The Impact of Geological Variation when Quality Controlling Precision
Consider a manufacturer of soft drinks who wants to quality control the precision of volume of liquids inserted into bottles. Their product is homogeneous. Sampling conditions are controlled. The number of samples they take is unlimited.
Contrast this with the challenges presented by the Mining and Exploration Industry. Samples and sampling differ markedly from that in manufacturing:
- Heterogeneous product (Geological variation is significant)
- Differing sample types (core, cuttings, stream etc.)
- Different sampling methods. Different sampling/preparation stages (field duplicates, crushing, pulverising, instrument repeats etc)
- Limit of two samples only (usually). Paired data (duplicates).
- Different analytical methods
Without a proper quality assurance program and supporting quality control, these errors would be missed. Geological variation is the largest consistent contribution to error.
A field duplicate, despite our best efforts, is not a duplicate. It is as close to a duplicate as we can make. However, the product (core, cuttings) may be so variable, because of lithology, mineralogy or alteration, that no effort can produce a ‘duplicate’ in the sense required for quality control. We must always be mindful of this when quality controlling analytical results. It may not be the laboratory that is at fault. It may not be not be a sampling issue on site. Returning to core descriptions, or viewing core, when we receive analytical results should always be considered.
How can we reduce the effects of geological variation, especially when analysing core sample duplicates? One way is take two quarter core (opposite quadrants) pairs and, together, crush these into two separate sample bags. This will produce a more homogenous product and reduce geological variation between duplicates. Costly. Yes. But what price quality?
From a statistical perspective, Statistical Quality Control (an established methodology in manufacturing) does have merit. But this for another discussion (QC Analytical Results Precision: Identifying Outliers. A Statistical Approach).
First published on LinkedIn, 1 November 2017. More on this topic: Duplicates.