Stress output of a polymer melt subjected to unaxial elongation force

Dear LAMMPS users,

I am subjecting a polymer melt system (Kremer Grest model) to uniaxial elongation shear using fix deform erate along with NVT/SLLOD. My objective is to calulate the tensile stress as a function of strain rate. I am attaching a small portion of the script here for reference.

units lj
boundary p p p

pair_style lj/smooth/linear 2.5
bond_style fene
special_bonds fene

pair_coeff 1 1 1.0 1.0 2.5
pair_modify shift yes
bond_coeff 1 30.00 1.50 1.0 1.0

variable lxs equal 16.47772811 ## The initial box length
fix f1 all nvt/sllod temp 1.0 1.0 0.1
fix 2 all deform 1 x erate 0.0327 y volume z volume remap v units box
compute press all pressure f1_temp
variable lx equal lx
variable strainr equal “(v_lx - v_lxs)/v_lxs”
variable press1 equal -c_press[1]
variable isopressure equal “(-c_press[1] + -c_press[2] + -c_press[3])/3” ## Trace of the stress tensor
variable ats equal “3*(v_press1 + v_isopressure)/2” ## Poisson’s ratio is 1.5

fix 6 all ave/time 62500 1 62500 v_ats v_strainr file response.txt

timestep 0.001
thermo_style custom step temp press pe ke etotal v_strainr v_ats v_lx
thermo_modify flush yes temp f1_temp

I am attaching the response.txt and the plot for easy reference.
The issue: The response file doesn’t show the expected v_ats vs. v_strainr relationship, the v_ats should increase as v_strainr increases. I go upto strain rate of 5 box length, and yet the v_ats (stress) fluctates around zero. Any suggestions are highly appreciated. Thanks for your time.
LAMMPS version: 29 Sep 2021 - Update 3

response.txt (1.2 KB)

Your computation of v_ats is correct but I just wanted to point out that the Poisson ratio of KG chains is ~0.5 not 1.5.

I am not entirely sure what is wrong with you data but I have a suggestion. Measures of stress/pressure in MD have very large fluctuations, which will complicate analysis without time averaging. I simulate similar systems and use fix ave/time with Nrepeat=Nfreq and Nevery=1 to obtain good statistical accuracy. For example, I use “fix 6 all ave/time 1 5000 5000 v_press1 file response.txt”. However, if you do this, you should shift the strain to be the strain value at the midpoint of every time averaging window. Therefore, if stress is output at strains = 0.1, 0.2, 0.3…, then the effective strains of every corresponding stress measure are = 0.05, 0.15, 0.25…


Thank you for the suggestions! I will have a look at them.