Selecting the Number of Increments for Daily Metallurgical Samples: Application to Goldcorp Eleonore Gold Ore Concentrator, Quebec, Canada

"The operators of mineral processing plants collect daily samples from strategic process streams with a view to prepare production balances to assess plant performances. The rules to minimize the risk of systematic sampling errors are known and applied in many mineral processing plants. The mass of the sample to be collected is calculated to yield a target measurement reproducibility considering the fundamental error associated to the sampling of heterogeneous material. This paper demonstrates that this practice is partly valid if one does not account for the time variation of the ore composition during the day. The concept is demonstrated using samples collected on a 30 minute basis from different streams of a gold ore concentrator. Results show that depending of the natural variability of the stream characteristics the mass of the daily samples may not be the critical variable to obtain reproducible measurements. On the other hand the number of cuts taken during the day could impact on the variance of the measurements obtained from daily samples.IntroductionMineral processing plants may treat several thousand tons of ore every day to produce concentrate or impure gold ingots. The operation and performances of these plants are monitored on a daily basis through the collection of ore samples from various streams of the processing circuit. The samples are prepared and assayed to obtain the concentrations of target elements that are subsequently used to calculate the mass flowrates in the streams of the circuit and ultimately the distribution of the target metals for the estimation of the plant performances (Morrison, 2009; Lachance and Leroux, 2016).The acquisition of representative and reproducible samples is therefore the first step in the characterization of plant performances. The term representative sample is used here to indicate an unbiased sample, i.e. a sampling procedure that does not introduce a systematic error. Rules for the collection of unbiased samples are reviewed in the literature by different authors (e.g. Morrison, 2009) and are not discussed in this paper. The term reproducible indicates the dispersion of the measurements around the average value of a characteristic of a stream meets an acceptable level of variance. The reproducibility of a measurement process (sampling and analysis) is assessed by the relative standard deviation or coefficient of variation of the measurement that is a function of the sampling and assaying errors (Morrison. 2009; Gy, 1979). Usually the focus is put on the fundamental error of sampling whose variance can be controlled by the choice of the primary sample mass or adequate particle size reduction during the preparation of secondary or tertiary samples extracted for the chemical analysis. When a sample is taken from a stream that exhibits a time varying characteristic, the obtained measurement exhibits a variability that is not only a function of the sample mass but also of the natural variability of the ore in the sampled stream. This paper examines this aspect of the sampling of a time varying process on the reproducibility of composite samples for a day of production."