Search Documents

By M Haest, O Dominguez, D Mittrup

Technological innovations and emerging analytical capabilities are transforming the way drill samples are analysed, with real-time mineralogical analysis made possible. In addition, new business/technical requirements are challenging long standing processes to incorporate more sources of information during planning in order to improve compliance to plan during mining and processing. Unfortunately, the new sources of information that are added do not always have adequate protocols in place to ensure quality data is collected.Spectral data covering the visible – near – short wave – midwave – longwave range are routinely collected by the Geoscience and Exploration department at BHP Billiton Iron Ore for accurate mineralogical identification. These data are used for geological interpretations (logging and modelling) and prediction of physical variables (metallurgical and geotechnical properties).This paper introduces how sampling and quality assurance / quality control concepts have been applied in the spectral data process. The described routine for sampling, data collection/management and reconciliation has been developed to track system precision and accurately and consistently ensure the data is of high quality before it is used across the mining chain.CITATION:Mittrup, D, Dominguez, O and Haest, M, 2017. A new approach to implement quality assurance and quality control to technological innovations – quality of spectral data capture and processes in the minerals industry, in Proceedings Eighth World Conference on Sampling and Blending , pp 265–270 (The Australasian Institute of Mining and Metallurgy: Melbourne).

May 9, 2017

By C J. Kruger, E le Roux

Anglo American Platinum has adopted a strategy of Best Practice Principles (BPP) for sampling, mass measurement, analysis and metal accounting for more than a decade now. When new plants are designed, the most suitable sampling equipment and sampling protocols are implemented and material flows and plant layouts are designed around sampling strategies. A few years ago, a crushing plant for a high-grade matte stream was commissioned.The design called for the high-grade material to be crushed to 95 per cent passing 2 mm and the design allowed for a single cutter primary Vezin sampler, increments taken from a gravity flowing stream at regular intervals, whilst filling a tanker for the pneumatic offload at our base metal refinery. This was a single point of sampling and analysis that would measure the wanted output from our smelting complex and the input to our refineries. In addition, the design incorporated the functionality to collect increments in an ‘interleaved’ fashion from the Vezin sampler for internal precision checks.The metal accounting strategy for Anglo Platinum is the ongoing measurement of all inputs, outputs and stock levels for all our operations for platinum group metal (PGM) and base metal (BM) content. Included are yearly stocktakes, usually done on 1 February every year, across the smelters and refineries (at least the base metal refinery). A trend was developing in the accounting period immediately following the introduction of the new crushing and sampling facility, whereby the absolute difference between the physical stock and theoretical stock (PGMs predominantly) increased for the two respective sites on an ongoing basis. The smelter (sender) showed an increase in physical stock compared to the theoretical stock and the base metal refinery (receiver) showed that their theoretical stock increased against their physical stock. Because of this anomaly, an additional stocktake was held in September for the period in question and it was confirmed through the physical stock measurement results, that the PGM accounting problem was indeed pointing towards the evaluation of the high-grade matte stream.This paper showcases parts of the process that took place to explain the poor accountabilities for the respective sites. Previous results from the sampler ratification process (of the interleaved sampling design) were analysed again, a physical inspection was carried out on the sampling system again and the sampler performance was monitored (quality control parameters) for the period.From these investigations, it was found that airflows present inside the Vezin introduced the sampling bias and a change in sampling protocol was introduced as well as physical changes to the sampling equipment were made to eliminate the airflow and thus the bias conditions significantly. The accounting period which followed, after a phased upgrading process, showed that the accounting between our smelter and base metal refinery had been restored to within statistical limits (difference between physical and theoretical stock below five per cent relative to theoretical stock).CITATION:Kruger, C J and le Roux, E, 2017. Pitfalls in Vezin sampling for finely crushed materials, in Proceedings Eighth World Conference on Sampling and Blending , pp 151–158 (The Australasian Institute of Mining and Metallurgy: Melbourne).

May 9, 2017

By P Goblet, S A. Séguret, E Cordier, A Galli

In salar-type deposits, lithium grades can be measured by bailers introduced at different elevations in the drill holes at locations where, later, the brine will be pumped out to recover the metal. When they are duplicated, these a priori rudimentary measurements may show large differences without indicating whether this is due to the procedure itself or to some physical causes linked to the dynamic behaviour of the salar (seasonality, rainfall, underground flow etc). In the reservoir presented in the paper, the problem is complicated by the double lack of stationarity of the lithium grades: the grades not only increase importantly with the depth but at the same time their fluctuations decrease, making it necessary to use non-stationary geostatistical techniques to simulate them. Reservoir simulation is developed in two steps: first, geostatistical simulations of the lithium grade create possible realisations of the reservoir; then, each geostatistical simulation is followed by a pumping simulation that calculates, at each drill hole of a given extraction scenario, the lithium produced each year over 40 years. For the set of 100 simulations produced in this way, a scenario reduction is made to choose the five most representative simulations where more complex calculations will be carried out (for example, changing hydrogeological parameters such as porosity and permeability, or the elevation of the filters, the number and the location of the pumping and the reinjection drill holes etc). The paper first presents an analysis of the geostatistical model sensitivity to the grade uncertainty when measured by bailer. Among the 100 measurements at our disposal covering the future production domain, 50 per cent are duplicates measured at the same place but at different times. They are randomly sampled to produce five subsets of 75 values, which will constitute the future 3D conditioning points for the geostatistical analysis and the simulations. For each subset, trend, standard deviation and variogram models are fitted, leading to five sets of 20 geostatistical simulations of grades. In this way, the reliability of the different parameters involved in the procedure is evaluated, as well as its impact on the pumping results. Then, a reconciliation study is conducted between the non-stationary nugget effects encountered in the 3D variograms of the lithium grade and a pure statistical analysis of 25 duplicates measured at different locations and/or different times. The result is that if the nugget effect reaches 20 per cent, which is the case of three conditioning subsets out of five, the reconciliation is good and the nugget effect of the variogram model represents the bailer uncertainty, ie the measurement error. Since the geostatistical simulation incorporates a nugget effect, it handles the bailer uncertainty and the impact on the produced metal is included in the evaluation. The final conclusion is that, after the hydrogeological pumping simulations, the initial differences between the five geostatistical models do not influence the final results: lithium grades measured by bailers can lead to a robust evaluation of the extractable resources, a conclusion that conflicts with conventional wisdom.CITATION:Séguret, S A, Goblet, P, Cordier, E and Galli, A, 2017. Bailer uncertainty evaluation in a lithium salar deposit, in Proceedings Eighth World Conference on Sampling and Blending , pp 395–402 (The Australasian Institute of Mining and Metallurgy: Melbourne).

May 9, 2017

By G J. Lyman, S A. Tittlemier

International regulations exist to manage the concentrations of mycotoxins such as Ochratoxin A (OTA) and deoxynivalenol (DON) in grains and prevent health risks from excessive exposure to OTA and DON. Since the permitted levels of these mycotoxins are very low (existing maximum levels are in the range of 5 ppb for OTA and 1–2 ppm for DON), sampling for these mycotoxins is potentially challenging. Given that the distribution of infected kernels in a consignment of wheat can be highly heterogeneous, detection by current sampling methods is subject to a high variance. Furthermore, the concentrations of mycotoxins in individual kernels vary over orders of magnitude, intensifying the heterogeneity of the bulk grain.In these circumstances, there is much interest in developing sampling plans that inform regulatory agencies, and mitigate the risk of misclassifying grain shipments as compliant or non-compliant within the global grain trade, when correctly implemented.Experimental data from laboratory and field studies describing individual wheat kernel concentrations of OTA and DON were combined with experimental data describing the variance due to the analytical methods used for OTA and DON to calculate the sampling distribution for wheat having various levels of contamination. This work has focused on quantifying the heterogeneity of samples at the point of submission for analysis and within subsequent sample preparation to arrive at total sampling distributions that include uncertainties due to the intrinsic heterogeneity and the analytical uncertainties. The sampling uncertainty that relates to the difference between the primary sample analysis and the true lot analysis is not included as it has been dealt with elsewhere.CITATION:Lyman, G J and Tittlemier, S A, 2017. The sampling characteristics of grains contaminated by mycotoxins, in Proceedings Eighth World Conference on Sampling and Blending , pp 281–294 (The Australasian Institute of Mining and Metallurgy: Melbourne).

May 9, 2017

By B W. Atkinson, J Docherty

An Australian Working Group has been compiling a guide for operators to use in order to improve the efficacy of their routine and non-routine sampling tasks within their coal preparation process. The guide focuses on unit operations within the process, and not conveyor belts or transfer stations that are adequately covered by the existing Standard ISO 13909.Representative sampling of unit operations is generally an after-thought following plant construction, whether that be a result of oversight, or more commonly, capital cost minimisation during initial construction. Minimising capital cost by avoiding adequate sampling systems is generally a poor decision, since the result is a process that cannot be adequately measured, meaning that it cannot be adequately managed and optimised.The use of the term ‘guide’ rather than ‘standard’ is intentional. Practical means of sampling unit operations within existing plants generally relies on a compromise between sample representativity (minimisation of bias) and sampling feasibility. The guide does not endorse poor sampling practice, but instead offers practical guidance to overcome, to the extent reasonably possible, real-life limitations.When it comes to sampling any particular process, it is almost always necessary to make a professional judgement as to whether or not the proposed samples are likely to be sufficiently representative such that the sampling will be worthwhile. Such judgement needs to be exercised by personnel who understand the principles of sampling, as they apply to the stream being sampled. There are many circumstances where available methods for sampling of a particular stream will be meaningless, and in such circumstances analytical data from that sampling will only serve to propagate poor decision-making.This paper outlines the approach for the guide, the areas it covers and the approaches adopted for different types of unit operations and associated equipment configurations.CITATION:Atkinson, B W and Docherty, J, 2017. A practical guide to sampling in coal preparation plants, in Proceedings Eighth World Conference on Sampling and Blending , pp 241–246 (The Australasian Institute of Mining and Metallurgy: Melbourne).

May 9, 2017

By R C. A Minnitt

The two methods for calibrating the parameters K and (alpha) for use in Gy’s equation for the fundamental sampling error (FSE), the duplicate sampling analysis (DSA) and the segregation fractionalised analysis (SFA) are described in detail. A case study using identical broken reef material from a Witwatersrand-type orebody is calibrated using the DSA and SFA methods and the results are compared. Classically, the form of Gy’s equation for the FSE raises the nominal size of fragments given by dN to a power of 3. Gy’s formula is a theoretical formula applicable in the case when fragments are randomly collected one by one, which is obviously never the case. This is because the fragment shape factor (f) and the granulometric factor (g) were introduced to take care of this anomaly. More recent formulae, such as those suggested by François-Bongarçon and Lyman, are empirically based formula. A later modification of Gy’s formula raises dN to the power ? (alpha), the latter term being calibrated with the coefficient K in the DSA and SFA methods. The preferred value of ? (alpha) for low-grade gold ores used by sampling practitioners in the mining industry is 1.5. A review of calibration experiments for low-grade gold ores using the DSA and SFA methods has produced values of K that vary between 70 and 170 and values of alpha in the range 0.97 to 1.30. The average value for alpha is found to be 1, rather than 3 as originally proposed in the classic form of Gy’s equation or the industry preference for 1.5. It is suggested that for low-grade gold bearing ores, the equation for the FSE should raise dN to a power of 1. Such an equation for the variance of the combined FSE plus the grouping and segregation error, now referred to as the short-term component of the quality fluctuation error (QFE1), greatly simplifies the characterisation of gold ores, which now only requires the calibration of K for a given mass and established fragment size. The implications for the heterogeneity test from the generalised equation are that provided the fragments have been screened to within a narrow size range, any particular size will return a value for K that is acceptable for use in the sampling nomogram.CITATION:Minnitt, R C A, 2017. A generalised form of Gy’s equation for gold ores – empirical evidence, in Proceedings Eighth World Conference on Sampling and Blending , pp 331–350 (The Australasian Institute of Mining and Metallurgy: Melbourne).

May 9, 2017

By Y Xie, L O’Connor, S C. Dominy

Sheeted vein gold deposits are characterised by multiple subparallel veins and often free milling gold. Variability in gold grade and recovery parameters are enhanced by poorly designed sampling and test work protocols. Poor quality samples generally equate to an enhanced nugget effect. A sample can be described as being representative when it results in acceptable levels of bias and precision. Total sampling variability can be quantified by the relative sampling variance, which measures the total empirical sampling variance influenced by the heterogeneity of the lot being sampled under the current sampling / test work protocol. Effort should be made to minimise the relative sampling variance through Theory of Sampling (TOS) and quality assurance / quality control (QA/QC) application. This contribution reports on a case study that exemplifies how high-quality grade and recovery data can be gained from a well-designed and planned drilling, sampling and test work program, followed by bulk sampling. Results across historical resource drilling, geometallurgical drilling, laboratory assay / test work, geometallurgical modelling and a bulk sampling/pilot processing study are reconciled. A whole core, entire-sample test work protocol was used to acquire fit-for-purpose gold grade and recovery data for inclusion into a Pre Feasibility Study reported in accordance with the JORC Code.CITATION:Dominy, S C, O’Connor, L, Xie, Y and Glass, H J, 2017. Geometallurgical sampling protocol validation by bulk sampling in a sheeted vein gold deposit, in Proceedings Eighth World Conference on Sampling and Blending , pp 185–196 (The Australasian Institute of Mining and Metallurgy: Melbourne).

May 9, 2017

By G J. Lyman

Sampling of grain for mycotoxins is challenging in that some mycotoxins, such as Ochratoxin A (OTA), have very low permitted levels and the kernels carrying the mycotoxin may be highly localised within the bulk of the grain as ‘hot spots’. Sampling for genetically modified organisms (GMOs) can be fraught with similar problems, especially if the contamination is adventitious and random. Incremental sampling under these circumstances can be shown to lead to a high variance in the sampling result.A new design of sampler for grain and other free-flowing particulates has been developed that is effective in practically eliminating the sampling variance due to distributional heterogeneity (DH).This paper introduces the design and argues that the sampler should be free of sampling variance due to DH.CITATION:Lyman, G J, 2017. A new sampler for grain and other free-flowing particulates, in Proceedings Eighth World Conference on Sampling and Blending , pp 277–280 (The Australasian Institute of Mining and Metallurgy: Melbourne).

May 9, 2017

By M D. Sussman

A large number of countries in the world now have some regulation for labelling the presence of genetically modified organisms (GMO) in food. For the last ten years, ISO/TC 34/SC 16 has been providing methods, guidelines and requirements concerning the evaluation of molecular biomarkers including GMOs. This activity makes sure that stakeholders on all sides of commercial transactions are satisfied and their marketing or regulatory needs are met. Developing and maintaining a transparent, science-based environment for users of horizontal methods for molecular biomarker analysis is a high priority. Sampling and analysis of molecular biomarkers has become critically important for food and agricultural organisations developing, producing, distributing and selling genetically engineered, non-genetically engineered, identity preserved and/or organic crops and their derived products and preventing their respective destructive diseases. To aid in international harmonisation, a new work item proposal to develop an ISO type A management system standard (MSS) for organisations including laboratories using molecular biomarker analysis and sampling has been submitted to ISO/TC 34/SC 16. The purpose of this MSS is ensuring that determination of the presence, absence or quantification of a particular biomarker with a relationship to a product quality determinant, a marketing claim, a regulation or a plant pathogen is carried out under a suitable quality assurance / quality control system. This MSS will provide an overarching system to apply and manage molecular biomarker analysis and sampling for any organisation large, medium or small in a private, government or academic setting.CITATION:Sussman, M D, 2017. ISO/TC 34/SC 16 horizontal methods for molecular biomarker analysis – molecular biomarker analysis and sampling management system, in Proceedings Eighth World Conference on Sampling and Blending , pp 295–302 (The Australasian Institute of Mining and Metallurgy: Melbourne).

May 9, 2017

By P Minkkinen

"Pierre Gy showed in his early works that the results in estimating the variance of the mean value of continuous lots depends on sampling mode (random, stratified or systematic), except in the theoretical case that the measurand is randomly distributed in the lot. The experimental variance estimate calculated from samples taken from the lot is independent of sampling mode but depends on sample size and the type of the lot heterogeneity. Variography can be used to estimate the variance of the lot mean for different sampling modes.Gy also defined weighting error, which is made if a simple average is calculated from the assays of samples taken from a continuous object (eg from a process stream) where the flow rate varies with time and is correlated with the concentration of the analyte in the stream.The effect on sampling uncertainty of the three sampling modes and their interaction with the following factors in estimating the uncertainty of the mean concentration is studied using simulated lots, and lots having heterogeneity patterns similar to those of real data sets, including the effect of: the type of heterogeneity (random, grading and clustered heterogeneity patterns) increment size and number employing composite sampling weighting error. CITATION:Minkkinen, P, 2017. Dependence of the variance of lot average on the sampling mode and heterogeneity type of the lot, in Proceedings Eighth World Conference on Sampling and Blending, pp 319–330 (The Australasian Institute of Mining and Metallurgy: Melbourne)."

May 9, 2017

By A C. Chieregati, A Heberle, F L. S P Villanova

Since the publications of Gy’s initial work, many authors have discussed the approaches to estimate the variance of the fundamental sampling error (FSE). Experimental tests suggested by Pitard and François-Bongarçon determine the sampling parameters K and a to calibrate Gy’s formula. The heterogeneity test and the sample tree method are well known and broadly used for many commodities. However, prior to the analysis, the collection of a representative sample is an intriguingly difficult task. The composite for the heterogeneity tests often consists of run-?of-?mine material or processed ore on a conveyor belt. In most cases the final sample does not have enough mass, mixes different geological units, represents a rare blend and does not characterise the deposit. In order to obtain a reliable composite for trace elements and low-grade ores, sampling practitioners suggest at least half a ton of material.To prevent the common representative pitfalls in the FSE, this case study focuses on a single gold-?bearing geological unit collected in situ. With this unit being structured by the mine geologists, but with the interaction between engineers and chemists, it was possible to validate the reconciliation suggested by Pitard (2015). The heterogeneity test and the sample tree method were performed using 500 kg of a smoky quartz from a single geological domain and the results were compared to Gy’s formula. The logging practice by the geologists was modified, emphasising the definition of the size of the gold nugget or its cluster, number of grains and association. Finally, a mineralogical study was conducted on the geological unit to define the liberation factor. All the information is compiled in this paper, showing the need to fully understand the mineralisation to correctly define representative samples and proper analytical techniques.CITATION:Villanova, F L S P, Heberle, A and Chieregati, A C, 2017. Heterogeneity tests and core logging – a final reconciliation, in Proceedings Eighth World Conference on Sampling and Blending , pp 107–114 (The Australasian Institute of Mining and Metallurgy: Melbourne).

May 9, 2017

By M Singh, R C. Steinhaus, J J. Rust

Metallurgical process control systems in mineral plants often employ state-of-the-art monitoring analysers in an attempt to optimise plant operations and enhance mineral recoveries. These systems are unfortunately invariably fed by simplistic ‘sampling’ designs consisting of various in-stream methods or pressurised probes. These are unfortunate compromises and provide a false picture of what a mineral flotation or other process can achieve. The situation is further aggravated by some manufacturers using misleading marketing to suggest discrete cross-stream methods for metallurgical accounting and unreliable continuous in-stream off-takes for control sampling and that they are of synonymous integrity.Reliable, absolute or representative samples are often absent in many modern day concentrator plants, due to compromises around sample taking, particularly in terms of delineation and extraction errors and thus leading to sample bias. For more reliable information from online analysers we need a continuous, but much more reliable sample taken, than is current practice.Online analysers require a fully proportional and continuous supply of sample for effective process control, otherwise maximised or benchmark recoveries will be nothing but a pipe dream. Large discrete cross-stream samples from mechanically less reliable equipment cannot be used as feed to online analysers, as it will negatively impact real-time feedback and efficient process control.It is possible to build correct sampling systems that could do both process controls online and metallurgical accounting at the same time. Multotec Process Equipment (Pty) Limited, in close association with experienced sampling consultant Francis Pitard, have jointly patented a concept to achieve this and which is embodied in the design of the unique Rotating Cone® Sampler (International Patent Application No. PCT/IB2016/057464). This versatile unit for dry and wet applications has high division ratios via two-step sampling and division.A prototype was built and tested to validate the design and levels of precision achievable when employing the unit to split dry mineral fines or even certain foodstuffs.CITATION:Steinhaus, R C, Rust, J J and Singh, M, 2017. Does process control sampling always have to be a compromise?, in Proceedings Eighth World Conference on Sampling and Blending , pp 197–202 (The Australasian Institute of Mining and Metallurgy: Melbourne).

May 9, 2017

By A Hollweg, K H. Esbensen, M Lischka

Two new splitter designs for intercepting moving streams of matter are presented, the variable cross-stream splitter (vCSS) and the rotary stream splitter (RSS), which are intended for implementation in a fully automated modus. The vCSS facilitates obtaining up to three subsamples with different splitting ratios from automated laboratory flow paths, while the RSS divides the intercepted sample into 16 equal analytical aliquots. The splitters can process a wide range of dry, granular and free-flowing materials (<5 mm). The required weight for each subsample can be set without mechanical adjustments in the case of the vCSS. Typical applications are subsampling/splitting for X-ray fluorescence pelletisation (or fused beads) or splitting for average shift composite sampling or grain size analysis. The vCSS combines the advantages of riffle splitting (three user-defined subsample masses) with miniature bed blending and can accept up to 15 kg in total incoming sample mass. In practice, the automated vCSS can deal with unlimited ingoing mass fluxes, while the RSS is designed to operate at a user-defined fixed subsampling rate (1/16th in the tested version). Both splitters have the potential to be optimised and evaluated according to an application/user-defined total sampling error (TSE) + total analytical error (TAE) threshold. Both splitters were tested under adverse conditions and with compositions involving three components with properties representing typical mass fluxes and components concentrations in industrial and technological systems: phosphate concentrate 89.00 per cent (<500 µm), slag ten per cent (>2 mm) and glass spheres one per cent (750–1000 µm). The slag particles vary widely in particle shape, aspect ratio and surface roughness, while the glass spheres are extremely prone to bouncing and segregation (spillage). This test system is extremely difficult to split and thus constitutes a realistic worst-case scenario. While the vCSS does not comply strictly with Theory of Sampling demands for correctness, both splitters may well be able to deliver fit-for-purpose representative results if/when tested properly. Average relative mass deviations ranged from 1.89 per cent for phosphate concentrate (relative standard deviations of 20-fold repeated splitting) to 4.72 per cent for glass spheres and 13.04 per cent for slag. For the RSS, the average replication experiment splitting variation was 2.89 per cent for phosphate concentrate, 4.82 per cent for glass spheres and 11.1 per cent for slag. Both new automated online splitters would therefore appear to be able to be tuned to fit-for-purpose representivity (defined here as maximum relative reproducibility <5 per cent) for many non-extreme materials; however, it is always imperative to test every new type/class of material empirically. There is a wide commercial potential for important materials and commodities sectors, including food, feed, drugs (pharmaceutical), secondary raw materials, recyclates, agricultural products and seeds.CITATION:Lischka, M, Hollweg, A and Esbensen, K H, 2017. New online/atline splitter designs for laboratory automation – feasibility results, in Proceedings Eighth World Conference on Sampling and Blending , pp 159–166 (The Australasian Institute of Mining and Metallurgy: Melbourne).

May 9, 2017

By B Chi, A Mahanta, R A. M Maddever, O Dominguez

BHP Billiton, in collaboration with Sodern and CSIRO, has developed a downhole logging tool for measuring geochemistry in situ. The FastGradeTM 100 (alternatively, FG100), exploits pulsed fast and thermal neutron analysis (PFTNA) to analyse the rock. This method is based on the detection and measurement of characteristic gamma rays emitted from isotopes in the material that become briefly radioactive when they are bombarded with neutrons. The tool outputs raw spectra arising from three types of neutron interaction, namely, inelastic scattering, neutron capture and neutron activation, which occur in successive time-windows (Smith et al, 2015). The spectra contain peaks at specific energies that are characteristic of the different elements interacting with the neutrons, the gamma count of which is related to elemental concentration. BHP Billiton uses in-house data processing software to convert the measured spectra to elemental concentrations.Much work has been done to understand the tool response under a wide variety of environmental and operating conditions, including logging speed, borehole rugosity, the presence of cavities, water etc. Effort has also been directed at understanding and quantifying the current sources of error in our conventional assays due to sample collection, preparation, and laboratory instrumentation in order to use this information as a baseline to compare against the FastGradeTM 100 results.To date, the FastGradeTM 100 primary use has been as a verification tool (JORC, 2012). However, beyond this, it has considerable potential to improve productivity in both exploration and production environments due to its ability to provide geologists with assay results soon after a hole is drilled.This paper presents initial test work and tool performance, including accuracy and precision under various conditions. Results are illustrated by comparing tool measurements to assays from laboratory samples collected from both reverse circulation and diamond-drilled holes.CITATION:Chi, B, Maddever, R A M, Mahanta, A and Dominguez, O, 2017. Near real-time assay with downhole assay tool (FastGrade™ 100), in Proceedings Eighth World Conference on Sampling and Blending , pp 137–144 (The Australasian Institute of Mining and Metallurgy: Melbourne).

May 9, 2017

By M Fritz, B Ziegelaar

"For moisture samples, the International Standard for iron ore sampling states the following: Moisture samples shall be processed as soon as possible and test portions weighed immediately. If this is not possible, samples shall be stored in non-absorbent airtight containers with a minimum of free air space to minimise any change in moisture content, but should be prepared without delay. ISO 3082 (2009)Prescribing what to do is of course easy. Putting it in practice is, however, a little more problematic.The iron ore industry has seen quite some changes in the last ten to 15 years, such as: a large increase in the volume of seaborne trade new ore types and blends coming onto the market new mines, processing plants and port facilities producing more and loading faster an increase in wet processed ore products a dramatic rise and then equally dramatic fall in the iron ore price followed by a recent recovery.All of the above have had an effect on sampling facilities. The large increase in volume coupled with ambitions such as labour efficiency, safer operations and higher precision has driven the installation and retrofitting of automated sampling and robotic sampling preparation plants in order to keep up with increased quality control.Designers of sampling facilities are invariably instructed to build plants that comply with ISO 3082. In general this means that the plants are compliant in respect of the number of primary increments, gross sample masses and sample division rules. The clause at the beginning of this abstract is mostly overlooked as it only appears in a section (4.1) describing the basic requirements of a sampling system.An error of 0.1 per cent in moisture changes the value of 1 Mt of iron ore at US$50/t by US$50 000. At 700 Mt/a of Australian iron ore exports, this is US$35 M. Frequently, errors from 0.5 to 1 per cent in moisture measurements are encountered.In comparison an error of 0.1 per cent Fe is US$0.081 (US$50/t and 62 per cent Fe base) per tonne, that is US$81 000 per Mt or about US$51 M for the same Australian annual iron ore exports.It is somewhat mystifying that we in the ISO Technical committees spend extraordinary energy (= money and time) on improving measurement techniques, whilst the big costs are in sampling errors. This is particularly true for moisture measurement. Our paper quantifies some of the sources of moisture measurements errors and suggests mitigating actions.CITATION:Ziegelaar, B and Fritz, M, 2017. Sampling for the percentage moisture in the iron ore industry, in Proceedings Eighth World Conference on Sampling and Blending, pp 271–274 (The Australasian Institute of Mining and Metallurgy: Melbourne)."

May 9, 2017

By C Haughy, F F. Pitard

Blue Mountain Minerals is a large supplier of calcium carbonate products used by various manufacturers as a major raw material. Very small amounts of free carbon in the rock can have a huge impact in carbon-sensitive applications. The marble in the deposit at Blue Mountain Minerals contains variable amounts of free carbon in the form of trace amounts of the mineral graphite.To make quality improvements, variography was used to measure the impact of non-ideal sampling practices at the Blue Mountain Minerals operations in Columbia, California in the United States. A major investment in correct sampling systems and sampling protocols was implemented over a one-year period. Subsequent variography was used to measure the improvement on the predictability of the graphite content in marble produced by the quarry and as it is processed as an acceptable raw material for extremely carbon-sensitive applications. The improvements of the sampling protocols, control of long-range changes and better knowledge of process cycles are easily noticeable on variograms when they happen to exist. The economic impacts are discussed.CITATION:Haughy, C and Pitard, F F, 2017. Use of variography to measure improvement of sampling practices for the quality control of graphite consistency in marble used for carbon-sensitive applications, in Proceedings Eighth World Conference on Sampling and Blending , pp 255–260 (The Australasian Institute of Mining and Metallurgy: Melbourne).

May 9, 2017

By D A. Vogel

"Sampling may be mandatory for safety or environmental reasons. It may be because the value of a sale needs to be determined or to provide important data for investments. Whatever the reason, an incorrect sample will come at a risk. Only the person that relies on sampling data can imagine what the potential cost of such risk is.Sampling itself has a cost too; the direct cost of a quality control department and indirect cost for interrupting a process, such as stopping a conveyor belt during loading. If all material is the same then only one small sample is enough. But if all the material is different, then a very large sample must be tested; in the conveyor belt scenario it must be stopped very frequently and the associated cost will be high.A good sampling plan finds the proper balance between risk and cost. A good sample must be small enough to handle yet large enough to be representative for what information is required. All good samples have three things in common: there was proper access to the material the sampling tool allowed equal probability of particles being selected nothing was disregarded, everything was included.This too is the core of Gy’s ideal sampling scenario, where all particles have the same probability of being selected and included in the sample. There is risk of bias when part of a lot is excluded from the sample. Extra care is needed when sampling particularly heterogeneous, segregated or layered materials. The sampling tool should never systematically leave out a portion; it shall not push aside large or hard particles. Wet samples shall not stick to the sampling tool. This extends to sample preparation too when reducing particle size and dividing the sample in the process.CITATION:Vogel, D A, 2017. Access, select, include – a review of the commercial sampling of traded bulk commodities in the context of Gy’s Theory of Sampling, in Proceedings Eighth World Conference on Sampling and Blending, pp 367–378 (The Australasian Institute of Mining and Metallurgy: Melbourne)."

May 9, 2017

By F F. Pitard

Observation from attending many conferences around the world reveals that many people are constantly shifting from the word ‘error’ to the word ‘uncertainty’ without a clear vision of what the subtle difference is. The same applies to a large swath of the scientific literature, in which these two concepts are unfortunately often used synonymously, which is a scientific flaw of the first order. In practice, as demonstrated in the Theory of Sampling (TOS), there are both sampling errors and sampling uncertainties. Sampling errors can easily be preventively minimised, while sampling uncertainty for a preselected sampling protocol is inevitable. P M Gy’s work, supported by G Matheron’s critical reviews, demonstrates beyond any possible doubt that sampling uncertainty cannot be achieved if sampling errors are not preventively minimised by applying very stringent rules that are too often ignored by sampling practitioners. A clarification, in line with the TOS, is presented to mitigate the confusion created by circles favouring the use of uncertainty versus the use of the word error. Both words are very different in nature and scope and they are both necessary in their due place and time.CITATION:Pitard, F F, 2017. From errors to uncertainty – a clarification for proper use by the Theory of Sampling, in Proceedings Eighth World Conference on Sampling and Blending , pp 355–358 (The Australasian Institute of Mining and Metallurgy: Melbourne).

May 9, 2017

By L Kempenaers, T van der Maten, U König, M G. C Zoontjes

Improving liquid blending operations in production plants is of paramount importance in many industries like mining, oils, petrochemicals and food. For accurate blending of different types of process streams, like homogeneous, heterogeneous, cold or hot streams, it is essential that the composition can be accurately monitored in real time. Advances in energy dispersive X-ray fluorescence (EDXRF) technology offers the possibility to apply XRF for online elemental analysis. In this paper we discuss an online EDXRF analyser for accurate blending control in many industries.EDXRF has the advantage that multiple elements can be measured simultaneously over a wide concentration range of wt per cent down to a few mg/kg, for elements between sodium (Na) and americium (Am) in the periodic table. To reach the desired accuracy and precision in an online XRF analyser, it is important to bring a representative and homogeneous sample of the process flow into the optical path of the analyser. The combination of a laminar flow cell design and advanced EDXRF technology leads to low detection limits and high performance. The obtained precision and accuracy allows for much tighter process control. In this paper we present two examples of industrial applications.CITATION:Zoontjes, M G C, van der Maten, T, Kempenaers, L and König, U, 2017. Online X-ray fluorescence spectrometer for accurate monitoring and improving blending sharpness in liquid production processes, in Proceedings Eighth World Conference on Sampling and Blending , pp 235–238 (The Australasian Institute of Mining and Metallurgy: Melbourne).

May 9, 2017

By T M. El Hajj, A C. Chieregati, M T. G C Marques, J M. Braga Jr, H Delboni Jr

"The estimate and control of sampling errors is essential for all mining operations, especially when dealing with strategic materials, such as niobium. This paper is based on a traditional quality assurance and quality control technique used to confirm mineralised intersection zones, and proposes the validation of sampling data using diamond drill twin holes, in order to evaluate and improve the quality of the current data set. The geological model of the niobium mine – located in Araxá, Minas Gerais, Brazil – was built including 77 diamond drill holes of a sampling campaign carried out during the 1970s. There are few records about the collection and preparation procedures of these past samples, making it difficult to estimate the errors associated with their analytical results. Therefore, 11 twin drill holes were drilled to compare their samples against the samples from the 1970s drilling campaign. The new hole locations were chosen as near as possible to the old holes location. The JORC Code, 2012 edition, relates the use of twin holes as a good sampling technique to assess sampling and assay data. Three parameters were considered to validate the 1970s drill hole campaign by the twin drill hole campaign: the difference of the coordinates between the paired holes the difference among the positions of the geological contacts to validate the geologic description the difference between the paired samples grades was calculated to validate the analytical results.The analysis of the grade range variation between sample pairs was based on three techniques: geostatistical interpretation of vertical semivariogram to obtain the nugget effect calculation of the relative difference between duplicates calculation of the error associated with the sampling protocol established according to Pierre Gy’s Theory of Sampling.The results were satisfactory for hole coordinates; however, the ore content showed a high deviation, although this divergence can be explained by the nugget effect. For the lithological intervals, the new results contributed to an effective identification of variations among lithologies, ie brown weathered ore and saprolite. The twin drill holes campaign successfully validated the geological model data, allowing effective mine planning and higher reliability of the reconciliation results.CITATION:Marques, M T G C, El Hajj, T M, Braga Jr, J M, Chieregati, A C and Delboni Jr, H, 2017. Theory of Sampling and geostatistics for twin drill holes analysis in a niobium mine in Araxá, Brazil, Australia, in Proceedings Eighth World Conference on Sampling and Blending, pp 97–106 (The Australasian Institute of Mining and Metallurgy: Melbourne)."

May 9, 2017