XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3 3227
To shed light on the materials flow behind the opti-
mal O1 mass recovery, the mass flow of NMC and graph-
ite were studied throughout the entire process. The Sankey
diagrams in Figure 6 depict the fractions of the input mass
exiting the process at intermediate output streams, before
the final separation into tailings and concentrate. Graphite
data illustrates such efficient separation in the last stage,
that more mass was output in intermediate stages than the
concentrate. Findings for NMC are similar: 8% of the mass
leaves the process in the intermediate stages. These find-
ings constitute the best materials flow outcomes among the
10000 we considered.
CONCLUSIONS
Although process simulation software has been a useful
tool in the evaluation of process designs for a long time,
the possibility to analyze process outcomes through a data-
driven approach has been long overlooked. As shown in
the work hereby presented, an automated computational
methodology can be used to explore thousands of operating
conditions and identify under which of them a target out-
come is possible, or even if it is not achievable at all. In the
specific battery recycling process analyzed, it is seen that
only 13 sets of operating conditions out of 10000 could
produce a recovery of graphite and NMC of 85% simul-
taneously. If these are considered conditions of acceptable
performance, they represent a “needle in a haystack” type
of solution, which would have likely been overlooked by
operators manually setting the conditions in a simulator.
This original approach opens an entirely new way of using
process simulation software that may facilitate the process
engineering work, particularly for complex designs.
REFERENCES
Aspen Plus. 2022. Leading Process Simulation Software.
AspenTech. 2023. https://www.aspentech.com/en/
products/engineering/aspen-plus.
Roine, A., and Kobylin. P., 2022. HSC Chemistry Package.
En. Metso. https://www.hsc-chemistry.com/.
Table 3. Process variable combinations that produce the top 3 optimal multi-objective recovery of NMC and graphite. O1, O2
and t data points are visualized in Figure 5.
Process Parameter O1 O2 O3 Unit
Pre-Milling Fixed SGE 5.08 5.54 5.6 kWh/t
Rosin-Rammler Slope 2.9 1.8 1.5 -
Magnetic Separator Magnetic Field Strength 0.007 0.003 0.01 Tesla
Interstitial Velocity 2.8 1.61 1.25 m/s
Screener Nominal Aperture Size 8 13 8 mm
Separation Sharpness α (alpha) 1.6 0.8 2.6 —
Milling Fixed SGE 1.55 0.27 4.98 kWh/t
Rosin-Rammler Slope 0.7 2 0.1 —
Flotation Cell 1 Fixed Residence Time Target 21.87 26.72 7.2 min
Flotation Cell 2 Fixed Residence Time Target 20.74 16.69 26.76 min
Tailings NMC Recovered mass 87.6 87.7 85.8 %
Concentrate Graphite Recovered Mass 85.9 85.4 86.4 %
Figure 6. Sankey diagrams of O1 process materials flow for a) graphite and b) NMC in all output streams
To shed light on the materials flow behind the opti-
mal O1 mass recovery, the mass flow of NMC and graph-
ite were studied throughout the entire process. The Sankey
diagrams in Figure 6 depict the fractions of the input mass
exiting the process at intermediate output streams, before
the final separation into tailings and concentrate. Graphite
data illustrates such efficient separation in the last stage,
that more mass was output in intermediate stages than the
concentrate. Findings for NMC are similar: 8% of the mass
leaves the process in the intermediate stages. These find-
ings constitute the best materials flow outcomes among the
10000 we considered.
CONCLUSIONS
Although process simulation software has been a useful
tool in the evaluation of process designs for a long time,
the possibility to analyze process outcomes through a data-
driven approach has been long overlooked. As shown in
the work hereby presented, an automated computational
methodology can be used to explore thousands of operating
conditions and identify under which of them a target out-
come is possible, or even if it is not achievable at all. In the
specific battery recycling process analyzed, it is seen that
only 13 sets of operating conditions out of 10000 could
produce a recovery of graphite and NMC of 85% simul-
taneously. If these are considered conditions of acceptable
performance, they represent a “needle in a haystack” type
of solution, which would have likely been overlooked by
operators manually setting the conditions in a simulator.
This original approach opens an entirely new way of using
process simulation software that may facilitate the process
engineering work, particularly for complex designs.
REFERENCES
Aspen Plus. 2022. Leading Process Simulation Software.
AspenTech. 2023. https://www.aspentech.com/en/
products/engineering/aspen-plus.
Roine, A., and Kobylin. P., 2022. HSC Chemistry Package.
En. Metso. https://www.hsc-chemistry.com/.
Table 3. Process variable combinations that produce the top 3 optimal multi-objective recovery of NMC and graphite. O1, O2
and t data points are visualized in Figure 5.
Process Parameter O1 O2 O3 Unit
Pre-Milling Fixed SGE 5.08 5.54 5.6 kWh/t
Rosin-Rammler Slope 2.9 1.8 1.5 -
Magnetic Separator Magnetic Field Strength 0.007 0.003 0.01 Tesla
Interstitial Velocity 2.8 1.61 1.25 m/s
Screener Nominal Aperture Size 8 13 8 mm
Separation Sharpness α (alpha) 1.6 0.8 2.6 —
Milling Fixed SGE 1.55 0.27 4.98 kWh/t
Rosin-Rammler Slope 0.7 2 0.1 —
Flotation Cell 1 Fixed Residence Time Target 21.87 26.72 7.2 min
Flotation Cell 2 Fixed Residence Time Target 20.74 16.69 26.76 min
Tailings NMC Recovered mass 87.6 87.7 85.8 %
Concentrate Graphite Recovered Mass 85.9 85.4 86.4 %
Figure 6. Sankey diagrams of O1 process materials flow for a) graphite and b) NMC in all output streams