Dear LAMMPS users
I am working indentation using “fix indent” command and have some questions about getting P-h curve.
I simulate indentation in MD using these scripts.
indentation.in (2.1 KB)
Above script is same condition as the paper which I consulted.
CuAg.eam.alloy.txt (1.7 MB)
The URL of this potential script shows below.
(https://www.ctcms.nist.gov/potentials/Download/2006--Williams-P-L-Mishin-Y-Hamilton-J-C--Cu-Ag/2/CuAg.eam.alloy)
After implementing this script, I make P-h curve, using v_IndentCoords and f_INDENTATION[3].
This graph which is compared from the paper’s P-h curve shows below.
I use “1eV/Å=1.602nN” for unit conversion to modify nN.
From the graph, my P-h curve is larger than one of paper’s.
LAMMPS version is latest version.
Why does the difference of curves result in ?
Please teach me how to get P-h curve.
I would be glad if you could answer my question.
This is overall off-topic, as LAMMPS appears to be working as expected, and you’re not missing any functionality.
In terms of correcting the science here, I would contact the authors of the paper directly and ask for their scripts. One thing I might suggest is running for a longer amount of time. 100k timesteps may be too fast, and the increase in stress may come from strain-rate-dependence.
Exactly. I would contact authors writing this paper.
I concern about this method because I haven’t match P-h curve in some papers after I start to work this research.
So, I think any problems exist in my script.
What you said is probably no problem because some papers implement 100k~130k timesteps.
Thanks for replying!
Yes, that must then be a “conceptual problem” and not a LAMMPS problem, unless you have some silly mistakes like typos on some parameters or not paying attention to “units box” versus “units lattice”.
Generally, to get reliable data, you need to conduct multiple runs with independently equilibrated starting geometries. That should give you somewhat different graphs and the spread between them should also indicate the variance and thus can be used to estimate the statistical error. Another item to look into would be finite size effects.
But you still need to consider the velocity with which you are moving the indenter as a possible source for differences.
I would try this to change indener velocity from 25[m/s] to 1[m/s].
Thank you for you adivices.
I’ll think about them.
