Design grade control

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/289007206
Design of grade control sampling programs for underground gold mines
Article · January 2011
CITATION
1
READS
7,538
2 authors:
Some of the authors of this publication are also working on these related projects:
Paper presented at the July 2013 SAIMM - Sampling in Africa conference View project
Mine value chain sampling optimisation View project
Simon Dominy
University of Exeter
96 PUBLICATIONS 841 CITATIONS
SEE PROFILE
Richard Minnitt
University of the Witwatersrand
79 PUBLICATIONS 497 CITATIONS
SEE PROFILE
All content following this page was uploaded by Simon Dominy on 13 July 2017.
The user has requested enhancement of the downloaded file.

Underground grade control protocol design:
case study from the Liphichi gold project,
Larecaja, Bolivia
S. C. Dominy*1,2,3
, I. M. Platten4
, Y. Xie5
and R. C. A. Minnitt6
Grade control programmes aim to deliver economic tonnes to the mill via accurate definition of ore
and waste. The foundation of successful grade control is high quality sampling supported by
geology and a suitable data management and modelling system. For underground operations
sampling methods include chip, channel and panel samples, grab/muck pile samples, and drill-
based samples. Grade control strategy is related to mining method and orebody type. Protocols
must be designed to suit the style of mineralisation in question. Holistic studies focusing on ore
mineralogy and gold particle deportment, size and distribution are required for sample collection
and preparation protocol optimisation through ‘Theory of Sampling’ application. Where possible,
characterisation programmes should be undertaken early in the life of a project. Appropriate
assaying procedures are also required. Programme implementation will require suitably skilled
individuals to train and mentor staff. On-going quality assurance/quality control monitoring and
review will allow protocols and staff to be updated as required. A case study of protocol design
from the Liphichi gold-antimony project in Bolivia is presented.
Keywords: Gold, Veins, Grade control, Sampling methods, Theory of Sampling, Ore characterisation, Bolivia
List of abbreviations
ACE accelerated cyanide extraction
AE analytical error
DE delimitation error
DSA duplicate series analysis
EE extraction error
ESD equivalent spherical diameter
FA fire assay
FSE fundamental sampling error
GRG gravity recoverable gold
GSE grouping and segregation error
HARD half absolute relative difference
HT heterogeneity test
PAL pulverise and leach
PE preparation error
P80/P90 80 or 90 per cent passing a given sieve size
QAQC quality assurance/quality control
ROM run of mine
SFA screen fire assay
TOS Theory of Sampling
Introduction
In any mining operation, ore and waste must be defined
effectively to ensure an economic mill feed. Grade control
is thus essential for efficient and economic mine opera-
tion, with key performance indicators including definition
of ore and waste, minimal dilution and optimal recovery.
The process of grade control comprises data collection,
data integration and interpretation, local resource/reserve
estimation, input into final stope design, supervision of
development and mining, and stockpile management
leading ultimately to reconciliation.
The requirement for quality samples and cost-effective
sampling protocols during exploration and mine produc-
tion has long been recognised (Sketchley, 1998; Vallée,
1992, 1998; Pitard, 2005; Dominy, 2007; Dominy, 2010a,
2010b; Minnitt, 2007, 2010). The consequences of poor
sample quality include erroneous resource/reserve esti-
mates, ore and waste misclassification, lost opportunities
and lower profitability (Carrasco et al., 2004; Minnitt,
2007).
This paper provides guidelines for the design of grade
control protocols for new and developing underground
gold operations, including the choice of appropriate
sampling method with respect to mining method and
orebody, and ore characterisation for sampling protocol
1
Snowden Mining Industry Consultants Pty Ltd, PO Box 416W, Ballarat
West, Vic 3350, Australia
2
Western Australian School of Mines, Curtin University, Bentley Campus,
Perth, WA 6102, Australia
3
School of Science, Engineering and Information Technology, University
of Ballarat, Mt Helen, Vic 3353, Australia
4
Snowden Mining Industry Consultants Ltd, The Magdalen Centre, Robert
Robinson Avenue, Oxford Science Park, Oxford OX4 4GA, UK
5
School of Civil and Environmental Engineering, University of Science and
Technology Beijing, Beijing 100083, PRC
6
School of Mining Engineering, University of Witwatersrand, Private Bag 3,
WITS 2050, South Africa
*Corresponding author, email sdominy@snowdengroup.com
ß 2010 The Australasian Institute of Mining and Metallurgy
Published by Maney on behalf of the Institute and The AusIMM
Received 12 October 2011; accepted 2 November 2011
DOI 10.1179/1743275811Y.0000000016 Applied Earth Science (Trans. Inst. Min. Metall. B) 2010 VOL 119 NO 4 205

design. A case study from the Liphichi Project is
presented.
Grade control considerations
Introduction
Underground grade control practices have evolved from
the use of paper-based methods, through to three-dimen-
sional (3D) modelling and geostatistical simulation. The
foundation of all programmes is that of geological
understanding led by sampling and geological mapping
(Rickard, 1907; McKinstry, 1948; Peters, 1987; Dominy
et al., 2009).
Sampling for grade control
Grade control samples should be reproducible, unbiased,
taken in a safe manner and operationally timely. A good
grade control sampling strategy should provide quality
information on gold grade and its relationship to geology.
Common grade control sampling methods are discussed
in Dominy (2010a, 2010b; see Tables 1 and 2).
Geological mapping
Geological mapping has a positive impact on grade
control and supports mine development. It is particularly
applicable where orebody geology and geometry are
variable and mine openings need to be planned accord-
ingly. The primary objective is to identify and locate the
ore zone or at least the section likely to contain the
ore with respect to the minimum stoping width. The
secondary objective is to determine the attitude and
position of relevant structures that control the 3D form of
ore shoots or gold carriers (Dominy et al., 2009, 2012).
Mining method and grade control
Grade control strategy is related to mining method,
which can be either an entry or non-entry type (Tables 3
and 4). Stopes provide 3D exposure, with entry methods
such as shrinkage and cut-and-fill stopes offering
opportunities for on-going mapping and sampling.
These are well-suited to complex, narrow (,3 m) sys-
tems that require strong geological control, and some
selectivity and flexibility. Non-entry methods such as
longhole open stoping can only be reliably mapped and
sampled in development drives, sub-levels and slot rises.
These methods are suited to simple, planar but thicker
(.3 m) structures where geological control is less critical
and bulk extraction of the vein is appropriate. Longhole
stoping has minimal in-stope flexibility if the structure
proves more complex than expected.
Sample spacing and size
Grade control sample spacing is usually controlled by
needs of the orebody and practicality. A small coarse-
gold, high-nugget operation will usually require close
spaced samples at 1 to 2 m to resolve small-scale
variability, though in bigger, less variable systems face
samples at a greater separation will likely suffice. Cross-
cuts through a wide orebody will require a continuous
sample in 1 to 2 m lengths. The separation distance
between cross-cuts along strike will tend to be a mining
decision, but should be responsive to geological require-
ments as well. Where enough sample data are available,
the construction of semi-variograms will assist in
determining optimum sample spacing.
It is important to understand that sample size (mass)
is related to the frequency and size of gold particles
(Dominy et al., 2008b). When the frequency within a
mineralised zone and within an individual sample is
high, then sampling problems are not usually significant.
When gold particles become rare, which could be
Table 1 Underground grade control sampling types and
methods
Sample type Sampling method
Linear Chip-channel
Chip-panel
Channel
Broken rock Grab
Drill hole Diamond core
Blast hole/sludge
Table 2 Positives and negatives of sample types available for underground gold mine grade control (for error deﬁnitions
refer to Table 5) after Dominy (2010a, 2010b)
Sample type Positives Negatives
Chip Moderately easy to collect and fast High EE and DE
(chip-channel) Reasonable number can be collected Generally poor quality
Relatively cheap
Chip Easy to collect and fast Very high EE and DE
(chip-panel) High number can be collected Generally very poor quality
Relatively cheap
Channel Requires substantial effort Moderate EE and DE
Moderate number can be collected Moderate quality
Time consuming
Channel Moderate number can be collected Relatively costly
(saw cut) High quality (low EE and DE) Requires diamond saw
Grab Easy to collect Very high FSE, GSE, EE and DE
Moderate to high number can be collected Poor–very poor quality
Moderately cheap Manual handling issues
Table 3 Application of different sampling methods to
activity in the mine production chain. Refer also
to Table 4 which integrates sampling method to
different stoping methods
Activity in the mining chain Sampling method
Development/pre-production Linear
Grab
Diamond core
Blast hole/sludge
In-stope Linear
Grab
Blast hole/sludge
Post-stoping Grab
Dominy et al. Underground grade control protocol design
206 Applied Earth Science (Trans. Inst. Min. Metall. B) 2010 VOL 119 NO 4

defined as less than 10 to 20 in any sample (based on
Poisson statistics – Dominy et al., 2010c), then sampling
problems are likely and the primary sample size needs to
be larger. Gold particle size and frequency within a
deposit and its domains may vary.
Optimising sampling protocols
Overview
Sampling protocols inclusive of sample type, collection,
preparation and assay methodologies should be
designed to suit the style of mineralisation in question,
designing the protocol to fit the mineralisation style is
critical though is frequently ignored. As previously
stated, the frequency and size of gold particles has a
direct impact on the effectiveness of sampling, sample
preparation and assay methods employed (Dominy
et al., 2008b). Within the coarse- to fine-gold spectrum,
the ability to sample a deposit ranges from relatively
simple for fine-grained disseminated particles, to more
difficult for low-grade coarse gold-dominated ores. Ores
with clustered gold particles may yield a pseudo-coarse
gold effect with enhanced variability (Dominy and
Platten, 2007).
Application of the TOS
The TOS deals with many issues related to sampling, but
principally defines a number of errors and how to
manage them (Gy, 1979, 1992, 1998, 2004; Pitard, 1993).
The principal errors are FSE, GSE, DE, EE, PE and
AE. A quality sample protocol is one where all sampling
errors are minimised. FSE and GSE relate to sample
precision and the others are bias generators. The errors
and their implications for grade control sampling are
summarised in Table 5.
A well-known part of TOS consists of the prediction,
estimation and minimisation of the FSE. It provides a
prediction for the value of the variance of FSE, based on
the heterogeneity of the population from which the
sample is taken using the well-known Gy Equation
(Gy, 1979, 1992, 2004; Pitard, 1993). Recent work has
questioned the validity of the equation (Geelhoed, 2011),
but to date there is no practical alternative and most
practitioners continue to have faith in the Gy Equation.
It must be remembered that this equation is a theoretical
application to estimate the variance of FSE, so should
be used with care and in context.
The FSE is the smallest residual error that can be
achieved even after homogenisation of a sample lot.
Effectively it can never be removed. When FSE is not
optimised for each sampling and sub-sampling stage, it
often becomes a major component of the sampling
nugget variance (e.g. nugget effect). FSE may represent
up to 50% of the total sampling errors. The FSE is
dependent upon sample weight, fragment size and shape,
liberation stage of the gold, gold grade, and gold and
gangue density. Experience shows that nugget effect is
often artificially high because sample weights are not
optimised. In addition, a high FSE may result from
over-splitting before crushing and pulverising.
The FSE equation addresses key issues of ore
sampling both in the mine and plant, key issues are:
Table 4 Sample type for different stoping methods during the mine production chain
Stoping method Stoping type
Development and
pre-production stage
Production
(stoping) stage
Post-production (bogging
and transport) stage
Shrinkage
Cut and fill Linear
Room and pillar Entry Linear Grab
Drift and fill Grab Drill hole Grab
Longhole (open and sub-level variants) Non-entry Drill hole Grab
Block and Sub-level caving
Table 5 Sampling error deﬁnition and relevance to grade control sampling
Error Error definition Implication for grade control
FSE The FSE results from the constitution (grade)
heterogeneity of the whole. 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 (e.g. high FSE)
Relates to the sampling of broken rock.
For example, muck piles, trucks or stockpiles.
Grab sampling prone to high FSE. An in situ
FSE exists when a channel sample is cut from a face
GSE Is the error due to the combination of grouping and
segregation of the fragments (distributional effect) in
the lot to being sampled. Once a rock volume is broken,
there will be segregation of particles
Refers to segregation within a broken rock pile, such
as fines at the base of a coarse stockpile or muck pile
DE The DE results from an incorrect shape of the volume
delimiting a sample
A channel sample, for example needs to be properly
delimited to provide a uniform volume
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
The delineated volume of a channel sample,
for example should be fully recovered
PE Refers to issues during sample transport, preparation
(contamination and losses) and unintentional human error
Emphasises the need for rigorous sample bagging,
tagging and handling after collection (e.g. effective
chain of custody)
AE Relates to all errors during the assay and analytical
process
Not directly related to the grade control process, but can
have a profound effect on final assay grades. Includes
laboratory human error, poor analytical machine
calibration, machine drift
Applied Earth Science (Trans. Inst. Min. Metall. B) 2010 VOL 119 NO 4 207

N what weight of sample should be taken from a larger
mass of ore, so that the FSE will not exceed a specified
variance
N what is the possible FSE when a sample of a given
weight is obtained from a larger lot
N before a sample of given weight is drawn from a larger
lot, what is the degree of crushing or grinding required
to lower error to a specified FSE.
The FSE is a loss function during sampling (Pitard, 1993).
The higher the FSE, the greater the likely financial loss.
For any process where the FSE escalates, there is an
associated loss due to uncertainty.
A number of difficulties have been encountered in the
practical application of the FSE equation to gold deposits
(François-Bongarçon, 1993, 1998; François-Bongarçon
and Gy, 2002; Dominy, 2007; Minnitt and Assibey-
Bonsu, 2009; Dominy et al., 2010a). These include:
N ease with which inconsistency in equation parameter
units can yield erroneous results
N difficulty in understanding the nature and meaning of
the various parameters and how to adequately set/
model their values
N difficulty in designing and performing the calibration
and characterisation experiments required to custo-
mise the key parameters.
A key technical issue with the use of the FSE equation is
the numerical value of the power in the liberation factor.
Gy (1979) proposed the value of 0?5 as a general case only
for all minerals. This value gives reasonable results for
various types of ore and at different grades; however, it
often gives unrealistic values when applied to low-grade
ores such as gold. François-Bongarçon and Gy (2002)
quote an example where an optimum sample mass of
13?9 t of blast-hole cuttings was required for a fine-
grained gold ore, clearly an unrealistically large sample.
Appropriate application of the FSE equation lies in
estimating the theoretical FSE made at successive mass
reduction stages. They may not be correct in absolute
terms, but are to some extent comparable and selection
stages introducing large errors can be improved and
redundant stages (i.e. those with minor errors) can be
eliminated.
We assert that proper ore characterisation is critical to
the application of the FSE equation and of the wider TOS
to ensure reality and practicability come together (Dominy
and Platten, 2007; Dominy et al., 2010c; Dominy et al.,
2008a, 2008b; Dominy et al., 2011b). There are instances
where impractical protocols are required to achieve correct
sampling particularly in the presence of coarse gold
(Dominy et al., 2000; Dominy and Petersen, 2005;
Johansen and Dominy, 2005; Cintra et al., 2007).
Further discourse of the TOS and FSE equation is
given in Gy (1979, 1992, 1998), François-Bongarçon
(1993), Pitard (1993, 2005), Frempong (1999), Minnitt
(2007) and Dominy et al. (2010a).
Defining the liberation diameter
The part of the FSE equation that reflects the nature and
heterogeneity of the ore type in question is the sampling
constant (K). It is dependent on the microscopic
geostatistical properties of the minerals, and varies with
gold grade and liberation size. As liberation size reduces,
so the value of K reduces; conversely, as grade reduces,
K gets larger. Large K values are related to samples with
a greater liberation size and low grade.
A key problem of defining K and applying the FSE
equation is the determination of the liberation diameter
(dL: Gy, 1979; Pitard, 1993; François-Bongarçon and Gy,
2002; Dominy, 2007; Minnitt and Assibey-Bonsu, 2009;
Dominy et al., 2010a). Using the original definition of dL
(Gy, 1979), very low sampling constant values are
generally defined for gold ores. Thus for gold deposits,
the dL value should be re-defined as dLmax to represent the
coarse most influential fraction. This is effectively the
screen size that retains 5% of gold given a theoretical
amount of liberated gold.
Approaches to dLmax characterisation range from
guesswork to the implementation of HT or DSA. The
results of both HT and DSA can be used to calibrate the
FSE equation, effectively defining dL through estimating
K (Gy, 1979; François-Bongarçon, 1993; Pitard, 1993;
Minnitt et al., 2007; Minnitt and Assibey-Bonsu, 2009).
The HT is most commonly applied calibration
method in the mining industry. The DSA approach is
both complex and time consuming to apply, so relatively
Table 6 Typical studies and outcomes from a ‘sampling’ characterisation programme (after Dominy et al., 2010c).
Outcomes from Stages 1 to 3 contribute to metallurgical characterisation. Multiple samples will provide a clearer
understanding of all outcomes and potential grade and spatial variability
Stage Study Outcomes
1 Geology and mineralogy [in situ and sample
studies – rock and core samples]
N Geological context of sample(s)
N Gold deportment and mineralogy (free gold
versus gold in sulphides)
N In situ gold particle size range and distribution
N Proportion of coarse versus fine gold
N Clustering effects
2 Metallurgical testing N Sample head grade
[bulk samples of .50 kg each] N GRG
[note does not include comminution studies] N Leach (gold in tailings/potential refractory gold)
N Partitioning between free gold and refractory gold
N Mineralogy of concentrate and tails
3 Coarse gold determination N Coarse gold particle size and abundance
[same sample as Stage 2 and based on gold
liberated during that stage]
4 Data integration N Gold particle size curve(s)
N Liberation diameters and variability
N Degree and impact of clustering

rare compared to the HT. The HT is prone to severe
precision problems, particularly when coarse gold is present
(Dominy, 2007; Dominy et al., 2010a; Dominy and
Minnitt, 2012). A further problem is that it provides values
for K only at the fragment size at which the calibration
exercise is carried out (often 10 mm). In such cases the
values for dL are far too low. In the presence of coarse gold,
the HT approach may require samples of hundreds of
kilograms in size. The method may not be a standard
industry approach, since it is likely only to provide a correct
value when mineralisation is disseminated.
As a result of these issues, the authors recommend a
staged approach to characterisation (Dominy et al.,
2008a, 2008b, 2010c; Table 6). The key outputs will be:
N realisation of gold deportment, in particular the
partitioning of gold as free gold, gold in sulphides and
refractory gold
N gold particle size curve(s), including effects of
clustering and relationship between gold particle size
and grade
N definition of key FSE relationship inputs (e.g. libera-
tion diameter versus clustered-liberation diameter)
and the sampling constant K
N recommendations as to optimum in situ sample mass
requirements.
As a result of undertaking a characterisation pro-
gramme, optimisation of sample collection and prepara-
tion protocols within the context of TOS will be
possible. Characterisation studies will allow the practi-
tioner to set the sampling expectations of the ore. Where
this requires specialist and potentially costly protocols,
then it will be possible to determine the level of risk
involved in using more practical methods. In essence a
gap analysis between the theoretical need and practical
reality of sampling can be undertaken. Finally, char-
acterisation will allow benchmarking with other deposits
and contributes to wider orebody knowledge (e.g.
geometallurgical parameters).
Cost of developing and operating a grade
control sampling programme
The investment required to design a sampling programme
is variable and dependent upon the orebody type. A full
characterisation study is likely to cost between AU$50k
and AU$250k per domain, though much of the data can
be gained and/or shared with metallurgical programmes.
Programmes are required during project evaluation or
during production when new orebodies and/or ore zones
are encountered.
On-going grade control sampling costs will relate to
orebody type and mining method. For a small operation
(y100 kt per annum) an annual cost could be in the
region of AU$350k, whereas for a much larger operation
(y1 Mt per annum) more likely around AU$2M per
annum. These figures include the cost of sample collec-
tion, sample preparation and assaying, staffing and
consumables, but do not include diamond drilling cost.
The cost of good grade control sampling is easily
justified against the potential loss generated through extra
dilution sent to the mill and/or ore/waste misclassification.
Monitoring a grade control sampling
programme
The ultimate monitor of grade control programme
efficiency is reconciliation. Some practitioners believe
that QAQC need not be applied to grade control
programmes – this view is flawed, not least that grade
control sample data often contribute to publicly
reported resource and reserve estimates. The general
issues of QAQC programmes are presented in Sketchley
(1998), Vallée (1998), Simon and Gosson (2008) and
Graindorge (2010).
Measuring and monitoring sampling quality is best
achieved by regular duplicate sampling at each sub-
sampling stage in the sampling regime. For more
complex ores in the presence of coarse gold, duplicate
sampling may be carried out as a batch each month or
on a day to day basis at a sufficient rate to provide a
suitable number of paired data to make a regular and
statistically significant estimate of precision. The mon-
itoring programme should test precision at each stage of
the protocol, but in a low variability deposit measuring
the total precision by duplicate field samples may be
appropriate. A case study of the duplicate approach is
given in Carswell et al. (2009).
Appropriate data analysis techniques should be used
and could include HARD, mean absolute pair difference
or Thompson–Howarth plots (Thompson and Howarth,
1976). A discussion of different precision metrics is given
in Stanley and Lawrie (2007) and Abzalov (2008).
Poor sampling precision can be reduced to an
acceptable level by maximising sample and sub-sample
size and by minimising the number of sub-sampling
steps. An acceptable precision will be dependent upon
deposit type and sample protocol.
Consideration of all these factors then leads to a
summary of QAQC needs and expectations (Table 7).
Implementation and operation of a grade
control programme
The most important issue during the implementation
and operation of a grade control programme will be
people. Acceptance of the programme value by manage-
ment will be critical and requires the business case and
deliverables to be promoted early on. A programme
manager or ‘champion’ is required to drive development
and ensure that all parties are engaged. The overall team
must include miners, managers, geologists, engineers
and metallurgists.
Programme implementation will be via proper equip-
ping and training of both geological and mining staff
supported by documented procedures. Safety issues
must pervade training. Suitable database and modelling
systems will also need to be in place.
Experienced staff will need to mentor junior and new
staff based on documented procedures including:
N sample collection, transport and security
N geological mapping and data recording
N laboratory submission, preparation, assaying and
reporting
N mine-based QA and QC
N data handling, checking and validation
N timely use of data to control the mining operation
N feedback and continuous improvement.
Documentation is an important part of the development
and implementation process and should include: proto-
col descriptions and roadmaps, QAQC procedures,
face, side wall and back mapping sheets, underground
development instruction sheets, sample tags and sample
submission sheets.

An overall programme roadmap should be produced
to ensure that all practical aspects are undertaken
effectively, data integrity maintained and that all
information is used and its implications communicated
to stakeholders.
Grade control innovation
It might be argued that other than through improved
computing capability (e.g. databases, 3D modelling,
grade estimation and reconciliation software), there is
minimal opportunity for innovation in grade control. To
some extent this may be true, however whilst sampling
methods are unlikely to be improved upon other than
via some kind of mechanisation, opportunities do exist.
There is scope to be smarter through the better use of
geology during the mining process. For example, at the
Bendigo Kangaroo Flat mine (Australia) face samples
were not taken with preference for a scorecard-based
grade calling (proxy) system. This utilised various
geological and mineralogical parameters in a fast and
simple to use form (Dominy et al., 2009). This approach
was chosen as face chip sampling was biased due to the
very coarse gold present (Johansen and Dominy, 2005).
Other operations have also developed methods to
support sampling based on the better use of geology
(Dominy et al., 2010b). In some operations, the efficient
use of geology during grade control can be considered
innovative.
Face or backs mapping can be made more efficient
through the use of field tablet computers where mapping
is directly onto the tablet. This results in an immediate
digital format that can be easily ported to other software
for editing and visualisation. Direct face mapping is
possible via the SirovisionH stereo camera rig which uses
two digital cameras to capture an image (Van Der
Merwe, 2009). The geologist can draw lines on the image
to represent geological features, which can then be
ported to suitable mining software fully registered in 3D
Table 7 QAQC programme and expectations for an underground gold mine grade control programme. Note that this
programme is generally applicable to all stages of gold project development
QAQC action Rate Instigator
Expectation
Fine-gold dominated deposit Coarse gold-dominated deposit
Field duplicates* 1 in 20 Mine 90%¡10–25% HARD 90%¡25–50% HARD
Coarse reject
duplicates{
1 in 20 Laboratory 90%¡10–20% HARD 90%¡15–50% HARD
Pulp duplicates{ 1 in 20 Laboratory 90%¡10% HARD 90%¡10–25% HARD
Standards1 1 in 20 Mine 2d–3d (‘warning’) re-assay
25% of batch
.3d (‘action’) re-assay
100% of batch
Blanks" 1 in 20 Mine Less than three times
detection limit (generally
,0.15 g/t Au)
Pulp quality** 1 in 20 Laboratory Sample to P90 275 mm
Contamination{{ 1 in 20–50 Laboratory 0% gold loss ,1% gold loss
Umpire{{ 1 in 20 Mine 90%¡10% HARD
Laboratory audit Annually Mine Ensure that the laboratory
is working to agreed practices
and performance levels.
*Applies to any sample type collected. For coarse gold/nuggety ores, very poor precisions are likely unless large samples are
collected and appropriate sample preparation and assay routes taken. For diamond drill core, the duplicate is the other half of the
core. In many cases, a quarter core will be taken as the ‘duplicate’. Operators might consider cutting core into thirds so that
‘duplicates’ match. Some mine operators take the entire core as the sample, particularly where coarse gold is present.
{Laboratory crusher reject split.
{Dependent upon nature of ore and assay method. For a coarse gold assayed via SFA, a good precision would be expected for
undersize fraction. For ACE-LeachWELL, pulp duplicate may not be possible if all sample used. For PAL technique no pulp duplicate
available, only duplicate available from coarse rejects.
1Recommendation to have around four to five standards at grades ranging from low-grade, through to cut-off, ROM and high-grades.
Note that by their very need to be homogeneous, standards do not bear coarse gold but they can be matrix matched by being quartz-
dominated and sulphide-bearing. Note that the laboratory will also insert standards. Standards used for SFA process will just be FA.
Where ACE used, mass of standard used usually less than that of actual sample, due to cost unless a bulk standard has been properly
defined for that mine. A 500g standard for ACE is generally fit for purpose.
"Blanks provide a measure of contamination. They can be specifically inserted after expected high grade samples. Laboratory will
also place blanks into the sample stream. In very high-grade coarse-gold dominated samples, contamination may be high and require
specialised protocols.
**Test involves screening of the pulp to ensure 90% passing or better. All samples should pass or the entire batch should be reground.
Maybe highly problematic in the presence of coarse gold, where loss and/or contamination are a risk.
{{Assaying of the pulveriser cleaning agent (e.g. sand), and could also include crusher cleaner (e.g. barren rock). For fine gold ores, a
rate of 1 in 50 is appropriate increasing to 1 in 20 for coarse gold ores. Careful management of coarse gold ores is required. It is
suggested that laboratories include a ‘wash’ after visibly high-grade (e.g. visible gold-bearing) samples. However, if the ore bears
notable coarse gold, then cleaning is best after each sample given that even low grade samples can bear coarse gold particles.
{{Monthly submission of samples (typically pulps), including standards and duplicates is sufficient to provide a check of primary
laboratory results. This is especially important where an on-site laboratory is being used as it provides independent confirmation of the
results. Where SFA, ACE or PAL used, there may be no pulp residues to submit. In this case, coarse rejects can be used. Umpire
samples (e.g. pulps or coarse duplicates) should be returned to the mine and submitted by mine staff to the umpire laboratory. In some
cases, the laboratory (mine or off-site) may submit umpire samples as part of their internal QAQC.

mine space. Other innovations could include portable X-
ray fluorescence spectroscopy units capable of in situ
gold assay of mine faces and the application of down-
hole radar to resolve mineralisation in wall rocks (Kemp
and Morgan, 2012).
Operational safety issues
Working underground is an intrinsically dangerous
activity. Grade control is no less onerous given that
geologists will be in active headings, raises and stopes
and around operating machinery. Mines will have local
approved codes of practice which will include guidelines
for where and how to undertake sampling. Geologists
must ensure that they carry all required personal
protective equipment, proper sampling equipment and
they should not work alone. In addition, there are issues
of manual handling of samples, working at height when
sampling backs and raises, and working close to heavy
machinery. Workplace hazard and risk assessments of
specific activities are recommended.
One of the greatest risks for the geologist is working
beneath unsupported ground. In many operations,
access to the face is not permitted and leads to the
application of grab sampling during development. This
tends to occur in the larger mines. Awareness of the risk
and proper training in barring-down is critical. High risk
also faces the geologist when sampling within raises.
Again awareness and proper training is required.
Liphichi gold–antimony–silver project,
Bolivia
Overview
The Liphichi property is located in the Andes of
Western Bolivia, 110 km to the northwest of La Paz.
It comprises four concessions within the Larecaja Province,
covering some 53 km2
and includes the Liphichi, Pro-
gresiva and Maria Luisa deposits. Liphichi is located in a
NW–SE trending, steeply-dipping deformation zone that is
traceable for some 14 km along strike (the Liphichi-Luisa
Belt). This zone locally hosts a 50 to 70 m thick zone of
quartz veining. The country rocks comprise argillites and
black shales crosscut by granitic intrusives. Mineralisation
is characterised by quartz-sulphide veins forming individual
narrow structures and multiple sets forming stockworks
and sheeted vein systems.
From 2004 to 2007, the area was explored by the former
Canadian listed entity, Luzon Minerals Limited who
undertook surface mapping, diamond drilling, under-
ground development and sampling. From 2004 to 2006,
studies were undertaken to design a resource development
programme which would include grade control sampling
for trial mining (Snowden, 2004, 2005a, 2005b).
From 2006 to 2007, Luzon operated an alliance with
Australian listed company, Republic Gold Limited. This
resulted in the public release of a Mineral Resource
estimate (Republic, 2006). At a 1?5 g/t Au cut-off, the
resource contained 325 000 oz gold, 47 000 t antimony
and 3 280 000 oz silver classified as an Inferred Mineral
Resource in accordance with the 2004 JORC Code
(Table 8).
This alliance terminated in mid-2007. At the current
time (October 2011), the project is believed to be in
private ownership and worked by artisanal miners.
Geology and mineralisation
Regional and local geology
North-western Bolivia lies at the convergence of the
South American and Nazca Plates, which were respon-
sible for the formation of the Andes. The Central Andes
consists of two sub-parallel mountain belts, the Palaeo-
zoic Eastern Cordillera and Mesozoic-Cainozoic
Western Cordillera. These belts are separated by the
Altiplano, a Cretaceous–Cainozoic intermountain basin
up to 200 km wide. Regional scale lineaments are
prominent, and are related to thrust faults, normal
faults and shear zones. The NW-trending, gently NE-
dipping thrust structures verge to the southwest, and
expose the folded Palaeozoic sequence. The attitude of
the sedimentary rocks is strongly influenced by the
thrusts and associated folding (Fig. 1).
Liphichi is located in the Eastern Cordillera, which is
made up of over 7 km of shale and siltstone. Regionally
distributed lithologies include Ordovician and younger
quartzite, sandstone, siltstone and shale with overlying
Permian carbonate. The Palaeozoic rocks have under-
gone intense folding, faulting and metamorphism related
to the emplacement of Permo-Triassic granite.
The mapped units around Liphichi are the Liphichi
Turbidite, North Liphichi Quartzite, North Liphichi
Turbidite and the Pacuni Turbidite. The most important
unit is the Liphichi Turbidite, which is characterised by
Bouma sequences. These are a dark grey to blackish in
colour and weather to a brown colour. The Bouma
sequence consists of interbedded shale and fine-grained
quartzite.
The sedimentary sequence is cut by fine- to medium-
grained porphyritic intrusive rocks that form sills and
dykes generally following the NW-regional tectonic
control.
1 Folded stratigraphy and thrusts at the Liphichi site.
The location of the two main mine levels are shown,
and the inset shows sheeted veins on the 3 Level
Table 8 Mineral Resource for the Liphichi gold project
reported in accordance with the JORC code at a
cut-off of 1?5 g/t Au (Republic, 2006). The entire
resource was classiﬁed in the Inferred Mineral
Resource category
Commodity Tonnage (Mt) Grade
Gold 4.2 2.4 g/t Au
Antimony 1.2% Sb
Silver 17 g/t Ag

Mineralisation
Mineralisation at Liphichi is of the turbidite-hosted quartz-
carbonate vein type. It is termed by local geologists as a
‘Bolivian vein deposit’, a type that is apparently peculiar to
the Bolivian Altiplano. It contains gold and antimony, but
also significant values of silver, copper, lead, zinc and
tungsten.
Four mineralisation styles are present:
N stratabound and stratiform mineralisation – this style
is present throughout the sedimentary sequence and
characterised by parallel to sub-parallel thin lamina
of pyrite with rare arsenopyrite
N saddle reefs – these quartz structures are observed
both on surface and underground. The saddles are
hosted within axes of anticlinal structures
N stockwork vein systems – these comprise quartz veins
in various orientations, ranging from a few milli-
metres to 30 cm in thickness (Fig. 2). The stockworks
are hosted within axes of anticlinal structures
N sheeted vein systems – these comprise parallel to sub-
parallel zones of individual quartz veins ranging in
thickness from a few millimetres to over 30 cm
(Fig. 2). The sheeted veins are hosted within axes of
anticlinal structures.
All vein types are auriferous, though the majority of the
inventory is hosted within the stockwork and sheeted vein
types. Saddle reefs are relatively unimportant. There is an
intimate relationship between stockwork and sheeted
veins, in most cases they form a combined chaotic vein
swarm.
Diamond drilling and underground sampling during
2004 to 2006 confirmed a broad zone of gold miner-
alisation ranging from 53 m at 5?3 g/t Au (surface
channel samples) to 62 m at 1?8 g/t Au (1 Level channel
samples). Core and channel samples also record short
higher-grade zones within the wider zone, for example
drill hole PG04 grading 125 g/t Au over 0?95 m and
channel sample 917 grading 102 g/t Au over 1 m.
The Liphichi veins contain variable amounts of pyrite,
arsenopyrite, pyrrhotite, chalcopyrite, sphalerite, stibnite,
wolframite, scheelite and carbonate. These minerals give
rise to high levels of silver, lead, arsenic and antimony
recorded in geochemical analyses. Visible gold is recorded
in oxidised near-surface material, drill core and under-
ground exposures. Geological mapping around Liphichi
indicated that mineralisation is related to a compressional
setting and is best developed in the hinges of anticlines.
The gross mineralised zone is 200 m thick by 2 km along
strike. Mineralisation is multi-phase as evidenced by
cross-cutting veins, differing vein mineralogy and tex-
tures, and overprinting of three alteration events.
Liphichi shows similarities to the Central Victorian
goldfields of Australia, Dolgellau gold-belt of Wales, UK
and Meguma Province of Nova Scotia, Canada. The
stockwork/sheeted veins are broadly analogous to the
‘spur’ zones of the Bendigo Goldfield, Central Victoria,
Australia.
Ore characterisation
Introduction
A characterisation study based on the approach outlined
(Table 6) was undertaken (Dominy et al., 2010c). Given
the presence of coarse gold and its potential impact and
dominance in terms of inventory value, characterisation
and optimisation focused on gold rather than silver or
antimony.
Three 120 kg panel samples were collected from three
locations on 2 Level (Figs. 3 and 4). These represented
one 1?5 m composite narrow vein and two 2 m sections of
stockwork/sheeted veining. The samples were subjected
to a test programme, including gold particle liberation
and analysis, HT study and mineralogical determination.
HT
An HT study was undertaken on forty 0?2 kg fragments
from sample #LMET02. The basis of the HT is given in
2 Stockwork (A) and sheeted vein style (B) mineralisation at Liphichi underground workings 2 Level
3 Panel sample preparation at the Liphichi gold project.
One-hundred kilogram samples taken for metallurgical
testing and grade veriﬁcation

Pitard (1993), Minnitt and Assibey-Bonsu (2009) and
Dominy et al. (2010a). This sample was chosen as the
original face sample grades indicated close to the
expected run-of-mine ROM grade. All samples were
pulverised to P90 275 mm and screen fire assayed
(100 mm screens) in their entirety. Triplicate fire assays
were taken on the undersize fraction. Coarse gold
(.100 mm) was observed in two high grade samples
(25?1 and 151?5 g/t Au), at 16 and 42% respectively. In
one lower grade sample (2?39 g/t Au), 11% of the gold
was coarse gold. All other values were below 10%
(mostly below 5%). Fragment grades ranged from 0?01
to 151?5 g/t Au with a mean of 5?7 g/t Au. A sampling
constant (K) of 145 g/cm1?5
and gold liberation diameter
(dL) of 50 mm was calculated for an equivalent nominal
fragment size (dN) of 8?3 cm.
The results of the HT where 40 sample grades are
ranked from low to high grade (Fig. 5). The plot profile
is typical of higher heterogeneity ores where there are a
large number of relatively low grades compared to few
high grades. Some 77% of contained gold is distributed
in just two samples. In a low heterogeneity ore, gold
distribution would be relatively even between samples. It
should be noted that the HT is usually undertaken on a
minimum of fifty fragments. A lesser number were used
in this case due to limited availability.
Specialised test protocol
The test programme was designed to investigate coarse
gold ores, principally from a perspective of gold particle
size determination (Dominy et al., 2010c). The protocol
is based on three stages, comprising preliminary
geological and mineralogical characterisation, followed
by two processes addressing metallurgy and gold particle
size determination after liberation.
Based on a limited number of samples, the Liphichi
ore contains some coarse gold (Table 9). After libera-
tion, between 12 and 37% of gold recovered was coarse
gold (.100 mm), though mainly below 500 mm and
rarely to 1000 mm. GRG values varied between 31 and
57%, indicating the potential application of gravity
concentration (Dominy et al., 2011a).
Mineralogical study
Fifty polished thin sections were studied for each of the
three samples (#LMET01 to 03). Gold particle mapping
was undertaken using an automated electron microp-
robe supported by reflected light microscopy. The
4 Collecting panel samples with hammer and moil at the
Liphichi gold project (see Fig. 3). Samples were chipped
from the drive side-walls onto a plastic ﬂoor sheet
5 Heterogeneity test results for 40 piece test on Liphichi
ore
6 Particles of visible gold within drill core from Liphichi.
Grey streaks are mainly stibnite
Table 9 Summary results from the specialist metallurgical test programme
Sample
Face grade
(g/t Au)
Laboratory split
grade (g/t Au)
Test head
grade (g/t Au)
GRG value
(%)
Per cent of
gold .100 mm
Au max
size (mm)
#LMET01 23.1 8.2 14.8 57 37 1000
#LMET02 1.8 1.2 2.7 45 22 295
#LMET03 0.8 2.8 1.4 31 12 225

shortcomings of 2D-based thin section studies are well-
known, but provided useful information. Gold particles
were exclusively hosted in quartz and often closely
associated with acicular stibnite or more rarely pyrite.
Visible gold was observed in both drill core and
underground exposure, but not commonly (Fig. 6).
The largest particle seen was 1?1 mm in size. A summary
of gold particle size distribution from the automated
mineralogical study is shown in Table 10. In agreement
with the metallurgical tests, the mineralogy indicates
gold mostly below 1000 mm and around 25 to 50%
coarse gold. Gold above 100 mm is generally up to
250 mm. The higher grade LMET01 sample bears more
coarse gold (50%) and gold up to 1?1 mm in size.
Both the visual (e.g. core and hand specimen, and
underground observations) and automated studies (thin
sections) identified gold particle clusters. These range
from groups of fifty 200 to 500 mm particles to one
hundred 200 to 500 mm up to 8 mm across. Composite
particle sizes range from around 1 to 3 mm in size.
Review of SFA data
Existing SFA data indicated that most samples con-
tained less than 10% of z105 mm gold – nominally less
than 10% coarse gold. This result differs from those
obtained to the metallurgical and mineralogical studies,
but such a divergence is not uncommon. Given that the
other studies have indicated coarse gold mostly between
100 and 250 mm, the pulverising process is working
relatively well. In addition, any gold clusters are
fragmented during pulverisation.
Review of duplicate chip-channel and diamond core data
During 2004, Luzon undertook a small re-sampling
programme underground and of diamond drill core. Out
of eight chip-channel samples, only one was within
¡20% of the original value. Variation included 0?5 g/t
Au versus 5?4 g/t Au and 102 g/t Au versus 45 g/t Au for
example. Three drill holes were re-sampled via quarter
core yielding 30 m at 2?3 g/t Au versus 9?8 g/t Au (hole
PG07), 44 m at 5?5 g/t Au versus 3?9 g/t Au (hole PG06)
and 74 m at 4?3 g/t Au versus 2?2 g/t Au (hole PG04).
These differences between field duplicates indicate a
level of natural variability that may be explained by the
presence of coarse gold. Some of the variability was
related to the comparison being between original half-
core versus duplicate quarter-core samples.
Gold particle size models
The studies enabled a generalised gold particle size
model for Liphichi run-of-mine ore (assuming a grade of
2?5 g/t Au) to be constructed (Fig. 7). Defining dLmax as
the 95th percentile (as noted previously, see Dominy
et al., 2010a), a mean value of 325 mm is given for
Liphichi. A minimum of 180 mm was defined and a
maxima of 1400 mm accounting for large particles and
possible clustering (dLclus). The minimum value repre-
sents the likely distribution at the cut-off grade at 1?5 g/t
Au and the maximum value for higher grade ore (.10
g/t Au). The mean K value for Liphichi was estimated to
be 5700 g/cm1?5
, in a minimum–maximum range of
3000 g/cm1?5
to 10 000 g/cm1?5
. K values below 200
g/cm1?5
likely pose no sampling challenge and above
5000 g/cm1?5
can lead to extreme issues. Liphichi ore sits
just above the boundary of the major to extreme class
and confirms that care is required.
The HT resulted in a dL of 50 mm being defined. This
understates the dLmax by over seven times and is a typical
result using the heterogeneity approach in a coarse gold
ore (Dominy and Minnitt, 2012). It is likely however to
represent gold particles associated with a background
grade. Preliminary evaluation of grade statistics indi-
cates a background gold grade in the range 0?3 g/t Au to
0?5 g/t Au. This background is likely to represent more
disseminated gold particles and yields a K value of
1400 g/cm1?5
.
Based on experience, Liphichi ore is classified as a
‘medium’ type coarse gold deposit (Table 11; Dominy,
1997). Like many orogenic vein systems it bears
7 Generalised gold particle size model for Liphichi ore based on metallurgical and mineralogical studies
Table 10 Summary of gold particle size data from the mineralogical study
Sample number
Per cent gold by size fractions
Per cent gold .100 mm
,50 mm 50–100 mm 100–250 mm 250–500 mm 500–1000 mm
#LMET01 32 18 33 17 ,1 51
#LMET02 43 27 19 11 0 30
#LMET03 34 41 25 ,1 0 26

moderate quantities (25 to 50%) of 100 to 300 mm gold.
In addition, there is likely to be some pseudo-coarse gold
effect where finer gold particles are in clusters. This is in
contrast to the Bendigo system which classifies as an
‘extreme’ coarse gold ore. At Bendigo, diamond drilling,
120 t development bulk sampling and 100 kg mini-bulk
samples were used during feasibility (Johansen and
Dominy, 2005) to assess grade. A proxy-based system
was developed for grade control (Dominy et al., 2009;
Dominy et al., 2010b).
Sample mass requirements
Based on the identification of coarse gold in Liphichi
ores, the Poisson approach was used to estimate likely
sample mass requirements (Dominy et al., 2010c).
Sample mass expectations based on various grade-gold
liberation diameter scenarios are in Table 12. The ROM
grade mass to achieve ¡20% precision is 16 kg. This is
considered a maximum mass value.
Existing underground sampling programme
Overview
During 2005 to 2006, the sampling strategy at Liphichi
involved collection of face (linear chip-channel), muck
‘grab’ pile (blasted development rounds) and NQ-sized
core samples. Every development face was chip sampled
and muck-pile sampled. Core samples were taken as 1 m
composites in mineralised zones. A two to three man
team using lump hammers and chisels cut an even
channel for face samples. Chippings from each channel
fall onto a tarpaulin and are bulked together to form a
total sample. The total sample of around 30 kg was then
reduced in size by so-called ‘fractional shovelling’ to
produce a final sample of 15 kg for laboratory submis-
sion. Muck pile samples were taken as a series of cuts
across the pile as the broken rock is bogged by hand.
Approximately 15 kg was submitted to the laboratory.
NQ drill core was logged and mineralised sections cut in
half on approximately 1 m lengths and submitted to the
laboratory (2 to 3 kg).
All samples were submitted to a commercial labora-
tory in Bolivia. On arrival, samples were dried, weighed
and crushed in one pass to P70 1?7 mm using a Rhino
crusher. After crushing, all samples were reduced to a
mass of 1 kg by riffle splitting and then pulverised to P90
150 mm prior to screening and fire assay. The screen
undersize was triplicate fire assayed. All items of
equipment were compressed air-blast cleaned between
samples and the pulveriser bowl cleaned using silica
sand. All sub-samples were shipped to North America
for fire assay.
Table 12 Indicative in situ sample size requirement for Liphichi ore based on different scenarios. Estimated using the
Poisson method – mass calculated to precision of ¡20% at 95% conﬁdence limits
Grade (g/t Au)
Liberation diameter
(dLmax or dLclus) (mm)
Indicated sample
mass (kg) Comment
2 1000 435 ROM ore assuming clustered particles
with composite size to 1 mm
2 325 16 ROM grade ore
0.3–0.5 50 1 Background grade material with fine-gold liberation diameter
Table 11 Key results from the characterisation study of Liphichi ore and comparison to Bendigo ore
Characteristic Liphichi Bendigo
Gold particle size dLmax: 325 mm dLmax: 10 000 mm
.100 mm: 40%; .1000 mm: ,1%;
.10 000 mm: 0%
.100 mm: 90%; .1000 mm: 60%;
.10 000 mm: 5%
Distribution As isolated particles and clusters.
More coarse gold and clustering in
higher grade ore
As isolated particles and clusters.
More coarse gold and clustering
in higher grade ore
Clustering observed, up to dLclus:
1400 mm
Clustering observed, up to dLclus:
10 000 mm
Deportment Free gold in quartz and locally
associated with stibnite
Free gold in quartz and locally
associated with arsenopyrite
and/or other sulphides
Grade ROM grade of around 2–2.5 g/t Au ROM grade of around 6–8 g/t Au
Sampling constant (K) 5700 g/cm1?5
[dLclus: 40 000 g/cm1?5
] 200 000 g/cm1?5
[dLmax and dLclus]
GRG .30% .65%
Indicative in situ
sample size
ROM: 16 kg [Clustered: 1.2 t] ROM: 22 t
Background grade Around 0.3 g/t Au and dominated
by ,100 mm particles
Around 1.2 g/t Au and dominated by
,100 mm particles
HT results 40 piece test 50 piece test
ESD58.3 cm; K5145 g/cm1?5
;
dL550 mm
ESD519.4 cm; K5350 g/cm1?5
;
dL5500 mm
Visible gold Observed up to 1 mm in size as
individual particles and clusters up
to 2 mm in (composite) size
Observed up to 10 mm in size as
individual particles and clusters up
to 10 mm in (composite) size
Comment Moderate-nugget (around 50–60%)
moderate-coarse gold system
Extreme-nugget (.90%) extreme-coarse
gold system
Sampling difficulty Low-moderate High-extreme
Source This study Johansen and Dominy (2005)

Review of protocols
Based on the defined dLmax of 325 mm, the total FSE
variance was estimated for each existing protocol
(Table 13) – FSE values where higher than acceptable.
In particular, the channel sample protocol had an
extremely high FSE due to the sample being split in
half prior to crushing. In general, the precision required
for a sample protocol should be less than ¡20%; above
this value, estimates become non-Gaussian and Poisson
distributed (Pitard, 1993, 2005).
Various issues needed addressing at site:
N the ‘fractional shovelling’ process applied was more
of a series of grab samples from the rock pile method
that was non-probabilistic and prone to bias (DE, EE
and GSE errors). An ‘alternate shovelling’ method
was deemed more appropriate based on twenty 1?5 kg
shovel loads alternately placed onto two heaps
and at the laboratory:
N during riffling (after crushing), the sample container
was moved from side to side on the splitter (thus the
cutters did not receive an even flow of material) – this
led to an EE error as the split was not a perfect 50/50
N a fixed device was used for screening which had the
potential for contamination (and/or loss) as a result
of gold particles lodging in the mesh (PE error)
N after screening, the coarse and fine fractions were
bagged and shipped to a North American laboratory
for assaying. For the coarse fraction in particular,
there was a strong risk of gold loss due to particles
lodging in the folds of the paper envelope (EE).
During transport, strong GSE was likely and this may
not have been removed depending upon how the sub-
sample was treated at the laboratory.
Potential mining operation
The 2005 plan was to undertake an advanced resource
development programme that would be supported by
underground development, diamond drilling and trial
stoping.
Small-scale narrow vein mining (y30 000 t to
60 000 t per year) could commence relatively quickly
at Liphichi and likely achieve a head grade of 5 g/t Au
or above. The deposit has greater potential as a bulk
mining operation (.400 000 t per year) at a grade of
between 1?5 and 2?5 g/t Au, with a cut-off grade of
between 1 and 1?5 g/t Au.
For a bulk mine operation, the tonnage yield per linear
metre of development and level interval (30 m) is y4800 t.
For a single narrow vein target, the same value is y120 t
per m. Assuming a gross strike extent of 600 m, the bulk
tonnage per vertical metre is y95 000 t. For a narrow vein
target of 100 m strike extent, the tonnage per vertical metre
is 400 t assuming a stoping width of 1?5 m. If the entire zone
was to be stoped, then a bulk-mine sub-level open stoping
method was deemed appropriate. If a selective operation
was appropriate, then shrinkage stoping or possibly cut and
fill stoping would be used. Both methods differ principally
through being non-entry and entry respectively. For bulk
stoping, the entire zone needs to be defined through both
diamond drilling, foot-wall and hanging-wall development
and regularly spaced linking cross-cuts. To some extent, a
similar approach would be required if narrow vein targets
were sought; however, once identified they would require
development.
Proposed grade control programme
Overview
For all options, development faces would be mapped
and chip-channel sampled. Cross-cuts would be 2 m
chip-channel sampled over their length. Development
drives (3 by 3 m) would be placed in a foot- and
hanging-wall position based on geological control, but
effectively where the stockwork veining began to reduce
in density. At approximately 40 m intervals, cross-cuts
would join the parallel drives. Where appropriate, an
additional drive could be placed on a narrow vein within
the ore zone. Vertical separation between main levels
was initially planned at 30 m with 15 m sub-levels.
Diamond drilling from underground and surface
locations was planned. Cross-cuts 15 to 25 m into the
hanging- and foot-wall rocks aimed to define miner-
alisation around 75 m up- and down-dip from the drill
bay. The intent was to stagger the drilling between levels
to give an effective 20 m drill spacing enough to
potentially define continuity for an Indicated Mineral
Resource.
Geological mapping
Geological mapping of all drive faces and cross-cut walls
via pro-forma face sheets and mapping sheets was
planned. All faces were to be photographed. Horizontal
plan maps would be produced from the face/wall
maps and from specific mapping where appropriate.
Geological maps would be digitised and photographs
registered into 3D modelling software.
Sampling
Faces were chip-channel sampled using a two-man crew
and hammer and moil. Broken material was collected by
a tarpaulin on the floor. Cuts of 15 cm width and 2?5 cm
depth were targeted, yielding y10 kg per m. Face
duplicate samples were initially planned at a rate of 1 in
10. It was recommended that a diamond saw was trialled.
Cross-cut walls were chip-channel sampled in 2 m
lengths with the aim of collecting a 10 kg lot.
NQ diamond core was logged and sampled close to 1 m
lengths, though matched to geology where appropriate.
In addition to the standard grade control sampling
methods, a bulk sampling programme was also planned
during resource development. The intent was to extract
a series of cross-cuts from foot- to hanging-wall which
would each total in the region of 1000 to 1200 t.
Table 13 FSE precision and absolute error for sampling
protocols (ﬁeld to assay) at Liphichi
Sample type Precision (70% CL) Grade error (g/t Au)
Channel samples ¡155% 2¡3.1
Grab/muck samples ¡60% 2¡1.2
Drill samples ¡50% 2¡1.0
Table 14 FSE precision for each stage of the revised
Liphichi channel sample protocol
Stage dN (cm) Mass reduction Precision (70% CL)
1 0.17 10 to 5 kg ¡19.5%
2 0.0075 5 to 2.5 kg ¡1.9%
Total – 10 to 2.5 kg ¡20%

Laboratory protocols
Given the errors encountered in the original protocols,
they were revised to minimise FSE.
For the face channel samples, the entire primary
sample was crushed to P70 21?7 mm, 2?5 kg riffle split
off, pulverised to P90 275 mm and screen fire assayed.
Table 14 shows the calculated FSE for each stage.
For the drill core samples, the entire sample was
crushed, pulverised and screen fire assayed. This effectively
yields a zero per cent FSE as the sample is not split. This
was deemed a sensible approach given that the data would
represent a new drilling programme and be used to support
a maiden resource estimate.
The bulk sampling programme required an on-site
sampling tower to reduce a 100 t lot to a final sub-
sample of 50 kg P80 25 mm (Table 15). The sub-sample
would be submitted to a laboratory for metallurgical
processing.
Quality Assurance/Quality Control
Standard QAQC would include field duplicates, blanks
and standards at a rate of about 1 in 20. Bulk sample
cross-cuts were to be drilled with horizontal holes drive
to drive, extracted and side-wall chip sampled to allow
the different sample supports to be compared.
Selected faces and/or cross-cut will be channel
sampled with both diamond cut and chipped channels.
These methods would be part of the trial mining
programme to investigate bias, and not part of any
routine production grade control. Standard operating
procedures were drafted for grade control activities and
integrated into the health and safety programme. An
annual independent review of grade control activities
was planned.
Data management, integration and interpretation
A major need was for the introduction of a proper data
management system beyond the use of spreadsheets. The
integration of mapping, logging and grade data into a
3D model is important for resource and reserve
estimation and to support final stope design. In
addition, the system needs to be able to integrate
external data such as stope survey, trucking and plant
data to facilitate reconciliation studies.
Staffing and implementation
For the resource development and trial mining phase, a
rostered-on two person underground geological team
was planned, supported by two geo-technicians for
seven day dayshift coverage. At production, the team
would likely be double in size. The existing geological
team had relevant expertise in underground grade
control. Additional training was given in sampling and
mapping, and data interpretation and integration into a
standard mining package. Training for the geological
team, in particular new starters was planned. In
addition, the relative merits of grade control were to
be included into the training programme of mining
crews.
Summary
Liphichi is an advanced exploration project well-
progressed into resource development. Its advantage is
that a substantial amount of underground development
exists. In early 2006, the resource estimate was
supported by 1500 m of underground development
(including 516 chip-channel samples) and 3100 m of
surface and underground drilling (including 1725 core
samples). The presence of underground works was
Table 15 FSE precision for each stage of the revised
Liphichi bulk sample protocol
Stage dN (cm) Mass reduction Precision (70% CL)
1 2.0 200 to 10 t ¡5%
2 1.0 10 to 0.5 t ¡8%
3 0.5 0.5 t to 50 kg ¡15%
Total ¡18%
Table 16 Summary of Liphichi sampling approaches by project stage and comparison with methods used at Bendigo
Stage Sample type Liphichi Bendigo
Resource
development
Diamond core NQ core. Half core taken for SFA HQ core. Half core taken for SFA
Face chip-channel 25–30 kg per face, split and half
submitted to laboratory for SFA.
Revised to 10 kg per m primary
sample submitted for SFA
3 kg per 1 m sample, mainly
assayed by FA but SFA used
if visible gold observed
Grab/muckpile Initially 10–12 kg samples collected from
development rounds. Discontinued
Minimal usage
Mini-bulk Not used 100 kg samples taken from individual
development stockpiles and processed
in small gravity plant
Bulk samples 200 t (four 50 t development rounds).
To be reduced to 50 kg for laboratory
assay (planned, mainly to be taken
along cross-cuts to enable comparison
between drill hole and channel sample
grades)
120–150 t development rounds
collected with 15% split after crushing
and processed in a purpose-built
gravity plant
Production grade
control
Diamond core NQ core. Half core taken for SFA. After
further testing, it is expected that SFA will
be replaced by FA
HQ core. Entire sample taken for SFA.
Developed a proxy-based system to
determine intersection grade and
support assay result
Face chip-channel 10 kg per m primary sample, with 2.5 kg for
SFA. After further testing, it is expected that
SFA will be replaced by FA
Not used. Developed a proxy-based
system to determine face grade
Source This study Johansen and Dominy (2005)
Dominy et al. (2009)

supported by the low operating cost in Bolivia where a
metre of development cost approximately 133% that of a
metre of diamond drill core in 2005. Mineralisation is
structurally controlled and requires moderate geological
supervision during development and mining. The place-
ment of the hanging- and foot-wall drives is crucial to
define the limits of the likely mining zone and is based
on both geological and grade criteria.
A characterisation study identified the presence of
coarse gold at Liphichi. The coarse gold fraction present
is not very coarse (e.g. over 1000 mm); however, clustering
is likely to result in higher variability on a local scale.
Liphichi classifies as a medium coarse gold type. This
generally indicates that some specialised protocol may be
required to achieve quality assay data. Continued SFA
was deemed appropriate for resource delineation drilling,
with its use for grade control to be evaluated as more data
became available. The programme included a comparison
programme between FA and SFA.
Unfortunately the planned resource development
programme was not undertaken, so there is no verifica-
tion of the protocols presented. The sampling methods
at Liphichi are compared to those used at Bendigo
(Table 16).
Conclusion
Grade control samples should be reproducible, unbiased,
safely taken and operationally timely. These points have
an impact on grade control and its ultimate effectiveness.
Key recommendations for the design and implementation
of a grade control sampling programme (Table 17). Mine
geologists and geotechnicians must strive to collect
samples within the framework of the TOS.
A critical issue during the implementation and operation
of grade control will be people. Acceptance of the
programme by all stakeholders will be important and
requires the business case and deliverables to be promoted.
A programme manager or ‘champion’ is required to drive
development and ensure that all parties are engaged. The
overall team must include miners, geologists, engineers and
metallurgists. Programme implementation will be via proper
equipping and training of staff, supported by documented
procedures. Integration between grade control activities and
stope planning and scheduling is required to achieve best
practice. The grade control process allows grade (sampling),
geological, geotechnical and metallurgical data to be
collected. This information should be integrated into the
final stope design process to ensure optimum design and risk
profiling.
This paper addresses the design of underground gold
mine grade-control sampling programmes. However, the
general approach presented is also applicable to the
design of open-pit grade-control programmes.
Acknowledgements
The authors acknowledge Luzon Minerals Limited and
various company staff who assisted with this study. The
use of mineralogical facilities at USTB is acknowledged.
Thanks are due to John Graindorge (Snowden) and two
AES reviewers, Dr Alwyn Annels (SRK Consulting, UK)
and Dr Jean-Michel Rendu (Newmont, USA) for helpful
reviews of the manuscript. AES Editor Professor Neil
Phillips is thanked for his input. An earlier version of this
paper was presented at The AusIMM Eighth International
Mining Geology Conference (August 2011). The opinions
expressed in this paper are those solely of the authors.
References
Abzalov, M. Z. 2008. Quality control of assay data: a review of
procedures for measuring and monitoring precision and accuracy,
Explor. Min. Geol., 17, 131–144.
Carrasco, P. C., Carrasco, P. and Jara, E. 2004. The economic impact
of incorrect sampling and analysis practices in the copper mining
industry, Chemomet. Intell. Lab. Syst., 74, 209–214.
Carswell, J. T., Yulia, K., Lesmana, D. and Steamy, K. 2009. Grade
control sampling quality assurance/quality control in a high-grade
gold mine – Gosowong, Indonesia, Proc. 7th Int. Min. Geol.
Conf., 283–290, Melbourne, The Australasian Institute of Mining
and Metallurgy.
Cintra, E. C., Scabora, J. A., Viegas, E. P., Barata, R. and Maia, G. F.
2007. Coarse gold sampling at Sao Francisco Mine, Brazil, Proc.
3rd World Conf. on ‘Sampling and blending’, 187–197, Porton
Alegre, Fundacao Luiz Englert.
Dominy, S. C. 1997. Sampling and Estimation of Gold Deposits,
unpublished research report, University of Greenwich and Welsh
Gold PLC, p. 230.
Dominy, S. C. 2007. Sampling – a critical component to gold mining
project evaluation, Proc. Project Evaluat. Conf., 89–96,
Melbourne, The Australasian Institute of Mining and Metallurgy.
Dominy, S. C. 2010a. Grab sampling for underground gold mine grade
control, J. S. Afr. Inst. Min. Metall., 110, 277–287.
Dominy, S. C. 2010b. Grade control sampling methods in underground
gold mine grade control, Proc. Sampling Conf., 7–20, Melbourne,
The Australasian Institute of Mining and Metallurgy.
Dominy, S. C., Annels, A. E., Johansen, G. F. and Cuffley, B. W. 2000.
General considerations of sampling and assaying in a coarse gold
environment, Trans. Inst. Min. Metall., 109, B145–B167.
Dominy, S. C. and Petersen, J. S. 2005. Sampling coarse gold-bearing
mineralisation – developing effective protocols and a case study
Table 17 Key stages in the design and implementation of a new underground gold mine grade control sampling
programme
Stage Aim Actions
1 Business case Stakeholder engagement
High-level programme review and business case
2 Characterise Preliminary review and design of characterisation programme
Undertake characterisation programme (see Table 6)
3 Design Review and interpret from Stage 2
Consider orebody, mine size and mining method
Design sampling protocols within framework of TOS
4 Implement Set-up systems and written codes of practice
Training of management, mine geology and production staff
5 Monitor On-going QAQC programme (see Table 7)
Annual internal and/or external peer review of systems
Annual internal and/or external peer review of individuals
6 Update On-going training/re-fresher
Revision of protocols if deemed necessary in Stage 5 – return to Stage 2 or 3 as appropriate

from the Nalunaq mine, Southern Greenland, Proc. 2nd World
Conf. on ‘Sampling and blending’, 151–165, Melbourne, The
Australasian Institute of Mining and Metallurgy.
Dominy, S. C. and Platten, I. M. 2007. Gold particle clustering – a new
consideration in sampling applications, Trans. Inst. Min. Metal.,
116, B130–B142.
Dominy, S. C., Xie, Y. and Platten, I. M. 2008a. Characterisation of in-
situ gold particle size and distribution for sampling protocol
optimisation, Proc. 9th Int. Cong. on ‘Applied mineralogy’, 175–
185, Melbourne, The Australasian Institute of Mining and
Metallurgy.
Dominy, S. C., Xie, Y. and Platten, I. M. 2008b. Gold particle
characteristics in narrow vein deposits: implications for evalua-
tion and metallurgy, Proc. Narrow Vein Mining Conf., 91–104,
Dominy, S. C., Platten, I. M., Fraser, R. M., Dahl, O. and Collier, J. B.
2009. Grade control in underground gold vein operations: the role
of geological mapping and sampling, Proc. 7th Int. Min. Geol.
and Metallurgy.
Dominy, S. C., Platten, I. M. and Minnitt, R. C. A. 2010a.
Heterogeneity, sampling errors and the nugget effect in gold
ores: implications for evaluation, exploitation and extraction,
Proc. Gravity Gold Conf., 3–17, Melbourne, The Australasian
Institute of Mining and Metallurgy.
Dominy, S. C., Platten, I. M. and Nugus, M. J. 2010b. Application of
geology to alleviate sampling bias during the evaluation of high-
nugget gold systems, Proc. Sampling Conf., 75–85, Melbourne,
Dominy, S. C., Platten, I. M. and Xie, Y. 2010c. Determining gold
particle size in gravity ores for sampling and metallurgical
characterisation: discussion and test protocol, Proc. Gravity
Gold Conf., 83–95, Melbourne, The Australasian Institute of
Mining and Metallurgy.
Dominy, S. C., Murphy, B. and Gray, A. H. 2011a. Characterisation of
gravity amenable gold ores: sample representivity and determina-
tion methods, Proc. 1st Int. Geometall. Conf., 281–292,
Dominy, S. C., Platten, I. M., Howard, L. E., Elangovan, P.,
Armstrong, A., Minnitt, R. C. A. and Abel, R. L. 2011b.
Characterisation of gold ores by x-ray computed tomography –
Part 2: applications to the determination of gold particle size and
distribution, Proc. 1st Int. Geometall. Conf., 293–309,
Dominy, S. C. and Minnitt, R. C. A. 2012. Application of heterogeneity
testing to coarse-gold ores: problems and solutions, Proc.
Sampling Conf., to be published, Melbourne, The Australasian
Institute of Mining and Metallurgy.
Dominy, S. C., Platten, I. M. and Cipriano, J. 2012. The value of
geological mapping as part of grade control in underground
operations, Proc. Resource Evaluat. Min. Conf., to be published,
Perth, The Australian Institute of Geoscientists.
François-Bongarçon, D. M. 1993. The practice of the sampling theory
of broken ore, CIM Bull., 86, 75–81.
François-Bongarçon, D. M. 1998. Gy’s Formula: conclusion of a new
phase of research, Aust. Inst. Geosci. Bull., 22, 1–10.
François-Bongarçon, D. M. and Gy, P. M. 2002. The most common
error in applying ‘Gy’s Formula’ in the theory of mineral
sampling and the history of the Liberation factor, J. S. Afr.
Inst. Min. Metall., 102, 475–479.
Frempong, P. K. 1999. The development of a robust sampling strategy
and protocol in underground gold mines, Proc. Aust. Inst. Min.
Metall., 304, 15–22.
Geelhoed, B. 2011. Is Gy’s formula for the fundamental sampling error
accurate? Experimental evidence, Miner. Eng., 24, 169–173.
Graindorge, J. 2010. Maintaining sample data quality through robust
quality assurance and quality control protocols, Proc. Sampling
and Metallurgy.
Gy, P. M. 1979. Sampling of particulate materials: theory and practice,
431, Amsterdam, Elsevier.
Gy, P. M. 1992. The sampling of heterogeneous and dynamic material
systems – theories of heterogeneity, sampling and homogenizing,
653, Amsterdam, Elsevier.
Gy, P. M. 1998. Sampling for analytical purposes, 153, Chichester,
John Wiley and Sons.
Gy, P. M. 2004. Sampling of discrete materials: parts I–III, Chemomet.
Intell. Lab. Syst., 74, 7–47.
Johansen, G. F. and Dominy, S. C. 2005. Development of sampling
protocols at the New Bendigo Gold Project, Australia, Proc. 2nd
World Conf. on ‘Sampling and blending’, 175–183, Melbourne,
Kemp, C. and Morgan, G. 2012. Using borehole radar to define
orebody interpretations in a narrow vein gold deposit, Proc.
Narrow Vein Mining Conf., to be published, Melbourne, The
Australasian Institute of Mining and Metallurgy.
McKinstry, H. E. 1948. Mining geology, 680, Englewood Cliffs, NJ,
Prentice-Hall.
Minnitt, R. C. A. 2007. Sampling: the impact on costs and decision
making, J. S. Afr. Inst. Min. Metall., 107, 451–462.
Minnitt, R. C. A. 2010. The state of sampling practice in the South
African minerals industry, Proc. Sampling Conf., 31–50,
Minnitt, R. C. A., Rice, P and Spanengberg, C. 2007. Experimental
calibration of sampling parameters K and alpha for Gy’s formula
by the sampling tree method, J. S. Afr. Inst. Min. Metall., 107,
513–518.
Minnitt, R. C. A. and Assibey-Bonsu, W. 2009. A comparison between
the duplicate series method and the heterogeneity test for
calculating Gy’s sampling constants, Proc. 4th World Conf. on
‘Sampling and blending’, 137–154, Johannesburg, Southern
African Institute of Mining and Metallurgy.
Peters, W. C. 1987. Exploration and mining geology, 685, New York,
John Wiley.
Pitard, F. F. 1993. Pierre Gy’s sampling theory and sampling practice,
488, Boca Raton, FL, CRC Press.
Pitard, F. F. 2005. Sampling correctness: a comprehensive guide, Proc.
2nd World Conf. on ‘Sampling and blending’, 55–66, Melbourne,
Republic. 2006. Intitial Inferred Resource Estimate for Luzon Minerals
Liphichi Property. Republic Gold Ltd ASX Announcment dated
23 March 2006, pp 12.
Rickard, T. A. 1907. Sampling and estimation of ore in a mine, 222,
London, Hill Publishing Company.
Simon, A. and Gosson, G. 2008. Considerations on QAQC and
sampling security, Proc. Sampling Conf., 135–140, Melbourne,
Sketchley, D. A. 1998. Gold deposits: establishing sampling protocols
and monitoring quality control, Explor. Min. Geol., 7, 129–
138.
Snowden, 2004. Liphichi gold project, Bolivia – review, unpublished
report for Luzon Minerals Limited by Snowden Mining Industry
Consultants Limited (Author: S. C. Dominy; Dated: November
2004), 31.
Snowden, 2005a. Liphichi gold project, Bolivia – project development
review, unpublished report for Luzon Minerals Limited by
Snowden Mining Industry Consultants Limited (Author: S. C.
Dominy; Dated: May 2005), 17.
Snowden, 2005b. Liphichi gold project, Bolivia – project development
plan, unpublished report for Luzon Minerals Limited by
Snowden Mining Industry Consultants Limited (Author: S. C.
Dominy; Dated: December 2005), 21.
Stanley, C. R. and Lawie, D. 2007. Average relative error in
geochemical determinations: clarification, calculation and a plea
for consistency, Explor. Min. Geol., 16, 267–275.
Thompson, M. and Howarth, R. J. 1976. Duplicate analysis in
geochemical practice part 1: theoretical approach and estimation
of analytical reproducibility, Analyst, 101, 690–698.
Vallée, M. A. 1992. Sampling optimisation, in Guide to the evaluation
of gold deposits, 45–62, Montreal, Canadian Institute of Mining,
Metallurgy & Petroleum.
Vallée, M. A. 1998. Sampling quality control, Explor. Min. Geol., 7,
107–116.
Van Der Merwe, H. C. 2009. SirovisionH: a proposed solution for the
implementation of a digital geological mapping system at
AngloGold Ashanti’s Moab Khotsong mine, Proc. World Gold
Conf., 183–191, Johannesburg, The Southern African Institute of
Mining and Metallurgy.
View publication stats
View publication stats

Design grade control

More Related Content

Similar to Design grade control (19)

Recently uploaded (20)

Design grade control