3306 XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3
significant if the p-value is less than 0.05. Correspondingly,
the p-value (0.0001) of the regression model is indicative
of the adequacy of the model at a 95% confidence level (as
shown in Table 6). The terms with p-values greater than
0.05 have insignificant influences on predicted responses.
The significance of the influence of each parameter in the
regression equation on the leaching recovery of the valuable
metals was determined by the F test. In addition, the regres-
sion model (R2) with values ranging from 0 to 1 (unity)
estimates the accuracy of the experimental outcomes, where
a higher R2 indicates a more accurate model. It is worth
mentioning that the systematic error was measured by the
lack of fit. The analysis of variance in this study is displayed
in Tables 6 for the recoveries of Ni.
The F-values obtained from the F test were 268.77 for
Ni, whereas the p-values of the models were all less than
0.05. Hence, the computed quadratic regression model can
be considered to be sufficient. It is worth mentioning that
the lack of fit value was insignificant (p-value greater 0.05)
relative to the pure error. The fit statistics are presented in
Table 7, encompassing the regression tests (R2) adopted
for evaluating the predicted outcomes from the model to
the experimental outcomes from leaching. R2 values typi-
cally ranges from 0 to 1, and they are essential parameters
for verifying any regression model. It represents how well
the regression model matches the experimental data. It is
worth mentioning that a minimum R2 value of 0.80 is a
good indication of a well-fitted model, according to (Garg
and Jain 2020), a conclusion reinforced by another studies.
Consequently, the correlation plot shown in Figure 6 dem-
onstrate the strong agreement between the predicted and
actual experimental recoveries.
Response Surface Plots and Analysis
The individual plots demonstrating the individual effects
of the various input parameters for the recovery of Ni from
the rougher tailings are shown in Figure 7. Notably, these
one-factor plots are visual representations that analyzes the
overall impact of a certain process condition on the response
variables, while assuming all other factors are constant. This
analysis of the results shows that increasing the sulfuric acid
Table 6. The analysis of variance for Ni recovery (%)
Source Sum of Squares DF Mean Square F-Value p-Value
Model 9070.46 19 477.39 268.77 0.0001 Significant
H2SO4 conc.: m1 2.31 1 2.31 1.3 0.2605
Catalyst dosage: m
2 259.75 1 259.75 146.23 0.0001
Leaching temperature: m
3 5972.71 1 5972.71 3362.55 0.0001
Leaching time: m4 991.52 1 991.52 558.21 0.0001
Catalyst type: m5 11.29 1 11.29 6.36 0.0156
m
1
2 4.57 1 4.57 2.57 0.1161
m
2
2 7.87 1 7.87 4.43 0.0413
m32 82.48 1 82.48 46.43 0.0001
m42 0 1 0 0 0.9881
m
1 m
2 53.51 1 53.51 30.13 0.0001
m
1 m
3 115.67 1 115.67 65.12 0.0001
m1 m4 3.58 1 3.58 2.01 0.1632
m1 m5 75.86 1 75.86 42.71 0.0001
m
2 m
3 518.42 1 518.42 291.86 0.0001
m
2 m
4 1.85 1 1.85 1.04 0.3129
m2 m5 133.79 1 133.79 75.32 0.0001
m3 m4 292.09 1 292.09 164.44 0.0001
m
3 m
5 289.79 1 289.79 163.15 0.0001
m
4 m
5 71.15 1 71.15 40.06 0.0001
Residual 74.6 42 1.78
Lack of fit 67.85 30 2.26 0.42 0.2807 Not significant
Pure error 6.75 12 0.56
Corrected total 9145.06 61
Table 7. The fit statistics of the experimental data and
regression model
Std. dev. R2
Adjusted
R2
Predicted
R2
Ni recovery (%)1.33 0.9918 0.9882 0.9772
significant if the p-value is less than 0.05. Correspondingly,
the p-value (0.0001) of the regression model is indicative
of the adequacy of the model at a 95% confidence level (as
shown in Table 6). The terms with p-values greater than
0.05 have insignificant influences on predicted responses.
The significance of the influence of each parameter in the
regression equation on the leaching recovery of the valuable
metals was determined by the F test. In addition, the regres-
sion model (R2) with values ranging from 0 to 1 (unity)
estimates the accuracy of the experimental outcomes, where
a higher R2 indicates a more accurate model. It is worth
mentioning that the systematic error was measured by the
lack of fit. The analysis of variance in this study is displayed
in Tables 6 for the recoveries of Ni.
The F-values obtained from the F test were 268.77 for
Ni, whereas the p-values of the models were all less than
0.05. Hence, the computed quadratic regression model can
be considered to be sufficient. It is worth mentioning that
the lack of fit value was insignificant (p-value greater 0.05)
relative to the pure error. The fit statistics are presented in
Table 7, encompassing the regression tests (R2) adopted
for evaluating the predicted outcomes from the model to
the experimental outcomes from leaching. R2 values typi-
cally ranges from 0 to 1, and they are essential parameters
for verifying any regression model. It represents how well
the regression model matches the experimental data. It is
worth mentioning that a minimum R2 value of 0.80 is a
good indication of a well-fitted model, according to (Garg
and Jain 2020), a conclusion reinforced by another studies.
Consequently, the correlation plot shown in Figure 6 dem-
onstrate the strong agreement between the predicted and
actual experimental recoveries.
Response Surface Plots and Analysis
The individual plots demonstrating the individual effects
of the various input parameters for the recovery of Ni from
the rougher tailings are shown in Figure 7. Notably, these
one-factor plots are visual representations that analyzes the
overall impact of a certain process condition on the response
variables, while assuming all other factors are constant. This
analysis of the results shows that increasing the sulfuric acid
Table 6. The analysis of variance for Ni recovery (%)
Source Sum of Squares DF Mean Square F-Value p-Value
Model 9070.46 19 477.39 268.77 0.0001 Significant
H2SO4 conc.: m1 2.31 1 2.31 1.3 0.2605
Catalyst dosage: m
2 259.75 1 259.75 146.23 0.0001
Leaching temperature: m
3 5972.71 1 5972.71 3362.55 0.0001
Leaching time: m4 991.52 1 991.52 558.21 0.0001
Catalyst type: m5 11.29 1 11.29 6.36 0.0156
m
1
2 4.57 1 4.57 2.57 0.1161
m
2
2 7.87 1 7.87 4.43 0.0413
m32 82.48 1 82.48 46.43 0.0001
m42 0 1 0 0 0.9881
m
1 m
2 53.51 1 53.51 30.13 0.0001
m
1 m
3 115.67 1 115.67 65.12 0.0001
m1 m4 3.58 1 3.58 2.01 0.1632
m1 m5 75.86 1 75.86 42.71 0.0001
m
2 m
3 518.42 1 518.42 291.86 0.0001
m
2 m
4 1.85 1 1.85 1.04 0.3129
m2 m5 133.79 1 133.79 75.32 0.0001
m3 m4 292.09 1 292.09 164.44 0.0001
m
3 m
5 289.79 1 289.79 163.15 0.0001
m
4 m
5 71.15 1 71.15 40.06 0.0001
Residual 74.6 42 1.78
Lack of fit 67.85 30 2.26 0.42 0.2807 Not significant
Pure error 6.75 12 0.56
Corrected total 9145.06 61
Table 7. The fit statistics of the experimental data and
regression model
Std. dev. R2
Adjusted
R2
Predicted
R2
Ni recovery (%)1.33 0.9918 0.9882 0.9772