Help in Green-Kubo thermal conductivity calculations

Dear Users

I am using the Green-Kubo approach to calculate thermal conductivity of nanowires. In most published works the range of simulation time is 6-18ns but mine is not converging even after 30ns and the values remained very large !. I tried to use different time steps (timestep=0.1,0.3,0.5,0.8 fs) and correlation time (100,300,500,800ps) but no change. The potential model I am using gives appropriate values of lattice and elastic constants and I also checked the structure of the nanowires is maintained with the minimum total energy after equilibration steps. I think I did some thing wrong in the G-K calculations ! Suggestion or ideas are of great help !!! Below is an extract from the input file.

dimension 3
boundary f f p # periodic BCs along the axis of the nanowire

read_data a.dat

equilibration and thermalization

velocity all create $T 1026 mom yes rot yes dist gaussian
fix NVT all nvt temp 300 300 1 drag 0.2
run 1000000

thermal conductivity calculation, switch to NVE if desired

unfix NVT
fix NVE all nve
run 1000000

reset_timestep 0

compute myKE all ke/atom
compute myPE all pe/atom
compute myStress all stress/atom NULL virial
compute flux all heat/flux myKE myPE myStress

variable Jz equal c_flux[3]/vol # the flux along the z-axis of the nanowire
fix JJ all ave/correlate $s $p $d &
c_flux[3] type auto file J0Jt.dat ave running # The correlation for the flux along the z-axis

variable scale equal {convert}/{kB}/$T/$T/$V*s*{dt}

variable k33 equal trap(f_JJ[3])*${scale}

thermo_style custom step temp v_Jz v_k33

run 30000000

variable k equal v_k33
variable ndens equal count(all)/vol
print “average conductivity: $k[W/mK] @ T K, {ndens} /A^3”


Calculation of transport coefficients can be a difficult task and only people who have done it many times and on many types of systems (not me) may feel comfortable providing general tips. I suggest you first look in the archives where links like the one that follows can be found:

You should always start with a much simpler case and built up your knowledge from it.