simulation about EDM

May 17th

Hello,

What I want to do is to simulate the process of EDM(Electrical Discharge Machining).The essence of EDM is a channel of plasma (ionized, electrically conductive gas with high

temperature) formed between the electrodes and develops further with discharge duration. So I use a heat source to simulate the discharge process in EDM. Firstly ,I created a box with Cu(copper) atoms ,like the figure1 shows below .

figure1 :simulation box

Secondly, I try to use the fix temp/rescale command to control

temperature at 7000K by velocity rescaling . The region of atoms that I use this command is always the surface of the simulation box. And it would be a irregular region , like the figure2

shows below . What the red lines show is the region that needs

fix temp/rescale command .

figure2:the surface region changes with time

What I want to ask is that whether lammps can define this

irregular region like the red lines in figure2.

I really appreciate your help~

My in.file is showed below.

units metal

boundary s s s

atom_style atomic

neighbor 2.0 bin

neigh_modify every 5 delay 0 check yes

#create geometry

lattice fcc 3.61

region box block -30 30 -30 30 0 32

create_box 3 box

create_atoms 1 box

#EAM potentials

pair_style eam

pair_coeff * * Cu_u3.eam

#define groups

region 1 block -30 -27.5 -30 30 0 32

region 2 block 27.5 30 -30 30 0 52

region 3 block -27.5 27.5 -30 -27.5 0 32

region 4 block -27.5 27.5 27.5 30 0 32

region 5 block -27.5 27.5 -27.5 27.5 0 2.5

region 6 union 5 1 2 3 4 5

group fixed region 6

region 7 block -27.5 -26 -27.5 27.5 2.5 32

region 8 block 26 27.5 -27.5 27.5 2.5 32

region 9 block -26 26 -27.5 -26 2.5 32

region 10 block -26 26 26 27.5 2.5 32

region 11 block -26 26 -26 26 2.5 4

region 12 union 5 7 8 9 10 11

group thermostat region 12

group mobile subtract all fixed thermostat

set group fixed type 2

set group thermostat type 3

#initial velocities

compute new thermostat temp

compute new2 mobile temp

velocity thermostat create 300.0 887723 temp new

velocity mobile create 300.0 887723 temp new2

velocity fixed create 300.0 887723 loop local

#fixes

fix 1 all nve

fix 2 thermostat temp/rescale 5 300.0 300.0 3.0 1.0

fix_modify 2 temp new

fix 3 mobile temp/rescale 5 300.0 300.0 3.0 1.0

fix_modify 3 temp new2

velocity fixed set 0.0 0.0 0.0

fix 4 fixed setforce 0.0 0.0 0.0

timestep 0.002

thermo 200

thermo_modify lost ignore flush yes

run 20000

region 13 cylinder z 0 0 8 31 32

#the beginning region of fix temp/rescale command and I donot

know how to define the next region

group kinetic region 13

compute new1 mobile temp/region 28

velocity kinetic scale 7000 dist gaussian

unfix 3

fix 5 mobile temp/rescale 10 7000.0 7000.0 10.0 1.0

fix_modify 5 temp new1

run 2400

unfix 5

run 15000

Nafens

mail1.pdf (270 KB)

                                                  May 17th

Hello,

What I want to do is to simulate the process of EDM(Electrical Discharge Machining).The essence of EDM is a channel of plasma (ionized, electrically conductive gas with high

temperature) formed between the electrodes and develops further with discharge duration. So I use a heat source to simulate the discharge process in EDM. Firstly ,I created a box with Cu(copper) atoms ,like the figure1 shows below .

figure1 :simulation box

Secondly, I try to use the fix temp/rescale command to control

temperature at 7000K by velocity rescaling . The region of atoms that I use this command is always the surface of the simulation box. And it would be a irregular region , like the figure2

shows below . What the red lines show is the region that needs

fix temp/rescale command .

figure2:the surface region changes with time

What I want to ask is that whether lammps can define this

irregular region like the red lines in figure2.

no, you cannot. you will have to write some
c++ code of your own, e.g. a separate fix
that adds kinetic energy to your atoms based
on their location and neighbors.

i would also like to point out some
concerns i have with your setup.

- temp/rescale is a very unphysical option
  to transfer kinetic energy. for a system like
  yours, i you want more randomness.
  (fix ttm mentioned below uses a langevin
   scheme instead, for example).

- i would not expect your plasma to be
  very sharply bounded and have the exact
  same "temperature" all over, but rather
  something more fuzzy with a gradient.

- modeling copper as classical particles,
  disregards the transport of kinetic energy
  through the electrons. with your extremely
  high temperatures, i would expect that to
  be a significant contribution. this can be
  handled in LAMMPS via fix ttm or fix atc
  from the AtC package.

axel.

I don't fully understand your picture, but if the atoms
in red are the ones you want to thermostat and they
are always the same atoms from frame to frame
(I'm guessing that's how you colored them), then
that is just a static group, and LAMMPS can apply
a thermostat to those atoms continuously. The
fact that that group of atoms becomes an irregular
region doesn't matter.

Steve