Abstract

Ball mills are normally used for grinding in the mining industry. The grinding operation is characterized by high energy consumption and low efficiency. This inefficiency lies mainly in the difficulty in identifying the level of steel balls inside the mill, which is essential to operate the mill at its optimal filling point. This work presents an alternative to enhance the existing ball level estimation model for semi-autogenous mills at the Doña Inés de Collahuasi mining company. The proposed approach corrects the model by employing a digital filter, which attenuates the deviations attributable to intrinsic noise sources in the current procedure. To validate the proposal, a Chebyshev type I infinite impulse response filter designed from the tools available in Matlab was implemented and tested using real plant data. The results obtained validate the proposal presented. The filter managed to reduce the average percentage error in the ball level estimation.