read_data for a stress strain curve

Hi,

I am following the simple tutorial 3 (https://icme.hpc.msstate.edu/mediawiki/index.php/Uniaxial_Tension) that does a tension test on a box of aluminum. As a learning exercise I am trying to read the data from an aluminum slab I made on Avogadro as opposed to creating it in the script. However, it is giving errors in the stress when it attempts to displace this box. Stresses are jumping from positive to negative when they should be slowly going up.

I replaced this:

----------------------- ATOM DEFINITION ----------------------------

lattice fcc {latparam} region whole block 0 10 0 10 0 10 create_box 1 whole lattice fcc {latparam} orient x 1 0 0 orient y 0 1 0 orient z 0 0 1
create_atoms 1 region whole

With this:

Atom Definition

read_data Alslabs2.lmpdat
replicate 1 1 1

The Alslabsmall.lmpdat file is attached

The program runs but the values are not making sense. Is this because I have all the bonds in the .lmpdat file? Should it just be coordinates? I know the rest of the script is fine as when I make my own box (as opposed to read data) it works.

Thanks

Liam

Alslab2.lmpdat (27.8 KB)

Hi,

I am following the simple tutorial 3 (https://icme.hpc.msstate.edu/
mediawiki/index.php/Uniaxial_Tension) that does a tension test on a box
of aluminum. As a learning exercise I am trying to read the data from an
aluminum slab I made on Avogadro as opposed to creating it in the script.
However, it is giving errors in the stress when it attempts to displace
this box. Stresses are jumping from positive to negative when they should
be slowly going up.

​your data file is bogus. it lists bonds, angles, and d​ihedrals for a
system, that should not use them.
in fact, this data file cannot be used with *only* the modifications to the
input listed below, as the data file is not compatible with atom style
atomic, as it is given in the input you quote.

in short, you are seeing the manifestation of the GI-GO principle.

axel.

Hi Axel,

Thanks for that. I have modified the data file as per attached and got rid of bonds/dihedrals etc. We then changed atom style to full and used replicate 5x5x5. The resulting stress strain curve had large oscillations. Where there is clearly something wrong is with the temperature which is jumping from 300 to 3000.

As a second attempt we built a much larger cell in Avogadro (125 atoms) and did not replicate it. Again, using full atom style, the temperatures were jumping from 300 to 3000. However, keeping it at atom style atomic (and remove charges) there were much less oscillations but temperatures were still off.

Why would these oscillations be present? Is it due to boundary conditions? I have attached the script and the two read data files.

Cheers

in.tensile.txt (2.06 KB)

Al.lmpdat (405 Bytes)

Hi Axel,

Thanks for that. I have modified the data file as per attached and got rid
of bonds/dihedrals etc. We then changed atom style to full and used
replicate 5x5x5. The resulting stress strain curve had large oscillations.
Where there is clearly something wrong is with the temperature which is
jumping from 300 to 3000.

As a second attempt we built a much larger cell in Avogadro (125 atoms)
and did not replicate it. Again, using full atom style, the temperatures
were jumping from 300 to 3000. However, keeping it at atom style atomic
(and remove charges) there were much less oscillations but temperatures
were still off.

Why would these oscillations be present? Is it due to boundary conditions?
I have attached the script and the two read data files.

​i don't have time to review your every input that doesn't work as
expected. you need to figure out ways to verify your input yourself. that
is just the process of teaching yourself a new tools and methodology.

you can easily check for close contacts using the delete_atoms overlap
option, similarly, visualization can help a lot, too.

you should really find yourself a proper tutor that you can discuss and
review your steps to set up systems with and also discuss how to validate
results. in MD simulations of condensed media, the difference between good
and bad can be very small and may not always be immediately obvious.

axel.