Compare Partition Separations

Demonstrate splitting a single sample and applying two different partition models along a dimension.

We will compare an ideal (perfect) partition with that with the partition model of Napier-Munn (1998)

This demonstrates why using only ore characterisation data to infer plant performance is dangerous. The robust method uses ore characterisation and process characterisation (partition) data

The two cases tested are:

• IDEAL - Perfect partition, represents using ore characterisation data only

• SIM - Simulated reality with Napier-Munn partition, represents using ore characterisation and process characterisation data.

Note

The Ep parameter injected to the Napier-Munn partition is speculative - for illustrative purposes only.

import plotly
import pandas as pd
from functools import partial

from elphick.mass_composition.flowsheet import Flowsheet
from elphick.mass_composition.utils.partition import perfect, napier_munn
from elphick.mass_composition.datasets.sample_data import size_by_assay
from elphick.mass_composition import MassComposition

# sphinx_gallery_thumbnail_number = -1

Create a mass-composition object

We get some demo data in the form of a pandas DataFrame

df_data: pd.DataFrame = size_by_assay()
df_data

# # since we intend to split the sample in half, we'll re-base the original mass to 200,
# for easier interpretation later.

df_data['mass_dry'] = df_data['mass_dry'] * 2

Create the object

mc_size: MassComposition = MassComposition(df_data, name='size sample')
print(mc_size)
mc_size.aggregate()

size sample
<xarray.Dataset> Size: 336B
Dimensions:   (size: 6)
Coordinates:
  * size      (size) object 48B [0.85, 2.0) [0.5, 0.85) ... [0.0, 0.045)
Data variables:
    mass_wet  (size) float64 48B 6.6 19.8 53.0 5.0 17.6 98.0
    mass_dry  (size) float64 48B 6.6 19.8 53.0 5.0 17.6 98.0
    H2O       (size) float64 48B 0.0 0.0 0.0 0.0 0.0 0.0
    Fe        (size) float64 48B 64.15 64.33 64.52 62.65 62.81 55.95
    SiO2      (size) float64 48B 2.04 2.05 1.84 2.88 2.12 6.39
    Al2O3     (size) float64 48B 2.68 2.23 2.19 3.32 2.25 6.34
Attributes:
    mc_name:            size sample
    mc_vars_mass:       ['mass_wet', 'mass_dry']
    mc_vars_chem:       ['Fe', 'SiO2', 'Al2O3']
    mc_vars_attrs:      []
    mc_interval_edges:  {'size': {'left': 'retained', 'right': 'passing'}}

	mass_wet	mass_dry	H2O	Fe	SiO2	Al2O3
name
size sample	200.0	200.0	0.0	60.09245	4.14753	4.27716

Visualise the data

fig = mc_size.plot_parallel(color='Fe')
fig

Split the Sample

mc_ideal_feed, mc_sim_feed = mc_size.split(0.5, 'ideal feed', 'sim feed')

Apply the Partitions

We partially initialise the two partitions The dim argument is added to inform the split method which dimension to apply the function/split to

part_ideal = partial(perfect, d50=0.150, dim='size')
part_sim = partial(napier_munn, d50=0.150, ep=0.1, dim='size')

Separate the object using the defined partitions

mc_ideal_coarse, mc_ideal_fine = mc_ideal_feed.split_by_partition(partition_definition=part_ideal,
                                                                  name_1='ideal_coarse', name_2='ideal_fine')
mc_sim_coarse, mc_sim_fine = mc_sim_feed.split_by_partition(partition_definition=part_sim, name_1='sim_coarse', name_2='sim_fine')


fs: Flowsheet = Flowsheet().from_streams([mc_size, mc_ideal_feed, mc_sim_feed,
                                           mc_ideal_coarse, mc_ideal_fine,
                                           mc_sim_coarse, mc_sim_fine])

fig = fs.table_plot(table_pos='left',
                     sankey_color_var='Fe', sankey_edge_colormap='copper_r', sankey_vmin=50, sankey_vmax=70)
# noinspection PyTypeChecker
plotly.io.show(fig)  # this call to show will set the thumbnail for use in the gallery

Note

The mass split and grades are different as shown in the table above. The difference in coarse yield (mass recovery) of the two cases is 7.4%. More work reviewing recovery of components in both cases would be prudent. This is illustrative only but demonstrates why using ore characterisation and ignoring process characterisation to capture the real world process inefficiencies is dangerous.