Modeling and fault detection of an industrial copper electrowinning process

Abstract

Copper electrowinning plants are where high purity copper (Cu) product is obtained through electrochemical reduction of copper from the leaching solution. The presence selenium (Se) and tellurium (Te) in copper sulphide minerals may result in contamination of the leach solution and, eventually of the copper cathode. Unfortunately, hydrometallurgical processes are often difficult to monitor and control due to day-to-day fluctuations in the process as well as limitations in capturing the data at high frequencies. The purpose of this work is to model key variables in the copper electrowinning tank and to apply statistical fault detection to the selenium/tellurium removal and copper electrowinning process operations. First principle modeling was applied to the copper electrowinning tank and partial differential equation models were derived to describe the process dynamics. Industrial data were used to estimate the model parameters and validate the resulting models. Comparison with industrial model shows that the models fit reasonably well with industrial operation. Simulations of the models were run to explore the dynamics under varying operating conditions. The derived models provide a useful tool for future process modification and control development. Using the collected industrial operating data, dynamic principal component analysis (DPCA) based fault detection was applied to Se/Te removal and copper electrowinning processes at Vale’s Electrowinning Plant in Copper Cliff, ON. The fault detection results from the DPCA based approach were consistent with the industrial product quality test. After faults were detected, fault diagnosis was then applied to determine the causes of faults. The fault detection and diagnosis system helps define causes of upset conditions that lead to coppercathode contamination.