Propagation Of Measurement Errors In Mass Balance Calculation Of Mineral Processing Data

Various computer packages are now available for balancing data obtained from complex concentration plants. It is well established that the filtered data generated by these programs are statistically better than the raw data. However the degree of improvement as well as the accuracy of the estimated variables depend strongly on the propagation of measurement errors through the estimation algorithm. Therefore, for the subsequent use of the filtered data, their confidence intervals are as important as the data themselves. Variance of filtered variables have been derived in a very general way for any arbitrary flowsheet and when data originate from various sources such as flowrates, chemical assays, particle size distributions and pulp percentages of solids. The theoretical formulae are incorporated in the BILMAT package for mass balance calculation and illustrated on various flowsheets involving grinding, classification and flotation. The validity of the proposed method is discussed by a comparison to a Monte Carlo approach, for a simple processing circuit.