VANCOUVER — When the “geological complexity” of their orebodies forced Rubicon Minerals’ (TSX: RMX; NYSE-MKT: RBY) to suspend development production at its Phoenix gold mine in Red Lake, Ont., and Goldcorp (TSX: G; NYSE: GG) to lower production at its Éléonore gold mine in Quebec, it brings to question whether miners today are fully prepared to handle the unexpected when it comes to mine geology.
Jean-François Couture, a corporate consultant from SRK Consulting, says the ability of mining companies to predict the nature of their orebody has “evolved tremendously” over the past decade — but a poor geological understanding can weaken or even break any resource model.
“We now have tools that are far more powerful than they ever were,” he tells The Northern Miner during a phone interview from his office in Toronto. “Aspects of projects are a lot less overlooked, geology is more scrutinized and models are built by multidisciplinary teams, rather than just one person operating out of their basement.”
He says if you carry out a forensic investigation on a failed project, for example, “it’d find a ton of causes and not just the geological ones. But in the case where the resource model is blamed for the inability to deliver what was expected, it’s usually because of poor geology.”
And unexpected geology is what Rubicon found soon after starting preliminary production at its Phoenix mine this year — an operation designed on a preliminary economic assessment (PEA) completed in 2013.
Trial mining during July and August delivered material grading 4.29 grams gold per tonne, which was in line with the predictions made in the PEA.
As the operation continued into September, the mining stopes highlighted enough “spatial variability” of the orebody that Rubicon suspended operations in November, effective until a new geological model is developed.
Goldcorp announced a similar situation in September at its Éléonore deposit, where locally intense folding and faulting of the vein-hosted orebody resulted in less-than-expected production from the mine.
As a result, Goldcorp revised its gold production guidance to between 250,000 and 270,000 oz. gold — down from 290,000 to 300,000 oz. gold — and is working to improve stope design to counter the unexpected dilution.
Couture says variability and continuity between sampling points are critical in narrow-vein deposits.
“Any situation where you need to be selective there’s always risk that ore and geology may not be as predictable in front of the headings underground,” he says. “With more infill and definition drilling you can reduce that risk to a level that’s bearable, or you go forward and learn from experience.”
The most critical part in any geological model, he says, is how the orebody is defined between observation points — whether it’s boreholes, a trench, an underground tunnel or samples from outcrops at surface.
To draw data into the empty space, he says, all aspects of the geology must be considered, including structures, distribution of alteration and mineralogy that either affect or control the mineralization.
Once a suitable level of confidence is reached, modellers like those at SRK Consulting use geostatistics to map the distribution of grade in the orebody, and build a resource block model that supports mine design and planning.
“Geostatistics is just a technique to try to see a reason in a cloud of data,” he says. “But it can be completely wrong if it’s based on the wrong geological interpretation.”
He says that a tighter density of boreholes is often needed when gold mineralization is nuggety, and confined to narrow vein systems cutting through barren country rock — like many deposits in the Abitibi region of Ontario and Quebec.
“You want the resource model to be a bit more conservative than reality to allow some flexibility to the mining team, and still achieve what is expected,” Couture says. “But you don’t want it make it so conservative that it kills the project — it has to be realistic.”
But understanding whether or not the model published in a National Instrument 43-101 technical report is correct poses a challenge for the average non-technical investor.
“It takes a fairly experienced technical person to recognize good from bad, so when sophisticated investors such as banks or funds make an investment to acquire an asset or participate in one, they hire groups like SRK to audit the technical work that was prepared by other persons, just to confirm that the geology and mining projections are reasonable,” he adds.
He says that depending on the author, experience and intent of the technical report, resource numbers can be expressed differently.
“There’s a set of rules on how a company can disclose technical aspects about a project, but there are no compulsory rules to satisfy the requirements. It’s just a disclosure,” he says.
Regardless, Couture says that resource models today are superior to what was done in the past, and regulators such as the B.C. Securities Commission or the Ontario Securities Commission are quick to flag any unrealistic estimates that come through the newswires.
One such occasion occurred in 2012 when the CEO at the time of Barkerville Gold Mines (TSXV: BGM), Frank Callaghan announced a 10.6 million oz. gold resource at Cow Mountain, southeastern B.C., in 69 million indicated tonnes grading 5.28 grams gold — an eyebrow-raising 2,400% increase from the previous resource estimate in 2006.
The press release also stated that the 6.4 km long trend has the “geological potential” to host 405 million to 684 million tonnes grading 4.11 to 5.49 grams gold, for 65 to 90 million oz. gold.
In response to the news, the BCSC slapped a cease-trade order on the company, which stayed in effect until Barkerville revised the resource estimate the following year to 1.04 million oz. in 17.7 million indicated tonnes grading 2 grams gold and 3.94 million oz. gold in 49.2 million inferred tonnes at 2.74 grams gold.
Since then, more drill-outs have raised the resource to 35.8 million indicated tonnes at 2.4 grams gold for 2.8 million oz. gold, plus 27.5 million inferred tonnes at 2.3 grams gold for 2 million oz. gold, using a 0.5 gram gold cut-off.
The resource was modelled using a technique called “multiple-indicator kriging” — a commonly used geostatistical tool in the industry — which estimates the grade within a block based on down-hole assay data.
But considering mineralization at Cow Mountain is contained within a number of rod-shaped, plunging structures in deformed sedimentary rocks, the company is working towards building an improved geological model to guide any further resource estimations.
“During the early stages of any project the geological model is preliminary, but by the time it gets to an economic analysis it definitely needs a geology model that integrates the entire volume of rock, and not just the mineralization. But often the geology model is not done, or if its done, it’s too general,” Couture says.
“You have to be careful, because sometimes you hear more about the bad stories than the good ones,” he adds. “There are plenty of cases where companies mine higher-than-expected grades, and that doesn’t make the news.”
“But it can be completely wrong if it’s based on the wrong geological interpretation.”
Or no geological interpretation …
The rash of recent failures is NOT derived from a lack of information it is the failure to capture and incorporate the available geological information into the resource model; this applies equally to the issuer and resource consultants.
All too often, and SRK was a leader in this approach, the resource model is constrained by a grade shell (e.g., Leapfrog shells imported into GEMs or other software, and used as the controlling domains). This has become the standard approach whereby geological controls have effectively been removed from the resource modelling process … a resource model must be built from a GEOLOGICAL MODEL not an assay cloud!
The original rationale behind the grade domains was simple, it was much quicker and simpler, no interpretation required and as a consultant you don’t need the expensive and time-consuming review of the geology of the deposit. And in many instances dealing with issuers the geological information was not collected or captured or was in such a state of disrepair as to be unreliable.
It was a practical shortcut that became the SOP and unraveled into the dark chasm of “implicit modelling” yes … not only can software create grade models (e.g., interpolated blocks of ore grade) but now it can create geological models. The end nigh!!!
The false sense of security that some derive from additional modelling algorithms is also a clear alarm for anyone reviewing resource models e.g., kriging is better than inverse distance … no multiple indicator kriging is better than ordinary kriging … no uniform conditioning is required on this special deposit! If the “mineability” of your deposit depending on some special solution or the gridding algorithm then you have another fatal flaw in the process.
Other key flaws occur in the parameters chosen such as capping, compositing, block size, search ellipse etc. … these factors are necessarily limited by the GEOLOGICAL controls (i.e. the 3D structural, stratigraphic, alteration, geotechnical, metal distribution etc.) NOT mining controls or arbitrary “expert opinions”. Once again there is no shortage of information and virtually all technical reports demonstrate the available data. However, in an amazing number of cases the data is subjugated based on variograms and mining plans (i.e. the 2500 tpd long-hole because it is cheaper and looks better in a PEA regardless if the deposit can handle it via mining loss and dilution!).
The “special case” of narrow vein deposits is yet another common fallacy necessitated due to the lack of structural controls incorporated into the resource (i.e. block) model … an example is a series of en echelon tension gash high-grade veins within some “panel” (ahh the famous “shear zone”). These dots of high-grade assays connect in a line and without field observations (i.e. a series of isolated sub-metre veins at some other orientation) the variography and grade shells create a completely false domain that when mined will never meet expectations with regards to head grade, dilution cost etc.
Capturing this information and incorporating into a proper geological model will allow expectations to be managed i.e. given this real scenario what is the appropriate mining method and with the concomitant financial implications do we still want to make this investment. Most deposits won’t pass this test … and that is a good thing for our industry.
Geology has been largely removed from the modelling process and the good (great) geologists are now starting to give up on the modelling process due to the lack of input and consistent failure to meet expectations. THIS is an understated and very alarming development! Instead of recognizing the problem and bringing more geological information and skill into the process the industry has devolved and now points the finger at exogenous factors … ye…
Poor geology is a poor, but valid, excuse for a mining company to use when describing why their ore deposits do not have the ore as expected.
For example, When a group describes their deposits as “orogenic gold” when they are actually “epithermal gold”, this would explain why their gold has very limited vertical extent. The only remedy for such mistakes is to use geologists who understand different deposits, particularly those that can differentiate between volconogenic, orogenic, and epithermal gold. There are sometimes subtle, but important differences, in these three types of gold deposits, and overlapping deposit types are also fairly common and need to be recognized.