In order to avoid effects of "non-additivity" of metallurgical recovery, Séguret et al proposed a method to estimate the recovery based on the "recoverable copper". The recoverable copper obtained from a metallurgical lab test is estimated, and the estimated Cu head grade of the orebody obtained from the exploration sampling is used. Then it is calculated the recovery in each block. Of course in this case is a concern the fact of having two head Cu grade population with different variability.

Other people estimate directly the Cu metallurgical recovery, assuming this parameter as additive and assuming subsequent errors. There is the option of adjusting of estimation during the process of escalating to the plant. In such "heretical" way, there is not relevant to have the mentioned different variability.

Yes we are applying a similar methodology wherein we prefer our sampling to capture the variability of the parent distribution (grades ICP etc, lithology, alteration and mineralisation) within the context of spatial sampling. A colleague has perfected the maths of the algorithm we are using and in essence spatial criteria are brought into play.

Aside: we do test blends outside of this variability testing, but that is a completely different topic.

We have been meaning to test this concept on one of our older mine sites where we have spatial lab test data at one fifth of the geo database. The ultimate proof for us would be for us to use the algorithm and to constrain it to the met data, and then check if our models still apply with the substantially smaller data set.

I am reminded occasionally that Geometallurgy is best defined as "ore variability analysis" comprised of multiple independent, possibly relational domains including geological, mineralogical, lithogeochemical alteration, metallurgical, ore, etc. With multiple populations within each domain there exists the need for relational comparisons within and across various developed data bases.

As a benchmark, even for existing operational process recovery improvements, "Applied Mineral Inventory Estimation" (Sinclair and Blackwell, 2001) is a very helpful text covering major themes in ore variability analysis.

Chapter 3 - Continuity, particularly Value Continuity (3.3) and Continuity Domains (3.4), discuss semivariograms - autocorrelation functions for horizontal and vertical directions within a deposit and the variability of grades into better defined ore domains.

Chapter 5 - Data and Data Quality, is most comprehensive. However, the section Improving Sample Reduction Procedures (5.6) and discussion of subsampling protocols is mandatory reading for drill-development geologists and sample prep/assay lab supervisors that are involved in large bulk sampling projects for advanced-stage exploration. Section 5.10, A Generalized Approach to Open-Pit-Mine Grade Control extends the discussion beyond development projects into mining issues, particularly valuable for new production geologists.

The previous summary of Seguret's "Spatial Sampling,..." comparing (metallurgical testing) laboratory samples and exploration samples (assayed) for reasons of spatial restriction, regularization, sampling density, and grade selection, continues to illustrate a problem in our industry where metallurgists and process mineralogists are brought into an exploration project at a very late stage with their applied testwork and advanced tools. I have been told by some junior explorer CEOs that met testing too early in a promising project can cast negative news in the public markets (TSXV) where they need to raise large amounts of money for the drilling campaign (ie: Carlin-style Au deposits with indicator As, Hg in the eastern Yukon, and no infrastructure to build or support an expensive autoclave).

Cross training more experienced exploration geologists with the basic principles of metallurgical recovery, or having the mining department and milling plant staff spending more time in each others "domain" understanding and responding to shared problems, may lead to more cooperation between the professions, reduced development timelines, and cost reductions at operations through process optimization tools such as geometallurgy.