I am working with a TMDC material in LAMMPS. I tried to introduce different defects like point defects, bi-vacancy and edge defects. Almost for all types of material I saw that point vacancy reduces thermal conductivity the most, then the edge defects and then bi vacancy. In my case I am getting the opposite trend. I found that thermal conductivity is highest for point vacancy and lowest for bi vacancy. Is it possible? If not what might be the possible source of error.

N.B: I created the mono layer with Atomsk. Then manually removed atoms from .xsf file to create defects.