GRAVITY GOLD 2010 21-22 September 2010 BALLARAT, VICTORIA The Australasian Institute of Mining and Metallurgy Publication Series No 8/2010 A u s i mm THE MINERALS INSTITUTE Edited by Dr Simon Dominy Published by: The Australasian Institute of Mining and Metallurgy Level 3,15 - 31 Pelham Street, Carlton Victoria 3053, Australia i ©The Australasian Institute of Mining and Metallurgy 2010 All papers published in this volume were refereed prior to publication. The Institute is not responsible as a body for the facts and opinions advanced in any of its publications. ISBN 9781 921522 291 Desktop published by: Kylie McShane and Olivia Tet Fong The Australasian Institute of Mining and Metallurgy Printed by: BPA Digital 11 Evans Street, Burwood Victoria 3125, Australia ii FOREWORD Multidisciplinary teams are a distinct feature of our global mining industry. Gravity Gold 2010 aims to bring together geologists, engineers and metallurgists. As a result of the Global Financial Crisis, there are now great corporate pressures to reduce costs and increase efficiencies and all within a framework of safe operations and environmental and economic sustainability. In the face of such increasing challenges, we need to redirect our work assiduityto cost-effectiveness and efficiency, technical competency and innovation. Optimisation of our activities, particularly of mining and metallurgical recovery has become ever more critical. Gravity Gold 2010 aims to explore the issues of mining operations that are characterised by coarse-gravity gold mineralisation with an often high-nugget effect. These deposits usually require strong geological control, selective mining methods and apply gravity-based processing technology. Key areas of common interest include: understanding the orebody; resource and reserve estimation; predicting grade and metallurgical variability; optimising mining and metallurgical extraction; definition of project risk and economics; tailings and waste handling; and environmental sustainability. This two day conference will share problems and learning experiences in a multidisciplinary environment within the rural setting of the Central Victorian goldfields. We have put together a program of some 20 papers covering issues related to flotation; gravity concentration; leaching; metallurgical testing; selective mining and stope cleaning; plant optimisation; orebody variability; safety and security; and sampling. Case studies include projects in Africa, Australia and North and South America. I hope that the proceedings volume, together with the oral presentations, discussions with the presenters and the open forum, will provide insights to challenge the way we do business. The organisation of a conference requires the support of a number of people. I would like to take this opportunity to thank everyone who has helped on the Organising Committee, without their dedication and enthusiasm the conference could not have taken place. On behalf of the Organising Committee, we would like to thank the authors and presenters of the technical papers for their high quality contributions. I would particularly like to thank the band of reviewers who undertook the task of maintaining the quality of the papers. Finally I would like to acknowledge the contribution provided by industry organisations and companies throughout Australia and overseas, for sponsorship, the trade exhibition and for supporting delegates to attend this meeting. I hope that all delegates will participate in the conference technical and social sessions and meet and engage with old and new friends, thus ensuring that the ideas, solutions and technologies presented will be of benefit to all. Dr Simon Dominy FAuslMM(CP), Snowden Group Conference Chair vii C O N T E N TS Keynote Addresses Heterogeneity, Sampling Errors and the Nugget S C Dominy, IM Platten 3 Effect in Gold Ores - Implications for Evaluation, andRCAMinnitt Exploitation and Extraction Where is the Risk in Coarse Gold Projects? WΒ Edgar 19 Gravity Gold - The Process of Change S Gray 21 Stope Cleaning - Historical Methods and Future M Tuck 25 Developments Papers Gravity Gold Recovery in Obuasi Mines J Κ Afidenyo, J Bernasko 29 and J Osei-Owusu Coarse Gold Separation at Bendigo Mining's M Braaksma and Ρ Gray 41 Kangaroo Flat Operation Development of the Sao Chico Gold Project, Brazil Κ Chappie 49 The Use of Nuclear Tracers to Evaluate the Gold R Clayon 51 Recovery Efficiency of Sluiceboxes Minimum Industry Gold Room Security Standards D Connelly 63 Selection Sizing and Developing the Optimum D Connelly 69 Gravity Gold Circuit for Your Project Redevelopment of the Morning Star Gold Mine G Curnow, Ρ Hepburn- 81 Brown and Ε O'Grady Determining Gold Particle Size in Gravity Ores for S C Dominy, IM Platten 83 Sampling and Metallurgical Characterisation - and Y Xie Discussion and Test Protocol Predicting the Benefit of Gravity Recovery Prior to M Fullam 97 Flotation Is the Python Right for Me? Python Amenability Τ Hughes, Κ Donaldson 101 Test Work and Β Murphy xiii Geology and Definition of the Wattle Dam Coarse R Hutchison 109 Gold Deposit The Role of Gravity and Intensive Cyanidation in Ν Katsikaros and Ρ Davies 119 Processing Preg-Robbing Gold Ores Effect of Gravity Recovery on Overall Plant CNaudι 121 Recovery, Sadiola - A Case Study Safety Culture and Resilience Engineering - M Pillay, D Borys, D Else 129 Exploring Theory and Application in Improving and M Tuck Gold Mining Safety Maximising Payable Gold in Base Metal C Potorieko, A Duran 141 Concentrators and R H Cuttriss Recovery of Gold from Gravity Tailings by Flotation J Rowe and S Sidrak 149 Modelling the Effect of Gravity Gold Recovery on W Ρ Staunton and AR Box 155 Leaching Performance Resue or Split Face Firing - Trial Results from M Tuck and Ρ Ganza 159 Hillgrove Mine In Plant Sampling for Gravity Recoverable Gold Β Watson 165 Author Index 167 xiv Keynote Addresses Heterogeneity, Sampling Errors and the Nugget Effect in Gold Ores - Implications for Evaluation, Exploitation and Extraction S C Dominy1-2,1 M Platten3 and R C A Minnitt4 ABSTRACT Sampling errors result from heterogeneities within the lot under study. Within a rock mass in situ heterogeneity results in the geological nugget effect, which accounts for most of the natural variability. In broken rock, constitution and distribution heterogeneities are important. A major result of the constitution heterogeneity is the Fundamental Sampling Error (FSE) which is the smallest residual error that can be achieved. It cannot be eliminated, but can be minimised and reflects the fact that the broken rock sampling process is never completely free of error. Together with other sampling errors, FSE contributes to the total nugget effect. The FSE for a sampling protocol can be evaluated using the so-called Gy equation. Its practical applications to low-grade gold ores have often been unsuccessful due to problems linking the FSE and minimum sampling mass and the definition of key parameters such as the liberation diameter. Well-designed sample protocols minimise the total sampling error and hence the nugget effect. Reduction of the nugget effect is a key requirement for all sampling programs whether related to exploration and resource evaluation, metallurgical test sampling, grade control or metallurgical plant sampling. This contribution reviews the relationship between heterogeneity, FSE and nugget effect in gold deposits; and emphasises the need for FSE management throughout the mine value chain. INTRODUCTION Poorly designed sampling protocols can result in elevated collection is followed by sample reduction in both mass and project risk by artificially increasing grade variability fragment size to provide an assay charge or test sample. (ie the nugget effect). Sample collection, preparation and The sampling process can be particularly challenging in the assay protocols that are based on the characteristics of the gold environment (Sketchley, 1998; Pitard, 1993) and no more ore type, together with Quality Assurance/Quality Control so than in the presence of coarse (gravity) gold ores (Royle, (QAQC) systems, will reduce the nugget effect. Reporting 1989; Dominy et al, 2000; Dominy and Petersen, 2005; codes (eg JORC, SAMREC, etc) require the Competent Person Cintra er al, 2007). From a sampling perspective, coarse gold to consider the quality and implication of sampling and is considered to be greater than 100 μπι in size (Dominy et al, assaying programs. High sampling errors beget higher project 2000; Dominy, Platten and Xie, 2010). Where it constitutes uncertainty; despite this fact sampling does not receive the more than ten per cent of all gold present, problems are likely attention it deserves. to be encountered during sampling and assaying. Sampling is thus a vital component to all stages of a mining The inherent nature of the mineralisation, in particular project and includes the sampling of in situ material and the nature and distribution of the gold particles, will control broken rock (Carrasco, Carrasco and Jara, 2004; François- sampling ease or samplability. Within the coarse- (>ιοο μπι) Bongarçon, 2004; Pitard, 2005, 2007; Dominy, 2007; to fine-gold (<ιοο μπι) spectrum, the samplability of a Minnitt, 2007; Holmes, 2008). Poor sampling produces deposit ranges from relatively simple for fine-grained hidden financial losses related to low quality resource/reserve estimates, the misclassification of ore and waste in the mine disseminated gold particles, through to difficult for coarse and poor mass balances in the plant (Carrasco, Carrasco and gold mineralisation. There is strong evidence to show that Jara, 2004; Minnitt, 2007). The aim of any protocol must relationships exist between increased gold grade and larger, be to extract a representative and reproducible sample that potentially clustered gold particles (Dominy and Platten, accurately describes the material in question. Primary sample 2007; Dominy, Xie and Platten, 2008b). 1. FAuslMM(CP), Executive Consultant and General Manager (UK), Snowden Mining Industry Consultants Limited, Abbey House, Wellington Way, Brooklands Business Park, Weybridge Surrey KT130TT, England. Email: sdominy<8>snowdengroup.com 2. Adjunct Senior Principal Research Fellow, School of Science and Engineering, University of Ballarat, Mount Helen, Ballarat Vic 3353. 3. Senior Principal Consultant, Snowden Mining Industry Consultants Limited, Abbey House, Wellington Way, Brooklands Business Park, Weybridge Surrey KT13 OTT, England. Email: [email protected] 4. JCI Professor of Mineral Resources and Reserves, School of Mining Engineering, University of Witwatersrand, Private Bag 3, Wits 2050, South Africa. Email: [email protected] GRAVITY GOLD CONFERENCE / BALLARAT, VIC, 21 - 22 SEPTEMBER 2010 3 S C DOMINY, I M PLATTEN AND RCA MINNITT Part of the total nugget effect is directly related to the throughout the mine value chain are presented in Dominy sampling process, that is, to the mass of the field sample (2010a and 2010b), Holmes (2005 and 2008), Minnitt (2010) taken; the effectiveness of collection; sample preparation (eg and Pitard (2005). crushing, pulverising and splitting); the mass of subsamples (samples for pulverising and assay); and analysis. THE NUGGET EFFECT This paper reviews the relationship between heterogeneity, Defining the nugget effect FSE and nugget effect; and emphasises the need to reduce FSE and hence nugget effect throughout the mine value chain. Nugget Effect is a quantitative t term describing Challenges are put in the context of evaluation, exploitation the inherent variability between samples at very small and extraction of hard rock deposits. This paper principally separation distances; though in reality has a wider remit than addresses the challenges of gold deposits, however the just differences between contiguous samples. It is effectively discussions are generally applicable to other precious metals, a random component of variability that is superimposed on native copper, molybdenite and galena bearing deposits that the regionalised variable and is defined in a variogram as the may show complex distribution and delayed particle size reduction. percentage ratio of nugget variance to total variance (the sill). Deposits that possess nugget effect values above 50 per cent SAMPLING ERRORS and particularly above 75 per cent are the most challenging to evaluate (Dominy et al, 2003). The magnitude of the nugget Sampling errors are defined in the Theory of Sampling ('TOS') effect is related to: as promulgated by the works of Pierre Gy (Gy, 1979,1992 and 1998). They are relevant to all sampling applications, including • in situ heterogeneity of the mineralisation, those used during evaluation, exploitation and extraction. The • sample support, key errors are (Gy, 1979; Pitard, 1993 - Table 1): • sample density, • Fundamental Sampling Error (FSE), • sample quality, and • Grouping and Segregation Error (GSE), • assay procedures. • Delimitation Error (DE), Through the mine value chain, optimised sampling protocols aim to reduce the total nugget effect thereby also reducing: • Extraction Error (EE), • Weighting Error (WE), • the total sampling variance, • Preparation Error (PE), and • the skewness of the assay distribution, and • Analytical Error (AE). • the number of extreme assay values. The FSE and GSE are irreducible random errors related to Composition of the nugget effect the inherent heterogeneity and characteristics of the material being sampled and can only be minimised through optimised The nugget effect reflects the proportion of nugget variance sampling protocols. The other errors arise as a consequence of present in a data set - in effect the random variability. This the physical interaction between the material being sampled variability has two principal components; (a) geological or in and the technology employed to extract the sample and result situ nugget effect (GNE) and (b) sampling nugget effect (SNE). in a sampling bias. Provided the correct sampling technology and procedures are employed the sampling bias can be The GNE component relates to the microscopic differences totally eliminated. All errors are cumulative and contribute in composition throughout the mineralised zone/domain - it to the total sampling error, which in turn contributes to the is in effect the in situ constitution heterogeneity (Platten and nugget effect. FSE and GSE are likely to contribute between Dominy, 2003). This principally refers to the distribution of approximately 50 per cent and 90 per cent and EE to AE up to single grains or groups of gold particles distributed through 25 per cent of the total sampling error. the ore to larger continuous zones. Variability is most serious Theoretical discussions on sampling errors can be found where there are very small-scale, low continuity structures in Gy (1979, 1998) and Pitard (1993). Practical implications such as high-grade gold carriers within the main structure TABLE 1 Definition of sampling errors. Sampling error Error definition Fundamental Sampling Error (FSE) The FSE results from grade heterogeneity of the broken lot. Out of all sampling errors, the FSE does not cancel out and remains even after a sampling operation is perfect. Experience shows that the nugget effect can be artificially high because sample weights are not optimised Grouping and Segregation Error (GSE) Relates to the error due to the combination of grouping and segregation of rock fragments in the sampled lot. Once a rock volume is broken, there will be segregation of particles whether this be in a surface stockpile or laboratory pulp Delimitation Error (DE) The DE results from an incorrect shape of the volume delimiting a sample Extraction Error (EE) The EE results from the incorrect extraction of a sample. Extraction is only correct when all fragments within the delimited volume are taken into the sample Weighting Error (WE) WE relates to collecting samples that are of a comparable support. Samples should represent a consistent mass per unit Preparation Error (PE) Refers to issues during sample transport, preparation (contamination and losses), and unintentional and intentional human error Analytical Error (AE) Relates to all errors during the assay and analytical process 4 GRAVITY GOLD CONFERENCE / BALLARAT, VIC, 21 - 22 SEPTEMBER 2010 HETEROGENEITY, SAMPLING ERRORS AND THE NUGGET EFFECT IN GOLD ORES or vein-lets within wall rocks (Platten and Dominy, 2003; HETEROGENEITY Dominy and Platten, 2008). The dissection and displacement of a previously continuous high-grade vein by later barren Defining heterogeneity events will also lead to a high GNE. A clear indication of the In situ heterogeneity GNE is where two halves of a drill core are assayed and show order of magnitude or more differences in assay grades. It is At the in situ level (eg collection of a sample as drill core or not impossible for the GNE to account for up to 50 per cent of from a mine development face) heterogeneity is based on the total nugget effect. the nature of gold particle size and distribution (Platten and Dominy, 2003; Dominy and Platten, 2007; Pitard, 2007; The SNE component is related to errors induced by Dominy, Xie and Platten, 2008b). Evenly disseminated inadequate sample size (FSE and GSE), sample collection mineralisation will have lower heterogeneity (eg Carlin style (DE, EE and WE), preparation methods (PE) and analytical mineralisation) compared to more erratically distributed procedures (AE). In some instances, the SNE is the dominant styles of mineralisation that have a higher heterogeneity (eg part of the total nugget effect and reflects a high FSE and non- mesothermal veins). In situ heterogeneity leads to the in situ optimal protocols. or geological nugget effect (GNE - Platten and Dominy, 2003; The contribution of GNE and SNE to the total nugget effect Pitard, 2007). provides an intimate link between sampling and geostatistics The GNE is reduced by collecting larger and more closely (François-Bongarçon, 2004). spaced samples (hence more samples). In situ sample mass requirements to achieve a given level of precision can be Nugget effect, sample support and estimated using a Poisson approach, where the rare gold volume-variance particles are assumed to have a Poisson distribution (Dominy, 1997; Dominy, Platten and Xie, 2010). The term 'support' embodies the notion that samples have a specific size, mass, shape, orientation or any other physical Broken rock heterogeneity characteristic that is common to them. Samples are usually In the broken rock condition, heterogeneity is defined as the collected at specific supports; for example 1 m lengths of nature of a lot under which all the contained fragments are diamond drill core. These samples represent average values non-identical. In this case, two types of heterogeneity are measured over specified geometrical shapes oriented in defined: (a) the Constitution Heterogeneity (CH) and (b) the specific directions. As the size of the sample increases, it will Distribution Heterogeneity (DH). CH and DH contribute to tend to encompass more of the structures that give rise to the sampling nugget effect (SNE), through the FSE and GSE. variability (eg GNE). In particular, it is likely to include more The CH is the heterogeneity inherent to the composition small-scale structures and/or gold particles. (eg gold grade, etc) of each fragment making up the lot. The The total nugget variance is inversely proportional to the greater the difference in composition between each fragment, the higher will be the CH of the lot (CH). Mixing and volume of the sample. In general, the variance (as nugget L homogenisation has no effect on CH. The CH is the principal variance) of samples decreases as their volumes increase, cause of the FSE, which occurs when a sample is selected from although not in a linear manner (the volume-variance a broken lot. Gold ores often show the ultimate heterogeneity, relationship). Other regionalised variable parameters such as where relatively few fragments in a lot contain large quantities spatial variance and geostatistical range remain unaffected. If of gold. the sample size (eg core diameter) is increased indefinitely the CH, relates to the differences in composition for a given limiting value of the variance is the spatial variance, this will number of fragments (N,.): remain virtually unaffected. Increasing the sample length will reduce the contribution from spatial variance, but only when CH = N„.Σίβ, - a,)2 M* / a,2. M,2 (2) the geostatistical range is small. When the geostatistical range L is large, then long samples are required to reduce the variance. where: However, a point is soon reached at which the benefit from any a is the content of a fragment reduction in variance is offset by the reduction in selectivity ( a, is the average content of the lot arising from long sample lengths. Mj is the mass of a fragment and M the mass of the lot L Relationship between nugget effect and The DH represents the difference in average composition of sampling errors the lot from one place to the next in the lot; it is responsible The total nugget effect (TNE) is composed of a series of for the irregular distribution of grade and values in groups of individual errors that are characteristic of both the in situ fragments of broken ore. The DH can be influenced by large gold distribution and sampling quality issues. It can be differences in density and fragment composition. DH of a summarised as: given lot (DN, ) is controlled by CH,, the spatial distribution of fragments and the lot shape. TNE = GNE + [£{FSE + GSE + DE + EE + WE + PE} For a given number of fragment groups (N), DH, is given G + AE] (1) as: The errors FSE to PE are applied to each sampling stage, DH, = N T(aj - a, )2 M2 / a,2. M,2 (3) G n from sample field collection to subsampling prior to assay - errors are cumulative. Good procedures in the laboratory do where: not cancel out a poor quality field sample or inappropriate a, is the content of a group, a, the average content of primary split. the lot GRAVITY GOLD CONFERENCE / BALLARAT, VIC, 21 - 22 SEPTEMBER 2010 5 S C DOMINY, I M PLATTEN AND RCA MINNITT M is the mass of a given group. 2 χ 104, it is important to determine its order of magnitude n DH leads to GSE, though it is also related to FSE since (Pitard, 1993). The value of HIL should be expressed with no a group of fragments will bear an inherent FSE. From a more than two significant figures (Gy, 1992). practical perspective, DH is not measured and GSE is kept to L Determining heterogeneity a minimum through accumulating as many small increments to form the composite sample as possible. The 'Gy 50/100 Piece' Experiment The determination of HI, is of major concern when applying Heterogeneity invariant the FSE equation. The method for its indirect determination A measureable approximation of CH, is the Heterogeneity is based on the 'Gy 50/100 piece' experiment, also known as Invariant (HI) for the lot, which is the product of the sampling the Heterogeneity Test (HT - Gy, 1979; Pitard, 1993; 2004). L The data can be processed to estimate the value of Κ and the constant (K) and the nominal fragment size d: n liberation size of the gold (d). L For coarse d ores (eg a stockpile or mill feed), 50 to 100 HI = CH.M/N (4) n L L L F individual large fragments of 1.5 kg to 2.5 kg are picked one by one and each assayed to extinction. For finer ores (eg blast hole and RC chips or laboratory jaw crusher product), 50 to HIL = K-D„B (5) 100 groups of 50 individual fragments are picked one by one. Each group of 50 fragments are each assayed to extinction. where: Figure 1 demonstrates the heterogeneity of a high-nugget M is the mass of the lot L gold system, through the large number of low-grade samples The term M,/N, is the average fragment mass. The HI, t (<3 g/t Au) and only a few high-grade samples (between 36 g/t is effectively the constant factor of CH and is a numerical Au and 1292 g/t Au). The calculated Κ value was 350 g/cmls. measure of how heterogeneous a sample lot is, having the In this case, whilst the HT indicated a variable ore, extensive dimensions of mass in g and is inversely proportional to grade processing of bulk samples proved that the gold particle (Gy, 1992). The value of HI, is defined as the Ί/iooth of the size was very coarse (~i cm) and the Κ value closer to sample mass that would generate a relative sampling standard 193000 grammes/cm15. deviation equal to ten per cent at the 95 per cent confidence The methodology is based on the hypothesis that the grade level' (Gy, 1992). Gy reports that values of HI, range from of a size fraction and the proportion M /M (where M is La L (/i 108 grammes for a finely ground high-grade concentrate to the mass of the size fraction and M, is the mass of the lot) 19 000 grammes for an alluvial gold ore (Gy, 1992). The of this size fraction in the lot, varies minimally between the fact that the major component of the sampling variance is size fragments (Gy, 1979; Pitard, 1993). It is then possible to proportional to HI leads to two conclusions. Firstly, it is calculate HI, from the formula: L necessary to rely upon effective estimation of the HIand L secondly, when a factor can take values ranging from 108 to HI = g [Σ (a - a)7a2. M* /M] (6) L q Q Q q Q FIG 1 - Heterogeneity curve resulting from a 'Gy 50 Piece' experiment (2.5 kg samples) from a high-nugget gold-quartz reef in Australia. 6 GRAVITY GOLD CONFERENCE / BALLARAT, VIC, 21 - 22 SEPTEMBER 2010