Thermal Conductivity of Liquid Argon Pressure, Temp, Density

Hello all,

I’m trying to calculate thermal conductivity of liquid argon near some values of pressure, temperature and density and compare with some experimental values.

I’m having problems in adjust the three parameters.

Inially I’ve performed an minimization of the system with NPT. After, lammps calculates the therma conductivity with Green-Kubo like in manual.

I intend to calculate thermal conductivity at:
T = 110.654 K
P = 10.1418 atm
density = 32.102 mol/L (the initial box size was calculated by this parameter)

Well, the minimization with NPT alters the box size. If i try minimization with NVT the initial pressure becomes far from my desired value.

In both cases the final value of pressure (after, thermal conductivity calculation) is very different of 10.1418 atm so as i can’t compare calculations with experimental data.

I don’t know what to do …

Above is my input file:

units metal
variable T equal 110.654
variable P equal 101.418

region box block 0 35.4876 0 35.4876 0 35.4876 units box
create_box 1 box
create_atoms 1 random 864 261086 box units box
mass 1 39.948
velocity all create $T 141283 dist gaussian loop geom

pair_style lj/cut 8.5
pair_coeff 1 1 0.010341 3.405

fix NPT all npt temp $T $T 0.1 aniso $P $P 0.15

minimize 1.0e-10 1.0e-12 1000 10000
thermo 0
run 20000
variable Pres equal press/0.986923267
print “Global Pressure = ${Pres} atm” (result: 99.03348615 atm)

reset_timestep 0
compute KEs all ke/atom
compute PEs all pe/atom
compute Strs all stress/atom virial
compute flux all heat/flux KEs PEs Strs

variable Jx equal c_flux[1]/vol
variable Jy equal c_flux[2]/vol
variable Jz equal c_flux[3]/vol

fix JJ all ave/correlate 5 20 1000 &
c_flux[1] c_flux[2] c_flux[3] type auto file J0Jt.dat ave running

variable k11 equal trap(f_JJ[3])
variable k22 equal trap(f_JJ[4])
variable k33 equal trap(f_JJ[5])

variable k equal ((v_k11+v_k22+v_k33)/3.0)1.602176460.001
variable Pres1 equal press/0.986923267
variable Dens1 equal (atoms/6.0221418)/(vol/1e4)
variable Temp1 equal temp
variable LLxx1 equal lx

run 30000

print “Average Thermal Conductivity: k W/m.K" result: 0.2973110338 W/m.K (more than double experimental data) print "Global Pressure = {Pres1} atm” result: 120.0020394 atm
print “Global Density = {Dens1} mol/L" result: 29.71249183 mol/L print "Global Temperature = {Temp1} K” result: 111.1647484 K
print “Box lenght = ${LLxx1} A” result: 40.1720394 A

My desired pressure is 101.418 atm. Not 10.1418. Input file is correct.

Things to try I think

(1) Use a larger system size than 864 atoms
(2) Use a larger cutoff for a LJ than 2.5 sigma (only 8.5 Angstroms)
(3) Measure dynamic properties microcanonically
(4) Equilibrate in NPT after minimization
(5) Run a simulation longer than 30 ps

Hi all,

I tried Joshua suggestions (1,2,4,5) and final state (p, T, dens) was very near the one i need. Thanks.

BUT thermal conductivity is more than three times experimental values…

Could you help me in doing a measure microcanonically (hint 3).

Thanks!!

Raph

Hi all,

I tried Joshua suggestions (1,2,4,5) and final state (p, T, dens) was very
near the one i need. Thanks.

BUT thermal conductivity is more than three times experimental values...

Could you help me in doing a measure microcanonically (hint 3).

where is the problem with that?

axel.

The idea of Joshua is just perform a NVE after NPT minimization and before a flux calculations?