I’m trying to reproduce the results of the paper (I attatched) based on flow boiling characteristics. The paper studies boiling phenomenon in flowing argon. But after a short run, I see my argon temperature exceeds the heating base copper temperature (which I set 180K). The structure is okay. The parent paper uses LJ potential for Cu-Cu, so did I. The structure file is okay, as I’ve run it and checked.
Brief from the paper (names are given by me):
Fluid Domain (3 regions): Contains liquid and vapor argon undergoing flow (caused by ‘addforce’, the paper didn’t mention)
1st region: Name: force (10mm of 370mm along the length). I applied force here.
2nd region: Name: reset (10mm to 17mm). Fluid temperature is always kept 86k after passing this region to create constant temperature inlet in the next region. Langevin thermostat applied.
3rd region: Name: study (17mm to 370mm). Data is collected from this domain.
Copper Wall Domain(3 layers):
1st region: Name: fixed (bottom most layer, no heating, just supports the substrate)
2nd region: Name: ghost (middle layer, heating is here. Temp is raised to 180k of this region to study boiling)
3rd region: Name: real (no heating, just used for conduction. Transfer heat from ghost to liquid argon at its surface)
Reflective upper wall at z-axis where vapor tries to escape.
Initial Conditions: Inlet temperature set to 86 K.
Study region (17–370 Å): natural convective flow observed.
Simulation Parameters:
Time step = 2 fs. The paper says in brief:
NVT run for 4 ns at 86 K for equilibration. System stabilized at 86 K.
NVE run for 0.2 ns to heat Copper wall to target temperature (180k)
Final NVE run for 4 ns for collecting data (my goal is temperature) to observe bubble nucleation and growth.
No mention of when to apply force. So I added it during equilibration.
My code involving main part of the procedure (My runs are much shorter, still temperature out of control):
Code:
#Initial Velocities
velocity integration create 86 12345
fix freeze fixed setforce 0.0 0.0 0.0
# Equilibrate the whole system
fix force_flow force addforce 0.02 0.0 0.0
fix equilibrate integration nvt temp 86 86 $(100.0*dt)
run 20000
unfix equilibrate
#step : heating
variable Th equal 180
fix NVE integration nve
fix temp_elevate ghost langevin 86 ${Th} 100 699948 # temp raise of copper substrate
fix cold_inlet reset langevin 86 86 100 122345 # cold inlet maintaining
dump dump1 all atom 10 dump_heating.equil
run 1000
unfix temp_elevate
undump dump1
#data collection
fix temp_maintain ghost langevin ${Th} ${Th} 100 699948
dump dump2 all atom 10 dump_final.equil
run 20000
Figuring out the origin is part of the job of a scientist. It is very hard to do with a complete and complex input, regardless of whether you are inexperienced or experienced. So the first step would be to start with an empty input and put in commands, and only commands that are necessary, piece by piece and check whether your system behaves as expected at every step.
Most importantly, you do this for the multiple sub-systems of a complex system separately to see, if your simulation settings for each are suitable. If they turn out unsuitable for one, then you can systematically change them to more conservative values.
If you need help with all of those steps, you need to find yourself a tutor. This is not what online forums are for. We can help with specific problems once you have identified them and it is clear the origin is a behavior that contradicts the documentation.
In my many years of experience I can tell you for certain, just posting your input, the paper and telling people that you don’t know what is causing the problem and don’t know what to do, is highly unlikely to get you any other response like the ones from @simongravelle and me. After all, you are asking that somebody else should do your job for you.
Sir, actually it’s been several days, and I promise I’m doing it exactly the way you’re telling. I first confirmed the structure, first gave it a heat to simulate basic boiling, then added a force separately, and did many things again and again to see and understand. But everything starts to fail when I combine them.
Sir I totally understand your point. I’ll again go through them.
I would start with a simpler bulk simulation without the copper using fix nve integrator and make sure your timestep is small enough to conserve energy (i.e. no artificial numerical heating). Then slowly add in more features.
Dear sir @akohlmey , I’ve updated my post sir. I’ll do as per you said earlier sir. For now, I’ve made it a bit more clear, sir. My main problem is not getting what the paper says and did I really do accordingly.
