or would it be better to prepare an Excel-Sheet with the Joint coordinates, Element joints and Loading on joints.

Thank you for your input in advance.

Re: Joint / Element / Loading generation with loops

Dear Mr. Kovacs,
I guess you have found the options to define a load gradient and also a projection?
Users with similar models found it often enough to define the load with 2 projections (each with a vertical gradient), one for the wind direction (e.g., X) and another in the lateral (e.g., Y; this one needs to be defined differently = with opposite signs - for the left and right parts).

If that is not enough, in GiD, it is possible to use Basic or Tcl/Tk to program your own routines. If you have some minimum programming experience, it should be possible to relatively easily prepare a simple dialog for generating the cone geometry based on a few parameters entered.

The load may need a bit more attention - you may either extend the Load Force for Surface dialog of ATENA, or define your own separate one - I don't know which is the easier, but I am quite sure this is a way to have the loads on the elements at the moment when the ATENA .inp file is being written from GiD. I also expect this to need less time compared to directly creating all finite elements and their loads from a spreadsheet.

If you send us more information about the load types you are interested in, we may also consider adding them into ATENA-GiD. I guess something like "double gradient" (i.e., vertical + depending on a horizontal coordinate) would be enough?

Regards.

Re: Joint / Element / Loading generation with loops

Dear Mr. Pryl,

thank you for your input. At the end, for me at least the most convenient, controllable and robust solution was to define an excel sheet with macros for the generation of a GiD Batch file. I aquired the GiD syntax through recording macros with the basic actions needed, then generating a section of the batch file for GiD with VBA. Then I can read the Batch File from GiD or simply copy-paste the generated spreadsheet with the commands. This way I can control the Node-Line-Surface-Volume ID-s and can select any of them for further commands. The execution in GiD is not fast (takes a minute for the geometry and meshing parameter), but as far my knowledge goes a GiD macro would do the same thing.

The resulting geometric modell for 3 segments out of approx. 20 looks like this:

https://www.dropbox.com/s/cxihupbqfqp9j … t.PNG?dl=0

The resulting numeric modell for 3 segments out of approx. 20 looks like this (link in next posts):

I would like to use quadratic CCIsoBrick elements for the stiffening rings and layered quadratic CCIsoShellBrick elements for the thin plate like sections in between.

As for the loading, for me the plate bending effects on the tower wall are of particular interest. These are present as a result of the temperature gradient through the wall and the wind suction on the sidewalls of the circular cross section as seen in the picture:

link in next posts

The value of "c.p0" is given through a logarithmic function depending on the "Z" coordinate where "Z" is for height above ground as illustrated in the picture:

link in next posts

I have a spreadsheet for the calculation of the wind loads for a specific height and azimuth, so I think with the excel-vba-GiD method I should be able to define a simplified Load Case consisting constant loads for a segmented tower with separate surfaces every 15° azimuth angle and a finite height. The X,Y,Z loads can be calculated and applied to the previously systematically numbered surfaces.

Please let me know if you think I have an error of reasoning somewhere in the process.

Re: Joint / Element / Loading generation with loops

The resulting numeric modell for 3 segments out of approx. 20 looks like this:

https://www.dropbox.com/s/8sdbi10l7phw6 … h.PNG?dl=0

Re: Joint / Element / Loading generation with loops

Wind suction on the sidewalls of the circular cross section as seen in the picture:

https://law.resource.org/pub/eu/eurocod … 070_01.svg

Re: Joint / Element / Loading generation with loops

The value of "c.p0" is given through a logarithmic function depending on the "Z" coordinate where "Z" is for height above ground as illustrated in the picture:

http://www.stadtentwicklung.berlin.de/u … 403e01.gif

Re: Joint / Element / Loading generation with loops

Dear a.kovacs,
sounds you have found a solution which suits you. I have just 2 comments/thoughts:

1. It may not be easy to check your load has been generated correctly (I agree this would be also valid when working with the GiD scripts or Tcl/Tk), although I guess in most cases, the symmetry would be broken when something goes wrong, and that can be checked easily in Postprocessing or even during Runtime.
And you can probably also quickly and easily check if the max. and min. nodal force in your model are the expected.

2. I understand the speed is acceptable (especially compared to analysis times).
While the GiD scripts are of comparable speed with your VBS, and transferring a special load condition into nodal forces at the time of mesh generation would probably take a similar time as your solution, the geometry would probably be simpler, possibly making some operations easier (like mesh refinement/optimization).