Incorrect results due to gaps at z axis

Dear all,

I have a simple water box and used spc/e model to compute the viscosity based on Lammps documents (Green-Kubo model). So, I got nice results and compatible to others.
However, there is a discrepancy for the output result when I create 2 Angstrom gaps at z=0 and z=60 Angestrom. And the viscosity increases as time goes (viscosity should not depend on the shape)
These gaps become important when we have two graphene sheets at z=0 and z=60, since these gaps avoid overlapping of atoms.

Hereby, could you please write any comments how to overcome this challenge for computing correct pressure tensor.

Thanks in advance

I have serious doubts whether any of the methods showcased in that examples folder is applicable to the systems you describe. I would expect that the simulated system needs to be homogeneous. I wouldn’t even know what a viscosity for a system with gaps or embedded graphene sheets would be measured like. Is there a macroscopic equivalent that you will be comparing to? if yes, what does that look like? and how would you measure viscosity in that case?