Incompatible mesh restart file

Submitted by mojgan on Sun, 02/10/2013 - 13:59

Hi,
I am modeling a test in three steps. first I generate some paricles and let them settle into a box without top plate. after that I make a restart file. then in the second step, by reading this restart file and placing a plate on top of the sample and by a fix move/mesh with a linear velocity in z direction, a prescribed normal stress is applied to the plate. then writing a restart file, lets further simulation by different boundary conditions. all of the walls are modeled as mesh.reading the restart file of step 1 in step 2 script does not make any error. but using the step 2 restart file in step 3, leads in the following log-file:

LIGGGHTS (Version LIGGGHTS-PUBLIC 2.2.4, compiled 2013-02-07-01:59:45 by mojgan based on LAMMPS 20 Apr 2012)

echo both
atom_style granular
newton off
communicate single vel yes

neighbor 0.002 bin
neigh_modify delay 0

read_restart restart.loading.*
orthogonal box = (-0.05 -0.03 -3.83061) to (0.145663 0.03 0.05)
1 by 1 by 1 MPI processor grid
5000 atoms

#Material properties required for new pair styles
fix m1 all property/global kn peratomtypepair 1 1e7
fix m2 all property/global kt peratomtypepair 1 1e7
fix m3 all property/global gamman peratomtypepair 1 0.7
fix m4 all property/global gammat peratomtypepair 1 0.7
fix m5 all property/global coefficientFriction peratomtypepair 1 0.5

pair_style gran/hertz/history/stiffness
pair_coeff * *

timestep 0.000001

fix gravi all gravity 9.81 vector 0.0 0.0 -1.0

# walls
fix y1walls all wall/gran/hertz/history/stiffness primitive type 1 yplane -0.01
fix y2walls all wall/gran/hertz/history/stiffness primitive type 1 yplane +0.01
fix botfacet all mesh/surface/stress file botfacet.stl type 1 stress on
ERROR: Incompatible mesh restart file - mesh has different properties in restarted simulation (multi_node_mesh_parallel_buffer_I.h:198)

I redefine fix mesh/surface/stress for all walls in all scripts. how can I fix this error?

thank you
mojgan

AttachmentSize
Binary Data Untitled Folder.tar__0.gz1.87 MB

mojgan | Mon, 02/11/2013 - 22:17

Hi,
the mentioned error would not arise when I unfix mesh walls at the end of each mentioned steps and then fix them again at the beginning of the next step. however, I am not sure about it.

besides, in any case ( I mean with or without restart files) when I try to output particle-wall interactions by the following commands, just the properties of the last mesh wall is included. and I do not know the meaning of "pos" for wall as the values in output seem irrelavant.
compute WallInter all wall/gran/local pos id force
dump Wallinteract all local 2000 post4/dumpWinteract*.shearing c_WallInter[7] c_WallInter[8] ....

It would be so kind of you if you help me.
thank you
mojgan

ckloss's picture

ckloss | Thu, 02/14/2013 - 09:32

Hi mojgan,

>>ERROR: Incompatible mesh restart file - mesh has different properties in restarted simulation
Whenever you have a fix mesh command in the script that writes a restrat file, you have to use exactly the same fix mesh command in the restart script to be able to continue

>>besides, in any case ( I mean with or without restart files) when I try to output particle-wall interactions by the following >>commands, just the properties of the last mesh wall is included.
Primitive walls cannot be used in this context, see doc

Cheers
Christoph

mojgan | Thu, 02/14/2013 - 14:25

dear christoph,
I do use the same fix mesh command in the restart script. but reading the restart file of the second step, in which, one of the mesh walls move in a constant velocity and stops, in the 3rd step leads in the mentioned error. it seems that moving the mesh results in the error as reading the restart file of the first step, in which walls have no movement, in the second step runs correctly.
I know that primitive walls cannot be used in this context. but I have 6 walls (2 primitive walls and 4 mesh walls). but, just the last mesh wall is included in the output.

cheers
mojgan

ckloss's picture

ckloss | Fri, 02/15/2013 - 00:34

Hi mojgan,

ok I think now I understand. The thing is that mesh movement creates a new mesh property (velocity). When restarting, the same properties have to be used. What you could try is to impose a surface velocity of 0 0 0 upon restart, that should proably do it

>>I know that primitive walls cannot be used in this context. but I have 6 walls
>>(2 primitive walls and 4 mesh walls). but, just the last mesh wall is included in the output.
Ok, if you post the case I can have a look

Cheers
Cheers
Christoph

mojgan | Sat, 02/16/2013 - 12:20

dear christoph,
thank you for your reply,
I have attached my input files plus their restart files and stls. I will be so gratefull if you take a look.
I want just emphasise that when I unfix mesh walls just before write_restart in in.loading and then fix them again at the beginning of in.shearing the error (below) is solved. but I am not sure what happens if I do so.

ERROR: Incompatible mesh restart file - mesh has different properties in restarted simulation (multi_node_mesh_parallel_buffer_I.h:198)

I have one more question as well.
how can I model a cylinder made up of stacked-rings filled with grains. I want the rings move because of the forces which are implemented on them by grains. do I have to couple DEM/FEM or it can be solved by liggghts?

cheers
mojgan