Bending

LAMMPS LAMMPS General Discussion
1

Hi! I am trying to simulate a 3-point bending test. I grouped the sample into 3 parts, unionized the left & right part, used a fix rigid on that; and applied a fix move linear on the middle part. But it doesn’t really feel right to me as the left & right parts are sitting still in the visualization (as expected) instead of curving in the opposite direction of the middle part like in a real bending setup. Any ideas how to replicate the real test? Also I need some guidance about what sort of boundaries should be used here. The code is attached below-

######### Three point bending test ##############
units 			metal
dimension 		3
boundary 		s s s
timestep		0.0001
atom_style  	atomic
atom_modify 	map array


############### Atom definition #############
lattice			fcc 3.52
region 			sample block 0 20 0 40 0 20 units box
create_box		1 sample
create_atoms 	1 region sample 


pair_style 		eam
pair_coeff		* * Ni_u3.eam
neighbor 		2.0 bin 
neigh_modify 	delay 10 check yes

reset_timestep 	0
minimize 		1e-6 1e-8 10000 100000

velocity 		all create 300 1256 mom yes dist gaussian
fix 			NVT all nvt temp 300 300 0.05 tchain 10
run 			2000
unfix			NVT

#fix 			NVE all nve
#run			1000
#unfix 			NVE

#fix 			NPT all npt temp 300 300 1 x 0 0 10 z 0 0 10
#run 			2000
#unfix 			NPT

region     		1 block 0 20 0 10 0 20 units box 
group      		left region 1 
region     		2 block 0 20 10 30 0 20 units box 
group      		middle region 2 
region     		3 block 0 20 30 40 0 20 units box 
group      		right region 3
group	   		good union left right


fix				1 good rigid single
fix 			2 middle setforce 0.0 0.0 0.0
fix 			3 middle move linear 5.0 0.0 5.0 units box
#fix 			4 middle move rotate 0 0 0 0 0 10 20


dump 			1 all custom 200 dump.nix.* xs ys zs fx fy fz

thermo 			100
thermo_style	custom step pxx pyy pzz temp etotal
run				5000
2

I see two main problems here:

  1. Mimicking macroscopic experiments on the atomic scale rarely makes any sense because length and time scales are very different. For the most part, the experimental behavior is dominated by bulk system behavior on the atomic scale. There are well established simulation protocols to determine the relevant properties, e.g. by using a combination of fix deform in one dimension and fix npt coupled to the other two dimensions. You need to study available tutorials and the published literature for descriptions.
  2. Your simulation set up makes no sense. a) You seem to be confusing the meaning of “rigid” (a group of particles moves around its center of mass without change of the relative distances) and “immobile” (atoms do not move at all). b) Your fix move command makes atoms follow a prescribed motion with complete disregard of the forces. There is nothing to be learned from that. What would be the purpose of computing force when you don’t use them?