XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3 1989
continuous flow model. All remaining domain boundar-
ies are assigned as fixed walls. The experimental procedure,
operating parameters and equipment specifications are
detailed in the literature (Starrett &Galvin, 2023).
An unstructured mesh is used for mesh construction.
Inlets and the inclined section are refined locally to ensure
structural integrity. Mesh sensitivity analysis was performed
in the early stage of study to determine the appropriate level
of mesh refinement, result accuracy and computational
time. The results presented in this paper were obtained
using a grid offering a reasonable trade-off between grid
resolution and simulation execution time.
The governing equations are solved in a three-dimen-
sional model using the finite volume method and a com-
mercial CFD software Star CCM+ is used for simulations.
The SIMPLE algorithm is incorporated in the CFD model
to facilitate pressure-velocity coupling and correction. The
second-order upwind scheme is used for discretization of
the convective terms. For the transient simulations, a time-
step of 1×10–4 s is implemented, allowing for a maximum of
20 iterations per time-step. Convergence of the algorithm
is confirmed via continuous monitoring of residuals dur-
ing simulations. The batch mode model is simulated over a
time period of 60 s, while the continuous flow simulations
extend for a duration exceeding 120 s each.
RESULTS AND DISCUSSION
Model validation
The CFD model was developed to study the RC following
the experimental study conducted by Starrett and Galvin
(2023) at the University of Newcastle, Australia. The exper-
imental setup involved a particle size distribution ranging
from 250 to 710 µm, while the CFD simulations focused
on mono-sized particles within this size range. Employing
mono-size particle fluidised bed CFD simulations enables
the examination of the hydrodynamic behaviour of individ-
ual particle sizes in isolation. However, it is acknowledged
that this approach neglects interactions between particles
of varying sizes. The future plan involves extending the
CFD model to incorporate multi-sized particles in a sub-
sequent study, allowing for the exploration of the impact
of interactions between particles of different sizes on bed
hydrodynamics.
A solid particle bed, initially set at a height of 14.2 cm,
was established in the same RC used for continuous flow,
excluding the side-water and feed flow for the fluidized bed
experiment. The experiment involved progressively increas-
ing the fluidization rate until reaching the maximum level
and then decreasing it to the minimum level. Initially, the
fluidization rate was increased from 0.0 L/min to 2.4 L/
min in 0.4 L/min increments, with steady-state bed height
and bed density measured using a pressure probe at each
Figure 1. Reflux Classifier (a) Schematic diagram (Starrett &Galvin, 2023), (b) 3D Computational domain
continuous flow model. All remaining domain boundar-
ies are assigned as fixed walls. The experimental procedure,
operating parameters and equipment specifications are
detailed in the literature (Starrett &Galvin, 2023).
An unstructured mesh is used for mesh construction.
Inlets and the inclined section are refined locally to ensure
structural integrity. Mesh sensitivity analysis was performed
in the early stage of study to determine the appropriate level
of mesh refinement, result accuracy and computational
time. The results presented in this paper were obtained
using a grid offering a reasonable trade-off between grid
resolution and simulation execution time.
The governing equations are solved in a three-dimen-
sional model using the finite volume method and a com-
mercial CFD software Star CCM+ is used for simulations.
The SIMPLE algorithm is incorporated in the CFD model
to facilitate pressure-velocity coupling and correction. The
second-order upwind scheme is used for discretization of
the convective terms. For the transient simulations, a time-
step of 1×10–4 s is implemented, allowing for a maximum of
20 iterations per time-step. Convergence of the algorithm
is confirmed via continuous monitoring of residuals dur-
ing simulations. The batch mode model is simulated over a
time period of 60 s, while the continuous flow simulations
extend for a duration exceeding 120 s each.
RESULTS AND DISCUSSION
Model validation
The CFD model was developed to study the RC following
the experimental study conducted by Starrett and Galvin
(2023) at the University of Newcastle, Australia. The exper-
imental setup involved a particle size distribution ranging
from 250 to 710 µm, while the CFD simulations focused
on mono-sized particles within this size range. Employing
mono-size particle fluidised bed CFD simulations enables
the examination of the hydrodynamic behaviour of individ-
ual particle sizes in isolation. However, it is acknowledged
that this approach neglects interactions between particles
of varying sizes. The future plan involves extending the
CFD model to incorporate multi-sized particles in a sub-
sequent study, allowing for the exploration of the impact
of interactions between particles of different sizes on bed
hydrodynamics.
A solid particle bed, initially set at a height of 14.2 cm,
was established in the same RC used for continuous flow,
excluding the side-water and feed flow for the fluidized bed
experiment. The experiment involved progressively increas-
ing the fluidization rate until reaching the maximum level
and then decreasing it to the minimum level. Initially, the
fluidization rate was increased from 0.0 L/min to 2.4 L/
min in 0.4 L/min increments, with steady-state bed height
and bed density measured using a pressure probe at each
Figure 1. Reflux Classifier (a) Schematic diagram (Starrett &Galvin, 2023), (b) 3D Computational domain