Dear all
I'm learning CFDEM. I started by analyzing the settling example. Then I modified this case more and more. At the beginning it seemed to work well. Any questions I tried to answer with the LAMMPS documentation and this forum. But now I'm standing in front of a big wall. My main problems are:
1) wall material in DEM: in the documentation i just found that the last number determines the wall material.
e.g.: fix xwalls all wall/gran/hooke 1 0 xplane 0 0.01 1
What kind of wall-types can I use? Where are they specified? How can I modify them?
2) In my simulation the particles penetrate the wall (see attachement). Why? After a few milliseconds there is not any particle left. But I have no openings or permeable walls set. This is inscrutable for me. How can that happen?
3) In CFD I usually use the CFL-Number to estimate the timestep to use. How can I estimate the timestep for DEM?
4) To adjust that the written timesteps from DEM and CFD match I use the formula: timestep(DEM)*dumpintervall(DEM) = writeIntervall(CFD)
Is that correct? It semms so, but eventually I forgot something.
Thanks for your help.
Claudio
| Attachment | Size |
|---|---|
 penetratedWall.png | 6.38 KB |
 cavityForChristoph.tar_.gz | 414.34 KB |
ckloss | Wed, 10/19/2011 - 14:38
>>What kind of wall-types can
>>What kind of wall-types can I use? Where are they specified? How can I modify them?
Materials are defined by means of material types, or "atom types"
See doc for pair/gran for more details on that.
>>2) In my simulation the particles penetrate the wall (see attachement). Why?
Not possible to answer this without further details. I would advise you to start with one of the tutorial cases, setp by step change them and identify the problem by checking after which step the problems start
>>How can I estimate the timestep for DEM?
see doc for fix check timestep/gran
>>timestep(DEM)*dumpintervall(DEM) = writeIntervall(CFD) Is that correct?
yes
Christoph
Claudio Wolfer | Wed, 10/19/2011 - 16:09
Not possible to answer
Hi Christoph
Thanks for your hints. I started changing the settlingTestMPI tutorial step by step to set up my case. But I can not remember after witch modification the penetration problem occured because sometimes I changed more than one thing.
>>Not possible to answer this without further details.
I attached the case in my post above. If you could spend some time checking the files and give me more help I would be very happy. I think it's a small case and with your expertise it wouldn't need a lot of time.
Thanks
Claudio
ckloss | Wed, 10/19/2011 - 16:26
>> because sometimes I changed
>> because sometimes I changed more than one thing.
i would recommend re-building the case just changing one step after another and then make a quick check of the results after each step
>>If you could spend some time checking the files
coupled cases are time consuming to debug so this is not possible. But if you can break down the problem to a pure LIGGGHTS simulation showing the same behavior, I could have a quick look
Christoph