Geology & GeoMetallurgy

Geology & GeoMetallurgy

  • To participate in the 911Metallurgist Forums, be sure to JOINLOGIN
  • Use Add New Topic to ask a New Question/Discussion about Geology and Geometallurgy.
  • OR Select a Topic that Interests you.
  • Use Add Reply = to Reply/Participate in a Topic/Discussion (most frequent).
    Using Add Reply allows you to Attach Images or PDF files and provide a more complete input.
  • Use Add Comment = to comment on someone else’s Reply in an already active Topic/Discussion.

Drill Core Sampling and Sample Preparation AKA Nugget Effect (12 replies)

Maya Rothman
8 years ago
Maya Rothman 8 years ago

Here is a problem of Poor Field Duplicate precision I first encountered when I was a young geologist working in a nuggety gold deposit, and now have recently encountered it again in a copper deposit. I would like to know if anyone has a suitable solution. The issue is that the copper mineralisation occurs as massive fillings in vughs (a small cavity in a rock or vein, often lined with crystals), such that there is poor precision in the field duplicates. (The primary assay samples are ½ core.) The (½) core is crushed to -2mm as part of sample prep, and the precision of duplicate samples at this stage improves dramatically. Of course ½ core needs to be retained for future geological reference so submitting all core for chemical assay to improve precision is not an option. I have toyed with the idea of suggesting crushing full core to -2mm and split this via a rotary splitter, place 50% back into the core trays and submit 50% to the assay lab for further preparation. I have never seen or heard of anyone doing this (of course crushing all the core to -2mm will mostly destroy any geological structure/texture). Has anyone seen this approach, or can suggest an alternative while retaining some reference core? This is also called the nugget effect, right?

(unknown)
8 years ago
(unknown) 8 years ago

Hello Maya,

you have a problem with particle size variations better known as the "nugget effect". As you know when grain size of the mineral of interest varies strongly in the rock, sampling is going to be highly variable. In your case, you know the cause and described it with high geological precision but your sample size is probably not adjusted to reduce that grade variation. You would have to increase the size of samples by diameter or length to reduce the sampling variance. You could include drill hole spacing and other parameters to reduce the sampling grade variation. Sampling grade variance can also be reduced by grouping existing samples into larger one or composites as we commonly do. The only other alternative is to grind the sample during preparation to destroy and delete all traces of geological evidence explaining the reason for this gross grade variation, i.e., the texture and form of the mineralisation. But making the assays repeatable will not suppress the problem.

JohnnyD
8 years ago
JohnnyD 8 years ago

If you take 2 sub-samples from crushed 1/2 core you observe a duplicate variance. If you take 2 similar sub-samples (with same sample mass) from the entire crushed core, you will have a LARGER duplicate variance due to the factor (1/MS - 1/ML), MS being the sample mass (not changed) and ML being the lot mass (doubled when the lot is the entire core compared to the 1/2 core).
Keep your 1/2 core for geology and use the duplicate variance results as an additional information about the heterogeneity of the ore. Reduce the duplicate variance can be a QA/QC objective, but it can have the disadvantage to hide such information.
If it is a true nugget effect, a Poisson distribution analysis can be done from these duplicates to estimate the nugget size and distribution.

Helena Russell
8 years ago
Helena Russell 8 years ago

I remember a study showing that the sampling error is commonly greater than sample preparation errors and analytical errors by several orders of magnitude. This applies in your case. With such a nugget effect, if you want your samples to be meaningful for estimation they'd better approach being representative of the area you will be drilling. Without a sufficient sample size (which in your case is the half-core), your sample analysis might be closer to drawing a ramdom value with a high standard deviation around your deposit mean grade.

I may suggest two things:

  • the sample size: the extra cost on sampling in a nuggetty deposit is always worth it (all technical experts agree on that), by increasing the drill diameter.
  •  a geological sample: a practice I have seen in some places is to keep a "skeletton", which consists in a thin slice of core cut along the edge (I'd say about 15mm thick depending on the skills of the person cutting core and the type of material). Then you have all the remaining core to crush and subsample, which would increase your sample size even further. And then if you want real field duplicates, closely spaced twin holes will give you a very good idea of your short distance variability.

(unknown)
8 years ago
(unknown) 8 years ago

I agree with increasing the core diameter if possible (i.e. HQ instead of NQ). Obviously its not ideal for handling however. Depending on the style of mineralization, increasing the sampling length can be less than ideal also, which favors increasing the diameter. We sample whole core due to the same issues you describe, but at a representative drill spacing sample half core to keep some physical core. We also ensure good quality photos of the core are taken to preserve the visible texture and structure. We have the advantage of exploring at a producing mine rather than being an exploration project, but you may want to consider whole core with representative holes sampled as half core. I believe there are other projects that do this also. The other sampling factor that can affect this is splitting the core versus sawing it if the mineralization is potentially being washed out. Cheers

(unknown)
8 years ago
(unknown) 8 years ago

Half core is effectively a specimen (not a sample in the true sense ) so I would not be too concerned with high variability between halves (or perhaps a quarter core repeat). Really you are confirming is the in situ nugget effect by doing these replicates. The fact that you get acceptable precision after crushing is more important. If you are concerned one side of the core may be leading to some sort of global bias (which I doubt) you can always alternate the side of core submitted for assay. Taking a bigger sample (whole or 3/4 core) may reduce the nugget effect somewhat but this may not be necessary for the ultimate end use. Regarding the fines-wash out problem is probably more important. The best solution is assay whole (unwashed) core or use dry cutting if can get away with that. Otherwise wrap the core sections up with cheap plastic packaging tape before cutting and do some saw sludge sampling to quantify grade of any losses.

(unknown)
8 years ago
(unknown) 8 years ago

Crushing all the core will never improve things beyond half the observed variance, and this may be still too high, while wasting valuable geological information. Increasing core diameter will improve things too, but also in a limited way if sticking to sensible diameters.

Averaging, compositing and other statistical tricks will reduce variance but not necessarily give more accurate results. As mentioned by Stephane, we must understand "how" grades vary and how far they behave nuggety. This depends on the scale of heterogeneity: is it at the rock grain size (typically pyrite-hosted Au) ? or at the structural scale (veins, fractures) which seems to be the case for your deposit ? did you measure the size or spacing of these mineralised "heterogeneities" ? do you record and analyse images of the half-core surfaces ?

There is an expensive tool, the core scanner with XRF sensor, which might help a lot.

JohnnyD
8 years ago
JohnnyD 8 years ago

You can really on deal with this by increasing the size of your initial sample through drilling larger diameter core, or making longer composites. If you maintain your sample size, and composite several samples into longer composites --treating each individual sample as an increment of the composite, you can decrease the initial sampling error. You could determine the appropriate sampling protocol by test-work to find the " small 'l'" as described by Pierre Gy. Alternately, if you have sufficient data, you could average together pairs of field duplicates to see if what it takes to reduce the variability to an acceptable amount. I'd be careful about crushing the whole core, as you may need it for other reasons.

(unknown)
8 years ago
(unknown) 8 years ago

Visualizing the surface of core sample 360 degree and taking high quality photograph is a must in case of copper, the host rock with chalcopyrite, malachite, bornite in specks, veins stingers or impregnated in pyrite can be observed. The most potential host rock has to be analysed as a priority , the method to be followed is splitting the core into half using saw. The half core must be retained as a representative samples for future observation. The half core should be broken into small pieces and surface observed for the trend of the ore body before grinding into -230 ASTM fraction. In case of fine grained rock the plastic wrapping suggested earlier is to be done. In case of fine grained ore without any evidence of mineralisation the whole of half core may be ground to -80 ASTM fraction, coned and quartered and sent for chemical analysis. The spacing of boreholes has to be reduced gradually in case of core having a longer host rock.

(unknown)
8 years ago
(unknown) 8 years ago

Nuggety gold is going to be a problem that can be solved only by taking larger samples to reduce the variance.  The calculation of the variance between two half core (perfect) assays leads to a variance of the assay difference equal to Ks/Ms where Ks is the sampling constant and Ms is the half core mass. Even though particles that do not go into one half go into the other, changing the statistics of the situation, the3 usual formula for variance applies.

In dealing with core situations, the issue of in-situ heterogeneity (a form of distributional heterogeneity) has to be considered. Thinking only of particulate heterogeneity (CH or IH) is a mistake. The texture of the ore matters. Even my statement above is based on the assumption that over the distance of the core diameter, the ore is homogeneous. This may be very wrong if the core has just 'nicked' a vein and the core is split ignoring this feature.

(unknown)
8 years ago
(unknown) 8 years ago

Folks who do not make a serious effort to understand the in-situ and particulate heterogeneity of their with respect to the elements of interest are putting themselves way out on a limb that represents a considerable risk in estimation of the deposit.

Victor Bergman
8 years ago
Victor Bergman 8 years ago

There needs to be a very real understanding of what it is you are trying to achieve.

Core drilling will give you:

  • Stratigraphy (rocks)
  • Ore orientation in 3D
  • Microscope work
  • Vein Set Mineralogy
  • Deep hole drilling

Core drilling is of little value to Resource Geostatistical Evaluation because:

  • It provides a small relative volume of material
  • It has extremely poor Duplicate Repeatability
  • The sample is not homogeneous
  • It is relatively slow & expensive

And the post drilling processing costs are very high - Assay data takes 3 times longer than RC due to additional sample preparation

"Balanced" RC Drilling/Sampling, which is widely misunderstood let alone carried out correctly, is your answer.

(unknown)
8 years ago
(unknown) 8 years ago

You have a problem and I assume that there is no bias between the field duplicates that suggests other than nuggets sampling? If not, then I think the suggestion of longer composites is the best first pass option. I guess you can do this in a computer with the data at hand.
I hope that, since BreEx, no one would really crush or process all core. I doubt that any independent auditor would accept 2mm crushed rejects as a validation sample, especially when there are questions over a database and its quality. I also doubt that digital photographs of core before crushing would substitute. Maybe you just need more drilling? Or some other validation sampling?

Please join and login to participate and leave a comment.