XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3 1561
elements such as Fe, SiO2, Al2O3, P, Mn, MgO, CaO, and
TiO2. Furthermore, thermogravimetric analysis provided
insights into the volatile content of the samples through
loss on ignition (LOI) measurements. Real density mea-
surements of solids were obtained using the Quantachrome
1200E Ultrapycnometer with helium gas, allowing for pre-
cise determinations of real density for individual samples
and the desliming underflow.
Mineralogical characterization focused primarily on
the desliming underflow and rougher flotation concen-
trates, involving detailed analyses of mineral liberation and
composition. Techniques such as quantitative evaluation
of minerals by scanning electron microscopy (QEMSCAN
Quanta 650) and optical microscopy were employed, pro-
viding valuable insights into the spatial distribution and
associations of minerals within the ore samples. This com-
prehensive approach to mineralogical and chemical charac-
terization establishes a solid foundation for understanding
ore behavior during mineral processing operations.
Geometallurgical Model Generation
The model generation involved a multi-faceted approach
combining process standard tests simulating the circuit of
the VGR2 at the bench scale with the chemical, mineral-
ogical, and granulometric characteristics of different ore
types, primarily compact itabirite (IC) and friable itabirite
(IF). Statistical models were constructed by integrating
these results. The study utilized multiple regression mod-
els to forecast metallurgical parameters based on geological
variables, employing a statistical approach that establishes
the relationship between a dependent variable (Y) and two
or more independent variables (X1, X2,…, Xn), ensuring
their correlation. Statistical treatment and spatial visu-
alization of the data were performed using software such
as Minitab, ioGas, and Leapfrog Geo, in addition to the
Notebook Colab platform. For the development of 3D
models, LeapfrogGeo software and Notebook Colab were
utilized. Multiple linear regression and regression tree
models were directly applied to each block of the ABO and
HZT models, facilitating a comprehensive analysis of ore
behavior and processing characteristics.
Validation of the Geometallurgical Model and
Comparison with Industrial Results
In order to validate the Geometallurgical Model and com-
pare the projected mass and metallurgical recovery data
with the industrial outcomes at VGR2, a comprehensive
reconciliation campaign was executed. Figure 2 illustrates
the ore flow process leading up to the feeding of VGR2
plant.
Utilizing the commencement and conclusion control
mechanisms for the formation and resumption of each feed
pile supplying the plant, coupled with a developed system
integrating dispatch system data with the geometallurgi-
cal model embedded within the geological block model,
enabled the retrieval of projected mass and metallurgical
recovery data for each feed pile at the VGR2 plant. This
system systematically retrieves truck data directed to the
plant, originating from both mining faces and stockpiles,
and provides geological and geometallurgical model details
for each loading point, as represented in Figure 3.
A comparative assessment was conducted between the
projected model data and the industrial outcomes at the
plant at two distinct periods:
1. Piles formed between August and October 2022,
encompassing Run of Mine (ROM) from the
ABO and HZT mines alongside a low proportion
of rich fine ROM.
2. Piles formed in December 2022, comprising ROM
exclusively from the ABO and HZT mines.
Figure 2. Diagram of the ore flow process preceding the supply to the VGR2 plant
elements such as Fe, SiO2, Al2O3, P, Mn, MgO, CaO, and
TiO2. Furthermore, thermogravimetric analysis provided
insights into the volatile content of the samples through
loss on ignition (LOI) measurements. Real density mea-
surements of solids were obtained using the Quantachrome
1200E Ultrapycnometer with helium gas, allowing for pre-
cise determinations of real density for individual samples
and the desliming underflow.
Mineralogical characterization focused primarily on
the desliming underflow and rougher flotation concen-
trates, involving detailed analyses of mineral liberation and
composition. Techniques such as quantitative evaluation
of minerals by scanning electron microscopy (QEMSCAN
Quanta 650) and optical microscopy were employed, pro-
viding valuable insights into the spatial distribution and
associations of minerals within the ore samples. This com-
prehensive approach to mineralogical and chemical charac-
terization establishes a solid foundation for understanding
ore behavior during mineral processing operations.
Geometallurgical Model Generation
The model generation involved a multi-faceted approach
combining process standard tests simulating the circuit of
the VGR2 at the bench scale with the chemical, mineral-
ogical, and granulometric characteristics of different ore
types, primarily compact itabirite (IC) and friable itabirite
(IF). Statistical models were constructed by integrating
these results. The study utilized multiple regression mod-
els to forecast metallurgical parameters based on geological
variables, employing a statistical approach that establishes
the relationship between a dependent variable (Y) and two
or more independent variables (X1, X2,…, Xn), ensuring
their correlation. Statistical treatment and spatial visu-
alization of the data were performed using software such
as Minitab, ioGas, and Leapfrog Geo, in addition to the
Notebook Colab platform. For the development of 3D
models, LeapfrogGeo software and Notebook Colab were
utilized. Multiple linear regression and regression tree
models were directly applied to each block of the ABO and
HZT models, facilitating a comprehensive analysis of ore
behavior and processing characteristics.
Validation of the Geometallurgical Model and
Comparison with Industrial Results
In order to validate the Geometallurgical Model and com-
pare the projected mass and metallurgical recovery data
with the industrial outcomes at VGR2, a comprehensive
reconciliation campaign was executed. Figure 2 illustrates
the ore flow process leading up to the feeding of VGR2
plant.
Utilizing the commencement and conclusion control
mechanisms for the formation and resumption of each feed
pile supplying the plant, coupled with a developed system
integrating dispatch system data with the geometallurgi-
cal model embedded within the geological block model,
enabled the retrieval of projected mass and metallurgical
recovery data for each feed pile at the VGR2 plant. This
system systematically retrieves truck data directed to the
plant, originating from both mining faces and stockpiles,
and provides geological and geometallurgical model details
for each loading point, as represented in Figure 3.
A comparative assessment was conducted between the
projected model data and the industrial outcomes at the
plant at two distinct periods:
1. Piles formed between August and October 2022,
encompassing Run of Mine (ROM) from the
ABO and HZT mines alongside a low proportion
of rich fine ROM.
2. Piles formed in December 2022, comprising ROM
exclusively from the ABO and HZT mines.
Figure 2. Diagram of the ore flow process preceding the supply to the VGR2 plant