Flotation control -- A multivariable stabilizer

This paper presents a stabilizing controller for flotation plants which uses a quasi-multivariable technique. The controller monitors all the levels in the plant, and by anticipating interactions between various parts of the plant, is able to stabilize the plant far more successfully than the normal plant control. Once stabilizing control has been achieved, optimization of the process becomes easier and more sustainable. An estimate of the improvement in metallurgical performance is made and a singular value analysis was conducted to verify that the multivariable algorithm will theoretically control better than a collection of individual PID loops. Metallurgical results are presented to show that the improvements are attainable in practice. Control by the Mintek algorithm was alternated with normal plant control, to show that the improvements are statistically significant.