PyGran is an open-source toolkit primarily designed for analyzing DEM simulation data. In addition to performing basic and custom post-processing, PyGran enables running DEM simulation with LIGGGHTS in Python. PyGran supports Python 3.X and is fully backward compatible with Python 2.2 and later versions.
The main features of PyGran:
- Interactive DEM simulation and/or analysis using Python
- Parallel "multiple parameter, single script" simulation for parametrization and sensitivity analysis
- Intuitive (matlab-like) syntax for particle manipulation and analysis (e.g. slicing, concatenating, etc.)
- Post-processing coupled particle-mesh DEM simulation with VTK
- Quick and easy plotting of DEM data with matplotlib
- Support for high-performance computing with MPI
Installing PyGran is quite straight forward on a Unix/Unix-like machine. Just fire up a terminal and then use pip (or pip3) to install PyGran:
pip install PyGran --user
Alternatively, you can clone the PyGran repsitory on github and run from the source code directory:
python setup.py install --user
You can download here a script (for Ubuntu 18.04 LTS) that compiles and installs LIGGGHTS-PUBLIC and PyGran from source code. For more options and information on setting up PyGran, see chapter I in the manual.
PyGran also provides an interface for running DEM simulation with LIGGGHTS. This is achieved by importing the simulation module as shown in the script below for simulating granular flow in a hopper.
from PyGran import simulation
from PyGran.params import stearicAcid, steel
# Create a DEM parameter dictionary
param = {
'model': simulation.models.SpringDashpot,
'boundary': ('f','f','f'),
'box': (-1e-3, 1e-3, -1e-3, 1e-3, 0, 4e-3),
'species': ({'material': stearicAcid, 'radius': 5e-5,},
),
'gravity': (9.81, 0, 0, -1),
'mesh': { 'hopper': {'file': 'silo.stl', 'mtype': 'mesh/surface', \
'material': steel}, },
}
# Instantiate a DEM class
sim = simulation.DEM(**param['model'])
# Insert 1000 particles for species 1 (stearic acid)
insert = sim.insert(species=1, value=1000)
# Evolve the system in time
sim.run(nsteps=1e6, dt=1e-6)
Using PyGran for doing post-analysis is also quite straight forward. Computing particle overlaps shown below for instance can be done in few lines of code:
from PyGran import analysis
# Instantiate a System class from a dump file
Gran = analysis.System(Particles='granular.dump')
# Instantiate a nearest-neighbors class
NNS = analysis.Neighbors(Particles=Gran.Particles)
overlaps = NNS.overlaps
For more examples on using PyGran for running DEM simulation, check out the examples.