The Australian handbooks on Leading Practice Sustainable Development Program in Mining are a good place to start. They are available in English, and some have been translated in Spanish, Mandarin and other languages. There is a section in the Acid & Metalliferous Drainage handbook on guidance for number of samples per waste lithology/strata to collect and analyse during the exploration, feasibility and mining operation stages.

If S and Ca are within the assay database then we calculate theoretical MPA, ANC and ABA. There are obviously several assumptions associated with this, but we find it is a good first pass at the exploration phase. We also work with the resource geologists to develop a waste rock block model that helps to develop a sense of potential waste rock issues. The same concept can also be used inside the resource to begin thinking about potential tailings issues.

Understanding lithological and mineralogical issues are also important. So some static tests, including XRD, are a good idea at this stage. Understanding that cost is always an issue, I tend to recommend doing all the recommended tests within GARD, including NAGpH, as it also lays the framework for understanding which tests are going to be helpful moving forward when a larger number of analyses are required. There is a reference I use that indicates 3-5 samples/lithology which I can't find early on a Saturday morning. I will dig that reference up and post to the group. As with everything, use judgement and understanding of your project to guide your thinking. A simple conceptual model at this stage pays dividends down the road, especially at the mine planning stage.

Hi - one way which can be used for undertaking lots of total S analysis and perhaps getting the S model up an droning is using the new portable XRFs which now include total S. I have not used it and would be interested to know if anyone has tried them out - at 30 seconds a sample you could get a whole lot of cores analysed in one day!

Lots of good advice here. I second the comments about using the assay data collected as part of the exploration stage. Many of our clients perform a great deal of analyses to support exploration, including ICP analyses of core and/or LECO analysis of total S, sulfide S, and total carbonate. Although they are not necessarily collected specifically for the purpose of environmental characterization, I have found it very useful to consider those data when selecting a subset of representative samples for more detailed analysis.

I've seen hand-held-XRF both work and not work for screening. I recently reviewed work where concentrations were so low that the noise was too great for any meaningful correlations - and this was from a very large data set. Where I have seen it be successful is where there are reasonable concentrations of major elements. It is still a screening tool requiring calibration and especially for ABA, S concentrations still need to be related back to acid generation/neutralization. As always, it can be a useful tool in the right circumstance, as long as there is a good understanding of site conditions and limitations. This is why I still recommend some ABA testing of all lithologies at the exploration stage - in addition to begin developing correlation between assay S and LECO S, calibration of hand-held XRF can also be done.

Here is the "Managing Acid and Metalliferous Drainage" Australian Department of Industry, Tourism and Resources http://www.industry.gov.au/resource/Documents/LPSDP/LPSDP-AcidHandbook.pdf

Table 1 (p15 ff) gives a good description of recommended investigations.

You probably know that I have a bit to do with the above mentioned handbook, and it nice to see that is is being used. Here is a simple formula I use to provide an indication of the number of samples for ABA and NAG testing at the feasibility stage:

No of Samples from EACH lithology = 25 x sqrt (X). X = Mt of material.

So, the minimum is 25 per lithology, but if the lithology is minor, I generally recommend less (based on geology and site specific factors)

For 100MT about 250 samples.

This is open for comment from others. It is based more on experience working with geostat people to produce ARD rock type models, rather than purely on geostats.

Actually I think the formula is:

No of Samples from EACH lithology = 25 x X/sqrt (X)

Where X is the mass of material in Mt

MEND 4.5.1-1 Review of Waste Rock Sampling Techniques (1994) uses N = 0.026M^0.5 where M is mass of geologic units.

I think this equation goes back to the BC AMD Task Force in the late 80's and I still find it appropriate. It is a minimum, and of course we need to go back and fill in any gaps once the mine shell is better defined. I use a Mt form of the equation, as I have find this easier to 'sell', especially at the feasibility stage.

Yes, I do note that my attempt at the equation was a bit confusing. Just to confirm and to use Bill's approach:

N = 25M^0.5

where M is in million tonnes - you could use 26, but then it looks too accurate?

I am sure you can find the BC AMD task force ref on the web but there is desktop review I need to get back onto!

Two questions need to be answered in order to determine how many samples will be sufficient:

1. What is the required level of confidence of the global mean and global variability (variance) of the parameters of interest for the rock type?
2. How important is the spatial (local) variability of the parameters of interest across the entire volume of each rock type?

Geostatistics can provide a sound basis for determining the number of samples and spatial distribution of samples characterised in relation to AMD at a particular site. Geostatistics is a set of statistical tools commonly used in the mining industry in resource estimation and therefore people with appropriate geostatistical skills are likely to be accessible to those responsible for geochemical characterisation.

The required level of confidence for the global mean and the variability depends on many factors including the stage of the project and the sensitivity of the receiving environment to AMD. The second question relates to the significance of spatial variation to the waste management strategy.

After geochemical characterisation of the waste samples there is often a need to estimate the mass of wastes that fall into the various classifications (e.g. PAF, NAF) for materials management purposes. However, one must be careful not to assume that the proportion of samples in a particular classification is directly proportional to the mass of mined material that should be given the same classification.

Probabilities and proportions based on sample size volumes can be somewhat misleading, as the waste will not be mined with equipment with the selectivity as small as that of a drill rig (i.e. the size of the drillcore that is generated and sampled). Mining selectivity is more likely to take place at the scale of an excavator bucket or truckload. The size of the selective mining unit should be taken into consideration as this will change the variability of the parameter under consideration.

A published a paper at the Life of Mine 2012 conference http://www.ausimm.com.au/lifeofmine2012/presentations.asp that illustrates and discusses the use of geostatistics in relation to AMD. I have placed an extended and modified version of the paper at the website http://www.srk.com.au/files/File/papers/adequacy_of_sampling_for_acid_mine_drainage.pdf

Excellent comments and paper and I agree that geoostats is useful in determining sample numbers and is of course essential for generating production schedules by ARD types. However, with respect the original question regarding sample numbers at the exploration stage, the necessary data for geostats may not be available and we need to make a call. The assay suite for all exploration samples should include S, Ca and Mg and ideally C (many companies now routinely include S, Ca and Mg in the exploration assay suite, and some include C). However, if you don't have these data at the exploration stage, than the simple formula referred to earlier is a good starting point.