[lammps-users] Fluctuation of torque variable

Dear lammps users,
I am running a torsion simulation on carbon nanotube. When the simulation is performed at 1 K, I could observe that torque is increasing as rotation is applied at one end of the nanotube. But when the same simulation is performed at 300 K, the torque is fluctuating a lot without any trend. Can anyone give any suggestions regarding this issue?



cnt.data_5_5_10 (25.4 KB)

in.cnt (1.2 KB)

log.lammps (23.6 KB)

Thanks Steve for your reply. I understand that fluctuations will be present as we increase temperature in MD simulations. But if you can look at the log file attached, torque variable is varying without any trend. I have performed similar simulations for tension, compression and bending of cnt at 300 K. Although fluctuations were there, the behaviour of stress ( in case of tension) showed an increasing trend as strain is applied. But only for torsion case torque is not increasing as rotation is applied at one end. Sorry for such a long reply

So torque seems like an odd thing to measure while twisting a CNT.
If I understand your model correctly, if the CNT is held steady in a twisted state,
there is no torque on a slice of atoms in the middle. They are in equilibrium. So if you are applying
some constraint to keep the CNT twisted, then the constraint is adding force and the
pairwise force between atoms is balancing it exactly. The atoms in the slice
are in equilibrium, hence no torque.

There is only torque if you are dynamically twisting the CNT. And it is constant if you
twist at a constant rotation rate. The more slowly you twist, the torque at any instant goes to zero.
Is that a correct analysis?

So my guess would be that at 1K you can see a small constant torque due to dynamic rotation.
But at higher T = 300K, it is swamped by thermal noise.

It would make more sense to me to apply a constant torque to the end of the CNT and see
how much the CNT twists, i.e. what is its final rotation angle when it comes to equilibrium.
That would give a measure of the resistance of the CNT to being twisted.


I suggest you talk with a local MD expert for advice/insight.

In general, everything in an MD simulation fluctuates more as you raise the temperature.