The Southern African Institute of Mining and Metallurgy

Diamonds - Source to Use 2010

G Popplewell and B Baeletse

ORAPA 3 PLANT CONCEPTUAL DESIGN EVOLUTION IN ACTION (LET

THE ORE DICTATE THE PLANT THAT YOU BUILD!!)

G Popplewell- Fluor Corporation

B Baeletse - Debswana Diamond Company

SUMMARY

Commencing with the Pre-Feasibility Study (PFS) conclusions, the Orapa 3 process design
evolved through a phase of value-engineering studies. An overall re-evaluation of the originally
proposed process design was necessary both in order to address the interim increase in target
throughput from 9.8 Mtpa to 12 Mtpa as well as to increase capital efficiency. In the interests of
clarity, the PFS process design for Orapa 3 has been omitted from this paper, as it is no longer
relevant. Recognition is however due to those engaged in earlier phases of the Orapa 3 project
with respect to fundamental mass~balancing modelling; and ore and diamond characterisation,
which fanned the basis for the process design that has evolved from feasibility study activities.

The background context for Orapa 3 is of an operation expected to yield operating utilisation and
revenue improvements relative to the Orapa 2 operation. Delivering these without undue
penalties to capital and operating cost required a shift in thinking, trading excess installed
capacity for flexible circuit configuration.

The process design adopted is "layered", with the purpose of preserving Run Of Mine (ROM)
throughput by reducing in-circuit arisings - particularly to the Dense Medium Separation (DMS)
section - rather than simply installing additional DMS capacity on the expectation of low DMS
availability.

Elsewhere, the ability to monitor and maintain critical sizing activities - particularly desanding -
without impact on overall plant throughput is intended to motivate operators to avoid the
temptation to trade quality for quantity.

At present, since an overall dynamic simulation of the Orapa 3 operation has still to be carried
out, the design mass balance has been based on a relatively onerous combination of worst case
feed type with 100% front-line process capacity in all plant sections. This means that, whilst the
installed DMS capacity is based on routing 100% of sized scrubbing section product directly to
the DMS, and with one DMS module always unavailable, the High Pressure Rolls Crushing
(HPRC) capacity is based on allowing for 50% of this stream to be instead routed first to the
HPRC section, at no more than 75% of maximum roll speed for the two units installed. This is
an obvious "belt and braces" approach. Following the dynamic simulation exercise (currently in
progress), it is likely that a less conservative approach will be taken. This will not affect the
conceptual design of the process plant, being mostly an exercise in refining the number of DMS
modules to be installed, and possibly reducing slightly the size of the HPRC roll units. Both of
these will have positive capital and operating cost impacts.

Page 215
The Southern African Institute of Mining and Metallurgy
Diamonds - Source to Use 2010
G Popp/ewell andB Hae/else

DMS capacity, despite being split into coarse and fiaes streams, consists of identical modules.
Two of the nine modules are set-up to receive either coarse or fine feed, the only difference
being that fines modules are rated at lower capacity than the same modules treating coarse feed.

Final recovery section capacity is based on entirely wet primary diamond recovery technology.
This greatly reduces both the cost of drying. a large amount of recovery section feed and the dust·
management issues associated with dry recovery technology.

In contrast to the Orapa 2 operations, a scavenginglauditgrease belt section is included as a

diamond recovery "goal-keeper" and to provide a process assurance function.

Key Features of Orapa 3 Plant

Key features ofthe Orapa 3 plant are summarised as follows, and discussed further in Section 6:

Main Plant Sections

• Parallel scrubbing and screening trains, which can also be fed asymmetrically to emulate
a primary/secondary scrubbing configuration
• Single secondary cone crusher, solely for HPRC feed preparation
• Studded roll HPRC units, in a combined tertiary/recrushing duty
• Desanding section separated from the scrubbing and screening section
• Degritting section provided to reduce overall water consumption and optimise thickener
performance and operability
• High-density thickeners of conventional design
• Simplified overall process water distribution system
• Pump fed DMS cyclones and dilute medium densification
• DMS capacity split into coarse, fines and swing modules, with however all modules of
identical design
• Steady process water head tank for DMS sinks screens sprays and pump flushing
• Wet primary X-ray sorter based recovery section, with all recovery sorters of same basic
design
• Automated grease belt back-up for fine X-ray sorter rejects

Recovery Section
•
• Streamlbatch system allowing high utilisation of units processes while not excessively
raising the height of the building.
• Modern but proven wet X-Ray sorters perform recovery of material with low pop-factors
and grams per ejection.
• Dry re-concenlTation systems fed by external volume feeders at low rate from 3 to 30kglh
produce high dianlOnd by weigh concentrates free of dust.
• Grease scavenging of prepared X-ray tails using advanced grease belt systems
incorporating automated control and cleaning systems.
• Control of unit processes from a control room to allow a hands off operation.

Page 216
The Southern African Institute of Mining and Metallurgy
Diamonds - Source to Use 2010
G Popplewell and B Baeletse

1.0 INTRODUCTION

The point of departure for the events discussed in this paper was a critical review of the
previously proposed design and related cost-reduction suggestions put forward by others. The
volutionary process represented by the Feasibility Study was undertaken by Fluor with the full
engagement of Debswana's project team. Workshops were held with key stakeholders, including
Orapa Mine management and De Beers Project Assurance, at critical decision points.
Value engineering studies were carried out to evaluate potential opportunities for capital cost
reduction, although at all times within the context of minimising risk to plant utilisation, which is
benchmarked against an international standard higher than that currently achieved by the Orapa 2
Plant. This presented a challenge and was a driving force towards the adopted "layered" design
philosophy where run of mine (ROM) capacity is protected at reduced DMS section capacity
through the ability to selectively crush DMS feed or reject coarse DMS floats.

2.0 CURRENT DEVELOPMENTS - JANUARY 2010

Following completion of the feasibility study phase conceptual process design, a dynamic
simulation exercise was carried out to confinn in-circuit material storage requirements and
indicate if further DMS section capacity reduction might be possible without compromising the
overall goal of Orapa 3 to achieve 12 Mtpa throughput of Orapa AK1 primary crushed ore. A
capital cost reduction workshop was held in December 2009 to review the outcome of the
dynamic simulation exercise and detennine which recommendations would be incorporated into
the plant design. As expected, significantly reducing the capacity of the feed silo, from 12,000
tons to nominally 2,000 tons, presented no throughput risk, nor did the removal of one DMS
module. These measures were adopted and engineering is currently in progress to detail the
changes to support cost estimating.

An additional option identified at the workshop, outside the scope of the dynamic simulation, is
to remove the dedicated Orapa 3 final recovery section, instead upgrading the existing CARP.
Although work is currently in progress to define the CARP modifications, which are based on
testwork carried out in parallel to the Orapa 3 design development, order of magnitude costing
indicates that it is an attractive way forward. Consequently, without changing the process
flowsheet, the overall plant plot-plan was modified to capture the benefits of excluding
constraints imposed by the final recovery section. A significantly more compact layout was
achieved without compromising access requirements. Along with the other cost reduction
concepts, this has now become the basis for the feasibility study engineering development.

Page 217
The Southern African InstitUle of Mining and Metallurgy
Diamonds - Source 10 Use 2010
G Popplewell and B Bae/else

I
I
I
I
I
I
I
I
__ _ _ _ _ _ _ _ _ _ _ _ _ _, - .
'~" . a- -'i!tl~..,.

I
I
I
"

-
......

I
I
I
I
I
I
I
I
,- - - --~:-:::;~:~-::;- ~- - - - - ~--~,~!"--:
I
I
......
w."""

I
I
L ___ _

Page 218
The Southern African Institute of Mining and Metallurgy
Diamonds - Source to Use 2010
G Popplewell and B Bae/else

3.0 MASS BALANCE DEVELOPMENT

3.1 Main Plant Mass Balance

Prosim is a windows-based dynamic flow sheet simulator which was developed for mine to mill
simulation by Metso Minerals in collaboration with Debswana Diamond Com pany. It uses
MinOOcad software. In ProSim, the ore characteristics are key inputs resulting in a suite of
treatment recipes as outputs. The following are important characteristics of Prosi m;

• Uses state-of-the-art popUlation balance models - applies law of conservation of mass.

• TIle models are calibrated with plant data and validated w ith other independent plant data
and equipment supplier specifi cations.
• Sometimes parameters can be calculated with the results from discrete e lement modelling
and computational fl uid dynamics simulations in High Fidelity Simulation (physics
based) which ensures high accuracy.
• Has the capability for running up to 200 times real-time, enabling the running of a year's
mine plan in a matter of hours.
• Iteratively reduce mine plan into treatment recipes and highlights production bottlenecks
or constraints in a flowsheet.

Mass balance modelling for the Orapa 3 process plant was undertaken by the Orapa Expansion
Project process team. During the pre-feasibility study~ models were developed using both Prosim
and the De Beers Diamond Wizard tool. The models were calibrated for the Orapa 2 flowsheet
using data obtained from Orapa 2 operations, including the results of the 100% basalt breccia
testwork carried out at Orapa Mine. Information from HPRC and scrubber testwork carried out
in Getmany was also incorporated into the model. Following an exercise to compare Prosim and
Diamond Wizard models, it was concluded that the model outputs were sufficiently convergent
for the Prosim platfonn to be used for further investigations, with however some concerns
regarding the prediction of thickener feed arisings, which the Prosim model under estimates as a
result of the relatively unsoph isticated modelling approach in this area. Fortunately sufficient
data was available from Orapa 2 operations to provide valuable input into the design for this
stream.

At the conclusion of the PFS, the model had been used to define the mass balance arisings for the
target of 9.8 Mtpa for the PFS proposed flowsheet and as a basis for process equipment sizing.
Dynamic simulation of the plant flowsheet and materials handling facilities was carried out to
detennine surge capacities required between process sections.

The Prosim model served from the inception of the feasib ility study as the basis for eva luation of
the increased target throughput of 12Mtpa., superimposed on tlowsheet development driven by
the need 10 minimise the capilal cost.

In addition to the usual envelope of plant feed types, represented in the case of Orapa 3 by the
lower (10%) and upper (47%) limits ofbasaJt breccia content of the ore feed, the proposed Orapa
3 tlowsheet introduces an internal recirculation stream, where a portion of the 6 to 25mm

Page 2 19
The Southern African institute of Mining and Metallurgy
Diamonds - Source to Use 2010
G Popplea.vell and B Baelelse

scrubber product 110nnally routed directly to the DMS section, can be bypassed to the HPRC
section. The purpose of this is to allow the process plant to maintain ROM throughput,
particularly of high basalt content feed, under such circumstances as lower than pl anned DMS
section availability, or alternatively to take advantage of planned available DMS capacity to
increase ROM throughput. This feature, together with the variable mid-cut-off size and the
option to route 4 to 6mrn material to either the fines or coarse DMS sections, added additional
dimensions to the mass .balance modelling. The Prosim model was revised accordingly to reflect
the range of scenarios being considered, so as to provide an understanding of the effectiveness of
the proposed i'nternal recirculation of 6 to 28mm material feed on decoupling the DMS feed
arisings from the basalt content of the feed.

The mass balance summary outputs shown in Tables 3.1.1 and 3.L.2 illustrate the range of
process pJant section feed arisings for progressively finer mid-cut-off sizes of 10 mm. 8 mm and
6 mm, and the limits of the 6 to 28mm material bypassed to the HPRC section.

Table 3.1.1- Mass Balance (dry t/h) for 0% bypa .. of 6 to 28 mm material to HPRC

Table 3.1.2 - Mass Balance (dry t1h) for 50% bypass of 6 to 28mm material
to HPRC

As can be seen from comparison of the total DMS feed arisings in matching columns, the ability
to bypass 50% of 6-28 mm scrubber product to tbe HPRC has a dramatic impact.

To narrow the mass balance and avoid costly over-design, a dec ision was taken by Debswana to
set the mid-cut-off size (i.e. the recrush size) at 8 mm. During the process flowsheet and
equipment selection exercise, it was also decided that the 4 to 6mm middlings fraction would be
routed to treatment in DMS modules set up to optimise separation of the 1.6 to 6mm feed size
range. The effect of a variable bypass ratio of 6 to 28mm material to HPRC on high basalt

Page 220
The Southern African Institute ofMining and Metallurgy
Diamonds-Source 10 Use20JO
G Popplewell and B Baeletse

breccia feed was modelled for these selections, and a benchmark added to represent low basalt
breccia feed for comparison. This mass balance is shown in Table 2.1.3. Worst case arisings
represented by this mass balance on a stream by stream basis have been used as the design
criteria for equipment selection for the Orapa 3 process plant.

As discussed above, the predicted thickener feed arisings are considered too low. and a value of
750 dry tIh has been specified in the design criteria. Similarly. the recovery section design
criteria feed rate has been based on worst case arisings of DMS concentrates rather than the
nominal values given in Table 3.1.2.

Table 3.1.3 Mass BaJance (dry tlh) for variable. -;8 bypass of6 to 28mm
malcrl.llo HPRC

-
ConiJIned CMS CQo.oeilbale
1413.2
947.1
1413.2
947
1413.2
9tiB ..,.
14'3.2 1413.2
946.3
14132
946
1931
128le

1t can be concluded from the tables above that the ability to bypass 6 to 28 mm scrubber product
to the HPRC seclion, within acceptable limits, can be used to eliminate the variation of scrubber
feed. HPRC section feed and DMS section feed arisings relative to basalt breccia content of the
plant feed. At least as important. is the fact that 1.6-28 mm carat content of DMS concentrate
from high basalt breccia feed is unaffected by bypassing 6 to 28mm scrubber product to the
HPRC, up to the imposed 50% limit. Note the basalt breccia content is approximately 0.27 ctlt
and kimberlite 0.8 ct/t. The headfeed carats are for the total content curve, from which carats •
< 1.65 mm are not recovered into DMS concentrate.

The wider process design implications of this feature are discussed in section 3.0 below.

3.2 Recovery Section Mass Balance

Mass and particle size distribution envelopes were generated based on production data for Orapa
I and 2 over several years. This generated a 10.4 tIh peak tonnage based 011 a OMS yield
equivalent to 0.61 % of ROM at a ROM feed rate of 12 Mtpa and a plant utitisation of 83%.
Particle size distributions from fine to coarse of:

Page 22 1
The Southern Aftican Institute of Mining and Metallurgy
Diamonds -Source to Use 2010
G Popplewell andB Baeletse

• 50%-27% +1.6 mm-3 mm

• 35%-30% +3 mm-8 mm
• 15%-43% +8 mm-32 mm

The plant design allows for the presence of -1.6 mm grit in the feed to the recovery as well as
grit developed in the recovery process but no reduction is applied to the recovery processing in
the X-Ray streams.

The material mass distribution is divided equally amongst the various available streams and is
recombined at reconstituting points such as the scrubber feed and tailings.

X-ray yields are based on luminescent material present but with allowable factors for the
presence of non diamond luminescent materiaL The factors used can indicate the maximum
presence of non diamond material based on the flow sheet without overloading the re-
concentrate circuit.

The water balance for the plant allows for a constant inflow of potable (desalinated) water to the
X-ray circuits but with raw water make up for level control and tapping of process water from
the DMS concentrate delivery. The grease process water circuits utilise recycled process water to
keep process water inside the temperature control envelope. The water balance makes use of
considerable recycled process water in transfer and spray systems with dirty (grits) process
watering being bled from the system via the scrubber product and safety screen.

Diamond size distributions were generated by Debswana operations and an additional 10%
allowance in total carats was made based on the extra liberation expected from the plant front
end. This equated to a head feed grade of 77 CPHT.

Metallurgical testwork on magnetic separation application to Orapa concentrates indicated that

there is potential for magnetic separation to achieve a significant reduction in X-ray sorter feed
arisings of a minimum 40% in all the size ranges, with no loss of magnetic diamonds (mainly
boart) from Drapa concentrates.

4.0 FLOW SHEET DEVELOPMENT

4.1 Baseline Flowsheet at commencement of Feasibility Study

The conceptual flowsheet developed during the PFS by Debswana adopted features that more
recently have become accepted within the De Beers group. In addition, lessons learned from the
operation of the Orapa 2 plant proved a valuable guide in flowsheet development. At a
conceptual level, these can be summarised as follows:
• Flexibility to manage changes in ore characteristics with minimal impact on ROM
capability and without excessive over-design.
• Advance HPRC to a combined tertiary/recrushing role, to reduce both the overall DMS
feed arisings and the impact of variations in ore competency thereon, and to liberate
diamonds as soon as possible.
• Combined primary and secondary scrubbing duties.

Page 222
The Southern African Institute of Milling and Metallurgy
Diamonds - Source to Use 2010
G Popplewell and B Baeletse

• General elimination of standby equipment and streams unless "mission-critical" .

• Minimising the number of drives required to function in a particular sequence, by use of
surge points and passive.standby.
• Inclusion of a degritting section to protect the thickeners from fast settling solids that
require no thickening, and to reduce overa ll water consumption.

4.2 Hooks and the Core Plant

Advancing from the general flowsheet concepts, the concept of the "Core Plant" was partially
considered at PFS stage. The "Core Plant" principle involves matching the priority of stream
processing to the value of the stream. This means that the flow sheet and associated layout should
allow for low value streams to be selectively either temporarily stored o r discarded where ROM
throughput is threatened. The principle examples of such streams are those containing:

• Small diamonds, in the case ofOrapa < 1.6 mm, which are relatively abundant but which
vary in value·significantly with trends in the global diamond market.
• Large diamonds, in the case of Orapa >28 mm, which are very rare and extremely
variable in value.
• Locked diamonds in the coarse DMS floats stream.
• Easily identifiable coarse waste that can be cost-effec tively removed to reduce
downstream processing of essentially barren material.

In each of the above, the contribution of the di amonds to overall revenue is relatively Jow, often
unpredictable. and the capital and operating cost associated with recovering the diamonds is
relatively high

With the exception of the coarse DMS floats stream, which was included in effect as a core
stream, the other low value streams were excluded from the intended Orapa 3 design. In each
case, consideration was given to the manner in which such streams could be generated at some
future point with minimal disruption to ongoing core operations.· Whether this will eventuate is
not possible to predict. However, in each case the relevant additions and modifications will only
be made if a business case can be proven. Since these streams will always remain relatively low
value streams compared to ROM ·ore, implementation should be such that bypassing such
sections can be easily accomplished if equipment failure etc threatens ROM throughput targets. ,
4.3 Capex Reduction - PrefeJIsibility Study

A Capex reduction initiative (Orapa 3 - Lean Option) carried out by Debswana during the
concluding period of the PFS evaluated a variety of potential options for simplifying the process
flowsheet and plant layout.

An additional series of value engineering studies was undertaken by Fluor during a basis of
design review at the beginning of the feasibility study stage. Alternatives were cons idered,
including those proposed in the O rapa 3 - Lean Option, and recommendations presented to
Debswana for approvaL

Page 223
The Southern African Institute of Mining and Metallurgy
Diamonds ~Source to Use 2010
G Popplewell and B Baeletse

Changes identified in the Orapa 3 - Lean Option which were adopted in the feasibility study
flowsheet following Debswana approval included:
• Reduce scrubber retention time to two minutes.
• Install a single stream comminution circuit. This referred to the elimination of the
previously proposed dual stream secondary crushing section, combining the secondary
crushing duty into a single unit, fed by >60 mm material from the scrubbing section and
discharging <60 mm crushed product onto the HPRC feed conveyor.
• Single HPRC feed bin. The feed from all sources to the HPRC section is combined in a
single bin, from which the two HPRC units are fed by variable speed belt feeders.
• Pump fed DMS cyclones instead of gravity fed DMS cyclones,
• Replacement of paste thickening with high-density thickening.
• Eliminate emergency DMS tailings node,
• Changes identified in the Orapa 3 - Lean Option which were not adopted in the proposed
flowsheet:
• Perform Desand screening in the scrubber section. This was not adopted because of the
inability of the proposed screens to achieve the process duty, and hence the proposed
concept was considered unworkable as presented. In addition, a philosophy was adopted
that critical screening duties should be upper-deck duties, promoting operator vigilance
and relatively simple maintenance, The duty proposed in the PFS was a bottom deck
duty on a double-deck screen.
• Scrubber feed stockpile simplification. This was partially adopted in that the
slewing/lufting stacker conveyor was eliminated. However, largely as a result of the
impact of dust control measures required to prevent the inundation ofOrapa 2 with dust
generated on the Orapa 3 stockpile, a 12,000 tonne concrete silo was adopted, similar to
the existing silo servicing Orapa 1 and Orapa 2 (and in future, also Orapa 3)

Changes identified in the Orapa 3 - Lean Option which were not applicable to the Fluor
proposed flowsheet:

• Eliminate one series ofDMS feed storage silos feed conveyor systems
• Reduction ofDMS feed storage silo capacity
• Eliminate the individual DMS feed conveyors

The above were rendered obsolete as a result of the decision to adopt a split DMS arrangement,
with separate coarse and fines DMS modules.

4.:4 Feasibility Study Flowsheet

Theflowsheet adopted as the basis of design for the Orapa 3 process plant differs in several
aspects from the flowsheet proposed at pre-feasibility study stage.
• Provision of the ability to bypass a portion of the scrubber product in the 6-28 mm size
range to the HPRC section. The reason for this is to trade-off, if necessary, a minimal risk
of damage to liberated diamonds against the ability to maintain ROM throughput with

Page 224
The Southern African Institute of Milling and Metallurgy
Diamonds - Source 10 Use 2010
G Popplewell and B Baeletse

reduced DMS section availability. Alternatively, this feature can be utilised to increase
ROM throughput capability, assuming DMS availability is as planned, up to the limit of
the capacity of the HPRC section. This is a somewhat more controversial application, as
it could be construed as over-design. A more rational evaluation would be to consider the
bypass facility as allowing the available capacity of the key plant sections to be balanced
to pennit the maximum ROM treatment possible under any particular set of
-circumstances.
• The DMS section has been configured with five coarse moduLes (6-28 mm), two fines
modules (1.6-6 mm) and two swing modules. The swing modules can be used for either
coarse or fines feed streams, but are configured as fines modules with regard to cyclone
operating parameters.
• The envisaged scenario, with respect to tbe capabilities provided in the tlowsheet, is
potentially as follows:
• Worst case ROM type (47% ROM}.t 1,715 dry t/h with no bypass of 6-28 mm scrubber
product can be accommodated using all five coarse modules, two fines modules and one
swing module operating on coarse feed. One DMS module (which would have to be
module 7, 8 or 9) is available for maintenance etc.
• If a second DMS module becomes unavailable. bypassing 20% of the 6-28 mm scrubber
product allows the ROM feedrate to be maintained at L,715 dry tIh. The second
unavailable rnodulecan be any module.
• If a third DMS module becomes unavailable, bypassing 30% of the 6-28 mm scrubber
product allows the ROM feedrate to be maintained at 1,715 dry t/h, In this case, of the
remaining six modules, three out of modules six to nine must be available, or ROM
throughput will fall towards 1,500 dry t/h to avoid compromi sing fines DMS mediwn:
ore ratio (or this could be temporarily compromised, which in practice is probably 'he
logical selection).
• The ultimate selection of 50% bypass of 6·28 mm scrubber product to HPRC potentially
allows 1,715 dry tlh of 47% basalt breccia feed to be accommodated in only five
modules, although if only two of these are available for fines treatment (i.e. only two of
modules six to nine), then a similar effect on fines DMS medium: ore ratio would be
encountered

As can be seen, the bypass facility affects almost eXClusively the arisings to the coarse DMS, and
will ultimately allow full ROM capacity to be accommodated in only three coarse DMS modules
(i.e. three out of modules one to seven). Beyond the 50% allowable bypass of 6-28 mm material
to HPRC, maintaining ROM capacity can tben only be achieved by selectively discarding coarse
DMS floats to rejects which is provided for on a module by module basis. This impacts largely
the fines DMS arisings, thus balancing the required fines modules versus available fines modules
six to nine. It is only at this point that revenue is potentially at risk, although ROM revenue
contribution is much greater than that from coarse DMS floats.

Page 225
The Southern African Institute of Mining and Metallurgy
Diamond.. - Source to Use 2010
G Popplewell and B Baeietse

4.5 Recovery Section

The flow sheet was developed following the basic unit process requirements as stipulated by
Debswana:

4.5.1 Sizing and Storage:

The concentrate from the DMS requires sizing into fractions suitable for further processing and
surge is required for both decoupling the recovery from the DMS as well as providing holding
for batch processing.

4.5.2 Transfer-to further treatment

This is achieved using jet pumplbooster pump hybrid systems due to the heights required, to
avoid excess tip speeds in the centrifugal pumps and to ensure reliable delivery from the bottom
of gravel filled bins. Slurry streams are dewatered prior to processing and weighing for
accounting purposes.

4.5.3 Bulk Magnetic Reduction:

l
BaFe and NdFeB magnetic separation units will be included in the flow sheet post the successful
bulk reduction testwork done on Orapa material.

4.5.4 Wet Primary X-ray systems:

X-ray sorters are per Debswana's requirement for wet and dry CWX1l8CD and CDX118CD
(ModRUP) sorters from Debtech. The X-ray sorters will be supplied with desalinated water to
avoid problems associated with dissolved salts crystallising on the sorter widows.

Wet X-ray treatment, in rougher/scavenger configured sorters, is split into a dedicated fines
stream, and batch middles/coarse streams.

X-ray tailings report to grease scavenging while the concentrates are re-concentrated to meet the
at least 50% diamond by weight (DBW) required for shipment.

4.5.5 Dry Re-concentration X-ray:

Dry re-concentration X-ray sorting was selected primarily as, unlike wet re-concentration, the
dry route was capable of achieving the required 50% DBW and hence eliminated the need for
further retreatment in a dry single-particle X-ray sorter.

This required drying and cooling step between the wet primary and dry re-concentration stages
was provided by vibrating pan infra~red driers with post-drier-cooling air to reduce the
concentrates temperature sufficiently to preserve diamond X-ray fluorescence response.

4.5.6 Weighing and packaging of X-ray concentrates

The X-ray concentrates will be collected and shipped to the Jwaneng Fully Integrated Sort House
(FISH) for final recovery. To fit the current system this will be by means of Debtech dock locks
and a system for optimally and reliably filling these is currently being developed.

Page 226
The Southern African inslitute of Mining and Melallurgy
Diamonds - Source to Use 2010
G Popp/ewel/ and B Bae1etse

4.5.7 Grease Scavenging of X-ray Tall,

Grease scavenging of X-ray fine tails is included, to increase overall diamond recovery and as an
auditing function on X-ray sorter performance.

Effective grease recovery of diamonds requires a light attritioning step to prepare the surface of
the diamonds for grease recovery. This is accomplished by using a conventional wet rotary
scrubber. The scrubbed product is scalped at 12mm, with the -12mm stream screened into two
size-fractions, which are fed separately to the grease belt to ensure acceptable material flow
control. Material larger than 12mm is returned to the HPRC section in the main plant both to
liberate any additional diamond revenue and, in the fare event that any large diamonds are lost to
X-ray rejects, to ensure that these are locked into the circuit.

Final degreasing of diamond-bearing grease concentrates will be either intental to the grease belt
unit or manually accomplished as an external operation. An internal automated inspection and
cleaning system is preferable to satisfY security protocols and is currently under development.

4.5.8 Material Transport and Transfer

Gravity will be used for the most part to transport material from unit process to unit process with
exceptions for entering feed into the plant for the first time and introducing material back to the
grease circuit All attempts were made to reduce impact velocities in gravity transport by
reducing drop heights by either height reduction and/or cascade chutes.

Jet pump systems are used either by themselves or as part of hybrid systems for transfer of
material from loaded bins with IiUle vertical height required under the bin and a reduced chance
of blockage.

Vacuum systems will be used for the return of re-concentration tails and spillage but not the
transfer of first stage X-ray concentrates due to the difficulty of contr_olling particle velocity
across the duty envelope and hence potential for unacceptable diamond damage.

4.5.9 Accounting

Mass accounting of material in the recovery will be achieved by either batch weighing systems
of variable accuracy depending on the stream to be weighed. Belt scales will be used for interim
accounting and process control and final tailings accounting.

Mass accounting in individual process streams for process control is achieved by monitoring the
throughputs of the volume feeders though these generate implied rates rather than rates
detennined from feedback.

Accounting of grease concentrates and to a lower degree X-ray concentrates will be handled by
the SCADA system. SADAS was omitted due to the functionality required not being that of a
full sort house.

Page 227
The Southern African Institute ofMilling and Metallurgy
Diamonds - Source to Use 2010
G Popp/ewell and B Bae/else

4.5. 10 Process stream requboernents were based on the following;

The total tonnage to the plant w ith the supplied particle size distribution envelope and utilisation
of 81 %. In addition to other details this limited the top particle size of 35 mm eliminating the
need for overs ize scalping in the recovery.

The feeding arrangement of a hybrid batch approach with a series of streams being allocated to
fmes only while middles and coarse shared a series of streams.

The diamond size, frequency distribution and grade which were used to determine yields and
load on final re-concentration sorters. Factors were applied to this based on sorter supplier
information and additionallwninescent material and fines.
Vendor supplied throughput data based on general throughputs achievable fOT the various unit
processes but without the benefit of specific material test work. An exception to this is the bulk
magnetic reduction test work which has just been completed and whose outcome will be used in
the detailed design phase.

5.0 B01T.L ENECK ANALYSIS

Detailed bottleneck analysis for the Orapa 3 process plant awa its the planned dynamic si mulation
exercise. However, a preliminary conceptual bottleneck analysis was undertaken to indicate the
achievable overall plant throughput as constrained by various sections of the process plant. The
key impacts on plant throughput are associated with the availability of DMS treatment capacity,
the extent to which the plant feed is infested with low grade, but hard, basalt breecia, and the use
made of the abi lity to bypass DMS feed to the HPRC section. The generalised bottlenecks are
shown in Table 4. 1 -Orapa 3 General Plant Capacity Constraints and the extended DMS specific
bottlenecks in Table 4.2 - Ompa 3 DMS Specific Plant Capacity Constraints.

Page 228
 I•,
c;)b~ ~•
':'l:l s· !'\l1;.00) ~
<::l ::!
~ cc' ....
~ ~ ~
(0; ..
~
~ I ...,
- a. ;:-. ,..
Table 4.1 - Orapa 3 ~ General Plant Capacity Constraints ~ ~ ~
- ~:... ~
§ :: ~ c
~ ~ (:;'
~ !'\l e u
-
b::I 0- .... ct..
~ 't
3
~~ S
!'\l ~
~. I Y1!
,~.
'm
,~
..,
~
,- H I'
et N r:
"" !'\l C f:) 't
,.... ~
..."",., ~
=~
,., " "N
,, ~
<:::l-Q., -g
,~.
,~
,~
"'" ,m
~
~,
,~
= ""'" ,""
-
'''. ~ 23. 2 ;;.
1:; "
-... ,-.... ..
.a <n, ,"'"
,,~
"" , ~.
."
,m 17.~
~.
~, ,~
.... ...
"" 11\$
""" ""
. - n ,- ""
Notes: _ aopOCttltf _ u.... lI ..... .,. _ _ _ • at. _ _ _ . with . n, ono OMS rnodlM "" m _ ",," ~
R..",.,.ry _ _ ~ lV dol...... b1 """'" of O,""-(l/0(8p. 2 _ ~_. 1* prO<I\ICIior1 .. "",<Is

1;
~ l2'
~
Bl Table 4.2 - Ora'pa 3 - DMS Specific Plant Capacity Constraints ~
~

2 500 2 426 4 926 594 11 33 942

3 750 2 426 5 1176 892 1541 141 2
3 750 3 639 6 1389 892 1700 141 2
4 1000 2 426 6 1426 11 89 1541 1804
4 1000 3 63. 7 1639 11 89 2266 1883
5 1250 2 426 7 1676 1486 1541 1804
5 1250 3 63. 8 1889 1486 2312 2354

n
The Southern African 1nstitute of Mining and Metallurgy
Diamonds -Source to Use 2010
G Popp/ewel/ and B Bae/else

The baseline design case assumes that a feed of 47% b..1salt breccia w ill be treated with no use of
the facility to divert the 6-28 mm portion of the DMS feed to the HPRC section. This has dri ven
the select ion of nine DMS modules, of w hich anyone is assumed to be unavailable at all times.
As can be seen from Table 8.5.1, the plant design is otherwise aligned quite well with the
required capac ity, noting that in the desanding section particularly the spare capacity is expected
to be t'imployed to allow individual screens to be stopped for re:gular inspection on a regular
basis.
Table 4.2 shows that, by employing the DMS feed diversion faci lity, it is possible to achieve the
design plant throughput even on ores with high basalt breccia content with as few as seven
modules. Following the dynamic simulation exercise this will be evalu ated more thoroughly and
the number of DMS modules to be installed confinned.

6.0 PROCESS DESIGN FEATURES OF THE ORAPA 3 PLANT

6.1 Scrubbing and Screening

The Orapa 3 scrubbing and screening section has been designed to incorporate the advantages of
parallel primary scrubbing with those of split primary/secondary scrubbing. Together with the
ability to divert 6·28 mm scrubber product to the HPRC circuit. this configuration provides a
hig hly flexible means of responding to changes in ore type and downstream equipment
ava ilability.

Convent ional scrubbing section arrangements in use in large diamond recovery plants follow one
of the following configurations (although there are variations in detail if not in concept):

6.2 Complex scrubbing, high DMS feed arisings, high scrubbing capacity required
All ROM ore is fed to primary scrubber(s). Washed ore in the size range for diamond recovery is
routed to the DMS circuit. Oversize from the primary scrubbers and coarse floats from the DMS
circuit are crushed in a combination of cone and HPRC crushers and the crushed products re-
scrubbed in separate secondary scrubbers. In this configuration, th e primary scrubbers are open·
circuit and the secondary scrubbers in closed-circuit. Typical examples are Orapa 2, Venetia, and
Diavik.

6.3 Simple Scrubbing, high DMS feed arisings, low scrubbing capacity required
All scrubbing duties are combined in a single scrubbing circuit, with crushed oversize and
recrushed HPRC product returning to join ROM ore. Washed ore in the size range for diamond
recovery is routed to the DMS circuit and coarse DMS fl oats reporting to the HPRC circuit. The
scrubbing section is in closed-circuit. A typical example· is Kimberley CfP.

6.4 Complex scrubbing, low DMS feed arisings, high scrubbing capacity required

ROM/Secondary crushed ore is fed to primary scrubbing with oversize routed to HPRC. Washed
ore in the size range for diamond recovery is also routed to HPRC and the HPRC product reports
to secondary scrubbing. Only washed ore in the size range for diamond recovery from the

Page 230
The Southern African ImIlitute 0/ Mining and Metallurgy
Diamonds - Source to Use 20 J0
G PoppJewell and B Bae/else

secondary scrubbing circuit reports to the DMS circuit. Coarse DMS floats are returned to the
HPRC. In this configuration, primary scrubbing is in open circuit. secondary scrubbing in closed-
circuit. The principle current example is Ekati.

The obvious configuration missing from the above is to combine simplified scrubbing with low
DMS feed arisings. This is the configuration embodied in the Orapa 3 plant:

6.S Simple Scrubbing~ low DMS feed arisings, low scrubbing capacity required

All scrubbing duties are combined in the same two scmbbers. Washed ore in the size range for
diamond recovery can be routed either to DMS or diverted to HPRC. DMS coarse floaffi are
normally routed to HPRC but can also be discarded to tails.

ROM ore can be preferentially fed to one scrubber and HPRC producl to the other scrubber. Any
combination of ROM orelHPRC product can be selected for feed to one scrubber, with the other
scrubber by default receiving the remainder. Therefore the scrubber receiving preferentially
ROM ore (pseudo-primary scrubbing) can be selected to deliver washed ore in the size range for
diamond recovery to HPRC, whilst the scrubber receiving preferentially HPRC product (pseudo-
secondary scrubbing) delivers washed ore in the size range for diamond recovery to the DMS
circuit. Either scrubber can assume either role, or the balance of HPRC product relative to ROM
ore can be assigned between scrubbers on the basis of maximising scrubbing efficiency.

The flexibility provided by the combination of asymmetric scrubbing feed and diversion of some
6-28 mm material to HPRC pre-DMS is shown in Table 6.1.1 - Effects of Asymmetric
Scrubbing and Bypass on 6-28 mm Ore to HPRC. The salient features are:
• 8.4% increase in scrubber feed (pseudo-secondary only)
• 7.5% increase in HPRC feed
• 34% reduction in feed to DMS

The appropriate capacity allowances are within the design safety margins for scrubbing and
HPRC. The ability to reduce DMS feed by 34% on the worst case ore type (47% basalt breccia)
whilst maintaining ROM ore throughput is key to delivering consistent plant perfomlance iD
periods of challenging DMS operation.

Page 231
 "~

"o,,
,
ClO~ 'C
- is- (!;, ~
'" ... V:l r
~ 50"
... . ..
f I} ~ ~
(!;, I ~ ~
- o· ::t:... ..
~ - ~ c
-'"
t,-
~ g B ~
Table 6.1.1- Effects of Asymmetric: Scrubbing and Bypass of 6-28mm Ore to HPRC - •
\:t:I o..... <
~
(!;, :::,-
a-~ s: ~ ~~
~ ~ s, _. ~
Q .Q., ~
1715 1868 3927 60 1029 935 0 ~
1715 186' 3927 ,1964
.... 70 1201 763 0 2212 1964 Ji
1715 1743 3955 1964 50 858 1106 10 2240 1992
1715 1743 3955 1964 60-----t 1029 10 22AO 1992
o •it
1715 1743 3S?5 1964 70 1201 - - 935
763 10 2240 1992
4'
30
." 1715 1617 3985 1964 - 1i5a o ~
50 f100 20 2270 2022 50
1715 161i 39-85 o 1::
~ 1964 60 1029 935 20 2270 2022 40
N 1715 1617 3985 1964 ro 1201 7E3 20 2270 2022 30
w 1715 14B9 4018 Jt
N 50 858 110' 30 2303 2055 50 -
,'" :o
1715 1489 60 1029 935 30 2303 2055
1715
'01' ,'" o
1489 4018
,'" 70 12'" 30 2303 2055 30 o
1715 1351 4il53 50 858 "" 40 2338 2000 - '"
50
,'" o
1715 135' 4053 '964 so-w:z9 - - " , "" 40 2338 2000 ., o
1715 '35' 4053 , ro 763 40 2338 2000 '0
'575
1715
'23' 4092
... 50
12'858" 1106 50 23n "29 o
' 231 4092 ,19"
... 00- ' 029 935 50 23n 2' 29
E-1
:1- 127'
1443 o
4092 1964 ro ,,,,, 763 ., zm
Scrubber 1 feedrate maintained constant, Scrubber 2 feedrate allowed to vary_
Total scrubbing section feedrate variation 165dtph: HPRC feedrate variation 165 dry tIh; DMS feedrate variation 637dry tIh

~
The Southern African Institute of Mining and MetoJJurgy
Diamonds - Source to Use 2009
G Popplewell alld B Boe/etse

6.6 Secondary Cone Crusher

At Orapa 3, HPRC has been promoted to the combined role of tertiary and recrush crushing.
HPRC delivers a much greater size reduction ratio than cone crushing, at an operating gap
that all but eliminates the risk of damage to diamonds present in the feed to the rolls.
However, the maximum feed particle size to the rolls must be restricted to avoid damage to
the studded surfaces and to reduce wear on the studs as a result of the inability ofdlC rolls to
draw in large particles efficiently. For this reason a single cone crusher has been selected to
generate a product with a 60 mm nominal top~size. Feed to the cone crusher is >60 mm
washed scrubber product. The cone crusher is in open-circuit, with the product joining <60
mm washed scrubber product and reporting to the HPRC section.

Although it is inevitable that >60 mm particles will occasionally be present in the cone
crusber product as a result of tramp bypass events etc., the small quantity reporting to the
HPRC poses very limited risk of meaningful stud damage and will have negligible impact on
HPRC throughput or product quality. The cost and complexity of closed circuiting the cone
crusher is therefore avoided.

Of importance, given only a single unit will be installed, is that the cone crusher product
quality is of litUe interest relative to tbe ability of the crusher to withstand a feed of up to
100010 basalt breccia without failing either mechanically or electrically, and the ability to
accept a large feed size on occasions. This means that the crusher can be over-designed with
wear parts selected for long life rather than to generate a product focused on a high "set-
under" value. Complexities such as grooved mantle and liner configurations etc. are less
attractive than a short crushing path and a coarse liner configumtion. The preliminary
selection of a Metso "HP800 with extra coarse cavity will accept feed particles up to 350 mm,
which would be a rare occurrence.

6.7 Studded Roll HPRC

The use of studded roll HPRC units has become common in diamond recovery opemtions.
Continual development of the studded surfaces, and more importantly protection of the roll
edges has resulted in units that have a significantly higher availability than was previously the
case with profiled rolls. More recently, the availability of studded roU HPRC wIits, which
inherently deliver higher specific throughput than non-studded rolls and are also more
tolerant of feed moisture, has allowed HPRC technology to be applied in a combined tertiary
and recrush crushing role. The combination of a large working gap (thUS protecting diamonds
from potential damage) and large feed size capability relative to rolJ diameter ideally suits
tbis duty. Unique at the present time within Debswana (and De Beers), the Orapa 3 HPRC
circuit is deSigned to be also fed with 1iberated diamonds contained within material of a size
range that is conventionalty routed to the DMS section for diamond recovery. Previously
(with the exception of Ekati in terms of major diamond operations) introducing liberated
diamonds into an HPRC was consi dered high risk. Whilst studies undertaken by a number of
operations had concluded that damage to diamonds less than approximately 80'% of the
working gap was economically inSignificant, the application of HPRC was still restricted to
feed stock containing locked diamonds. Thus the ability to greatly reduce the DMS feed
arisings by crushing and then re-screening the DMS feed was excluded. However, simple
consideration of the mechanics of the roll crushing process leads to the conclusion that, even

Page 233
111e Southern African Ins/ilule ofMining and Metallurgy
Diamonds - SOllrce fo Use 2009
G Popplewell and B Baelelse

if a diamond is locked within a piece of kimberlite when fed into the roll gap... it is not
effectively liberated at the instant of exit from the rolls, but at some point between entering
and leaving the roll gap. At that point it becomes identical in circumstance to a free diamond
of the same size fed into the roll along with kimberlite particles to which it is not related.
Hence, if the risk to the fonnerly locked diamond is economically insignificant as long as it is
less than 80% of the roll gap jn size, then the risk to the fonnerly free diamond of the same
size is logically also economically insignificant.

The Orapa 3 flowsheet takes advantage of this conclusion in a very conservative way. That is,
the installed DMS capacity is sufficient to permit the target ROM ore throughput to be
attained with no diversion of DMS feed to the HPRC. However, given the challenges
reported by Debswana in maintaining DMS module availability, particularly during periods
of unstable power supply, the ability to reduce the volume of feed reporting to the DMS by
up to 34% without reducing ROM ore throughput, represents a means of preserving operating
revenue vastly greater than that at risk through any notion ofHPRC damage to diamonds less
than 80% of the rolls working gap. The bypass of DMS feed to the HPRC section in a
sequential and controlled manner simply aims to match the available DMS capacity to the
target ROM ore throughput rate on a continuous basis.

As discussed in Section 8.4 Process Design, the HPRC section product can be split between
the two scrubbers to suit a range of operational requirements. Each HPRC is capable of a
minimum of 133% of design capacity, and significantly more if the DMS section is at
planned availability and hence no DMS feed is being bypassed to the HPRC section. The
impact of temporary loss of one HPRC is therefore greatly reduced. Since the product from
the two HPRCs is combined before being distributed to the scrubbers, individual HPRCs and
scrubbers are isolated from each other.

6.8 Oesanding

In the context of the Orapa 3 process plant, desanding applies to the process of removing
from downstream diamond recovery stages the ore particles and accompanying diamonds that
are smaller than the size range of diamonds to be recovered. From this point onwards,
diamonds removed by the desanding· section are essentially irretrievably lost. The desanding
section is therefore one of the most critical sections of the process plant given the AKl
diamond content curve. Desanding is carned out on single·deck multislope screens, using low
pressure wash sprays to increase screening efficiency. Holed, or excessively worn, screen
panels wlll result in loss of economically recoverable diamonds to the tailings conveyors. It is
very unlikely that these could be cost effectively salvaged at a later time. For this reason, the
design philosophy for the Orapa 3 plant is aimed at facilitating operation and maintenance of
the installed screens and screening surfaces. Of the four desanding screens installed, any
three can be used to satisfy the design duty. When available, the fourth screen can be also
utilised 10 enhance overall screening efficiency. At other times, the avai lability of an
" installed spare" allows periodic inspection of all the screens, with time available to replace
holed or WOnt panels without impacting on overall ROM throughput. Isolating the desanding
duty from all other screening duties facilitates operational control as access to screen panels
installed on a single deck screen is simple and quick, and there is no time lost to inspection of
bottom·deck screen panels.

Page 234
The Southern African Institute of Mining and Metallurgy
Diamonds - Source to Use 2009
G Popplewell and B Baeletse

As each Desand screen underflow pump discharges to a dedicated degritting cyclone,

inspection and maintenance of degritting cyclones should be synchronised with maintenance
of the associated Desand screen or its underflow pump.

6.9 Degrltting

Degritting. in the context of the Orapa 3 plant. refers to the removal of particles greater than
nominally 300 ",rn from the feed to the thickening section. If fed to the thickening section,
these fast-settling particles create potential instability in the thickeners, absorb flocculant
unnecessarily, and increase overall water consumption. The degritting section consists of a
cycloning section followed by dewatering screens. Each degrit cyclone is fed by a dedicated
desand screen underflow pump. The cyclones are either rubber or ceramic lined for extended
life. and will be inspected for spigot and internal wear to the same schedule as desanding
screen pane l inspections. As the degrit cyclone performance is not a critical process duty,
cyclone replacement, when required, will be synchronised with desanding screen panel
replacement.

The underflQw from all operating degrit cyclones is combined before being distributed to the
degrit screens. Similarly to the desand section, there are four degrit screens of which any
three can be used to satisfy the design duty. This avoids· the necessity to stop any of the
desand screens to maintain one of the degrit screens (or associated underflow pwnp).
Degrit screens are not designed with the intention of achieving an efficient size separation.
The purpose is rather to capture as many solids as possible into a deep bed, which is
progressively dewatered as it climbs the inclined degritscreen in the direction of discharge.
The screen panel aperture is somewhat larger than the nominal size cut of 300 ",rn, as finer
partides are captured within the deep bed rather than on the screening surface itself. The
screen discharge is conveyable and is co-disposed with the fines DMS floats material. Water
and fine solids passing through the degrit screen are conventionally returned to the degrit
cyclone feed which in the case of Ompa 3 would require returning this stream to the
desanding section feed distributor. However, by delivering this stream to the thickener feed
launder, the circuit is simplified. Small amounts of coarse material are not detrimental to the
thickener perfollllance. The installed spare degrit screen encourages operations personneito
maintain screening surface integrity without wldue impact on ROM plant throughput This is
arguably a superior outcome to recycling degrit screen underflow, which has no economic
value, and inherently assumes that operations personnel will act negligently.

6.10 High Density Thickeners

An evaluation of the advantages and disadvantages of the initially proposed paste thickening
installation led to all interim selection of an alternative of using deep-bed compression
thickeners of Ultrasep design. However, to achieve the process duty, at least 10 thickeners
were required of the largest standard design available. The alternative of using modified
conventional thickeners, with a ,higher wall height to increase underflow density, was adopted
based on operational experience with the conventional thickeners at Orapa 1. Crucially, and
as per the proposed Ompa 3 circuit, Orapa 1 has a degritting section to remove fast- settling
grits from the thickener feed.

Page 235
The Southern African Institute qf Mining and Metallurgy
Diamonds - Source to Use 2009
G Popplewell and B Baeletse

Basic design data derived from Orapa operating experience specified as follows:

Conventional thickener
2
Flux Rate 0.18 tlm .h
Rise Rate 2m1h
3
Underflow Density 1.25 tlrn
Flocculant Consumption 60gltonne head feed
High Rate thickener
2
Flux Rate 0.2 tlm .h
Rise Rate 4m1h
3
Underflow Density 1.3 tlrn
Flocculant Consumption 60gltonne head feed

Maximum thickener feed slimes arisings was estimated, based on modelling work, at 760 tIh.
The flux rate and feed rate were used to calculate the thickener sizes and the actual rise rate
checked from the overall process balance.

The merit of using high rate thickening is the higher underflow density. The water
consrnnption of the plant is principally involved with the pumping of slimes to the dam and
the water losses associated with deposition final moisture and net evaporation.

However, the relative disadvantage of the lower densities of conventional thickeners may be
overcome by improving the design of sedimentation on the slimes dam to maximise water
recovery from the dam.

A value·engineering study carried out during the feasibility study, inclusive of feedback from
Orapa operations management, concluded that an installation consisting of three high-density
thickeners of 45 metre diameter installed above ground would provide the best balance
between perfonnance and cost. High-density thickeners are simply thickeners of
conventional design with an increased wall height to generate greater compression. Three
thickeners were chosen instead of two, larger (60 metre diameter) thickeners for the
following reasons:

The advantages were considered to be as follows:

• Able to operate at higher capacity when one thickener is on maintenance or underflow
lines are blocked as there are more operational units. ,
• Smaller thickeners can be above ground steel tanks saving civil works.
• Good access to underflow valves, pipes and instruments for maintenance
• No issues for tunnel flooding
• These were considered to outweigh certain disadvantages, such as:
• The capital cost is higher than for above ground thickeners on concrete.
• Additional prunps, drives, valve work, instruments and pipe work to maintain.
• Additional height of feed distributor system and access to each thickener.
• Process Water System

The Orapa 3 plant process water distribution system is designed to reduce complexity.
Internal recirculation of flows has been largely avoided, particularly the concept of recycling
DMS effluent flows to the scrubbing and screening section. The reason for this is that the

Page 236
The Southern African lnslihlte of Mining and Melall:lrgy
Diamonds - Source 10 Use 2009
G Popp/ewell and B Baeletse

DMS plant consists of nine modules, any of which may be operating, at any moment. This
presents a potentially very wide variation in recirculated flows which must accordingly be
accommodated in a parallel water supply system. In the past the driving factor has been the
desire to maintain locked into circuit any }XJtentially valuable diamond bearing material
arising from worn screen panels etc in the DMS. The result has however been a complex
water management system with a lot of additional instrumentation, adding to potential circuit
instability and consequent process interruptions.

At Orapa 3, this has been avoided by routing all of the DMS eflluent to a settling tank, with a
bottom cone section. Whilst potentially containing misplaced diamond bearing material as
discussed above, DMS erouent also contains grits (nom >0.3 mm) as a result of screening
inefficiencies and in-circuit degradation. Such particles settle into the cone section and are
colltinuouslyextracted along with approximately 10% of the incoming flow. which is then
discharged to the desand feed distributor.

The slimes loading of the DMS effluent is usually very low. TIlis means that the effluent
water can be used without flocculation and thickening~ and hence is nommlly routed directly
to the process water tanks. For process upsets, and for periodic cleaning of the process water,
the complete DMS effluent settler overflow can be routed to the thickener feed distributor.
More usually, a small fraction only of the effluent flow would be routed to the thickeners for
this purpose. This can also be isolated if required, since the settler underflow itself represents
a bleed of water to the thickeners and hence will act as a continuous purge for cleaning
purposes.

Two process water storage tanks have been provided, manifolded together. These are of the
same, self-cleaning. design as those for Orapa 2.

Start-up after full plant shutdown always requires care in order to avoid water hammer. To
assist with this,a small process water pump has been provided in parallel with the main
process water supply pumps to serve the purpose of maintaining the system piping full of
water during shutdown. This pump would also be used on initial plant start-up for the same
purpose, with the first main supply pump being started once system pressure indicates the
piping is full.

6.H DMS PI.nt

6.11.1 Pump-Fed DMS Cyclones and Dilute Medium Den!lification

,
The use of pump-fed DMS cyclones versus gravity fed cyclones was the subject of much
debate during each phase of the Orapa 3 plant evolution. The selection of pump-fed cyclones
was driven by the reduced capital cost associated with lower plant height. Whilst the fact that
the cyclones are the highest point of a pump-fed DMS plant whereas gravity fed cyclone
configuration has an additional 'JO metre structure above this point is an obvious driver of
capital cost, there are other more subtle advantages that translate into capital cost advant.1ge:
• Cyclone size and operating head are divorced from plant height.
• The same module design can be used for cyclones of a different size. This is
important in the context ofOrapa 3 as the 5.10 mm diameter cyclones may be changed
for 420 mm diameter cyclones if the bottom size for diamond recovery is lowered
significantly.
• The same cyclones can be used at different operating head.

Page 237
The Southern African Institute of Mining and Metallurgy
Diamonds - Source to Use 2009
G Popplewell and B Baeletse

• Therefore the same module design can be used for different duties.
• Different duties are interchangeable on-line in the same module design.

Other relative features of pump-fed versus gravity-fed cyclones are quite well known and
can, depending on the relative importance assigned to each of these, be used to justify a
selection either way.

The use of dilute medium densification is unusual in large plants. It has become more
common to install densifiers fed with either circulating medium, or occasionally floats drain
medirnn. However, once a decision has been made to adopt pwnp-fed dense medium
separation cyclones, this requires an additional pump to provide sufficient pressure for the
densifiers to function. The alternative, since the dilute medium is pwnped to the magnetic
separators, is to install the densifiers on this stream instead. This has the advantage of
reducing the overall magnetics loading on the separators, leading to lower ferrosilicon losses.
Adequate densification is maintained by diverting medium into the dilute medium circuit via
a spiitter on the underflow of the floats drain panel, and medium cleaning is ensured by
diverting part of the densifiers underflow into the magnetic separator. At present this has
been configured such that of the three densifiers installed, one is essentially a dummy unit, as
both overflow and underflow products report to the magnetic separator. Further refinement is
possible at the detailed design stage if considered justified.

6.11.2 DMS Capacity Configuration

The Orapa 3 DMS circuit is configured as a split coarse/fines circuit. Feed to the DMS (1.6-
25 mm) is in two streams, 1.6-6 mm (fines DMS) and 6-25 mm (coarse DMS).

Since the design of the DMS modules for both fines and coarse fractions is identical, only the
medium ore ratio differs, the cost impact of designing for separate size fractions is to reduce
the overall capital cost. This is because modules designated as coarse modules have a
nominal feedrate 25% higher than those designated as fines modules. If a single size range is
treated, the module capacity is reduced to optimise perfonnance on the fine end of the size
range, since this is where the majority of the revenue occurs. The net result is that more
modules are required for a particular feedrate if this is treated as a single stream.

Based on the mass balance, the maximum DMS feed arising is 1,868 tIh. This would require
ten DMS modules if the throughput rate were limited to 200 tIh to satisfy the medium to ore
ratio requirement of 7: 1 for the finer feed fraction. If evenly fed, the ten modules would be
fed at 187 tIh. At a reduced medium to ore ratio of6.7:1, nine modules would be adequate,
feeding at 208 tfh.

For a split DMS, the requirement is maximum nine modules. Of these, six are coarse modules
operating at 241 tlh, and three are fines modules operating at only 142 tIh. At a slightly
reduced medium to ore ratio of 6.5:1, only two fines modules would be required. Of nine
installed modules, one module is therefore,at small risk, a spare module.

Of the nine modules installed, two modules can be fed with either fines or coarse feed size-
ranges.

The DMS capacity can therefore be flexibly matched to the arisings in the coarse and fines
size ranges by assigning several modules to address either duty. With the majority of Orapa

Page 238
The Southern African Institute of Mining and MetaJJurgy
Diamonds - Source to Use 2009
G Popplewell and B Baeletse

diamonds in the <6 mm size range, capacity dedicated to the 1.6-6 mm size range at-enhanced
medium : ore ratios would reduce the risk of diamond loss. TIle size split between fines and
coarse can be optimised in the range 4-6 mm by changing screen panels in the scrubbing
section, to maximise overall diamond revenue without impact on the overall DMS capacity
required, as the swing modules will absorb the mass balance impact.

The swing modules (modules 6 and 7) are of the same design as all the other modules.
Designation as a fines or coarse module simply implies that the feedrate is changed to suit the
designation.

6.11 .3 Steady Water Head tank for Critical DMS Duties

Power failure has unpleasant conseq ue.nces in a DMS plant. Mostly these cannot be easily
avoided. Two duties which are important in facilitating restart are sinks screen spmy water
and medium pump flushing.

On general power failure in a pump-fed DMS module, the feed pressure to the DM cyclones
is lost. The entire contents of the four cyclones are consequently discharged to the sinks
screen, which ensures no loss of diamonds, whilst at the same time the sinks screen is
stopping. Although the sinks screen is supplied by the emergency power circuit, this is
unlikely to restart the sinks screen in time to be effective in restoring the screening
performance within the critical time period. The drain section of the sinks screen will be less
effective than normal in removing medium, which will report to the rinse section in greater
quantity than normal, along with a much greater volume of un-separated gravel than is
nonnally fed to the sinks screen. If the spray water has also ceased as a result of the general
power failure, the net result is a large volume of settled medium and gravel sitting on the end
tltird of the screen. When the screen restarts, even if the spray waler bas been restored, this
material is discharged into the concentrates handling system, which is not equipped to
efficiently handle this. Even if the transfer to emergency power is, sufficiently rapid to
maintain a reasonable screen performance, without wash 'sprays a large amount of
ferrosilicon will still enter the concentrates handling system.

By providing a steady head tank to supply water to the sinks screen sprays for a period
following failure of the process water pumps, washing of the combined ferrosilicon and
gravel is maintained. This will minimise the amount of ferrosilicon reporting to the
concentrates handling system.

The cyclone feed and circulating medium pumps stop on power failure and the drain valves
open. Most ferros ilicon reports to the spillage containment area, along with gravel from the
cyclone feed pump. Some ferrosiliconlgravel inevitably remains in the pumps, and standard
practice is to flush the pumps with water after a time delay. Again however, if pressure has
been lost in the process water system as a result of the power failure, this will be ineffective,
although some water will be avai lable at a steadily falling pressure as pipes higher in the
building drain down to the pumps. By supplying fl ushing water to the circulating medium
and cyclone feed pumps from the steady head tank, the water pressure is maintained for
sufficient time to provide adequate flushing of the pump casings. This greatly assists with
restarting the DMS module once power is restored,

Page 239
The Southern African Institute ofMining and Metallurgy
Diamonds - Source to Use 2009
G Popplewell and B Baeletse

6.12 Recovery Section

6.12.1 Wet Primary X-ray Sorting

For some considerable time, recovery sorter technology in De Beers family of companies has
been based on dry sorting. A decision was taken in the early 1990's that development of wet
X-ray sorters would be discontinued in favour of dry X-ray sorters. The implications of this
were that the entire feed to the recovery plant had to be dried. Coupled with a philosophy of
maximising gravity flow, this led to the extensive use ofpneumodriers to dry and elevate the
recovery plant feed simultaneously. The use of pneumodriers has however not been without
controversy, from a cost perspective as well as through concerns as to risks of diamond
damage and the generation of significant dust.

In parallel with the decision within De Beers to adopt a dry X-ray sorting route, elsewhere
new diamond recovery operations, ofa comparable sophistication to those of De Beers, were
adopting wet X-ray sorting technology (Ekati, Diavik). The benefits of successful
implementation of wet X-ray sorting, most particularly primary sorting, are obviously
associated with the elimination ofthe bulk drying stage, with it's perceived disadvantages as
noted above.

A new direction in sorter design development was taken more recently, with the evolution of
new wet X-ray diamond sorters at Debtech. The current version, designated the Modular
Recovery Unit Process (ModRUP), has been extensively tested in pilot plant operation
against the only commercially available alternative, the DP6D supplied by Ultrasort Pty Ltd
of Australia. Following a review of the results of this exercise, using a balanced scorecard
approach to evaluation of the characteristics and perfonnance of the two sorters, it was
decided to select the ModRUP as the basis for design of the Orapa 3 recovery section.

The ModRUP unit can be configuredas a dry sorter for reconcentration duties in a two-pass
configuration. The design criteria for the recovery section calls for the product transferred to
the FISH to be 70% diamond by weight CDBW) for the fines fraction and 50% DBW for the
middles and coarse fractions. By using two~stage dry reconcentration, it was considered
possible to eliminate the need for single~particle-sorting using De Beers Raven technology.

This confines the X~ray sorter technology to a single base model, with only minor differences
between the wet and dry versions.

6.12.2 Grease Scavenging

Grease scavenging of X-ray rejects is not currently a feature of either of the CARP
installations at Orapa or Jwaneng Mines. Historically, at least at Jwaneng, a fines grease plant
was used for the primary recovery of diamonds smaller than 2mm, as the X-ray sorters
available at the time were considered to perform poorly below this size, particularly wet X-
ray sorters. A contemporary installation at Venetia used the same philosophy and in both
operations the result of attempting to fully automate the diamond grease recovery process
was expensive and operator/maintenance intensive. The advent of more advanced X-ray
sorting technology resulted in later installations (viz the CARPs at Jwaneng, Orapa and CTP)
becoming based on dry X-ray sorters, with no grease applied for any duty. More recently, a
complete revision of the implementation of technology for grease recovery of diamonds, as
well as a pragmatic response to the limitations of a single diamond recovery technology, in

Page 240
The SOl/them African Inslittlle of Milling and Melolltlrgy
Diamonds - Source 10 Use 2009
G Popplewell and B Baeletse

this case XRF, has brought grease as a scavenging method back into vogue. This has
accordingly been adopted for the Orapa 3 recovery plant flowsheet. Recovery plant rejects
will therefore have been subjected to two stages of XRF, followed by attritioning and grease,
prior to being deposited on the secure recovery rejects stockpile. The sequence of processing
events is characteristic of an auditing process. Second~pass X-ray sorting ejection counts are
an indication of first pass·X~ray sorter effectiveness, and the number of diamonds captured
on the grease belts. batched by size fraction. is an indication of the overall X-ray sorting
section efficiency.
AcknowJedgements

The authors would like to thank Debswana Diamond Company for the pennission to prepare
and present this paper, which seeks to docrnnent the evolution of the Orapa 3 process plant
conceptual design.

Thanks is a lso due to the. Debswana process team for access to the Prosim simulation model
developed for Orapa AKI ore. and who assisted with modelling of the Orapa 3 fl owsheet
options as they were generated.

TbeAuthor

Graham Miles PoppJewell, Technical Director - Diamond Processing, Fluor Corporation

Graharn has over 31 years experience in the design, operation and management of diamond,
industrial minerals and base metals recovery plants . For the last 22 years, Graham has been
involved in process design and project commissioning, focusing on diamonds particularly in
Africa and Canada. Major assignments have been as lead process engineer for the 800tph De
Beers Venetia Main Treatment plant, the Final Recovery Plant for Alexkor. and the 420tph
Jwaneng Recrusb DMS Plant for Debswana. Numerous more minor assignments have been
completed in a variety of countries. Graham managed the process engineering, layout
development and commissioning of the BHP Ekati diamond project in 1998 and since that
,
time has been leading study work to define expansion strategies at Ekati, in addition
perfonning a similar role for the Diavik diamond mine in 2004. Graham was process
consultant and commissioning manager for the De Beers l000tph CIP diamond recovery
facility in Kimberley, South AtTica. Recent assignments have been as study manager and
process lead for the Diavik Small Diamonds Project (Rio Tinto), and Project Manager for the
Ekati New Thickeners Project (BHP BiUiton). Currently Graham is managing the Orapa
Expansion Project FS (Debswana), being undertaken by Fluor in Melbourne. As one of only
two independent cons ultants to the Debswana " Experienced Persons Group" throughout
2006, Grabarn was instrumental in the company undertaking a new direction in conceptual
design, which is currently manifesting itself in the evolution of the 12Mtpa Orapa 3 plant in
Botswana. Graham was appointed a Fluor Senior Fellow in 2009.

Page 24 1
The Sol/them Africa" institule oJMining and Melallurgy
Diamonds - SOllrce 10 Use 2009
G Popplewell and B BaeJefse

Page 242