Integrity Monitoring Of Xstrata Copper's Kidd Metallurgical Division Mitsubishi 3-Line Furnaces Using Multivariate Methods

Extending the campaign life of the Mitsubishi furnaces at Xstrata Copper's Kidd Metallurgical Division Copper Smelter has been a major focus of attention for many years. One aspect of this effort is real time monitoring to detect abnormal conditions and safely shut the furnaces down before significant damage occurs. Traditional real time methods of monitoring the integrity of refractory furnaces containing molten metal include alarms on individual temperatures of the refractory or cooling water. The need to set alarm limits which are not exceeded across the range of normal operating conditions leads to a system that only provides a limited warning, if any, before a failure occurs. Integrity monitoring systems that look at many different signals and alarm only changes in the relationships between the signals have been implemented on the smelting and converting furnaces. The systems are based on principle components analysis (PCA) models and implemented with the monitorMV software package. These provide much more sensitive and selective alarms so that changes in integrity are alarmed well in advance of any event. At a critical alarm level, the converting furnace is set to automatically shut down so that operating and technical personnel can review the situation. The configuration of the systems and operating experience with them will be discussed.