Sag mill power draw simulation using variance propagation approach

Most mineral processing system models include variables that have both systematic errors (bias) and associated randomness. Therefore, in analyzing these models it is essential to incorporate their uncertainties. In general, there are a couple of different methods of variance or uncertainty analysis used by analysts such as sensitivity analysis, Monte Carlo simulation, etc. In this paper, the law of propagation of errors (LPE) and one of the favorite analysis tools, the Monte Carlo simulation using Latin hypercube sampling, are presented. Limitations of the LPE are emphasized and simple applications of these two methods in mineral processing are also briefly shown. A case study of evaluating the variance of semi-autogenous (SAG) power draw in a grinding circuit is presented as well. Using Monte Carlo simulation showed that the mean power draw of the Sarcheshmeh SAG mill is about 9349.75 kW with a standard deviation of 983.2 kW, with the highest frequency occurring between 9058.5 and 9862.5 kW. Comparing the results of LPE and Monte Carlo simulation for the Sarcheshmeh SAG mill power draw showed that the mean is slightly higher than calculated, (9441.1kW), while its standard deviation is lower than the other one.