Harnessing color: predicting copper recovery in bioleaching processes with RGB measurement

Purpose Our study focuses on a monitoring station with a digital red, green, blue (RGB) sensor, capable of detecting subtle color changes in solutions. This aids in real-time analysis of metal dissolution and recovery at the bioreactor station. The setup includes a color sensor and an ESP32-based programmable logic controller, compatible with Arduino tools, for data processing and transmission. Prior research highlights Random Forest Regression as the preferred algorithm for estimating bioleaching recovery rates. We introduce a novel application of this algorithm, using RGB values to predict copper recovery. Design/methodology/approach This research aims to forecast copper recovery using data from a smart color sensor, guiding the bioleaching plant to halt the process at an optimal threshold to conserve energy. Currently, operators rely on sensor readings, experience and occasional analysis. We use root mean square error Pearson correlation coefficient and coefficient of determination to evaluate performance. Four bioleaching trials were conducted to establish a relationship between Fe2+/Fe3+ color and concentration. Each trial used 16g of shredded cables, 10g of 3D-printed filler and 1.3 L of Fe3+ solution. Multiple samples tracked Fe2+/Fe3+ values and their color, with a pH sensor on-site. Findings The proposed method can well-estimate the concentrations of Cu(II) and Fe(III) ions from the color of the solution and pH on the training set, both when working with PCBs, with determination coefficient equal to 0.97 for Cu(II) and equal to 0.98 for Fe (III), respectively. Therefore, it was determined that the model predicts the concentration of Fe(III) more accurately than that of Cu(II). In summary, the results indicate a promising research direction concerning the correlation between the color sensor and the recovered material's value. Originality/value It is advisable to augment the dataset used for model training and explore alternative independent variables. For instance, one could explore excluding the blue value from the sensor data. Subsequent to this adjustment, conducting a comparative analysis of the resulting values is essential for refining the experiment. Furthermore, optimizing all facets of the real-time bioleaching monitoring system and control infrastructure, including color sensors, harbors the potential for substantial advancements in large-scale bioleaching plants.