# Inconel model nanowire under tension

## Goal
Compare the tensile response of a random Ni–Cr–Fe wire at 100 K and 800 K. The paper examines nanoscale Inconel-718 structures under mechanical loading. This model keeps the simplified Ni 60%, Cr 21%, Fe 19% composition and uses a smaller FCC wire. Random selection gives approximate, not exact, percentages.

## Potential file
Download `FeNiCr.eam.alloy` from the corrected [2011 Bonny Fe–Ni–Cr NIST entry](https://www.ctcms.nist.gov/potentials/entry/2011--Bonny-G-Terentyev-D-Pasianot-R-C-et-al--Fe-Ni-Cr/) or [direct file](https://www.ctcms.nist.gov/potentials/Download/2011--Bonny-G-Terentyev-D-Pasianot-R-C-et-al--Fe-Ni-Cr/3/FeNiCr.eam.alloy). Save it here. The paper cites this 2011 model and a 2013 model. The 2011 model was developed around Fe-rich stainless steel; its accuracy for Ni-rich wire requires validation.

## Requirements and run
MANYBODY package. Run commands from this directory. Replace `lmp` with your LAMMPS executable. See [LAMMPS_SETUP.md](../LAMMPS_SETUP.md). The input uses metal units: lengths in Å, time in ps, and pressure in bar. `rate` is in inverse ps; the default 0.01/ps is 10^10/s. The output stress is in GPa. Python requires NumPy and Matplotlib.

All runtime estimates are unmeasured. They depend on the processor, LAMMPS build and potential. Check atom count and wall time in the LAMMPS log. The 15-minute target is not verified.

```sh
lmp -in in.inconel.lmp -var T 100 -var out tension_100.txt
lmp -in in.inconel.lmp -var T 800 -var out tension_800.txt
python analyze.py
```

## What to expect and differences
The script plots both curves and prints each measured peak. No numerical result is prescribed. The wire is periodic along z and has free lateral surfaces. Engineering stress uses the initial circular area and the global pressure including the vacuum correction. The wire radius is 14 Å. Surface relaxation changes the effective area, so compare area conventions carefully. The small wire, short equilibration and faster loading differ from the paper. Estimated runtime: unmeasured; 17,000 MD steps per run.
