Resumen de Quality control and public reporting in industrial minerals
Andrew Scogings, Jacqui Coombes, Industrial Minerals
The 1997 Bre-X gold project scandal and consequent credibility gap in resource classifications resulted in an increased usage of Quality Assurance (QA) and Quality Control (QC) procedures in mineral exploration programmes. Publicly traded companies are now required by the Australian Securities Exchange (ASX), and other bourses, to release data that is accompanied by an outline of sampling and QAQC procedures used during the collection and analysis of exploration samples. Many financial institutions now require an impartial audit of geological and analytical data as part of the 'due diligence' process when raising funds for mining projects. Such audits usually include a detailed review of QA/QC procedures undertaken during exploration, including proof of accuracy and precision. A positive outcome of a QA/QC audit is the database's 'stamp of approval' and without this, all the effort put into geological modelling and resource classification is called into question. Poor QA/QC practice results in generating resource estimates 'in the dark'.
The classic analogy used to describe the differences between precision, accuracy and bias is throwing darts at a dartboard. If the arrows are clustered within the bull's-eye, they show good accuracy and precision ( Figure 1A ) whereas if they are clustered away from the bull's eye they show poor accuracy but good precision and high bias ( Figure 1B ). In reality, samples invariably incur a degree of imprecision and inaccuracy and by using proper sampling practices it can be ensured that the errors are as small as possible. Sampling errors occur at each stage of the sampling process and the reader is referred to, i) Pierre Gy's Seven Sampling Errors and, ii) to the concept of the Fundamental Sampling Error which is the "error that remains when the sampling operation is perfect," ([Pitard FP], 1993; [Jacqui Coombes], 2008).
Residues of samples may be collected at all stages of the sampling protocol. Experience suggests that so-called duplicate core samples are often not true duplicates, due to the common practice of using an 'original' 1/2 core and a 'duplicate' 1/4 core. Clearly the 1/4 core duplicate is not a duplicate as it is half the volume of the original and, furthermore, it is spatially different from the original 1/2 core. It is preferable to prepare core duplicates after the initial coarse crushing stage. Abzolov (2008) suggests that, for reliable control of sample precision, approximately 5-10% of field duplicates and 3-5% of pulp duplicates should be tested.
