A range of freely-available tools and learning resources to assist comminution engineers. %

#### Output

 Accuracy (+/-) % Number of samples Two-sided t-Score Coefficient of variation %

To learn more, watch Session 2 of the SMC Masterclass series that demonstrates a worked example.

Sample coefficient of variation (cvs): Variability as defined by the standard deviation divided by the mean of the hardness values obtained from the n samples. In a greenfield design situation where there is no pre-existing hardness variability data, a reasonable starting assumption is that the cv is equal to mean value of the other deposits containing your ore type in the SMC Testing database of 1900+ deposits. This can be selected by using the "SMC Preset" option. The overall mean value of all ore types is 31%.

Accuracy (%): this is the range that the mean of the n hardness values will fall in with the probability defined by the confidence level. The mean and associated accuracy are vaild only for the volume of the orebody that the n samples were taken from, providing the samples were evenly distributed throughout this volume. This volume, divided by the treatment rate of the comminution circuit (expressed as cu.m/hr) gives the time that it takes for the volume to be processed. This time period is known as the resolution associated with the accuracy. Hence, if the n samples were taken from a volume which will take 1 year to process then the resolution would be annual and the accuracy and coefficient of variation refer to the mean annual hardness for that year only. If, say, the same accuracy was required for estimating the mean monthly hardness then the resolution would be monthly and the n samples would have to be taken from monthly volumes.

Confidence level (cl): probability that the mean of the n hardness values will fall within the range indicated by the accuracy value. It can be thought of as being related to the risk of getting the wrong answer, in this case getting a mean hardness value from the n samples which is further from the true value than expected from the indicated accuracy. For example with a confidence level of 95% the risk of getting the wrong answer is 5% and with a 90% cl the risk is 10% etc.