[lammps-users] prombelm using fix langevin together with fix press/berendsen

hi all,
I’m simulating hydrate with bulk methane aqueous solution. During the equilibrium stage, after using fix langevin to control the fluid temperature while hydrate was fixed, fix langevin and fix press/berendsen were used to control the temperature and pressure of the whole system. However, the system just “explode” after few steps.
Besides, I tried nose-hoover thermostat and barostat, and the simulation went on. And I checked the mail list and found that some users have used the combination of fix langevin and fix press/berendsen successfully.
I’m using the 2017 version of LAMMPS. The log file of simulation named free.pbs.o2763 and the input file named free.in are attached. According to the log file, during the process when fix langevin controls the fluid temperature, the system pressure is very low. Is this factor causing the simulation result?
Hope someone could help.
Best regards
yun

free.pbs.o2763 (15 KB)

free.in (2.6 KB)

the berendsen barostat will not be stable for any choice of Pdamp for any system. if it is chosen too short, then you get too large volume fluctuations and those will lead to numerical instabilities due to a resonance between the pressure fluctuations and the rescaling from the fix. you may also need to adjust the value of the bulk modulus for your system since the default setting of “10.0” is representative of a “soft” Lennard-Jones fluid. on top of that, pressure fluctuations will be larger for smaller systems and thus the aforementioned instability is more likely, since you are more likely to get into a resonant state.

the nose-hoover barostat is less sensitive (but not insensitive) to such issues since it uses a far more indirect coupling of the volume to the pressure.

axel.

p.s.: please also note that there are potential inconsistencies and a definite inefficiency in your coulomb handling from using a hybrid pair style.
there is no need for a hybrid style for your system since you can use lj/cut/tip4p/long for all interactions.