Investigating an approach to parameter fitting for the development of a semi-empirical electrowinning model

A steady state electrowinning model was developed from first principles to predict copper yield, current efficiency and specific energy consumption. The electrochemical reactions incorporated were copper reduction, water oxidation and the cyclic reduction and oxidation of iron as impurity. This paper details a parameter fitting approach for this model which involves calibration of the model to bench-scale electrowinning data. A factorial experimental design, in which the electrolyte composition and current density were varied, was used. Each electrochemical reaction required parameters for rate expressions, and a current loss parameter was included to account for stray currents, other hardware losses and side reactions. Average current loss for the bench-scale experiments was 0.145 A (1-5% of total current). The kinetics parameters fit relatively well to the experimental data, with R2adj values of 0.864 for copper reduction, 0.739 for water oxidation, 0.724 for iron reduction and 0.661 for iron oxidation. The average relative errors between the predictive model and experimental data were 3.2% for current efficiency, 3.0% for specific energy consumption and 7.0% for copper plating rate. The results demonstrate the ability of the model to predict performance of a system in which the parameter fitting approach has been specifically applied. The parameters could be used as is for operator training and as a decision making tool. Industrial data varies substantially between plants, and therefore unique parameters would need to be fit specifically for a system should the model be used for performance prediction.

Automated summary

A steady state electrowinning model was developed to predict copper yield, current efficiency, and specific energy consumption, with parameters fitting relatively well to experimental data. The model demonstrated the ability to predict system performance and could be used for operator training and decision making. Unique parameters would need to be fit specifically for different industrial plants for accurate performance prediction.