On-Line Fault Detection on a Pilot Flotation Column Using Linear PCA Models

Abstract On-line fault detection, for instrumentation and process operation, has become important part of industrial programs leading to improve process operation and therefore product quality over time. Multivariate statistical projection methods, such as Principal Component Analysis (PCA), have been proposed to effectively deal with these situations. In this work, a pilot flotation column is operated under distributed control of froth depth, gas hold up and bias, to experimentally collect operation data at steady state, to build a PCA model. The basic control is implemented in a PLC, and all data is communicated to a PC network for displaying and further processing, under Intouch software. The column is operated in a hybrid form, for the air water system, while concentrate and tailing grades are obtained by on line predictions by using a static metallurgical model. A steady state on-line detector has been implemented on the PC to test when the collected data met the requirements to be used to build a PCA model. Several examples are discussed, detecting both particular instrumentation failures and abnormal operating conditions, and how using the system suggestions the metallurgical objectives of the process are met again.