PomPy is a NumPy based implementation of the puff-based odour plume model described in
Farrell, J.A., Murlis, J., Long, X., Li, W. and Cardé, R.T., 2002. Filament-based atmospheric dispersion model to achieve short time-scale structure of odor plumes. Environmental fluid mechanics, 2(1-2), pp.143-169. Index page (springer.com) :: PDF (semanticscholar.org)
An example simulated concentration field generated by the package is shown below
PomPy allows simulation of dynamic 2D odour concentration fields which show some of the key characteristics of real chemical plumes in turbulent flows including short term intermittency, diffusive effects and longer term variations in spatial extent and location, while being cheaper to run than a full fluid dynamics simulation.
PomPy requires a Python 2.7 or newer environment. For usage of the pompy.models and pompy.processors modules NumPy and SciPy are required. For the demonstrations in the pompy.demos module Matplotlib is also required. The requirements.txt file lists versions of NumPy, SciPy and Matplotlib known to work with PomPy. A Jupyter notebook server installation will also be required to run the example notebooks locally.
To install the package run
pip install pompy
Documentation of the PomPy API is available at Read the Docs.
Two Jupyter notebooks showing examples of using the package are included in the notebooks directory. The Farrell et al. (2002) example.ipynb notebook illustrates an example of using PomPy to generate an animation of an odour plume using the simulation parameters described in Farrell et al. (2002) and includes keys to match the symbols used to define the parameter in the paper with the relevant PomPy class attributes. The Demonstration.ipynb notebook and accompanying module pompy.demos give several other examples of setting up plume models using PomPy and visualising the simulations.
The below script will generate a 20 second MP4 animation (see above for animated GIF version) of a generated plume with model parameters consistent with those proposed in Farrell et al. (2002).
frompompyimportmodels, processorsimportnumpyasnpimportmatplotlib.pyplotaspltfrommatplotlib.animationimportFuncAnimation# Seed random number generatorseed=20180517rng=np.random.RandomState(seed)
# Define wind model simulation regionwind_region=models.Rectangle(x_min=0., x_max=100., y_min=-50., y_max=50.)
# Define wind model parameterswind_model_params= { 'n_x': 21,
'n_y': 21,
'u_av': 1.,
'v_av': 0.,
'k_x': 10.,
'k_y': 10.,
'noise_gain': 20.,
'noise_damp': 0.1,
'noise_bandwidth': 0.2,
'use_original_noise_updates': True
}
# Create wind model objectwind_model=models.WindModel(wind_region, rng=rng, **wind_model_params)
# Define plume simulation region# This is a subset of the wind simulation regionsim_region=models.Rectangle(x_min=0., x_max=50., y_min=-12.5, y_max=12.5)
# Define plume model parametersplume_model_params= {
'source_pos': (5., 0., 0.),
'centre_rel_diff_scale': 2.,
'puff_release_rate': 10,
'puff_init_rad': 0.001**0.5,
'puff_spread_rate': 0.001,
'init_num_puffs': 10,
'max_num_puffs': 1000,
'model_z_disp': True,
}
# Create plume model objectplume_model=models.PlumeModel(
rng=rng, sim_region=sim_region, wind_model=wind_model, **plume_model_params)
# Define concentration array (image) generator parametersarray_gen_params= {
'array_z': 0.,
'n_x': 500,
'n_y': 250,
'puff_mol_amount': 8.3e8
}
# Create concentration array generator objectarray_gen=processors.ConcentrationArrayGenerator(
array_xy_region=sim_region, **array_gen_params)
# Set up figurefig=plt.figure(figsize=(5, 2.5))
ax=fig.add_axes([0., 0., 1., 1.])
ax.axis('off')
# Display initial concentration field as imageconc_array=array_gen.generate_single_array(plume_model.puff_array)
conc_im=ax.imshow(
conc_array.T, extent=sim_region, vmin=0., vmax=1e10, cmap='Reds')
# Simulation timestepdt=0.01# Run wind model forward to equilibrateforkinrange(2000):
wind_model.update(dt)
# Define animation update functiondefupdate(i):
# Do 10 time steps per frame updateforkinrange(10):
wind_model.update(dt)
plume_model.update(dt)
conc_array=array_gen.generate_single_array(plume_model.puff_array)
conc_im.set_data(conc_array.T)
return [conc_im]
# Animate plume concentration and save as MP4anim=FuncAnimation(fig, update, frames=400, repeat=False)
anim.save('plume.mp4', dpi=100, fps=20, extra_args=['-vcodec', 'libx264'])