Bonding model (parallel bond)

Submitted by Soo-Min Ham on Fri, 07/24/2020 - 08:12

Hi,

I'd like to modelling the cement bonds by using parallel bonds which consist of a finite-sized beam between the two particles. I do not know the structure of LIGGGHTS system, so here is the light questions.

#1. Is there any method that I can use the codes provided by someone else in github or somewhere? and non-specialist can do that? I think it is a quite difficult work because I have to change from the source code.

#2. I've found the LIGGGHTS-With-Bonds ! To use this package, do I have to re-install LIGGGHTS? And I also want to know that this also work with CFDEM/

#3. I've also found the Aspherix program, successor of LIGGHTS. It seems like to be easier to use and has lots of options, but the ouput file from this program can be used as the input file of CFDEM?

#4. Or is there any method that I can model the parallel bonds?

Thanks in advance!

mschramm | Fri, 07/24/2020 - 18:00

Hello,
I have been working with the bond code and have been making headway in the coupling with CFDEM.
https://github.com/schrummy14/LIGGGHTS_Flexible_Fibers

This is an updated version of LIGGGHTS-WITH-BONDS and is working with LIGGGHTS 3.8.
There are examples on how to use the bond code with a shear cell, uniaxial compression, 3-point-bending, and cantilever beam test.

I am currently using this to run CFDEM simulations in agricultural equipment. The drag models/coupling needs to be adjusted to take
into account that there is a bond but I have gotten reasonable results to some benchmark simulations.

If you have any questions, please let me know.

As to your questions.
1) Both LIGGGHTS-WITH-BONDS and LIGGGHTS_Flexible_Fibers model the same bond forces between atoms based off of D.O.Potyondy P.A.Cundall's 2004 paper. The difference between the codes (other than LIGGGHTS version) is that LIGGGHTS-WITH-BONDS did not implement velocity dependent damping while LIGGGHTS_Flexible_Fibers uses the work of Guo et al. 2013. This damping value can have a drastic effect on the time step. There is a pdf document that goes over the time step associated with bonds.

2) You would need to use that specific version of liggghts. As an example, on my computer I have a sybolic link to liggghts -> the public release of liggghts, and liggghts_bonds -> my fiber implementation in liggghts.

3) I can't answer this one very well. I would assume that this would be the easiest method and that their program already couples well with OpenFOAM. This would also give you access to more sophisticated drag models that will take particle geometry into account (something that I still need to do with the public releases...).

4) If you would like to make your own model, you can edit the bond_gran.cpp -> compute function. This would allow you to make any bond model you would wish for (as long as you do not need anymore history values than what is currently being used....).

Again, if you have any questions, please let me know.