Minerals Engineering 65 (2014) 916

Contents lists available at ScienceDirect

Minerals Engineering
journal homepage: www.elsevier.com/locate/mineng

A process mineralogy approach to geometallurgical model renement

for the Namakwa Sands heavy minerals operations, west coast of South
Africa
C. Philander , A. Rozendaal
Department of Earth Sciences, University of Stellenbosch, Private Bag X01, Matieland 7602, South Africa

a r t i c l e

i n f o

Article history:
Received 4 February 2014
Accepted 7 April 2014
Available online 4 May 2014
Keywords:
Heavy minerals
Geometallurgy
Ore characterisation
Quantitative mineralogy
Mineral processing

a b s t r a c t
The original geometallurgical model for the Namakwa Sands deposit was modied to accommodate ore
blends in addition to the various single ore types. A process mineralogy approach was followed in a structured and systematic manner to evaluate the integrity of the adjusted model, particularly for ilmenite and
zircon, the minerals of highest intrinsic value. This study reproduced recovery relationships predicted by
the geometallurgical model for each of the key process functions, and as a result the integrity of the geometallurgical model is validated. Overall, the recovery potential determined for ilmenite and zircon are
well adjusted to model estimates. Poor mineral liberation, an anomalously high abundance of garnet
and pyroxene and variation in particle chemistry are recognized as the key recovery penalties. The gangue content is the most signicant constraint to ilmenite recovery, whereas zircon chemistry is the most
important negative factor in the production of a premium quality zircon product. Results of this study
contributed to the renement of the current geometallurgical model and also identied opportunities
to optimise mineral resource utilisation in the future.
2014 Elsevier Ltd. All rights reserved.

1. Introduction
Tronox Mineral Sands is presently operating the world-class
Namakwa Sands heavy minerals deposit which is located along
the west coast of South Africa. Estimated pre-mining mineral
resources are in excess of 1 100 Mt of ore with in situ grades of
7.9 wt.% total heavy minerals, 3.16 wt.% ilmenite, 0.85 wt.% zircon,
0.50 wt.% leucoxene and 0.21 wt.% rutile. The bulk of this megadeposit represents cemented high-grade ore that hosts a diverse
heavy mineral suite.
Geometallurgical challenges related to the heterogeneity of the
ore prompted systematic intervention, endeavouring to improve
mineral resource intelligence (Philander and Rozendaal, 2008,
2009, 2011). As a result, a geometallurgical template model was
developed for the Namakwa Sands deposit (Philander and
Rozendaal, 2013). This model demonstrated a promising ability
to predict the valuable mineral recovery potential for the seven
ore types individually, but was not previously evaluated for ore
blends. Since then Namakwa Sands abandoned ore-type campaigning in favour of consistent blending, which provides a greater

Corresponding author. Tel.: +27 27828742525.

E-mail address: carlo.philander@za.tronox.com (C. Philander).
http://dx.doi.org/10.1016/j.mineng.2014.04.006
0892-6875/ 2014 Elsevier Ltd. All rights reserved.

balance between life of mine sustainability, mine development,

throughput, mineral recoveries and product quality.
The primary objective of this study was to modify this starter
geometallurgical model in order to quantitatively evaluate its ability to determine mineral recovery potential for ore blends. In addition, it was envisaged that the outcomes of this study would help
to rene the geometallurgical model and highlight opportunities
to improve mineral resource utilisation. A process mineralogy
approach was preferred since mineralogy denotes an integral
building block of the original geometallurgical model (Adams,
2007; Evans et al., 2011; Lotter, 2011). In the current study emphasis is placed on the minerals of greatest economic interest, namely
ilmenite (FeTiO3) and zircon (ZrSiO4).
2. Methodology
The original geometallurgical model describes selected relationships between ore characteristics and mineral recoveries that were
determined from controlled sample populations. These ore characteristics manifest as bulk properties, for example oversize contents
(+1 mm particle size), nes contents ( 45 lm particle size), mineral grade and heavy mineral composition, or as particle attributes
such as size, shape, density, surface exposure, mineral liberation
and particle chemistry. Model indications are that these mineral

10

C. Philander, A. Rozendaal / Minerals Engineering 65 (2014) 916

characteristics control mineral recoveries to variable degrees

depending on the ore type processed. Mineral grades and the particle chemistry of the valuable fraction were identied as the key
recovery drivers. All seven ore types are individually accounted
for in the original geometallurgical model, but for this study the
model was statistically transformed in an attempt to accommodate
specic ore blends. As a consequence, all the recovery relationships
that were established in the starter model were copied into the
current blend model.
The ore type compositions of monthly ore blends were determined from geological block model depletions and used as inputs
in the geometallurgical model to calculate the mineral recovery
potential at key processing stages over an 18 month period.
Monthly composite samples representing key sampling points
at several processing stages were collected with automated samplers. Samples were investigated using a variety of analytical techniques. Optical microscopy assisted with mineral typication, Xray uorescence spectroscopy provided bulk chemical assays,
QEMSCAN quantied mineral composition, and electron microprobe and laser ablation techniques were employed to determine
the chemistry of selected heavy minerals.
3. Process mineralogy
The Namakwa processing ow sheet starts with two primary
concentrators, PCP East and PCP West, which are fed by separate
mining operations (Fig. 1). Their resulting heavy mineral concentrates are blended into the Secondary Concentration Plant (SCP)
where magnetic and non-magnetic concentrates are produced. At
the Mineral Separation Plant these concentrates are further
upgraded by removing unwanted contaminants to produce the
saleable products. In the following sections the adjusted geometallurgical model is systematically evaluated with respect to the relevant process.
3.1. Primary concentration
The key purpose of the primary concentration process is to
remove the low density minerals from a dressed ore (45
1000 lm fraction). PCP West houses a semi-autogenous mill that
mills the cemented ore to 1 mm. Conventional wet spirals are
used in a four stage duty to recover the heavy minerals (density
greater than 29 g/cm3). The starter geometallurgical model relates
zircon and ilmenite recoveries mainly to their spiral feed grades,
because the zircon grades of both the concentrate and tails are
stringently controlled within xed specications. By comparison,
particle characteristics such as size, shape and liberation were previously determined to have lesser effects on ilmenite and zircon
recoveries (Philander and Rozendaal, 2013).
Actual grade-recovery data traverse the model grade-recovery
curves relatively well (Fig. 2). Zircon recoveries for PCP East are
markedly better than for PCP West across a wide feed grade range.
Ilmenite recoveries for the two primary concentrators mirror the
zircon recovery trends, although at lower levels because the current mine grade control philosophy targets zircon. PCP East recoveries remain consistent across a broad feed grade range, but by
comparison, the extensive spread in zircon recoveries at a xed
feed grade effectively implies that feed grade is not the only driver
of PCP West recoveries (Fig. 2). Therefore, the recovery performance of the two primary concentrators was systematically reassessed with reference to the basic fundamentals of spiral
separation.
Theoretically, the recovery of heavy mineral particles on a spiral
is dependent on their hydraulic behaviour, which is mainly a function of their particle density (Pascoe et al., 2007; Grobler and
Bosman, 2011). A density cut-point of 3.4 g/cm3 is in use at the

two primary concentrators (Fig. 3). Mineral particles with greater

densities, such as zircon (q = 4.7) and ilmenite (q = 4.7) would
essentially report to concentrate (Fig. 3). Valuable mineral losses
to tail would be limited, yielding ilmenite and zircon recoveries
typically approaching 98% as achieved by PCP East. The inconsistent zircon losses to PCP West tails, which is an order of magnitude
greater by comparison, prompted further investigation.
Previous studies showed that wet spirals have limitations in
recovering heavy minerals reporting to the 45 lm and +250 lm
fractions and that spiral separation is particularly hampered by
polymodal particle size distributions (Burt, 2000; Richards et al.,
2000; Mohanty et al., 2002; Pascoe et al., 2007; Walklate and
Jeram, 2007). All seven ore types that constitute the Namakwa
Sands deposit exhibit unimodal particle size distributions for all
the heavy minerals analysed, although there are slight differences
in median particle sizes amongst ore types (Table 1; Philander and
Rozendaal, 2013). In addition, the total proportion of ilmenite and
zircon in the 45 lm and +250 lm fractions constitute less than
5 wt.% in both the tails and concentrates of PCP East and PCP West.
In agreement with the previous study, the current ndings conrm
that the particle size distribution of valuable heavy minerals has no
meaningful bearing on their recovery in the current primary concentration process.
Detailed QEMSCAN analysis revealed that up to 35% of the zircons reporting to PCP West tails are poorly liberated. Evidently,
the cemented nature of the ore fed to PCP West reduces the effective particle density of the heavy minerals, resulting in increased
losses of valuable minerals to tailings (Laplante and Spiller,
2002). Poor liberation impairs the recovery of zircon to a lesser
degree compared to ilmenite, because the latter has a more conducive surface template for the cementing agent to interlock. The
geometallurgical model imposes a linear penalty on total zircon
recovery to account for liberation effects (Fig. 4). Actual data straddle the calculated line prominently, which conrms that recovery
penalties owing to poor liberation could be as high as 10% for zircon and even greater for ilmenite.
The balance of the zircon and ilmenite in PCP West tails is properly liberated, but are notably ner compared to zircon and ilmenite reporting to concentrate (Table 1). This is a clear indication of
entrainment. Unlike PCP East, PCP West treats ore blends with a
variable heavy mineral composition that could contain more than
50% gangue of which garnet and pyroxene constitute the major
part (Fig. 1). The abundance of garnet, a silicate mineral that
reports essentially to the heavy mineral concentrate, because it
has a density (q = 4.3) above the density cut-point complicates primary concentration and other processes downstream.
Wet spirals appear particularly effective in rejecting the bulk of
lower density gangue heavy minerals such as pyroxene (q = 3.4),
apatite (q = 3.2) and aluminosilicates (q = 3.2). This however
comes with an inadvertent penalty as it appears that pyroxene
facilitates the entrainment of zircon and ilmenite to tails, due to
its comparatively larger particle size (Table 1). The data suggest
that the recovery of valuable heavy minerals is strongly a function
of the pyroxene feed grade, which could impose recovery penalties
of up to 20% for ilmenite and zircon (Fig. 5). Ilmenite and zircon
feed grades, which are inversely correlated with the pyroxene
grade (r > 0.9) however, remain good predictors of mineral
recoveries.
The present study indicates that the starter geometallurgical
model incorrectly highlighted feed grade as the chief recovery driver for the primary concentration process. Instead, the recovery of
ilmenite and zircon appears to be related to their degree of liberation and the pyroxene feed grade (Table 2). These two recovery
drivers have a signicant impact on the recovery performance of
PCP West, which treats variable ore blends that exhibit striking
variations in the degree of cementation and heavy mineral compo-

C. Philander, A. Rozendaal / Minerals Engineering 65 (2014) 916

11

Fig. 1. A simplied illustration of the Namakwa Sands processing ow sheet. The average mineralogical composition of the total heavy mineral fraction at each sampling
point is depicted as pie charts. IRMS = induced roll magnetic separator, HAL = hot acid leach; Zirkwa = secondary zircon product; Tiokwa = secondary rutile product.

12

C. Philander, A. Rozendaal / Minerals Engineering 65 (2014) 916

Table 1
Median particle sizes of selected minerals for key processing stages. Values are in lm.
Reclaim refers to reject streams that are re-treated at a later stage.
Stream

Fig. 2. Actual zircon grade-recovery data as a data density plot, which is draped
onto modelled grade-recovery feed curves for the two primary concentrator plants.
PCP East displays consistent recoveries across a wide feed grade range, whereas PCP
West recoveries are notably scattered for a xed feed grade.

sition. By contrast, the conspicuously better recovery performance

of PCP East is related to the favourable processing character of its
ore, which portrays highly uniform, virtually unconsolidated sands
with extremely low quantities of pyroxene. The recovery of ilmenite and zircon at PCP East is primarily determined by their level of
liberation. For PCP West however, ilmenite and zircon recoveries
are more controlled by the abundance of pyroxene than the degree
of liberation. The liberation data suggest that the current SAG communition process could benet from optimisation to improve
ilmenite and zircon recoveries downstream (Fig. 4; Table 2).
3.2. Secondary concentration
The secondary concentration process utilises wet high intensity
magnetic separators (WHIMS) followed by wet spirals to split the

Ilmenite

Garnet

Pyroxene

Primary Concentration Plant East

Feed
110
Concentrate
112
Tails
113

Zircon

116
124
124

144
132
143

83
82
120

Primary Concentration Plant West

Feed
103
Concentrate
105
Tails
97

108
114
100

126
118
137

120
114
121

Secondary Concentration Plant

Feed
Magnetic concentrate
Non-magnetic concentrate
Reclaim

108
98
105
111

115
103
136
127

133
113
145
142

124
113
127
151

Mineral Separation Plant

Ilmenite product
Zircon product
Rutile product
Ilmenite reclaim
IRMS reclaim
Quartz reclaim
Zircon reclaim
Rutile reclaim

63
105
105
99
102
130
113
95

102
61
109
109
143
168
114
106

114
31
12
117
158
164
129
108

78
84
36
112
115
124
107
102

primary concentrate blends into magnetic and non-magnetic concentrates. Both concentrates are subjected to mechanical attritioning to reduce surface related contamination such as duricrust
cement, ferruginous and siliceous coatings as well as clays. A
reclaim route serves the re-treatment of reject material (Fig. 1).
The geometallurgical model predicted that feed grade and particle size affect zircon recovery the most, whereas magnetic susceptibility is less important (Philander and Rozendaal, 2013).
Actual zircon recoveries average around 90%. Zircon losses to magnetic concentrate amount to approximately 3% of total feed. Trace
quantities of this particular zircon population either have ironbearing inclusions or are highly coloured, a zircon type known
for containing elevated amounts of iron, which increases its
magnetic susceptibility (Philander and Rozendaal, 2008). The bulk

Fig. 3. Heavy mineral density proles for the two concentrator plants PCP West and PCP East in relation to the densities of key heavy minerals. Note that a density cut-point
of 3.4 g/cm3 is used at the primary concentrator plants.

13

C. Philander, A. Rozendaal / Minerals Engineering 65 (2014) 916

Fig. 4. Actual data straddle the modelled liberated total zircon recovery trend
line, illustrating the important effect of liberation on PCP West recovery.

Fig. 5. The inverse relationship between pyroxene feed grade and ilmenite and
zircon recoveries.

of the zircon reporting to magnetic concentrate however is represented by ner particles that were probably entrained by coarser
ilmenite or garnet (Table 1). Particle overcrowding in the middling
section of the WHIMS circuit was found to occur frequently due to

a high abundance of garnet in the feed, causing signicant entrainment of zircon to magnetic concentrate. This type of entrainment
highlights the inherent weakness of conventional WHIMS to
recover ne non-magnetic rutile and zircon in the presence of high
quantities of magnetic minerals (Dobbins et al., 2009).
The majority of zircon not recovered to non-magnetic concentrate, register in the reclaim. These zircon particles are slightly
coarser than those in the non-magnetic concentrate (Table 1).
Coarser garnet is suspected to entrain the bulk of the zircon to
reclaim. The garnet content of the spiral feed, which is inversely
related to the zircon abundance, emerged as an important zircon
recovery driver (Fig. 6). Garnet, the most abundant gangue constituent in the feed has the ability to complicate the secondary concentration sub-processes, degrade both the magnetic and nonmagnetic concentrates and contribute to valuable mineral losses.
Almandine (Fe2+
3 Al2[SiO4]3), the chief garnet-type identied, has
the problematic tendency to respond positively across a range of
magnetic eld strengths due to variable iron and manganese content (Fig. 7). Nearly a third of total garnet fed to SCP reports as a
non-magnetic, whereas the rest contains sufcient iron and manganese and behaves magnetically similar to ilmenite. The heavy
mineral concentrate produced by PCP East usually contains less
than 2 wt.% garnet, whereas the ore treated by PCP West is highly
garnetiferous by comparison, and as a result heavy mineral concentrate could contain up to 30 wt.% garnet. Therefore, a 180 lm
vibrating screen is used to remove coarse garnet from PCP West
concentrate at the SCP. Less than 20% of garnet, however, is
removed in this process, whereas corresponding losses of ilmenite
and zircon are usually less than 3%.
Ilmenite recoveries to magnetic concentrate usually reach 80%.
Approximately 14% of ilmenite not recovered reports to reclaim
and another 4% ends up in the non-magnetic concentrate (NMC),
of which nearly 40% is altered, or is intimately associated with
other low iron bearing phases (Fig. 8). The magnetic susceptibility
of these particles is considered insufcient to be recovered by
WHIMS. A signicant proportion of the ilmenite in the non-magnetic concentrate is noticeably coarser, a known disadvantage of
WHIMS to recover coarse magnetic mineral particles (Table 1;
Dobbins et al., 2009). Similarly, a considerable proportion of the
ilmenite reporting to reclaim is coarse and an almost equal quantity is unaltered (Fig. 8). The recovery of ilmenite in the presence
of almandine, which exhibits similar magnetic behaviour, occurs
with a substantial penalty (Table 2).
In summary, ilmenite and zircon recoveries in the secondary
concentration process are essentially inuenced by the quantity
of garnet in the feed and not the ilmenite or zircon feed grades
as the geometallurgical model predicted. Other recovery penalties
such as magnetic susceptibility and particle size are conrmed to
be subordinate. The latest ndings highlight the difculty of conventional WHIMS to effectively separate heavy minerals that display extreme differences in magnetic behaviour due to variations
in chemical composition.

Table 2
Ilmenite and zircon recovery penalties determined for the four process areas. Values are expressed as percentages.
Recovery Penalties

PCP East

PCP West

SCP

MSP

Zircon

Ilmenite

Zircon

Ilmenite

Zircon

Ilmenite

Zircon

Ilmenite

Pyroxene feed grade

Garnet feed grade
Mineral liberation
Magnetic susceptibility
Particle size
Particle chemistry

0.2

0.5

0.0

0.2

2.0

0.0

12.2

4.6

0.0

10.9

7.5

0.0

3.1

0.3
2.9

9.4

3.0
4.3

2.1
12.7

4.2

Recovery potential

99.3

97.8

83.2

81.6

93.7

83.3

85.2

95.8

14

C. Philander, A. Rozendaal / Minerals Engineering 65 (2014) 916

Fig. 6. The inverse relationship between garnet feed grade and zircon recovery in
the spiral circuit of the Secondary Concentration Plant.

Fig. 7. Single grain chemistry of garnet particles illustrating that the magnetic
susceptibility of garnet is related to its iron and manganese content.

3.3. Mineral separation

The mineral separation process has the purpose of upgrading
the SCP concentrates to pure saleable products by removing
unwanted contaminants. Ilmenite is beneciated with magnetic
and high tension separators whereas the production of zircon follows a separate and more elaborate processing route that in addition includes wet separation stages (Fig. 1). Hot sulphuric acid
leaching is a paramount step to reduce zircon surface contamination. Ilmenite and zircon recoveries are mainly governed by product quality and to a lesser extent the feed grade (Philander and
Rozendaal, 2013).
Actual ilmenite recoveries are usually above 85% and were
found to be unrelated to feed grade. The current data conrm that
ilmenite recovery is mainly controlled by the intricate deportment
of SiO2, a key product quality penalty (Table 2). About 45% of total
ilmenite product SiO2 is hosted by liberated mineral contaminants
such as garnet and other silicates (Fig. 9). Garnet, the mineral

Fig. 8. A simplied ilmenite characterisation scheme, showing that altered

ilmenite, which is iron decient and display weaker magnetic susceptibility is
preferentially not recovered to magnetic concentrate. The total iron content
averages 32.7 wt.% and 44.1 wt.% for altered and unaltered ilmenite respectively.

contaminant targeted for removal is usually slightly coarser than

ilmenite in the product (Table 1). This type of coarse non-conductor misreporting is typical of high tension roll separators (Ziemski
and Holtham, 2005; Tripathy et al., 2010). Similarly, entrainment
explains the presence of other ne silicate contaminants. The bulk
of the ilmenite product SiO2 is intimately locked with the ilmenite
host as surface coatings and silicate inclusions (Figs. 9 and 10). A
signicant proportion of ilmenite has to be sacriced to reduce
SiO2 levels, creating a sensitive trade-off against recovery
(Fig. 10). The geometallurgical model attributes the contribution
of unliberated silica sources to process inefciencies and instead
utilises a product conformance measure, which accounts for the
Fe2O3, TiO2 and SiO2 specications to determine ilmenite recovery
potential. Current data indicate that the geometallurgical model
overestimated this quality aspect, perhaps being unable to simulate unspecied processes that could have helped with the reduction of coating levels (Table 2).
Actual zircon recoveries generally average around 70%, which is
signicantly lower than ilmenite recoveries. Titanium and iron, the
two most important penalty elements related to zircon product
quality, present major recovery constraints. Titanium deportment
is mainly in the liberated form of very ne rutile and FeTi oxides
that were most likely entrained. The deportment of iron, a more
signicant product quality penalty is relatively more complex.
About 25% of the iron is in the form of nely liberated garnet
and FeTi oxides that co-report with zircon to the product
(Fig. 9). The rest of the iron is locked as part of the crystal structure,
Fe-rich inclusions and surface coatings. Coupled substitution of
Fe3+ for Zr4+ causes the majority of the iron in the zircon product
to be structurally locked with the zircon lattice. Optically clear zircon, which by denition is colourless and free of inclusions is lower
in Ti, U, Th, rare earth elements and particularly Fe, compared to
coloured varieties that host these penalty elements in signicantly
higher concentrations above product specications (Fig. 11). Coloured zircon generally comprises about 30% of the zircon feed,
whereas less than 5% of it can be tolerated in the product. The zircon cleaner circuit is congured and operated to target the recovery of on-specication zircon to product, whereas the rest is
gradually rejected downstream. Collateral zircon losses could
impose a recovery penalty of signicant impact (Table 2). Clear zircon recovery usually approximates 90%, whereas the recovery of
coloured zircon is limited to less than 20%. The new geometallurgical model uses a zircon product conformance measure, dened by

C. Philander, A. Rozendaal / Minerals Engineering 65 (2014) 916

15

Fig. 9. The deportment of key penalty elements for the ilmenite and zircon products.

Fig. 10. Ilmenite grains showing common siliceous surface coatings and silicate mineral inclusions.

the Fe2O3 and TiO2 specications to determine recovery potential

(see Fig. 12).
In conclusion, this study proved that feed grade is not a recovery driver in the mineral separation process as the starter geometallurgical model assumes. Particle chemistry controls the recovery
of ilmenite fundamentally and imposes an even larger penalty on
the recovery of zircon.
4. Geometallurgical evaluation
The original geometallurgical model, which characterises individual ore types, indicated that only mineral grade, liberation,
magnetic deportment, particle size and particle chemistry

Fig. 11. Negative relationship between locked SiO2 content of ilmenite and
recovery.

meaningfully inuence ilmenite and zircon recoveries (Philander

and Rozendaal, 2013). Feed grade and particle chemistry were
ranked the most important recovery drivers. The current geometallurgical model that was adjusted to accommodate specic ore
blends reects similar relationships.
The present study conrmed all the ndings mentioned above.
Feed grades, however, were found to be proxies, rather than true
determinants of ilmenite and zircon recoveries in the primary
and secondary processes. Ilmenite and zircon recoveries are evidently controlled by the abundance of gangue minerals, particularly garnet and pyroxene in the feed, which are inversely related
to ilmenite and zircon grades. Overall, the gangue content represents the most detrimental constraint to ilmenite recovery. The
gangue content and zircon chemistry impose an equally signicant
penalty to zircon recovery (Table 3).

Fig. 12. Average iron content of the two varieties of zircon sampled from different
MSP streams in relation to the zircon product specication.

16

C. Philander, A. Rozendaal / Minerals Engineering 65 (2014) 916

Table 3
Observed and model recovery penalties for ilmenite and zircon in the overall context.
Values are expressed as average percentages. npd = not previously determined.
Recovery penalties

Feed grade
Gangue content
Mineral liberation
Magnetic susceptibility
Particle size
Particle chemistry
Recovery potential

Zircon

Ilmenite

Model

Observed

Model

Observed

11
npd
3
2
2
8
74

9
3
1
3
9
76

20
npd
7
5
3
14
51

17
6
3
5
3
66

The recovery potential observed for zircon is close to model

estimates, although there are slight deviations for the various
recovery penalties highlighted (Table 2). Some of these differences
are also discernible at process level. A few explanations are considered plausible for these deviations. Firstly, the methodology that
was used to adjust the original geometallurgical model to accommodate ore blends most likely was not optimised. Probably of
greatest signicance, though, are the recent ndings related to
the recovery impact of ilmenite chemistry, which was overestimated, mainly because of a model shortcoming.
The geometallurgical model under consideration is validated as
the observed data are convincingly aligned to the recovery relationships depicted by the model. Furthermore, these recovery relationships were implicitly explained in terms of the established
behaviour of heavy minerals in key separation processes. Deviations from the model are explainable and ndings from this study
represent the foundation of continuous efforts to improve the geometallurgical model.
It is recognised that actual ilmenite and zircon recoveries are
lower than determined potential. This presents enticing opportunities to improve mineral resource utilisation by creating a better t
between mineral characteristics and related behaviour, processing
technology and market requirements. The current ndings serve as
a valuable guide to help focus strategic interventions that aim to
optimise mineral resource utilisation for various blending scenarios. An extra comminution stage is considered a feasible concept
to improve mineral liberation. Separate ne and coarse spiral circuits and the latest developments in wet density and magnetic
separation are being investigated to deal with the negative effects
of gangue minerals. More effective alternatives to dry screening for
the removal of coarse gangue are being pursued. Opportunities to
address the issue of zircon quality are situated in optimised grade
control and value in use applications such as product
diversication.
5. Conclusions
The process mineralogy approach followed in this study provided a structured and systematic means to validate the integrity
of the Namakwa Sands geometallurgical model, which was
adjusted from a starter single ore type model to accommodate
ore blends. Overall, the recovery potential determined for zircon

is well aligned with model estimates. The impact of particle chemistry on ilmenite recovery, however has been overestimated in the
model, and as a result the recovery potential for ilmenite proved
higher than previously determined. The ndings of the current
study present substantial incentives to further rene the geometallurgical model, which could nd application in the broader mineral
sands industry.
Mineral liberation, gangue content and particle chemistry are
conrmed as the key recovery penalties. Several proposals to
address these constraints are mainly directed to better align mineral resource characteristics, processing technology and market
dynamics. Geometallurgy remains integral to Namakwa Sands
mine planning strategies that endeavour to optimise mineral
resource utilisation that is sustainable across life of mine. Currently, there is a methodical focus to investigate alternative technologies and research the latest developments in heavy mineral
separation to help narrow the gap between actual mineral recoveries and determined recovery potential.
References
Adams, M.D., 2007. Towards a virtual metallurgical plant 2: application of
mineralogical data. Miner. Eng. 20, 472479.
Burt, R.O. 2000. Gravity Concentration. Ullmanns Encyclopedia of Industrial
Chemistry.
Dobbins, M., Dunn, P., Sherrell, I., 2009. Recent advances in magnetic separator
designs and applications. In: Proceedings 7th International Heavy Minerals
Conference, What next, The Southern African Institute of Mining and
Metallurgy, Drakensberg, South Africa, pp. 6369.
Evans, C.L., Wightman, E.M., Manlapig, E.V., Coulter, B.L., 2011. Application of
process mineralogy as a tool in sustainable processing. Miner. Eng. 24, 1242
1248.
Grobler, J.D., Bosman, J.B., 2011. Gravity separator performance evaluation using
Qemscan particle mineral analysis. J. South Afr. Inst. Min. Metall. 111, 401
408.
Laplante, A.R., Spiller, E., 2002. Bench-scale pilot test work for gravity concentration
circuit design. Min. Proc. Plant Des., Practice Control Process 1, 160175.
Lotter, N.O., 2011. Modern process mineralogy: an integrated multi-disciplined
approach to owsheeting. Miner. Eng. 24, 12291237.
Mohanty, M.K., Palit, A., Dube, B., 2002. A comparative evaluation of new ne
particle size separation technologies. Miner. Eng. 15, 727736.
Pascoe, R.D., Power, M.R., Simpson, B., 2007. QEMSCAN analysis as a tool for
improved understanding of gravity separator performance. Miner. Eng. 20, 487
495.
Philander, C., Rozendaal, A., 2008. The geometallurgy of Namakwa Sands A South
African TiZr mine. In: Proceedings ICAM 2008 Conference. Automated
Mineralogy, Brisbane, Australia, pp. 459464.
Philander, C., Rozendaal, A., 2009. Mineral intricacies of the Namakwa Sands
mineral resource. In: Proceedings 7th International Heavy Minerals Conference,
What next, Drakensberg, South Africa, pp. 99105.
Philander, C., Rozendaal, A., 2011. The contributions of geometallurgy to the
recovery of lithied heavy mineral resources at the Namakwa Sands mine, West
Coast of South Africa. Miner. Eng. 24, 13571364.
Philander, C., Rozendaal, A., 2013. The application of a novel geometallurgical
template model to characterise the Namakwa Sands heavy mineral deposit,
West Coast of South Africa. Miner. Eng. 52, 8294.
Richards, R.G., MacHunter, D.M., Gates, P.J., Palmer, M.K., 2000. Gravity separation
of ultra-ne ( 0.1 mm) minerals using spiral separators. Miner. Eng. 13, 6577.
Tripathy, S.K., Ramamurthy, Y., Kumar, C.R., 2010. Modeling of high-tension roll
separator for separation of titanium bearing minerals. Res. Powder Technol.
201, 181186.
Walklate, J.R., Jeram, H., 2007. The development of a metallurgical owsheet to treat
tails. In: Proceedings 6th International Heavy Minerals Conference, Back to
Basics, Hluhluwe, South Africa, pp. 9197.
Ziemski, M., Holtham, P.N., 2005. Particle bed charge decay behaviour under high
tension roll separation. Miner. Eng. 18, 14051411.