[lammps-users] Simulation of tensile test


I’m studying the tensile behavior of a polymer using LAMMPS(10 Nov 2005).

What is the best method for simulation of tensile test using LAMMPS?

Is there anybody who has experience in tensile test simulation?

Best regards,


�����! ����� ��Ÿ��! �ZDMB�� 32,000��

You can expand the simulation box (see fix volume/expand) or you
can pull on the atoms. See an example of applying a strain (shear not
tensile, but same idea) in the examples for a metal. The LAMMPS WWW
page has pictures and citations for work people have done pulling on a polymer
melt to watch the chains disentangle and break.


In my polymer tensile tests, I use a combination of the NPT ensemble and
fix_volume/rescale. Here is a specific example of the fix commands I
use for a uniaxial tensile test simulation:

fix 1 all volume/rescale 10 x true 0.000001
fix 2 all npt 300. 300. 50. aniso NULL NULL 0. 0. 0. 0. 1000.

Here, the simulation cell is being pulled in the x-direction. The NPT
ensemble is used to maintain zero pressure on the faces normal to the
draw direction (y- and z- directions). The pressure on the draw faces
is not regulated by the NPT ensemble, but used to determine the
stress-strain relations. The parameters listed above for the
volume/rescale fix are modified from the normal LAMMPS distribution.
Normally, after the x parameter, the low and high boundaries in the
x-direction are given. I modified LAMMPS so that the first parameter
can also be defined as "nom" for nominal (engineering) strain or "true"
for true (logarithmic) strain followed by the strain rate. I can post
the modifications, which are relatively minor, if anyone is interested.
They have worked well for my needs, but I have not validated the
robustness of these changes.

-Jim Shepherd

I am trying to use a combination of fix_volume/rescale and NVT ensemble to do stress strain test of my polymer system. First i rescale the volume and there by strain the system . Then i keep the volume fixed and run MD to find the equilibrium stresses developed in the strained system. However this is only good for tensile stress/strains as it is not possible to give shear strain to the system.


Jim Shepherd wrote:

Note there is also a fix uniaxial in LAMMPS contributed by Carsten Svaneborg
that alters all 3 box dimensions in a self-consistent way as a tensile
strain is applied …