Problems of Mesh generation Soil+Foundation+Steel mast tower

[Mek_Greek]

Hi! Nice to meet you to all.
By the way, I am the newest member on this forum right now .
GID + Open Sees is helping me a lot with my master thesis project, so thank you in advance to all the developers and users who made it better.
My topic is: the model is a semi-cylindrical volume of soil homogeneous sand, over the soil there is the semi-conical volume of the superficial concrete foundation of the wind energy mast tower that I modelled with a simple beam element. The structure is symmetric but the forces antisymmetric so the system is antisymmetric.
I am getting troubles meshing the model. I am using:
-Hexahedra elements for both the volumes (foundation + soil) and I tried to mesh as a structured one;
-Linear element and a structered mesh for the beam (mast tower).

What is the proper way according to you to mesh this model?

I do not refer to a specific problem cause I am getting many problems on my mesh because is the first time I do this so I'd rather ask how would you mesh it!
I attached a photo and also the GID model.
Bye! I hope to get you soon!

[escolano]

Hi Mek,
Unfortunately this question must not be asked to GiD developers, but to OpenSees developers
In fact I've never calculated with OpenSees and I don't know its features

Maybe it is not possible to mix for this simulation beam lines and volume elements, it depends on its internals.
e.g. usually in FEM structural simulations beams have 6 degrees of freedom (unknowns) by node: 3 displacements and 3 rotations and solid elements have only 3 displacements,
and are not directly compatible: the lower node of the bar can have its 3 displacements set, because is joined with the solid, but become a mechanism (singular linear system K.x=f without solution)
but maybe can have special formulations to link the rotations of the beam with the displacements of more nodes

Concluding, must ask OpenSees users or read its documentation.

[Mek_Greek]

Ue Escolano, thanks for your nice help!
Ofc you are right! The analysis combining the beam element and the volume elements does not run. I appreciate your comment on it, I have to solve this problem with the OpenSees developers for this kind of element combination.
By the way, I've been able to mesh the model without the tower (beam element), meshing only the volumes in this way: combining structured mesh for the outer 6 faces volume of soil and semi structured volume for the inner core of soil and the foundation volume (it can not be meshed as a structural volume for its geometry).
I post a picture.
Would you have any raccomandation of how to mesh it better and improve it?
The way I meshed the problem seems to be effective, the elastic analysis gives reasonable result.
But I am still doubting, for example the top surface of the inner volume of soil, the one that is the interface between soil and foundation is not meshed radially...
How can I mesh in order to have a radial mesh in that direction?

[escolano]

The first options seem ok,
a radial pattern is a bad option because in the pole is not possible to draw a quadrilateral, it degenerates to a triangle (a prism in your 3D extruded shape)

Note: in theory is possible to mesh a surface with triangles and a radial pattern with
Mesh->Structured->Surfaces->Center structured
(and if the automatic center is not the one you want can set it with Mesh->Structured->Surfaces->Set center)
but this structured-centered option is rarely used, I do not recommend at all.

If do you want, you can divide your geometry in other fully structured pattern, like this.

fully_structured_pattern.png (21.97 KiB) Viewed 15207 times

Extruding it in 3D you can create volumes that can be meshed with hexahedra fully structured (structured in the 3 directions)

And off course, for not so-simple shapes can generate unstructured meshes of tetrahedra

[Mek_Greek]

Hi Doctor Escolano!
I've got your help and it works for the fully structured pattern of a semi-cylindrical shape to mesh.
But there are troubles...
When I impose the structured mesh to all the volume elements it works, but then when I mesh it compares a warning (attachment).
Why? I imposed a structured mesh and it talks about semi ones. it seems no sense...
By the way the volume 19 is the one in your figure is colored grey.

[escolano]

You have not attached the model that complain when meshing

[escolano]

If the volume is topologically the extrusion of a 'master' surface:
-Has a pair of 'opposite' surfaces, with a 1-1 relationship of its point and curves
-And the lateral surfaces have 4 sides and connect pairs of lines
then you can generate a semi-structured volume mesh, setting this type and the number of divisions in the extrusion direction

and then can be meshed with hexahedra (unstructured quadrilateral in the base surface, and structured in the extrusion direction)

By your description is seems that could be your case

And if the master surface is topologically like a square (4 lines and 4 point corners) then also this surface can be meshed structured with nXm divisions on each direction, then the volume could be meshed fully structured with nxmxl divisions

[Mek_Greek]

Sorry , this is the screenshot to attach and show you.

Bye, Marcello

[Mek_Greek]

As I explained in the message, I imposed a fully structured mesh with the geometry you showed me, and then the software answers talking about semi-structured volume.
Why? Seems to be no sense

[escolano]

Are you sure that all lateral surfaces of this volume number 18 are surfaces with 4 lines connecting master-slave pairs of curves of both tops?

Attach a zip with the model.