Rotating drum

Submitted by riegera on Tue, 05/05/2015 - 23:56

Dear all,

attached to the post you find an input deck with my attempt at creating a rotating drum which is just a cylinder of a fixed diameter and length. In order to achieve wall friction I inserted a fix,namely

fix cyl_wall1 all wall/gran model hertz tangential history primitive type 1 ycylinder .09 0.0 0.0

The computation runs smoothly, but doesn't show what I expect. For those of you with greater experience my error might be obvious, but let me explain my thinking:

As I understand the fix wall/gran command, by my attempt I create a cylindrical wall that is aligned with the y axis at x and z coordinates (0 0). I gave the wall the properties of atomtype 1. I don't see any reason to use an imported stl-mesh for the cylindrical wall, since there is a primitive that doesn't do anything different. To rotate the wall I add the following fix

fix 3 all move rotate 0. 0. 0. 0. 1. 0. ${T_r}

which rotates all together. If I change the fix to

fix 3 cyl_wall1 move rotate 0. 0. 0. 0. 1. 0. ${T_r}

I get an error telling me "ERROR: Could not find fix group ID (../modify.cpp:754)". I assume I cannot apply the fix to rotate the wall to the wall/gran fix as defined on above. Is this in general impossible or are there any syntax errors I am not aware of.

Any help or adivce you can give me, is greatly appreciated.

Best regards,
André

PS: I use the most recent version of LIGGGHTS Public.

AttachmentSize
Plain text icon in.rotdrum.txt2.21 KB
richti83's picture

richti83 | Wed, 05/06/2015 - 14:11

1. I don't see any reason to use an imported stl-mesh for the cylindrical wall, since there is a primitive that doesn't do anything different.
There is a difference between a wall and a particle.
primitive walls can't be moved, the move command you tried is only usable for "atoms" = particles. Offcourse you can make your drum out of particles, search the forum I posted an example somewhere.
To make a moving geomety you have to make a STL drum (you can use paraview->source Cylinder, Save Data as drum.vtk and than remesh it with:

gmsh -2 -o drum.stl drum.vtk

(because PV does not export the drum as STL file because it's internally handled as quads instead of triangles for some stupid reasons).
Than use the code from examples/movingMeshGran/ to import the generated stl file as cad1 and fix move rotate cad1 to rotate the triangles of the stl over time.

Cheers,
Christian.

I'm not an associate of DCS GmbH and not a core developer of LIGGGHTS®
ResearchGate | Contact

riegera | Thu, 05/07/2015 - 10:14

Dear Christian,

thank you for your quick help. As I am quite impatient I already tried all you suggested myself and succeeded. It works really well. I just was not aware that primitives cannot be moved in the same fashion as an stl mesh can.

Maybe you allow for another question or two: I actually work on that rotating drum to simulate mixing of spherical and non-spcherical particles. The latter will either be cylindrical or chip like (flat cuboid). As I understand from reading I can produce these by sticking several spheres together to form what a want.

My problem is that I have actually no clue how to do that. Can you give me a hint or is there any tutorial on that. Also from what I read in the forum and documentation I have to build each individual non-spherical particle myself. So I do not create a template that is than duplicated to my willing, but have to create an appropriate number of particles, have I not?

Any help of yours is greatly appreciated.

Best regards,
André.

richti83's picture

richti83 | Thu, 05/07/2015 - 12:03

Hi Andre,
you need to know that there are two ways of making clumps out of particles:
1st: buy the premium version, you get a new integrator and a fix insert command based on prototype files:
http://www.dcs-computing.com/premium-versions-liggghtsr-and-cfdemrcoupling
I made this simulation with the premium package: https://www.youtube.com/watch?v=Mzo9vdEFtuk
2nd: you use the PUBLIC fix rigid integrator.
Therefore you need to activate MOLECULE Package by make yes-MOLECULE (or copy all files from src/MOLECULE to src) and recompile liggghts.
Than use a hybrid granular molecule pair style and the rigid integrator based on molecules.
After that you create your clumps, either by read_data command (take care that the columns in the data-file match the hybrid pair style, see manual !)

atom_style hybrid granular molecular #needs make yes-MOLECULE !!
...
read_data or create_atoms and than assign mol ids for each rigid body
...
fix integr all rigid molecule
...

I made this simulation using the PUBLIC approach:
https://www.youtube.com/watch?v=W_bRl4wmkY8

For a little donation I can share a full working example with you.

Best,
Christian.

I'm not an associate of DCS GmbH and not a core developer of LIGGGHTS®
ResearchGate | Contact

riegera | Thu, 05/07/2015 - 13:07

Dear Christian,

firstly thanks again.

Regarding the donation I like to write to you in a more private environment. Since the forum has no pn option (Or maybe I just cannot finde it.) is there a way of sending you an e-mail?

Also is it possible with the fix rigid command to simulate particle clumps mixing with spherical particles?

And could I create a hollow sphere by clumping togerther two spheres with the same center point but different diameters (one smaller than the other) and densities? If the inner sphere has a very low or potentially negative mass density (would liggghts allow that), I could create a real hollow sphere, that would not only have the correct density but also correct moment of inertia. To my understanding just correcting the density to a lower value, as suggested several times in the forum, would result in the correct mass but not in the correct moment of inertia, meaning a hollow sphere of density rho has not the same inertia as a full sphere of equal mass (so reduced effective density). But to account for angular momentum and rotational energy correctly, the correct moment of inertia is required and crucial.

Best regards,
André

CarlosRJ | Mon, 03/18/2019 - 21:45

Hi,

I´m new on the software and I like to simulate a rotary drum, but I´m using v3.0, so some code lines are different, so I want some guide to import the stl file, and insert the atoms, inside the drum.

Thanks

richti83's picture

richti83 | Tue, 03/19/2019 - 09:49

Please have a look in examples\LIGGGHTS\Tutorials_public\movingMeshGran and adapt this to your needs. (eg. changing meshfile, use only rotational movement)

I'm not an associate of DCS GmbH and not a core developer of LIGGGHTS®
ResearchGate | Contact

CarlosRJ | Fri, 03/22/2019 - 20:48

Thanks for the help I read about the documetation, and change some codelines, but I have some questions like how can I change the region of insertion, and insert the particles inside the drum, also the stl file must be a surface closed or not to show the lifters. Can I got another file with instrcutions or something else.
Thanks for the support and help
CR