Q: What do you see as the greatest issue facing the mining industry in present times?
The fact that mines fail to live up to their production expectations and to provide adequate returns to investors is in part why money for exploration is so scarce these days. It’s unreasonable to assume that investors will continue to provide funding when there is such a high risk that new mining operations will fail to deliver the returns promised by mine developers.
In spite of many examples from the past of failed mining developments, our industry has not learned the “how not to” lessons of mine development. There are many recent examples of projects that failed to meet the promises of the feasibility study. I believe these pitfalls can be avoided in future and the confidence of investors in mining projects can be restored.
Q: What are the factors contributing most to successful mine development?
Some factors apply to almost every industry, such as a sustainable market demand for the product, and excellent corporate, technical and project management. That starts with a thorough technical, socio-economic and environmental assessment of the project and includes project engineering and construction management and operations start-up planning.
Other factors are more esoteric, and investors from outside the mining industry do not always appreciate the impact these risks can have. I’m thinking primarily of issues relating to the estimation of mineral resources and reserves, which are a primary cause of mine failure. These issues can be grouped in to four main classes: Orebody Characterisation; Top-Cutting Outlier Assays; Geology in Mineral Resource Estimation; and Grade Interpolation. Let’s take each of these in turn:
In our experience the single greatest cause of mine failure is inadequate study of the orebody. Recent improvements in the reporting requirements for exploration and mining projects, particularly the JORC Code, CIM standards and NI 43-101 all make perfectly clear the quality of exploration data required to define mineral resources. What is difficult to regulate is how, and how much, exploration data is used for the systematic characterisation of mineralisation and the estimation of mineral resources. Unfortunately, not everyone agrees about the density of data required to define a given category of mineral resource.
A gold deposit located in Yukon, Canada provides an excellent example of what can go wrong. Thousands of density measurements were recorded for the mineralisation, but unfortunately these were derived from sample assay pulps. These samples failed to represent the porosity of the ore and, upon further analysis, ore reserves were reduced by 24%. The subsequent reduction of reserves combined with unexpected difficult mining conditions led to a significant write-down of assets.
Top-Cutting Outlier Assays
A common problem encountered during the assessment of orebody character is the treatment of outlier assay data. Outlier assays are generally thought to represent sampling errors and the problem is whether or not to top-cut high assays. Sampling errors can occur when the volume of a sample is small and coarse gold particles exaggerate the grade of a sample – the ‘nugget effect‘. The methods commonly used for identifying outlier assays are not entirely robust and most often involves looking at the top end of a log-probability plot of assay or composite data, and attempting to identify a critical deviation in the distribution that marks a distinct, high-grade population. A more scientific method is the Kogan method, developed in the Soviet Union and introduced in North America by Irv Parrish (following a visit to Russia). Many examples of arguable top-cutting practice could be cited, especially amongst projects operated by exploration companies that believe their role in the industry is to bring early-stage projects along to the feasibility stage and ultimately sell them to a mine developer. Micon is aware of one example where the high top-cut was defended on the basis that Russia’s GKZ would not accept the resource model if it contains less gold than the reserves estimated using the traditional method. Generally, if a large proportion of the metal in an assay database can be attributed to a very small proportion of assays, something needs to be done. If the data can support a non-parametric grade interpolation method such as indicator kriging, then top-cutting may not be justified. However, many mineral resource estimates employ variations of ordinary kriging and therefore consideration of top-cutting is warranted. As far as we’re aware, mining projects don’t fail due to excessive top-cutting, yet the most vociferous antagonists of top-cutting tend to be exploration companies: they don’t like to see their metal resources (and consequently, their project’s NPV) ‘arbitrarily’ reduced.
Outliers can also occur when the mineralisation is not well understood and high assays represent a high-grade phase of the mineralisation. An excellent example of this problem was evident at a gold mine located near Wawa, Ontario. The shear zone was drilled on 15 m centres and the alteration was sampled at 1 m intervals from hangingwall to footwall. Assays were top-cut following the Abitibi rule-of-thumb to one ounce per short ton, or 34.28 grams per metric tonne. The average grade of resources was 8 g/t Au and the average grade of the reserves was 6 g/t Au. After 18 months of production mill heads averaged 4.2 g/t Au and rarely exceeded 5 g/t Au on a daily basis. The SAG mill liners were removed and extra security cameras were installed. Miners were blamed for excess dilution. Eventually, underground mapping revealed the presence of high-grade ladder veins that traversed the shear zone. The veins were up to 15 cm thick and 10 m in length. Drill core was resampled recognising more detailed geological contacts. The grade of the ladder vein material was typically more than 30 g/t Au and some assayed in excess of 300 g/t Au. After the ladder veins were top-cut to 30 g/t Au, the average grade proved to be nearly 20% lower and reserve grades matched the mill head grade. The uniform one-metre sampling had masked the outliers, since 30 g/t Au over 15 cm became 5.35 g/t Au when diluted over 1 m with 1 g/t Au material. Furthermore, since most gold particles were under 25 microns, no nugget-effect had been expected. However, gold in very fine-grained clusters gave a high degree of heterogeneity, essentially creating a nugget effect in the ore. Only after the uniform one-metre drill core sample assays were cut to 17 g/t Au did the reserve grade match the mill head grade.
Geology in Mineral Resource Estimation
There many examples of mine failures that result from misuse or misinterpretation of mineral exploration data and it is here that the greatest follies occur.
Grid drilling complex deposits can provide a lot of assay and geology data. However, in complex geological environments from time to time we still see the children’s dot-to-dot approach to connecting assay data within wireframes that are assumed to host relatively homogeneous mineralisation. If you don’t understand the geology of the mineralisation, according to JORC Code guidelines, you are not ready to estimate mineral resources.
A gold mine in California provides another excellent example of what can go wrong. The operation was built as a combined heap-leaching and milling operation. The process plant was built, but the high-grade ore proved to be thinner and much less continuous than originally interpreted. The reserves were reduced from 8.7 Mt of mill ore, plus 32.4 Mt of leach ore to just 20 Mt of leach ore and the milling plant was closed. Gold production that was originally planned to be 145,000 ounces per year was reduced to 80,000 ounces per year from the heap leach. Consequently, much of the original investment was written off.
Block model grade interpolation is the greatest pitfall for mine production-reserve model reconciliation. Often grade interpolation can introduce an element of bias by grade smoothing, exacerbated by a lack of geological knowledge and geological domain definition, and failure to adequately deal with outliers assays.
Whilst a carefully controlled kriging protocol can produce a reliable mineral resource model too often bias is introduced as a result of smoothing. Grade smoothing occurs, for example, when high-grade assay composites are used to interpolate the grade of low-grade mineralisation, and vice versa. Under-estimation of the grade of mining project may prevent project development but it is hard to imagine how it would kill a mine. Conversely, over-estimation of grade often leads to mine failure.
The following data was derived from a nickel laterite project in Turkey. Initially, the mineralisation was outlined using kriging to identify material with grade greater than 0.6% Ni. Blocks with grade greater than 0.6% Ni were subsequently kriged using only composite data greater than 0.6% Ni. The result was a most unusual distribution of block model grades! Unsurprisongly, this resource model failed a reserve audit and a new mineral resource estimate was later generated by an independent consultant.
Q: What approach does Micon take to the prevention of these problems?
Whether we’re acting for the project proponent or an investing institution, we want to ensure that the project has fundamentally sound economics, by which we mean that it is sufficiently robust to survive adverse changes within the range of accuracy of the estimates of grade, tonnage, recovery and cost.
For example, consider a feasibility study for a 2.2 Mt/a gold operation in a relatively remote area. The capital cost is relatively high due to the lack of infrastructure and complex metallurgy. The LOM average grade is forecast to be 2.3 g/t Au but some higher grade material is available early in the mine life.
The Base Case start-up capital includes a $3.6 million investment in operational readiness. The operation expects to treat 2 Mt in Year 1 and plans to achieve its design throughput from Year 2 onward. The base case NPV is positive, but is most sensitive to revenue and gold grade: a 10% change in the gold price has a significant impact on the project value.
Now consider a scenario in which the $3.6 million operational readiness capital is not spent, in order to preserve cash. As a result, instead of 2 Mt in Year 1, only 1 Mt is treated. Year 2 achieves 1.5 Mt of ore processed and full production is achieved in Year 3. Grade control failures in Years 1 and 2 result in a head grade 10% lower than planned. As a consequence, the NPV is now negative, and the risk of default is large. As well as highlighting the importance of training and start-up planning, this example shows the importance of de-risking the reserves. A modest investment in definition drilling, bulk sampling and pilot testing can help to avoid financial disaster.
On a more positive note, the next scenario considers an additional $1 million capital is invested in an enlarged laboratory and $1.1 million in opex is added to increase the number of grade control samples and hire a larger grade control crew. Extensive training prior to start-up is undertaken to ensure that everyone understands the principles and problems of sampling. For each consequent 2% improvement in head grade, net revenue increases by $3 million, more than compensating for the higher up-front and ongoing costs.
An improved head grade leads to lower unit costs and higher margins, since less dilution (waste rock) is processed. In our example, each 2% rise in head grade equates to 2% reduction in cost per ounce.
Q: In closing, could you summarise the key points for us, Stan?
Make sure you have:
- Excellent, experienced management;
- High quality resources and reserves to ensure economic survival;
- Used best industry standards for exploration data collection, including Quality Control/Quality Assurance;
- A realistic geologic interpretation and block model. Too much optimism is dangerous;
- Comprehensive metallurgical test work for ore characterisation, bulk sampling and pilot plant testing;
- An optimised mine planning to maximise cash flow and use robust grade control;
- A thorough feasibility study, using realistic market assumptions, capital and operating costs from first principles, strong social and environmental commitment and a comprehensive, independent review;
- An appropriate EPC/EPCM contract type with critical negotiations up front to nail everything down;
- Investment in people and operational readiness to hit the ground running and attain production targets; and
- A bit of luck always helps!