Skip to content

Commit

Permalink
Merge pull request #27 from parhamdee/main
Browse files Browse the repository at this point in the history 
minor updates to the IHT module
  • Loading branch information
@parhamdee
parhamdee authored Sep 18, 2024
2 parents d2ef0ef + 5ff3b60 commit 612d298
Showing 1 changed file with 1 addition and 35 deletions.
36 changes: 1 addition & 35 deletions Scripts/IHT.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -18,11 +18,8 @@
import subprocess
import gc
from tkinter import *
# from skimage.restoration import denoise_tv_chambolle
from scipy.ndimage import gaussian_filter



DEFAULTS = {
"initial temperature": 286.15,
"surface emissivity": 0.94,
Expand Down Expand Up @@ -96,12 +93,10 @@ def rho_metal(T_g):

def Grashof(dT_w, T_film, T_amb):
L = 1
# print(11111,dT_w , nu_gas(T_film - 273.15), T_film)
return 9.81*(1/(T_film))*abs(dT_w)*L**3 /(nu_gas(T_film - 273.15))**2

def natural_conv_correlation(Gr):
Pr = 0.701
# print(111111,Gr , (0.825 + (0.387*(Gr*Pr)**(1/6))/(1+(0.492/Pr)**(9/16))**(8/27))**2)
return ((0.825 + (0.387*(abs(Gr)*Pr)**(1/6))/(1+(0.492/Pr)**(9/16))**(8/27))**2)

def calculate_gradients(T_values_2d, i, j, element_width, element_height, use_higher_order=False):
Expand Down Expand Up @@ -180,7 +175,7 @@ def inverse_heat_transfer(time_series_data, element_width, element_height, time_
hfc[i,j,k] = DEFAULTS['convective hc exposed']

q_conv = (DEFAULTS['convective hc exposed'] + DEFAULTS['convective hc unexposed']) * (T - Tamb[k])
q_rad = 2 * DEFAULTS['surface emissivity'] * STEFAN_BOLTZMANN_CONSTANT * (T**4 - Tamb[k]**4)
q_rad = 2 * DEFAULTS['surface emissivity'] * STEFAN_BOLTZMANN_CONSTANT * (T**4 - 0.5*Tamb[k]**4)
q_storage = cp_metal(T) * rho_metal(T) * DEFAULTS['wall thickness'] * temp_grad_time[i, j, k]
estimated_flux[i, j, k] = (q_storage + q_conv + q_rad - q_cond) / (1000 * DEFAULTS['surface emissivity'])
return estimated_flux , hfc , T_film , Tamb
Expand Down Expand Up @@ -308,9 +303,6 @@ def process_directory_and_plot(source_dir, dest_dir, element_width, element_heig
elif input_file_type == "h5py":
temp_data = h5py_input_file[file_path][:]
temp_data = temp_data.astype(np.float64)
# temp_data = denoise_tv_chambolle(temp_data, weight=0.2, eps=0.0002, max_num_iter=1000, channel_axis=None)
# temp_data = savgol_filter(temp_data, window_length=21, polyorder=2, axis=0)
# temp_data = savgol_filter(temp_data, window_length=21, polyorder=2, axis=1)
time_series_data.append(temp_data)

time_series_data = np.stack(time_series_data, axis=-1)
Expand All @@ -328,31 +320,14 @@ def process_directory_and_plot(source_dir, dest_dir, element_width, element_heig
elif input_file_type == "h5py":
additional_temp_data = h5py_input_file[file_path][:]
additional_temp_data = additional_temp_data.astype(np.float64)
# additional_temp_data = savgol_filter(additional_temp_data, window_length=11, polyorder=2, axis=0)
# additional_temp_data = savgol_filter(additional_temp_data, window_length=11, polyorder=2, axis=1)
additional_data.append(additional_temp_data)

additional_data = np.stack(additional_data, axis=-1)
combined_data = np.concatenate((additional_data, time_series_data), axis=2)
# combined_data = savgol_filter(combined_data, window_length=window_length, polyorder=2, axis=2)
time_series_data = combined_data#[:, :, -batch_files:]

# else:
# time_series_data = savgol_filter(time_series_data, window_length=window_length, polyorder=2, axis=2)
# sigma_values = [0.2, 0.2, 0.4]
# Apply Gaussian Filter with different sigmas for each axis
# time_series_data = gaussian_filter(time_series_data, sigma=sigma_values)
estimated_flux, hfc, Tf, T_inf = inverse_heat_transfer(time_series_data, element_width, element_height, time_step, convection_method)

# estimated_flux, hfc, Tf, T_inf = inverse_heat_transfer(time_series_data, element_width, element_height, time_step, convection_method)

# if DEFAULTS['output_smoothing']:
# for k in tqdm(range(estimated_flux.shape[2]) , desc = 'Applying Median Filter: '):
# estimated_flux[:, :, k] = median_filter(estimated_flux[:, :, k], size=15)
# denoise_tv_chambolle(np.array(estimated_flux[:, :, k]), weight=0.6, eps=0.0002, max_num_iter=2000, channel_axis=None)

# Define different sigma values for each axis
# sigma_values = [0.1, 0.1, 4]
sigma_values = [7, 7, 15]

# Apply Gaussian Filter with different sigmas for each axis
Expand Down Expand Up @@ -468,8 +443,6 @@ def export_pngs(root, batch_size):

# Initialize the progress bar for the total number of frames
pbar = tqdm(total=total_frames, desc="Generating PNG frames")
# def format_func(value, tick_number):
# return f'{value:.1f}'
# Process each batch
for batch_index in range(total_batches):
batch_start = batch_index * batch_size
Expand Down Expand Up @@ -502,25 +475,18 @@ def export_pngs(root, batch_size):
axs[1].set_title(f'Incident Radiative Heat Flux {frame_index}')
surface_width_m = float(global_state["constants"]["surface width"].get())



xmin, xmax = axs[0].get_xlim()
ymin, ymax = axs[0].get_ylim()
axs[0].set_xticks(np.linspace(xmin,xmax,labels_x.shape[0]) , labels_width, rotation='vertical')
axs[1].set_xticks(np.linspace(xmin,xmax,labels_x.shape[0]) , labels_width, rotation='vertical')
axs[0].set_yticks(np.linspace(ymin,ymax,labels_y.shape[0]), labels_height, rotation='horizontal')
axs[1].set_yticks(np.linspace(ymin,ymax,labels_y.shape[0]), labels_height, rotation='horizontal')
# axs[0].xaxis.set_major_formatter(FuncFormatter(format_func))
# axs[1].xaxis.set_major_formatter(FuncFormatter(format_func))
# axs[0].yaxis.set_major_formatter(FuncFormatter(format_func))
# axs[1].yaxis.set_major_formatter(FuncFormatter(format_func))

plt.tight_layout()
fig.savefig(os.path.join(output_dir, f"batch_{batch_index}_frame_{frame_index:06d}.svg") , dpi = 150, bbox_inches='tight')
pbar.update(1)

plt.close(fig)
# Clear memory
del source_data_frames, dest_data_frames
gc.collect()

Expand Down

0 comments on commit 612d298

Please sign in to comment.