problems about "restart+fix deform"

LAMMPS LAMMPS Mailing List Mirror
1

Dear all,

I simulated the tensile behavior of a material with the “fix deform erate” command. In the first simulation, the intermediate
configurations in a relative large time interval. In order to obtain some details of fracture, the task was restarted from a
intermediate step with an adjusted value of the paramter “erate” (the details are illustrated in the attached file of “tensile.in”) to keep
the same strain rate in these two simulations. In the “restart simulation”, the strain of every step is indeed kept at the same value as
that in the first simulation, while the material fractured at different time (the deviation is about 0.3ns). I do not know why there is a
such large difference between the first simulation and the “restart simualtion”, and the difference also brings some difficulties in data analysis.
Please give me some instructions, thank you in advance. (attachments instruction: tensile.in is the input file, tension26350000.dat and tension26400000.dat are the configurations before and after fractrue in the first simulation, tension_restart_26950000.dat and tension_restart_27000000.dat are the configurations before and after fracture in the “restart simulation”).

从QQ邮箱发来的超大附件


tensile.in (915B, 无限期)
进入下载页面


tension26350000.dat (1.94M, 无限期)
进入下载页面


tension26400000.dat (1.86M, 无限期)
进入下载页面


tension_restart_26950000.dat (1.95M, 无限期)
进入下载页面


tension_restart_27000000.dat (1.81M, 无限期)
进入下载页面

2

two comments:

  • the cleaner way to restart would be to use the “start” and “stop” options to the “run” command. no other changes to the input would be needed. please refer to the documentation for details
  • since fracturing is an activated process, some variance/hysteresis would be expected unless you take some provisions to explicitly induce or promote fracture (e.g. see the use of neigh_modify exclude in the “crack” example bundled with LAMMPS)

axel.