• To participate in the 911Metallurgist Forums, be sure to be Logged-IN
  • Use Add New Topic to ask a New Question/Discussion about GeoMetallurgy or Geology.
  • 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.

RC sample weight comparisons (9 replies)

Dizzy Flores
9 months ago
Dizzy Flores 9 months ago

When taking a 'A' and 'B' sample from a cone splitter on an RC rig and comparing those weights, what differences in weights between the two samples would you start to think there was a problem. Is this also commodity dependant e.g. gold vs. iron ore.

Alan Carter
9 months ago
Alan Carter 9 months ago

Excellent question. My 2c worth as follows. A static cone splitter? For a (supposedly) equal splitting device I would hope the differences are within 5%-10% at most and I would be concerned if a correct and properly used riffle splitter had 5% (should be more like 1%or 2%). The other things to look at are the size distributions of (a sample of duplicates) and also the on average weight differences. If using a static cone - more fines can accumulate in one split compared to the other (likely bias issue) as this device collects samples from just two parts of the sample stream (not across the whole stream) - a bit like a riffle splitter with all but two of the slots blocked. You probably realize I'm not a fan of static cone splitters.

9 months ago
OberstGruppen 9 months ago

My experience with cone splitters has been frustration with heterogeneity and segregation effects, I do not recommend using them.

When properly used, the Jones riffle will give weights within 1%, the key point is proper use. Training and monitoring staff performance is very important for achieving good splits with minimal heterogeneity and bias issues. A spinning riffle is superior - but must be set up and used properly with properly prepared feed.

Ace Levy
9 months ago
Ace Levy 9 months ago

I agree whole heartedly with comments, observations and opinions. You have answered the initial question about the expectable weight difference between the A and B sample correctly. You have made a good point about it.

9 months ago
Unterstarm 9 months ago

A QQ Plot will tell you if you have a systematic bias and, as such, if you have something to worry about.

Helena Russell
9 months ago
Helena Russell 9 months ago

Whilst I am comfortable with RC sampling provided it is done properly. In my experience the cyclone fitted with a static cone splitter is commonly not operated to the manufacturer's recommendations. It is common to see the knife gates left open and the sample continuously trickled over the cone. Great for achieving metres and enhances the drillers bonus or driller’s company profit. I have even seen recommendation for KPIs by an internationally recognised accounting/audit firm recommend metres drilled in a month as a measure of the performance of a geology department.

I have seen recommendations from QA consultants that for HARD 70% of the field duplicate samples to be no greater than 10% difference. I also like to generate QQ Plots and scatter plots (with 5%, 10% & 30% sample weight differences bounds) as well as % frequency and accumulative frequency plots of the sample weights. I would expect the sample weights; frequently measured by the laboratory to form a bell-shaped distribution. Commonly the weight distribution is skewed, with larger weights recorded at rod changes reflecting contamination build up. It is interesting to watch the reaction of contract drilling company managers when confronted with the graphs and pointing out potential biases. This is the only way I find to manage the driller. The driller has a difficult job to read the ground. Generally at the drill rig the least experienced people are employed. Training is critical for geologists what to look for and maintain creditability with the driller.

The design of static cone splitters does not conform to TOS rules. The knife gates are drawn from one side, or may be modified to draw to either side, momentarily dumping a rectangular shaped sample over a cone. They are not designed for sampling dynamic particulate sample streams. Duplicate samples do not take into account dust losses.

Riffle splitters I rarely see used correctly.

I think the industry needs to focus on the quality of the drill sample as well as the laboratory preparation; otherwise the sample is stuffed before it even reaches the laboratory. RC drilling is cheap until you find out the consequences when you mine it.

Carl Jenkins
9 months ago
Carl Jenkins 9 months ago

My experience with static cone splitters is that it is all about how the sample is delivered to the splitter. Ideally the drill cuttings should be delivered via a drop box (under the cyclone) with a fast acting knife valves. It should not be delivered directly from the cyclone as issues with segregation can result. Similarly, butterfly valves above the cone splitter can cause issues as when the valve is opened the cuttings tend to flow more to one side of the splitter (at least initially while the valve is being opened).

To answer your question regarding duplicate sample weights, I would say they should be within 5-10% for dry drilling conditions (which you should be able to achieve using a blow-down valve). For a 3kg sample this would be no more than +/- 0.3.

Alan Carter
9 months ago
Alan Carter 9 months ago

As a follow up question - does anyone know where these static cone splitters originated and what was the thinking in fitting them to RC rigs?

I recall that they suddenly appeared in the 1990s and not sure from where - perhaps as alternative to much (and rightly) maligned chisel splitters that were fitted to RC rigs for wet sampling. As far as I can tell the cone splitter seems to be modelled on similar static devices fitted to fine powder storage bins. I understand here the goal is to reduce to rolling/sifting segregation of materials as they flow into the bin.

9 months ago
Unterstarm 9 months ago

We pioneered the first commercially available static cone splitter at Metal Craft in Perth in the late 90's and worked closely with Anglo Gold at Sunrise Dam to test it with a known sample against a tiered riffle splitter. I have to say, the tiered riffle splitter did not compare at all well and the testing was exhaustive & independent. A paper was produced by AG on the results.

We then started developing the Rotary Cone Splitter (rotating sample ports) to further improve representivity [against flow bias through the throat to the sampling ports & to help negate effects of levelling] and this was later perfected at UDR (Universal Drill Rigs) in 2004/5 with the release of the RotaPort Cone Splitter. Since then, there have been many copies of cone splitters and sampling systems modelled on the original Metal Craft system. Buyers beware! While it may resemble a cone splitter or a sampling system, it will likely not take into account design fundamentals that are critical to making it work. This is not just cone splitting principals, it's "air pollution control" design principles (cyclonics, etc), materials handling design and flow dynamics principals that all play a critical part in the result. Sadly, I see these nasties time & time again these days and it is crucifying the reputation of RC drilling.

Other CRITICAL factors is the delivery of the bulk sample to the sampling system from the ground, and that comprises a multitude of things all the way back to the right selection of rig, compressor, hammer, drill rod, use of a blowdown valve, the sample collection device, storage device, release mechanism ...etc, etc, all of which have been pointed out previously. It always makes me grimace when I hear the poor old splitter taking the rap simply because this is where the sample is delivered from.

Drill sampling is a lot like preparing a plate of food. The meal doesn't matter (iron ore bolognese or gold wellington), the principals are the same, if you don't have the right utensils; you won't be able to eat it, let alone sell it. Ask yourself the question, would you open a restaurant and not employ a qualified chef? Would you provide him with a kitchen, equipment & utensils? If you're just making burgers you might be okay without this entire but don't look for sushi profits.

That said, I wouldn't recommend using a cone splitter or tiered riffle anymore anyway. Things have moved on considerably since the advent of cone splitters 17 years ago and riffle splitters 200 odd years ago.

Facts to consider:

If you're getting wet RC samples, you are doing something very wrong - and it is little surprise that your splitter is blocking, contaminating and biasing.

Tiered Riffle Splitters are fundamentally flawed by design and will NOT provide representative or even repeatable samples ever. Only a 50/50 riffle splitter (single tier) will ever work as required and, even then, it is dependent on even sample delivery/spread.

A core sample will not give you duplicate repeatability - after all, the two halves of the core are different pieces of ground. And, even if it did, core is only a tiny portion of the ground it is meant to represent.

Done properly, for the same cost, RC drilling will provide 20 times the volume of homogenised, duplicate samples in half the time of core.

Fines loss skews the result a lot more than you may think. We've just done a study in RC grade control where the collection of the fines for sampling increases the gold content by 6%, silver by 7% and copper by 9%.

Incidentally, the predicted RC grade control result for the entire year 2012/13 with this "Balanced RC" package was 6.13g/tonne gold. Actual gold poured for the year was 6.14g/tonne. The mine produced 1.1 million ounces Au last year and it is a highly complex sulphide gold deposit with massive ground water issues at pH1.5

Jean Rasczak
9 months ago
Jean Rasczak 9 months ago

For one specific programme, we made an RDP plot ranking difference in weights between duplicates and setting them out against variance in gold grade to see if the variance would significantly increase with increasing difference in weight. There was as much as 8kg difference between original and duplicate. There was more variance, but less than I expected, probably due to the relatively low IH