def insert_boundary(self,
temperature,
depth_range,
location,
material='boundary'):
"""
Insert boundary layers in the model.
:param temperature:
:param depth_range:
:param material:
:return:
"""
console.event("Inserting boundary ({}) into the box!".format(material),
verbose=self.verbose)
t_start = time.time()
subdf = self.mesh.query('{} <= z_vals <= {}'.format(
depth_range[0], depth_range[1]))
coords = np.array(subdf['coords'])
temperatures = np.array(self.mesh['temperature'])
objects = np.array(self.mesh['object'])
object_ids = np.array(self.mesh['object_id'])
for coord in coords:
if depth_range[0] <= coords[2][self.dimension -
1] <= depth_range[1]:
index = backends.predict_index(coord=coord,
max_x=self.max_x,
max_y=self.max_y,
max_z=self.max_z,
spatial_res=self.spatial_res,
verbose=self.verbose)
console.nominal("Inserting boundary ({}) at {}...".format(
material, coord),
verbose=self.verbose)
temperatures[index] = temperature
objects[index] = material
object_ids[index] = backends.generate_object_id(
object_type='boundary', id_val=self.id_val)
self.id_val += 1
self.mesh['temperature'] = temperatures
self.mesh['object'] = objects
self.mesh['object_id'] = object_ids
self.boundary = True
# the upper-most usable portion of the model is the matrix immediately below the boundary
if location.lower() == "top" or location.lower() == "t":
self.upper_model = round(depth_range[1] + (2 * self.spatial_res),
self.spatial_sigfigs)
elif location.lower() == "bottom" or location.lower() == "b":
self.lower_model = round(depth_range[0] - (2 * self.spatial_res),
self.spatial_sigfigs)
console.event(
"Finished inserting boundary ({}) into the box! (task took {}s)".
format(material,
time.time() - t_start),
verbose=self.verbose)
def build(self, spatial_res, x, y, z=None):
"""
Constructs the mesh based on specified parameters.
:param spatial_res:
:param x:
:param y:
:param z:
:return:
"""
if z is not 3:
self.dimension = 3
self.max_x, self.max_y, self.max_z = x, y, z
else:
self.dimension = 2
self.max_x, self.max_y, self.max_z = x, y, None
self.spatial_res = spatial_res
console.event("Constructing box...", verbose=self.verbose)
console.nominal("Building mesh...", verbose=self.verbose)
# instantiate the mesh
m = mesh.Mesh(spatial_res=spatial_res,
x=x,
y=y,
z=z,
verbose=self.verbose)
self.spatial_sigfigs = m.get_spatial_sigfigs()
console.nominal("Assigning mesh to dataframe...", verbose=self.verbose)
# construct the mesh
m.build(df=self.mesh)
console.nominal("Exiting box construction...", verbose=self.verbose)
console.event("Box constructed!", verbose=self.verbose)
console.event("Fetching nearest neighbors...", verbose=self.verbose)
coords_range = range(len(self.mesh['coords']))
# create all of the columns for the instantiated dataframes
backends.set_columns(mesh_df=self.mesh,
object_df=self.objects,
coords_range=coords_range)
self.x_coords = np.NAN
self.y_coords = np.NAN
self.z_coords = np.NAN
spatial_sigfigs = m.get_spatial_sigfigs(
) # find the number of sigfigs defined by the spatial resolution
neighbor_count = 1
total_count = len(self.mesh['coords'])
t_start = time.time()
arr = np.array(
self.mesh['coords']) # an array of all the mesh coordinates
x_plus = []
x_minus = []
y_plus = []
y_minus = []
z_plus = []
z_minus = []
# find nearest neighbors in 3D
for coords in arr:
if self.dimension is 3:
console.nominal(
"Finding neighbor for ({}, {}, {}) ({}/{} points)".format(
coords[0], coords[1], coords[2], neighbor_count,
total_count),
verbose=self.verbose)
else:
console.nominal("Finding neighbor for ({}, {})".format(
coords[0], coords[1]),
verbose=self.verbose)
n = neighbors.get_neighbors(verbose=self.verbose,
coords=coords,
max_x=x,
max_y=y,
max_z=z,
spatial_res=spatial_res,
spatial_sigfigs=spatial_sigfigs)
x_plus.append(n[0])
x_minus.append(n[1])
y_plus.append(n[2])
y_minus.append(n[3])
if self.dimension is 3:
z_plus.append(n[4])
z_minus.append(n[5])
neighbor_count += 1
self.mesh['xplus_index'] = x_plus
self.mesh['xminus_index'] = x_minus
self.mesh['yplus_index'] = y_plus
self.mesh['yminus_index'] = y_minus
if self.dimension is 3:
self.mesh['zplus_index'] = z_plus
self.mesh['zminus_index'] = z_minus
console.event(
"Finished finding nearest neighbors! (task took {}s for {} points)"
.format(time.time() - t_start, total_count),
verbose=self.verbose)
return self.mesh
def update(self, auto_update=True, timestep=False, animate_model=False):
"""
Update the model over one time step. Has the ability to run the model to completion.
:param auto_update:
:param timestep:
:return:
"""
t = time.time()
# extract values that are not updated during the loop from the dataframe as lists
coords = np.array(self.mesh['coords'])
x_plus = np.array(self.mesh['xplus_index'])
x_minus = np.array(self.mesh['xminus_index'])
y_plus = np.array(self.mesh['yplus_index'])
y_minus = np.array(self.mesh['yminus_index'])
z_plus = np.array(self.mesh['zplus_index'])
z_minus = np.array(self.mesh['zminus_index'])
object_ids = np.array(self.mesh['object_id'])
conductivities = np.array(self.mesh['conductivity'])
viscosity = np.array(self.mesh['viscosity'])
density = np.array(self.mesh['density'])
diffusivities = np.array(self.mesh['diffusivity'])
temperatures = np.array(
self.mesh['temperature']
) # load in current temperatures across the mesh
dT_dts = np.array(self.mesh['dT_dt'])
mesh_indices = np.array(self.mesh.index)
len_coords = len(coords)
# calculate the timestep based on the maximum conductivity of material in the box
self.delta_time = backends.override_timestep(
timestep=timestep,
conductivities=conductivities,
spatial_res=self.spatial_res,
spatial_sigfigs=self.spatial_sigfigs,
diffusivities=diffusivities,
verbose=self.verbose)
real_delta_time = backends.override_timestep(
timestep=False,
conductivities=list(conductivities),
spatial_res=self.spatial_res,
spatial_sigfigs=self.spatial_sigfigs,
diffusivities=diffusivities,
verbose=self.verbose)
# will run model to completion while the remaining time is above 0
while auto_update is True and self.evolution_time > 0:
console.nominal("Model time at: {} (timestep: {})...".format(
self.evolution_time, self.delta_time),
verbose=self.verbose)
# temperatures = np.array(self.mesh['temperature']) # load in current temperatures across the mesh
# perform actions on objects inside of the model but independent of the mesh
object_coords, nearest_indices, cell_indices = objects.object_actions(
matrix_temperatures=temperatures,
objects_df=self.objects,
spatial_res=self.spatial_res,
spatial_sigfigs=self.spatial_sigfigs,
evolution_time=self.evolution_time,
delta_time=self.delta_time,
initial_time=self.initial_time,
matrix_densities=density,
matrix_viscosities=viscosity,
x_plus=x_plus,
x_minus=x_minus,
y_plus=y_plus,
y_minus=y_minus,
z_plus=z_plus,
z_minus=z_minus,
max_x=self.max_x,
max_y=self.max_y,
max_z=self.max_z,
lower_model=self.lower_model,
upper_model=self.upper_model,
verbose=self.verbose,
conduction=self.conduction,
matrix_conductivities=conductivities,
matrix_ids=object_ids,
coords=coords,
matrix_diffusivities=diffusivities)
if self.conduction:
# finite central difference conductivity across entire box
conduction_t = time.time()
conduction = heat.conduction(coords=coords,
len_coords=len_coords,
x_plus_indices=x_plus,
x_minus_indices=x_minus,
y_plus_indices=y_plus,
y_minus_indices=y_minus,
z_plus_indices=z_plus,
z_minus_indices=z_minus,
temperatures=temperatures,
conductivities=conductivities,
spatial_res=self.spatial_res,
delta_time=self.delta_time,
object_ids=object_ids,
mesh_indices=mesh_indices,
num_workers=self.num_workers,
multiprocess=self.multiprocessing,
real_delta_time=real_delta_time)
# conduction will return tuple: temperature at index 0 and dT/dt at index 1
temperatures = conduction[0]
dT_dts = conduction[1]
console.nominal(
"Finished modeling conduction! (task took {}s)".format(
time.time() - conduction_t),
verbose=self.verbose)
# plot the model's dynamic components, remove in the operational version
self.plots.plot_cell(object_coords=object_coords,
nearest_coords=nearest_indices,
vertex_indices=cell_indices,
mesh_coords=coords,
max_x=self.max_x,
max_y=self.max_y,
max_z=self.max_z,
temperatures=temperatures,
spatial_res=self.spatial_res,
model_time=self.evolution_time,
save=animate_model,
show=False,
heat=self.conduction)
# update the new time in the model
new_evolution_time = round(self.evolution_time - self.delta_time,
self.spatial_sigfigs)
self.evolution_time = new_evolution_time
self.iterations += 1
console.event(
"Model time is at 0! (task took {}s ({} iterations (1 iteration = {}s), {} timestep)"
.format(time.time() - t, self.iterations,
(time.time() - t) / self.iterations, self.delta_time),
verbose=self.verbose)
self.return_mesh(mesh=self.mesh,
temperature=temperatures,
dT_dt=dT_dts,
conductivity=conductivities,
density=density,
viscosity=viscosity)
# will create animations of models if specified
if animate_model is True:
self.plots.animate(initial_time=self.initial_time)
return None
def insert_object(self,
material,
temperature,
radius,
conductivity,
density,
drag_coeff,
cp,
x,
y,
z=None):
"""
Insert an object in the the objects dataframe.
:param material:
:param temperature:
:param radius:
:param conductivity:
:param x:
:param y:
:param z:
:return:
"""
console.event("Inserting object ({}) into the box!".format(material),
verbose=self.verbose)
if z is None and self.dimension is 3:
console.error("Box dimension is 3 and z is not specified!",
verbose=self.verbose)
sys.exit(1)
if self.dimension is 3 and z is not None:
requested_coord = (x, y, z)
elif self.dimension is 2 and z is None:
requested_coord = (x, y)
t_start = time.time()
objects = self.objects['object'].tolist()
object_ids = self.objects['object_id'].tolist()
temperatures = self.objects['temperature'].tolist()
radii = self.objects['radius'].tolist()
locs = self.objects['coords'].tolist()
densities = self.objects['density'].tolist()
velocities = self.objects['velocity'].tolist()
conductivities = self.objects['conductivity'].tolist()
drag_coeffs = self.objects['drag_coeff'].tolist()
cps = self.objects['cp'].tolist()
console.nominal("Inserting object ({}) at {}...".format(
material, requested_coord),
verbose=self.verbose)
objects.append(material)
object_ids.append(
backends.generate_object_id(object_type='object',
id_val=self.id_val))
radii.append(float(radius))
locs.append(requested_coord)
temperatures.append(float(temperature))
conductivities.append(float(conductivity))
densities.append(float(density))
velocities.append((0.0, 0.0, 0.0))
drag_coeffs.append(float(drag_coeff))
cps.append(float(cp))
object_headers = [self.objects.columns.values]
self.objects = pd.DataFrame(
) # reset the dataframe to avoid length issues
self.objects['object'] = objects
self.objects['object_id'] = object_ids
self.objects['temperature'] = temperatures
self.objects['radius'] = radii
self.objects['coords'] = locs
self.objects['conductivity'] = conductivities
self.objects['density'] = densities
self.objects['drag_coeff'] = drag_coeffs
self.objects['cp'] = cps
self.objects['velocity'] = velocities
self.objects['nearest_index'] = np.NAN
self.objects['cell_indices'] = np.NAN
self.objects['cell_vertices'] = np.NAN
self.id_val += 1
self.object = True
console.event(
"Finished inserting object ({}) into the box! (task took {}s)".
format(material,
time.time() - t_start),
verbose=self.verbose)
def insert_matrix(self, material, temperature, conductivity, density,
viscosity, depth_range, heat_capacity):
"""
Insert a matrix into the model.
:param material:
:param temperature:
:param conductivity:
:param depth_range:
:return:
"""
console.event("Inserting matrix ({}) into the box!".format(material),
verbose=self.verbose)
t_start = time.time()
subdf = self.mesh.query('{} <= z_vals <= {}'.format(
depth_range[0], depth_range[1]))
self.mesh.drop(['z_vals'], axis=1)
coords = np.array(subdf['coords'])
temperatures = np.array(self.mesh['temperature'])
objects = np.array(self.mesh['object'])
object_ids = np.array(self.mesh['object_id'])
conductivities = np.array(self.mesh['conductivity'])
densities = np.array(self.mesh['density'])
viscosities = np.array(self.mesh['viscosity'])
diffusivities = np.array(self.mesh['diffusivity'])
self.conductivities.append(conductivity)
len_coords = len(coords)
# insert the matrix into coordinate positions defined by the user's z-range
# TODO: we can limit recursion time by using the index prediction equation here
for coord in coords:
if depth_range[0] <= coords[2][self.dimension -
1] <= depth_range[1]:
index = backends.predict_index(coord=coord,
max_x=self.max_x,
max_y=self.max_y,
max_z=self.max_z,
spatial_res=self.spatial_res,
verbose=self.verbose)
console.nominal("Inserting matrix ({}) at {}...".format(
material, coord),
verbose=self.verbose)
temperatures[index] = float(temperature)
conductivities[index] = float(conductivity)
diffusivities[index] = float(conductivity /
(density * heat_capacity))
objects[index] = material
densities[index] = float(density)
viscosities[index] = float(viscosity)
object_ids[index] = backends.generate_object_id(
object_type='matrix', id_val=self.id_val)
self.id_val += 1
self.mesh['temperature'] = temperatures
self.mesh['object'] = objects
self.mesh['object_id'] = object_ids
self.mesh['density'] = densities
self.mesh['viscosity'] = viscosities
self.mesh['conductivity'] = conductivities
self.mesh['diffusivity'] = diffusivities
self.matrix = True
console.event(
"Finished inserting matrix ({}) into the box! (task took {}s)".
format(material,
time.time() - t_start),
verbose=self.verbose)