[lammps-users] rapid increase in volume

Dear all,
I am trying to equilibrate a nanotube/polymer system. I declare the nanotube as rigid. I constraint C-H bond lengths in the matrix by SHAKE. Then I integrate the system with fix NPT aniso partial where I apply pressure constraints in x and y for the matrix alone. I did not employ constraints in z since the nanotube stays as it is eventough the nanotube does not span through the box, it is short and in the middle.

The volume of the system immediately enlarges by almost 7 times. Then it decreases only slightly. I ran my simulation for 100 ps and the volume reached to a steady level but still is huge.

I visualized the atoms. The atoms are squeezed in on corner of the big box and few atoms are on the other end due to PBC. But the relative changes of coordinates are minor. I can not explain why this happens. My initial system is in the main box will image flags all zero. When I try soft potential instead of lj/cut I have the same problem, worse this time. Any ideas why this may be happening?

Thank you for your time,

       I believe your system is stuck in a local minima. You can try building the system with a very low density and then compress it in x and y direction (as NT's are in z direction) and run NVT in between and keep doing it till you reach a very compressed system (around the density you think your composite should be). Then run NPT with no constraint and see where the system goes. Hope this works out.


Burcu Eksioglu wrote:

Sounds to me like you might have some confusion about scaling of the atom coordinates. Are all of the atom coordinates in your dump file between 0 and 1? You might try “dump_modify your_dump scale no”. Read this page:



Thanks Paul and Arnab,
I used dump custom unscaled coordinates (x y z) when visualizing so I guess this is the same for dump_modify scale no ? My coordinates are not between 0 and 1.
I will try to use NVT as Arnab suggested.

Thanks again,

Are you saying the initial config of your system has all the atoms
in the corner, or that they rapidly move to one corner. Either way,
that's not good and should give you something to debug.