Homec4science
Diffusion akantu (features/68-automatic-differentiation) Browse

rAKA/examples/python/custom-material/custom-material.py
2b3e65032fedfeatures/68-automatic…

Last Change
Raw File

custom-material.py
View Options

#!/usr/bin/env python3
from __future__ import print_function
################################################################
import os
import subprocess
import numpy as np
import akantu
################################################################
class FixedValue:
def __init__(self, value, axis):
self.value = value
self.axis = axis
def operator(self, node, flags, disp, coord):
# sets the displacement to the desired value in the desired axis
disp[self.axis] = self.value
# sets the blocked dofs vector to true in the desired axis
flags[self.axis] = True
################################################################
class LocalElastic:
# declares all the internals
def initMaterial(self, internals, params):
self.E = params['E']
self.nu = params['nu']
self.rho = params['rho']
# First Lame coefficient
self.lame_lambda = self.nu * self.E / (
(1 + self.nu) * (1 - 2 * self.nu))
# Second Lame coefficient (shear modulus)
self.lame_mu = self.E / (2 * (1 + self.nu))
# declares all the internals
@staticmethod
def registerInternals():
return ['potential']
# declares all the internals
@staticmethod
def registerInternalSizes():
return [1]
# declares all the parameters that could be parsed
@staticmethod
def registerParam():
return ['E', 'nu']
# declares all the parameters that are needed
def getPushWaveSpeed(self, params):
return np.sqrt((self.lame_lambda + 2 * self.lame_mu) / self.rho)
# compute small deformation tensor
@staticmethod
def computeEpsilon(grad_u):
return 0.5 * (grad_u + np.einsum('aij->aji', grad_u))
# constitutive law
def computeStress(self, grad_u, sigma, internals, params):
nquads = grad_u.shape[0]
grad_u = grad_u.reshape((nquads, 2, 2))
epsilon = self.computeEpsilon(grad_u)
sigma = sigma.reshape((nquads, 2, 2))
trace = np.trace(grad_u, axis1=1, axis2=2)
sigma[:, :, :] = (
np.einsum('a,ij->aij', trace,
self.lame_lambda * np.eye(2))
+ 2.*self.lame_mu * epsilon)
# constitutive law tangent modulii
def computeTangentModuli(self, grad_u, tangent, internals, params):
n_quads = tangent.shape[0]
tangent = tangent.reshape(n_quads, 3, 3)
Miiii = self.lame_lambda + 2 * self.lame_mu
Miijj = self.lame_lambda
Mijij = self.lame_mu
tangent[:, 0, 0] = Miiii
tangent[:, 1, 1] = Miiii
tangent[:, 0, 1] = Miijj
tangent[:, 1, 0] = Miijj
tangent[:, 2, 2] = Mijij
# computes the energy density
def getEnergyDensity(self, energy_type, energy_density,
grad_u, stress, internals, params):
nquads = stress.shape[0]
stress = stress.reshape(nquads, 2, 2)
grad_u = grad_u.reshape((nquads, 2, 2))
if energy_type != 'potential':
raise RuntimeError('not known energy')
epsilon = self.computeEpsilon(grad_u)
energy_density[:, 0] = (
0.5 * np.einsum('aij,aij->a', stress, epsilon))
################################################################
# main
################################################################
spatial_dimension = 2
akantu.initialize('material.dat')
mesh_file = 'bar.msh'
max_steps = 250
time_step = 1e-3
# if mesh was not created the calls gmsh to generate it
if not os.path.isfile(mesh_file):
ret = subprocess.call('gmsh -2 bar.geo bar.msh', shell=True)
if ret != 0:
raise Exception(
'execution of GMSH failed: do you have it installed ?')
################################################################
# Initialization
################################################################
mesh = akantu.Mesh(spatial_dimension)
mesh.read(mesh_file)
mat = LocalElastic()
akantu.registerNewPythonMaterial(mat, "local_elastic")
model = akantu.SolidMechanicsModel(mesh)
model.initFull(_analysis_method=akantu._explicit_lumped_mass)
# model.initFull(_analysis_method=akantu._implicit_dynamic)
model.setBaseName("waves")
model.addDumpFieldVector("displacement")
model.addDumpFieldVector("acceleration")
model.addDumpFieldVector("velocity")
model.addDumpFieldVector("internal_force")
model.addDumpFieldVector("external_force")
model.addDumpField("strain")
model.addDumpField("stress")
model.addDumpField("blocked_dofs")
################################################################
# boundary conditions
################################################################
model.applyDirichletBC(FixedValue(0, akantu._x), "XBlocked")
model.applyDirichletBC(FixedValue(0, akantu._y), "YBlocked")
################################################################
# initial conditions
################################################################
displacement = model.getDisplacement()
nb_nodes = mesh.getNbNodes()
position = mesh.getNodes()
pulse_width = 1
A = 0.01
for i in range(0, nb_nodes):
# Sinus * Gaussian
x = position[i, 0] - 5.
L = pulse_width
k = 0.1 * 2 * np.pi * 3 / L
displacement[i, 0] = A * \
np.sin(k * x) * np.exp(-(k * x) * (k * x) / (L * L))
################################################################
# timestep value computation
################################################################
time_factor = 0.8
stable_time_step = model.getStableTimeStep() * time_factor
print("Stable Time Step = {0}".format(stable_time_step))
print("Required Time Step = {0}".format(time_step))
time_step = stable_time_step * time_factor
model.setTimeStep(time_step)
################################################################
# loop for evolution of motion dynamics
################################################################
model.assembleInternalForces()
print("step,step * time_step,epot,ekin,epot + ekin")
for step in range(0, max_steps + 1):
model.solveStep()
if step % 10 == 0:
model.dump()
epot = model.getEnergy('potential')
ekin = model.getEnergy('kinetic')
# output energy calculation to screen
print("{0},{1},{2},{3},{4}".format(step, step * time_step,
epot, ekin,
(epot + ekin)))
akantu.finalize()
c4science · Help