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.

Sampling Procedure Error (10 replies)

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

I have been looking at a project where substantial coarse sample masses were generated (>20kg) which were beyond the capacity of the sampling teams to effectively handle. Due to various reasons and decisions, no QAQC samples were inserted into the sample batches when they were submitted for screen-fire assay (AU).

Now, several weeks later, the question has been raised around the validity of taking the coarse rejects from the laboratory (now crushed and more easily handled), and adding in blanks, standards and duplicates and then re-submitting the whole mess as a new set of samples.

Would this be valid and acceptable? My initial thinking is no it would not be due to various possible errors that could be introduced by the crushing and splitting at the Lab.

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

Are the coarse rejects you describe the remainder of the original 20 kg samples or are they the coarse rejects from some smaller field sub-sample of the original 20 kg? How were the samples crushed and split in the lab? What equipment?

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

These would appear to be the coarse rejects from the 20kg samples that were crushed using a jaw crusher and then split using a 10mm riffle-splitter into the 500g sub-sample for the screen fire assay.

It must be noted that the project geologists would also use a riffle splitter to split the samples if they were to proceed with the “re-submission” of the samples.

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

If you can assume (or better confirm) that the crushing and splitting was carried out in a reasonable manner in the laboratory, then the coarse rejects should be acceptable for testing accuracy via standard reference materials. I would recommend that you insert say standards at least a 1:20 ratio and use more than one SRM. I'd also recommend that you take some duplicate splits as well (again say 1:20). The only other issue I see is that the use screen fire assay suggests you have coarse gold? In this case a 500 g sub sample may be too small - do you have any idea of the coarse gold maximum size and grade expected. You can use Gy's theory to get an estimate of the size of sample you would need to take to get acceptable precision between successive samples.

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

The lab that performed the work is certified and certainly seems to have carried the work out in a reasonable manner.

There is coarse gold and I do believe that the 500g sub-sample is way too small. However, convincing people of this is proving to be very difficult. I will have a look at the application of Gy's theory when I have a 'moment spare'.

The feeling is that the often substantial differences in the grade between duplicates (field & lab) is due to the nugget of the orebody and that nothing more is needed as this is simply what it is.

The expectation that I have is that in every 10-15 samples there would be a blank, duplicate, and one of 2 or 3 standards as was done for the RC drilling campaign.

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

Your plan sounds good. If you want to convince people of the poor precision than duplicate samples will demonstrate this most effectively. I did this once by taking a 30 kg sample in a gold mine and splitting the whole lot into 30X1 kg samples and then assaying all the splits. This clearly demonstrated that a 1 kg sample from the same lot would have a very (unacceptably) wide range of values (basically the grade of sample varied by +/- 50 percent at the 95 percent confidence level).

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

Thanks for the Feedback. Going to try and solve this 'problem' this week.

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

I note the recommendation about using Gy's theory to estimate the proper sample size. I don't think that is wise. This is because Gy's sampling equation derives directly from the binomial theorem, and thus doesn't apply to rare grains. It does work great for base metals and bulk products such as coal, Fe ore, bauxite, etc, but I do not understand why all of the Gy acolytes want to push its application into gold sampling, when it is clearly not appropriate to do so. An analogous approach using Poisson statistics is theoretically justified. Several methods have already been described in the literature, going back as far as 1969 (see Clifton et al., USGS Professional Paper).

S
Standartenfurer
8 years ago
Standartenfurer 8 years ago

I've been following this discussion with great interest. I look at this from a completely different perspective. What you have described suggests to me that there is a fundamental error in procedure, protocol and equipment selection at the drilling and drill sampling stage. A high concentration of coarse particulate in both the bag and bulk sample suggests a substantial loss of fines at the drill. This can be caused by a number of things and a combination thereof:

Losses of fines in the hole during drilling due to excessive use of water washing the fines away from the bit face. Often done to keep dust down but at the expense of sample quality. Note: There is no need to use water to keep dust down. We have equipment available to maintain a dry sample with 100% dust control around the rig.

Excessive use of compressed air, to increase drill penetration, but with a corresponding loss in hole integrity causing substantial loss of sample to outside return, which will carry the fines with it and further destabilises the hole. If we continue to pay drillers on metre rates, they will always look for ways to drill faster albeit at the expense of the sample quality unless their ability to beat the system is controlled.

Poor quality sampling system design produces low collection efficiency and, as a result, fines that are lost to atmosphere/dust collector. The new Progradex Sampling System provides 100% recovery (coarse and fines) for sampling.

Automation of sampling system or lack of, at the drill introduces human error. The Progradex Sampling System is automated to the point where no operator involvement is required at all during drilling.

Use of multi-tiered riffle splitters at the drill. They are fundamentally flawed in their design and those alone will create substantial bias. In coarse gold, this will be especially prevalent.

I can only imagine the resistance you will get if you suggest re-drilling to correct this, not to mention, if you say to do it but only after scrapping and replacing much of the drilling and drill sampling equipment on site, retraining everyone, modifying protocols/procedures, adjusting sample size, introducing 1:20 duplicates, etc. However, in my view, this is exactly what you need to do. I would go as far as to suggest that the existing samples you have are totally unrepresentative, as I think you suspect, and no amount of theory can/will correct them.

Assuming this is a grade control application, I for one would be really keen to see how this all turns out. If, as I suspect, you are forced to go with the samples you have, it would be really educational for me, and I'm sure the wider audience, to know how they match with the actual grades that go through the mill and then we can extrapolate some simple numbers to highlight what these decisions cost.

Rahil Khan
8 years ago
Rahil Khan 8 years ago

Although my intervention might be out of place due to the timing of the decision making, I do not clearly have the sample quantity left at the field site as backup when the 20kg was forwarded to the lab. If a substantial sample is still left as can provide another batch of sample, I think it to be of utmost necessity that this exercise be conducted at the project site. You may get very surprising things happening at the lab. I have noted in more than one occasion where well recognized labs get mixed up even with standards. Separately, getting it done on the field gives you a chance of correcting the sampling protocol: getting field coarse duplicates added, blanks (if you need them inserted at the field site). This may also provide the opportunity for you to get the issue of the sample size resolved.

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

The current state is that the client has taken the samples in hand. Verified (as much as possible) that they are intact and valid. Then, and this is all hopefully because I have not had site of his actual procedure, sent the samples back to the lab to split into multiple splits of -1kg, selected from the subset of samples a number of duplicates and basic samples. The procedure should have included the insertion of multiple duplicates from the same parent sample to test for appropriate sample size. 3 different grade range standards are going to be inserted according to the batch size of the lab.

So roughly speaking, the samples have been re-split, duplicated, had standards inserted and resubmitted to the lab for screen fire assay.

Please join and login to participate and leave a comment.