1882 XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3
process circumstances is shown by the model equation.
These empirical formulas can be used to predict the tailings
leaching recoveries at any chosen parameter level and deter-
mine the relative importance of the components according
to their coefficients.
Ni recovery (%)=12.17 +2.302n1 +1.060n2 +50.59n3
+3.12n4 – 1.623n5 – 0.1492n12
– 0.03873n22 – 23.50n32
– 0.1503n42 +0.01747n52
+0.06248n1n2 – 0.646n1n3
– 0.0448n1n4 – 0.01099n1n5
– 1.1617n2n3 +0.07824n2n4
+0.016360n2n5 – 0.197n3n4
+0.2497n3n5 +0.00182n4n5 (2)
Cu recovery (%)=–16.96 +5.069n1 +2.056n2 +62.65n3
+7.41n4 – 1.570n5 – 0.1670n12
– 0.002631n22 – 16.10n32
– 0.2725n42 +0.01972n52
+0.05037n1n2 – 1.238n1n3
– 0.1248n1n4 – 0.03993n1n5
– 1.7871n2n3 +0.06359n2n4
+0.00673n2n5 – 2.736n3n4
+0.6376n3n5 – 0.001195n4n5 (3)
The analysis of variance (ANOVA) was utilized to assess the
model this method looked into how different components
interacted with one another and how each of them inde-
pendently affected the set response (Owusu et al., 2022
Tang &Steenari, 2016). The significance, insufficient fit-
ness, regression, and dependability of the RSM model are
all statistically measured by the analysis of variance (Arshadi
et al., 2016 Nazari et al., 2014). The model’s significance
is usually calculated using probability values denoted by
F and P, and it is based on the relationship between the
expected and actual experimental outcomes (Abioye et al.,
2023 Owusu et al., 2022).
Consequently, a model that is deemed acceptable has
lower P-values (≤0.05) and higher F-levels at a 95% confi-
dence level. Furthermore, the accuracy of the experimental
results is estimated using the regression model (R2), whose
values range from 0 to 1 (unity) a greater R2 denotes a
more accurate model (Abioye et al., 2023 Owusu et al.,
2022).
It is noteworthy that the lack of fit was used to calcu-
late the systematic error. Table 6 present the results of the
study’s analysis of variance. Focusing on the recovery of Ni,
the model’s F-value of 288.24 suggests that it is significant,
with noise having a mere 0.01% probability of explaining
such a high F-value. In addition, the p-value was less than
0.05, demonstrating the importance of the factors that were
modelled. It is important to note that the pure error was far
larger than the lack of fit value. Tables A1 present model
of Cu. This model also showed lower P-values (0.0001)
and higher F-values, indicating that they are significant
and well-fitted models that closely agree with experimental
leaching results with a 95% confidence level.
The fit statistics, which include the regression tests (R2)
used to assess how well the model anticipated the leach-
ing experiment’s results, are displayed in Table 7. R2 values,
which normally fall between 0 and 1, are crucial metrics for
confirming the validity of any regression model. It shows
the degree to which the regression model and the experi-
mental data agree. It is important to note that, as per (Garg
&Jain, 2020), a minimum R2 value of 0.80 is a reliable
Table 5. The central composite response table for the
leaching of Ni and Cu
Standard
Order Run
Ni Recovery,
%
Cu Recovery,
%
4 1 28.12 32.47
20 2 72.57 90.10
8 3 43.69 61.10
23 4 42.81 51.16
25 5 26.70 52.70
9 6 5.56 8.36
1 7 55.94 86.21
27 8 42.83 54.14
10 9 66.84 78.97
11 10 40.92 56.61
6 11 28.34 40.36
18 12 48.23 69.85
17 13 32.75 43.83
12 14 43.04 54.54
21 15 92.57 99.87
29 16 32.91 44.43
3 17 38.25 52.65
2 18 46.22 56.96
16 19 2.61 6.02
22 20 42.65 54.91
5 21 57.70 97.52
14 22 27.27 31.66
7 23 42.30 76.41
28 24 12.99 33.69
26 25 41.97 51.88
32 26 38.07 48.51
13 27 4.09 4.60
19 28 1.32 4.93
15 29 61.62 100.00
30 30 42.87 53.85
24 31 18.12 27.82
31 32 75.63 98.43
process circumstances is shown by the model equation.
These empirical formulas can be used to predict the tailings
leaching recoveries at any chosen parameter level and deter-
mine the relative importance of the components according
to their coefficients.
Ni recovery (%)=12.17 +2.302n1 +1.060n2 +50.59n3
+3.12n4 – 1.623n5 – 0.1492n12
– 0.03873n22 – 23.50n32
– 0.1503n42 +0.01747n52
+0.06248n1n2 – 0.646n1n3
– 0.0448n1n4 – 0.01099n1n5
– 1.1617n2n3 +0.07824n2n4
+0.016360n2n5 – 0.197n3n4
+0.2497n3n5 +0.00182n4n5 (2)
Cu recovery (%)=–16.96 +5.069n1 +2.056n2 +62.65n3
+7.41n4 – 1.570n5 – 0.1670n12
– 0.002631n22 – 16.10n32
– 0.2725n42 +0.01972n52
+0.05037n1n2 – 1.238n1n3
– 0.1248n1n4 – 0.03993n1n5
– 1.7871n2n3 +0.06359n2n4
+0.00673n2n5 – 2.736n3n4
+0.6376n3n5 – 0.001195n4n5 (3)
The analysis of variance (ANOVA) was utilized to assess the
model this method looked into how different components
interacted with one another and how each of them inde-
pendently affected the set response (Owusu et al., 2022
Tang &Steenari, 2016). The significance, insufficient fit-
ness, regression, and dependability of the RSM model are
all statistically measured by the analysis of variance (Arshadi
et al., 2016 Nazari et al., 2014). The model’s significance
is usually calculated using probability values denoted by
F and P, and it is based on the relationship between the
expected and actual experimental outcomes (Abioye et al.,
2023 Owusu et al., 2022).
Consequently, a model that is deemed acceptable has
lower P-values (≤0.05) and higher F-levels at a 95% confi-
dence level. Furthermore, the accuracy of the experimental
results is estimated using the regression model (R2), whose
values range from 0 to 1 (unity) a greater R2 denotes a
more accurate model (Abioye et al., 2023 Owusu et al.,
2022).
It is noteworthy that the lack of fit was used to calcu-
late the systematic error. Table 6 present the results of the
study’s analysis of variance. Focusing on the recovery of Ni,
the model’s F-value of 288.24 suggests that it is significant,
with noise having a mere 0.01% probability of explaining
such a high F-value. In addition, the p-value was less than
0.05, demonstrating the importance of the factors that were
modelled. It is important to note that the pure error was far
larger than the lack of fit value. Tables A1 present model
of Cu. This model also showed lower P-values (0.0001)
and higher F-values, indicating that they are significant
and well-fitted models that closely agree with experimental
leaching results with a 95% confidence level.
The fit statistics, which include the regression tests (R2)
used to assess how well the model anticipated the leach-
ing experiment’s results, are displayed in Table 7. R2 values,
which normally fall between 0 and 1, are crucial metrics for
confirming the validity of any regression model. It shows
the degree to which the regression model and the experi-
mental data agree. It is important to note that, as per (Garg
&Jain, 2020), a minimum R2 value of 0.80 is a reliable
Table 5. The central composite response table for the
leaching of Ni and Cu
Standard
Order Run
Ni Recovery,
%
Cu Recovery,
%
4 1 28.12 32.47
20 2 72.57 90.10
8 3 43.69 61.10
23 4 42.81 51.16
25 5 26.70 52.70
9 6 5.56 8.36
1 7 55.94 86.21
27 8 42.83 54.14
10 9 66.84 78.97
11 10 40.92 56.61
6 11 28.34 40.36
18 12 48.23 69.85
17 13 32.75 43.83
12 14 43.04 54.54
21 15 92.57 99.87
29 16 32.91 44.43
3 17 38.25 52.65
2 18 46.22 56.96
16 19 2.61 6.02
22 20 42.65 54.91
5 21 57.70 97.52
14 22 27.27 31.66
7 23 42.30 76.41
28 24 12.99 33.69
26 25 41.97 51.88
32 26 38.07 48.51
13 27 4.09 4.60
19 28 1.32 4.93
15 29 61.62 100.00
30 30 42.87 53.85
24 31 18.12 27.82
31 32 75.63 98.43