Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Geometallurgical Block Model vs Geometallurgical

Units

Bch, Ing, MSc, Samuel Canchaya


General Manager of SAMPLING OK SAC and Associate Professor at Universidad Nacional de
Ingeniería & Pontificia Universidad Católica del Perú

ABSTRACT

Following traditional geometallurgical modelling, a geologist first identifies GUs: Geological


Units (Unidades Geológicas: UGs). “Representative” composites of each GU are then acquired
and subjected to metallurgical testing. The results define the respective GMUs:
Geometallurgical Units (Unidades Geometalúrgicas: UGMs); which in turn inform
plant and metallurgical process design.

However, due to a deposits heterogeneity, representativeness of these composites is very


questionable. For this reason an alternative approach; the so-called Geometallurgical Block Model
(GMBM) is proposed. This is an improvement on traditional geological block model where, in
addition to assays, lithology and alteration, also critical geometallurgical variables such as:
percentage of clays; acid, lime or cyanide consumers; penalty elements, hardness, density, etc.,
are being modelled.

With recent technological advances of infrared spectrometers, it is already currently possible to


obtain quick semi-quantitative mineralogical analysis, less than a minute per sample. The
addition of systematic quick measurements of uniaxial point load and density; make it possible
to generate thousands of data required to implement a probabilistic GMBM; which by far
constitutes currently the best deliverable product of Geometallurgy.

INTRODUCTION
In the XIV Peruvian Congress of Geology and XIII Latin American Congress of Geology, the
author laid the foundations for what is to be understood as a Geometallurgical Model: GMM
(Canchaya 2008). Unfortunately, each specialist is designing and understanding GMM in
different ways, sometimes very unique and special.

Corresponding author: Samuel Canchaya Moya, General Manager of SAMPLING OK SAC, and Associate
Professor at the Universidad Nacional de Ingeniería and Pontificia Universidad Católica del Perú. Urb.
Portales de Surco; Mz. A Lote 1; Psje 2. 3er piso. Lima 33. Phones: +51 988021992; +51 947686149. Email:
canmoysa@gmail.com.

1
It is the aim of this paper to properly define the Geometallurgical Block Model (GMBM), such as
a result of probabilistic modelling based on quantitative data. To obtain such data, it is necessary
to follow the sampling procedures, established by the "Theory of Sampling" (TOS) after Pierre Gy
(1992 and 1999).

THE “TOTAL ROCK CONCEPT”

For the purposes of this discussion, the basis of Geometallurgy is the TOTAL ROCK CONCEPT
(TRC), as defined by Canchaya (2008). This concept considers the rock as a whole and that
designations of ore and waste are circumstantial and established as a result of human being based
on classification juncture requirements or current needs. Therefore, the differentiation between
ore and gangue will be only a valid exercise if considered as parts of a whole. The TRC considers
that gangue minerals are the main actors in metallurgical circuits, such as crushing, grinding,
flotation, cyanidation, acid leaching, etc. Additionally the main problems occuring in plants are
related to gangues, especially due to their relative abundance, which in some types of deposits,
especially in the low-grade, are largely predominant.

For example, in the Cu-(with Mo and/or Au) porphyritic deposits, ores (chalcopyrite,
molybdenite, chalcocite, bornite, etc.) do not exceed the 2 or 3 % of the total, while gangue
minerals are above 96 % of the total rock. Even more critical is the case of high sulfidation
epithermal gold deposits (such as Yanacocha and Pierina in Perú; or La Coipa and Can-Can in
Chile); where gold ore is usually below 1gr/ton (1 ppm), which in percentage means only: 0.0001%
of the rock; it means that the rest: 99.9999 % is gangue: different types of quartz, eventually
accompanied by smaller amounts of alunite, clays, pyrophyllite, jarosite, Fe-hydroxides, etc.
Therefore, it is easy to deduce that any type of problem, like cyanide or lime consumption,
comminution resistance, or reduction of the recovery percentage or extraction by cyanide, etc. is
mainly due to the gangue minerals. Of course, we cannot ignore the case where sulphides,
sulfosalts or the same native sulphur, are also cyanide consumers; but even in these cases, we are
speaking of gangues.

Although it is obvious and clear that gangue minerals are the protagonists in low-grade deposits;
the same applies to virtually all types of deposits, even at the higher grades, as the proportion of
ore is usually less than gangue.

GEOMETALLURGICAL CHARACTERIZATION (GMC)

In order to implement a Geometallurgical Model it is first necessary to perform a


Geometallurgical Characterization (GMC) exercise. Nothing more related and committed to the
TRC than GMC, which is “global” because includes the following four analysis (Figure 1):

2
Figure 1 Analysis types covered by the geometallurgical characterization, based on the total rock concept.

I. Chemical characterization: Apart from the traditional assays of both, economic and
deleterious elements, this should be included from exploration phase, the Au and Ag
cyanide analysis; as well as the sequential Cu analysis: total Cu analysis (CuT), Cu soluble
in acid (CuSAc), soluble in cyanide (CuSCN) and the residual Cu (CuR). For example,
this is a practice already established in the Explorations Department of Cía. de Minas
Buenaventura-Perú.
II. Mineralogical characterization: "Bulk" mineralogical analysis by X-Ray Diffraction
(XRD) is the most commonly used technique in geometallurgical characterization;
however, it is expensive and time consuming. Of course, the alternative methods such as
optical microscopy and QEMSCAN are more expensive (Table 1). On the contrary,
FTNIR analysis (Fourier Thermal Near Infra Red) is the more suitable because it is faster
and cheaper (Canchaya 2016). FTNIR analysis is explained in Figure 2.
III. Structural and textural characterization: In Table 2 is described the structural-textural
characterization for every sampling domain at different scales; it is also mentioned the
specific tools and main applications of every type of characterization. In all cases and
scales, the performance of this analysis has been enhanced with the recent appearance of
software for automated image analysis (AIAS), which allows rapid qualitative and
quantitative characterization of: fracture patterns and their frequency (fracture density);
mineral and alteration assemblages, modal analysis of grains and particles (grain
intergrowths); as well as their degrees of liberation. Anyway, MLA (Mineral Liberation
Analyser) and QEMSCAN (Quantitative Evaluation of minerals by Scanning Electron
Microscopy) are the most evolved equipment; which provides virtually the entire service,
that is to say: micro-chemical analysis, mineralogical analysis and intergrowth analysis;
being the only capable to analyse “invisible Au” (as solid solution, nanometric or
structural inclusions in other minerals or colloidal masses).

3
IV. Physical-mechanical characterization: After Canchaya (2012) and Canchaya et al. (2013),
the most important and common physical-mechanical variables in a geometallurgical
database are: bulk density and mechanical resistance (“hardness”). The traditional way
for density determination is using paraffin, with recent more practical variations using
lacquer-spray or plastic thin film; however for a systematic sampling of the density it is
better the direct register of wells using gamma radiation. The mechanical competence
should be obtained from systematic point load tests (PLT) of decimetre DDH core pieces
of "intact rock"; which are reported in megapascals (MPa). In addition should be also
considered "Rock Quality Designation" (RQD), that together with the density of
fracturing and type of fracturing, obtained from the geometallurgical logging, are the
parameters that allow the optimization of primary crushing and blasting. On the other
hand, mechanical resistance of intact rock (PLT), serves to design and management
comminution circuits, especially the secondary and tertiary crushing.

Currently there is a wide range of methods of instrumental analysis that allows GMC. However,
in order to be aligned with the TRC, the GMC should be carried out, preferably, using those
analyses that are "bulk" or global in nature, as for example: whole chemistry analysis, microscopic
studies of polished-thin sections, bulk mineralogical XRD analysis, whole mineralogical analysis
by near infrared spectrometers, etc.

Execution time 20,000


Method Deliverable results
per sample samples

Semi-quantitative mineralogical
FTNIR Spectrometry analysis (clays, micas sulphates, Tens of seconds 15 days
carbonates, limonites, etc.)

Quantitative “Bulk” mineralogical


X-Ray diffraction Tens of minutes 420 days
analysis.

Optical microscopy Modal mineralogical analysis of ore


Several tens of
with automatic image and gangue; intergrowths analysis, 560 days
minutes
analysis liberation degree, etc.

Full automatized chemical, 1000


LMA or QEMSCAN Hours
mineralogical and textural analysis days

Table 1 Comparison between the most common methods of mineralogical analysis used in
geometallurgical characterization. It is clear that the modern infrared spectrometric method (FTNIR) is the
most appropriate for the implementation of a geometallurgical database and consequently for generating
a GMBM, because it is fast, cheap and does not require additional sample preparation; since it can measure
up to -10 mesh particles or less. Modified after Canchaya (2016).

4
Figure 2 FTNIR Semi-quantitative mineralogical analysis. A: FTNIR Bruker Spectrometer, model Matrix I.
B: Example of calibration curve for FeS2 using DRX as primary analysis. C: Example of calibration curve for
swelling clays using Cation Exchange Capacity (CEC) as primary analysis. D: Example of a results chart
showing percentages of different minerals.

GENERAL PROCEDURE FOR IMPLEMENTATION OF THE


GEOMETALLURGICAL BLOCK MODEL (GMBM)

It is an important consideration that metallurgists understand “modelling” in a different way to


geologists. Metallurgical modelling aims to find a deterministic relationship between input
variables (mainly mineralogical) and output variables (mainly metallurgical); which are usually
performed testing composites. On the contrary, the GMBM as proposed in this work, is very
similar to Geological Resource Modelling. The difference being that, in addition to traditional
variables like: coordinates, assays, lithology and alterations, "critical" geometallurgical
variables, like: clays, acid consumers, cyanide consumers, penalty element carriers, etc. should
be also considered.

There are five steps for the implementation of a GMBM (Canchaya 2008); see Figure 3:

Phase I: Pilot geometallurgical characterization, the primary objective of which is to


identify and characterize critical variables that will influence the success and cost
of mining and metallurgical processing.

5
Scale Sampling Tools Main applications
domain

Geological mapping by cells.


Macro-images processed by
ROCK MASS
MEGASCOPIC automatized image analyzer Blasting optimization
software (AIAS)

DDH CORE %RQD

Blasting and comminution


MACROSCOPIC DDH OR RC Geometallurgical logging optimization.
Metallurgical treatment

Optical microscopy with Mineralogical modal analysis;


HAND AIAS. Electronic microscopy mineral intergrowths, liberation
MICROSCOPIC SPECIMENS (SEM/EDS o WDS), degree of ore and minerals carriers
AND LMA/QEMSCAN of penalty elements
POLISHED-
THIN Electronic microscopy For submicroscopic, structural,
SUB-
SECTIONS (SEM/EDS o WDS), solid solution and colloidal
MICROSCOPIC
LMA/QEMSCAN/TESCAN occurrence of Au, Ag, As, Sb, etc.

Table 2 Domains, tools and main applications, at different scales of the structural-textural characterization.

Figure 3 Phases for the implementation of the Geometallurgical Block Modell. Modified after Canchaya (2008).

6
Phase II: Metallurgical testing of samples with known content of the critical variables, to
model their metallurgical behaviour. This is the phase where the specification
limits of every critical variables is outlined.
Phase III: Systematic sampling and quantitative analysis of the critical variables, using
analytical methods, that are fast and cheap. Sampling must be carried out in
accordance with the principles of the Theory of Sampling (Gy, 1992 and 1999).
Optimal sampling distances should be determined by geostatistical methods.
Phase IV: Geometallurgical block modelling of every critical variable, which should be
done by kriging or simulations using geometallurgical databases, in the same
way as is done for ore assays (Au, Ag, Cu, Pb, Zn, Sn, etc.) or for penalty or
undesirable elements (As, Sb, Hg, etc.).
Phase V: Updating the model whenever a substantive change in the mineralogical
assemblies or in textural and/or structural patterns occurs. In this way is entirely
feasible to have, in a single ore deposit, different sub-domains with different
critical variables and therefore different geometallurgical databases and models.

The GMBM is in essence similar to the Geological Block Model, where every block contains: the
estimated assays of economic elements (% Cu, % Mo, Au ppm, % Zn, etc.), lithology and type of
alteration. If we additionally include in every block through the respective estimation of critical
geometallurgical variables, we will be implementing a GMBM.

THE TRADITIONAL METHOD: FROM GUs to GMUs

In traditional modelling methodology, geologists first divide the deposit into Geological Units or
GUs (Unidades Geológicas: UGs), and obtain “representative” composites of each GU, for
metallurgical testing to define the respective GMUs or Geometallurgical Units (Unidades
Geometalúrgicas: UGMs); figure 4. This GMUs enables plant and metallurgical process design.

The main criticism of this method is related to representativeness of the composites, which
according to the Theory of Sampling (TOS of P. Gy) only become specimens, no samples, due to
natural heterogeneity (mineralogical, textural, lithological, etc.) of the deposits. This error is
enhanced when the results of the metallurgical tests, obtained from these composites, are
extrapolated daring and recklessly to million tonnes (GMUs), resulting in the well-known errors
traditionally associated with this methodology.

On the other hand, some metallurgical traditional tests like Bond Work Index (BWI), Sag Power
Index (SPI), or similar, cannot be incorporated into a probabilistic database, mainly because they
are based on specimens and not samples systematically obtained. Additionally their acquisition
is both very expensive and time consuming (Canchaya 2012 and Canchaya et al. 2013).

7
Figure 4 From Geological Units (GUs) to Geometallurgical Units (GMUs). The main error of this method is
their low representativeness of the composites; this error is incremented when the results of metallurgical tests
of composites are extrapolated to millions of tonnes; assuming erroneously that the GUs are homogeneous.

DISCUSION

Until a few years ago, the main "bottle-neck" that prevented the implementation of a probabilistic
GMBM was the lack of fast and cheap quantitative analytical methods. As already mentioned
(Table 1) QEMSCAN analysis are very expensive, tedious and take several tens of minutes per
sample, or more depending of the mode of analysis and its complexity; XRD mineralogical
analysis are also expensive, although less so than QEMSCAN, depending on the angular range
required to register and plot the respective difractogram. Neither of these methods is sufficiently
fast and cheap to implement a geometallurgical database, which normally requires thousands to
tens of thousands of quantitative mineralogical analyses. Fortunately, today we already have
alternatives, e.g. infrared spectrometric analysis (FTNIR), which can deliver cheap, semi-
quantitative mineralogical analysis in only tens of seconds per sample.

Critical variables determined in Phase I of figure 3, to be incorporated into a geometallurgical


database, should be sampled and analyzed systematically; therefore, their optimal sampling
distance should first be determined by geostatistical methods, as is done with assays of economic
elements.

8
Figure 5 The Geometallurgical Block Model. The geometallurgical database is just an upgrade of the
geological database. The algorithms for block estimation of every critical variable are the same as that used
for classical assays estimation. Modified after Canchaya (2008)

Short term modelling


Everything mentioned so far in this work is limited to the predictive or long-term GMBM, based
on diamond drills holes; however, the possibility of quick and inexpensive FTNIR mineralogical
determinations from blast holes samples allows optimization of the selection of the mineral (ore
vs gangue). It is not even necessary to make any additional conminution, since the sample as it is
obtained from the detritus promontories of blast holes can be placed directly into the FTNIR
spectrometer sample holder and be analyzed in less than a minute.

CONCLUSIONS

Currently there are still a few cases of application of GMBM and in general only partially: Cu-Mo
porphyry Cerro Verde (Fennel et al. 2005), Cu-Mo porphyry Trapiche (3,227 MPa and 1,050
density determinations), the mesotermal deposits San Gabriel (Canchaya et al. 2013; 3064 MPa,
5200 density determinations, 3030 FTNIR analysis, and thousands of RQD) and Marcapunta
(Huallpallunca & Zapata 2017). On the other hand, Shahuindo (high sulfidation Au deposit in

9
Cajamarca-Perú) is a pioneering example of a project where blasting has been optimized, with
the criteria mentioned in this paper, and therefore do not need crushing facilities.

With recent technological advances of infrared spectrometers, it is already currently possible to


obtain rapid semi-quantitative mineralogical analysis, less than a minute per sample. If we
add systematic quick measurements of uniaxial point load, RQD and density; it is now possible
to have thousands of data required to implement a probabilistic GMBM; which by far constitutes
currently the best deliverable product of Geometallurgy.

REFERENCES

 CANCHAYA S. (2008) ‘El Modelo Geometalúrgico’.- XVII Congreso Peruano de Geología y XIII
Congreso Latinoamericano de Geología; 29 de Set. a 3 de Oct. 2008; 6 p.

 CANCHAYA S. (2012) ‘Density and Mechanical Competence, main physical and mechanical
variables in Geometallurgical Models’.- Proceedings GEOMET 2012 International Seminar on
Geometallurgy; Santiago de Chile 5 al 7 Dic. 2012: 24-25.

 CANCHAYA S. (2016) ‘Espectrometría FTNIR semi-cuantitativa, herramienta principal para


implementar la data geometalúrgica’.- Proceedings GEOMET 2016 International Seminar on
Geometallurgy; Lima-Perú; 11 al 13 Dic. 2016.

 CANCHAYA S. & BAUMGARNTNER R. & GAIBOR A. & TRUEMAN A. (2013) ‘Bulk density
for Resource Estimation and Geometallurgical Purposes- Canahuire Au-Cu-Ag Deposit,
Southern Peru’.- 6th World Conference on Sampling and Blending (WCSB6); Lima 19 al 22 Nov.
2013; 1 p.

 FENNEL M. & GUEVARA J. & CANCHAYA S. & BAUM W. & VELARDE G. & GOTLIEB P.
(2005) ‘QEMSCAN Mineral Analysis for Ore Characterization and Plant Support at Cerro
Verde’.- XXVII Convención Minera; Arequipa-Perú; 11p.

 GY P. (1992) Sampling of Heterogeneous and Dynamic Material Systems. Theories of


Heterogeneity, Sampling and Homogenizing.- Elsevier, New York; ISBN 0-444-89601-5.

 GY P. (1999) Sampling for Analytical Purposes.- John Wiley & Sons, New York; 153 p.

 HUALLPALLUNCA I. & ZAPATA J. (2017) ‘Aplicación de la geometalurgia en la optimización


de costos, Tajo Norte-Mina Colquijirca-Cerro de Pasco-Perú’.- Proceedings Procemin-GEOMET
2017; Sgo. de Chile; 4 a 6 Oct. 2017.

10

You might also like