[lammps-users] Measuring yield stress, cont'd

Hi, many thanks to those of you who donated indent examples earlier in the year. I want to calculate the value of the yield stress of a lammps material. I'm open to more than one method, but particularly focused on using an indentation right the moment.

I have the indenter coming in along one axis. I output the component of the force on the f_indent. What is this force? The documentation helpfully explains that it is extensive, that is, it varies with the number of atoms in the simulation. What does this actually mean?

I just want to calculate the force at yield, and thus be able to calculate the yield stress by combining it with the area, does this mean I have to divide this force by the number of atoms in the simulation? This of course is not very useful, as not all atoms have to be involved in the stress field and it is not clearly defined how many atoms are.

Anthony

Hello Anthony,

To obtain a measure of yield stress without worrying about indentation area
or other such complicated issues, I suggest running a simple uniaxial strain
or uniaxial stress simulation on a rectangular parallelepiped region.
Uniaxial stress would be ideal as then the stress along the direction of
creation. However, such a simulation may be difficult to do in LAMMPS as I'm
not certain you can change system volume (i.e. Enforce a barostat of zero
perform a uniaxial strain simulation with the minimize routine, but then
will have to convert the system stress state {-pxx, -pyy, -pzz, -pxy, -pxz,
-pyz} into the equivalent or Von Mises stress in order to estimate yield
stress at failure.

Jon Zimmerman

The energy and force that the fix reports are the total
energy/force on the indenter. If you want to divide it by area
you'll have to do that, e.g. via a variable that stores
an equation. "Extensive" simply means that if you
scaled up your system (more atoms, bigger indenter),
the reported force would be larger. Unlike temperature
which would be "intensive". Note that when you ouput
these values with thermo output, they may or may not
be scaled by N = # of atoms in simulation depending
on the normalization option you have set.

Steve