Hello Ronald,

It would certainly be nice to have a probe that measures the degree of segregation for powders in bulk bags. Even if such a system were ever developed, it would be difficult to calibrate. In fact, the composition and distribution (segregation) components of the sampling variance are easy to estimate with on-stream analyzers. Ironically, the very process of determining these components under dynamic conditions reduces the segregation variance.

Kind regards,

Jan W Merks ■

Hi Ron,

I refer to on-stream analyzers when XRF modules are used to measure one or more random variables in a bulk solid such as coal or cement, or in a slurry flow at a mineral processing plant. In the latter case, different sample flows may be measured with one XRF module, or each sample flow may have its own.

When you refer to "segregation", do you mean the separation of particles due to differences in density or size? I try to avoid this term for several reasons. Visman's sampling theory defines the variance of the primary sample selection stage as the sum of the composition and segregation variances. Visman did agree that "segregation" suggests that the sampling unit was more homogeneous at an earlier time, and that "distribution variance" is a more intuitive measure for the degree of heterogeneity of a random variable in a sampling unit.

For example, if the distribution variance is statistically identical to zero, the sampling unit is homogeneous, and the composition variance determines the variance of the primary sample selection stage. If some random variable in a homogeneous sampling unit could be determined with an on-stream analyzer, the ordered set of measured values would not display a significant degree of spatial dependence. In other words, the first variance term of the ordered set would be statistically identical to the variance of the randomly distributed set. Under such conditions, a sampling variogram cannot be constructed.

So what does the term "segregation" mean in your work? And what is the random variable of interest in your powder

Kind regards,

Jan W Merks ■