Apply constant stress

Dear LAMMPS users,

I am trying to mimic a creep test on metals, that is to apply a constant stress along one direction at certain temperature. Firstly, I use "nvt" command to relax the sample at a desired temperature, then use "npt" command to do the relaxation in three directions at 0 pressure. To apply a constant stress, I use "npt" to set the pressure in x direction from 0 to a desired pressure (pressures in the other two directions remain zero). At last, the simulation runs the relaxation at the desired pressure in x direction. However, the output of strain as function of time shows a linear shape (constant strain rate), which is not the case for creep behavior. I cannot find out the problem. Do I misunderstand the simulation procedure? I also wonder if there is another way to apply constant stress in LAMMPS. Thank you very much.

Best regards,
Jane

The strain is linear in time because the target pressure is linear in time. If you follow this simulation with an npt simulation with fixed target pressure, the strain will remain fixed (with small random fluctuations), for sufficiently small tension. If you make the tension larger, you will eventually rupture the sample in a complicated way, creating voids, and the strain will grow without bound. If you look in the literature, you will find lots of papers on this, for example, see the work of Norman and Stegailov.

What does this have to do with creep? Unfortunately, not much. If you open any undergraduate text book on material mechanics, you will read that creep is a very slow process that is mediated by migration of dislocations and grain boundaries on very long timescales. Dislocations, grain boundaries, long timescales are three things that are absent from your simulation.

Dear Aidan,

Thank you very much for your reply. Now I understand the problem of linear results. I will consider the three things you mentioned in my simulation.

However, I still feel puzzled at applying stress on LAMMPS. I found literatures on “LAMMPS Publications” about creep study (eg. ACTA MATERIALIA, 99, 382-391 (2015); COMPUTATIONAL MATERIALS SCIENCE, 110, 254-260 (2015)), in which “a constant stress was applied” for simulation. So I try npt simulation at first, and wonder whether this idea is correct or not.

Best regards,
Jane