Thicken Er Design and Control
Thicken Er Design and Control
net/publication/330131872
CITATIONS READS
0 1,561
1 author:
Mike Miller
SNC Lavalin
2 PUBLICATIONS 0 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Mike Miller on 04 January 2019.
By
Mike Miller
Mike Miller
mike.miller@snclavalin.com
ABSTRACT
The common thickener underpins the performance of most hydrometallurgical facilities. Thickeners
are simple, reliable, effective and efficient – under the right conditions. For the rest of the time,
however, thickener performance (or lack of it) seriously affects the rest of the operation. In many
plants, thickener problems are a regular, day-to-day operational event. So much so, in fact, that
these issues seem to be accepted as a normal part of doing business. Given recent developments in
thickener design and modelling, this should not be the case.
The purpose of this paper is to present a generalised understanding of the various inputs that
combine in the design, control and operation of a modern thickener. This paper does not seek to
provide all the answers in thickener design and control, but hopes to give the reader a greater
understanding of the questions that need to be asked in the course of preventing or controlling
thickener problems.
Starting with the basics of thickener theory, key elements of the sedimentation process and options
for thickener selection are explored, as well as testwork requirements and the features of critical
thickener components. Thickener control strategies and shortcomings in current thickener operations
are discussed, including the strengths and weaknesses of the given field instrument options and their
application in the overall control strategy. . Many of these can be considered as “growing pains”,
associated with changes introduced by process improvements in other areas.
Finally, options for design and operational improvements (for both greenfield and brownfield
operations) are discussed.
INTRODUCTION
Thickeners are used in slurry applications where large-scale separation of fine-grained solids and
liquids is required and where a dry filter cake product is either technically or economically impractical.
Thickener products, either as high density underflow slurry or as clean overflow liquor (or ideally,
both) are produced to enhance the efficiency of the downstream circuits. To the extent that the
associated process circuits assume a specific product parameters from the thickener, then thickener
performance is a necessary condition in achieving overall metallurgical efficiency.
As dewatering devices, thickeners lie somewhere between cyclones and filters in that they perform a
solids-liquor separation duty that is inferior to a filter, but better than a conventional cyclone.
However, particularly when handling fine solids, the cyclone is constrained by its separation efficiency
and the filter is constrained in its volumetric capacity, in which case a thickener will often be installed
in association with these options.
For some circuits, a series of thickeners is used to displace soluble species from the product
underflow slurry as a countercurrent decant (CCD) washing circuit.
The purpose of this paper is to present a generalised understanding of the various inputs that
combine in the design, control and operation of a modern thickener. It will discuss the roles of the
various thickener components and aspects of thickener control, concluding with a discussion of some
concepts for future thickener development.
Process Definition
It is commonly accepted that there are three basic solid-liquid stages taking place in a thickener:
Free settling,
Hindered settling, and
Slurry Compaction.
These processes take place in the progressively lower sections of the thickener. “Free Settling” refers
to a zone where the individual particles fall through the supernatant solution without inter-particle
interference. The particle settling rate approaches the Stokes Law terminal velocity of the solid
through the liquor phase and is governed by particle size, particle shape, liquor viscosity and the
density differences between the solid and liquid phases. The supernatant solution is displaced
upwards, and reports to the overflow collection system. In the “hindered settling” zone, interactions
between particles become significant and the solids settling rate reduces as these interactions
become more significant. Inter-particle liquor release becomes more difficult due to liquor upflow from
the region between particles and the differential between the solids and the fluid (slurry) medium
decreases. Slurry densification continues in the slurry compaction zone, often with the assistance of
the thickener rake, which generates localised compression zones to “squeeze” excess supernatant
liquor from the partially densified slurry. Some authors prefer to differentiate between gravity and
rake compaction mechanisms; this can be regarded as separate subject in itself.
In any event, the perfect thickener would be one that delivers a clean, solids-free overflow liquor and
a compact underflow slurry product (at minimal operating cost) in a capital-efficient manner. In the
real world, however, the actual thickener design is a compromise between these criteria.
Flocculant or Coagulant?
Both flocculants and coagulants are surface-active reagents for enhancing thickener performance,
however they serve different purposes and have different modus operandi. Coagulants improve
overflow clarity by neutralising the solids’ repulsive surface charge to permit short-range attractive
forces to come into effect. They target the ultrafine slimes, which have a low settling velocity and a
high ratio of surface charge to particle mass. Coagulation is frequently achieved using mineral salts
such as alum, ferric sulphate or lime, especially in the domestic water treatment industry. Synthetic
polymer coagulants are also produced.
Flocculants, on the other hand, provide inter-particle links to produce fast-settling agglomerates. By
agglomerating the fines, a clean overflow is often produced, besides a dense underflow. Where
flocculant alone cannot achieve a clear overflow, pre-dosing of the slurry with coagulant is sometimes
used to condition the solids ahead of flocculation.
The development of polymer flocculants in the 1960’s (e.g. McCarty, 1959) provided the impetus for
most of the thickener improvements that we have witnessed in recent years.
Thickener Types
Different thickener sizes, shapes and configurations reflect their differing design intent and extent of
technology adoption. Thickeners come in recognisably different profiles but there are no hard and fast
rules to differentiate one type from another. With the obvious distinction between raked and un-raked
thickeners, design seems to fall into a spectrum of options.
Conventional thickeners were developed following the pioneering work by Hazen (1904) and Dorr.
Conventional thickeners are characterised by low solids loadings, low flocculant requirement and
shallow feedwell configuration. Though associated with high capital cost and a large plant footprint,
they do have the advantages of being relatively stable, low flocculant rates, and being simple to
operate. In cases where flocculant might be exorbitantly expensive or deleterious to the downstream
process, the conventional thickener remains as a viable option.
As an example of a conventional thickener, Taggart (1945) records the Mt Lyell copper concentrate
thickener as being 6 metres diameter by 3 metres high, and processing 150 ton per day to give a
solids flux rating of 0.2 t/m2/h. In a more challenging duty, Moa Bay’s original leach feed thickening
was performed by a parallel pair of Dorr-Oliver thickeners to produce a 45% solids underflow. Each
thickener was 128 metres in diameter, with a depth of 9.8 metres. At a combined solids feed rate of
5500 tons per day – without flocculant - this equates to 0.01 t/m2/h. This compares with ~0.4 t/m2/h
(typically) for modern flocculated thickeners in similar duty today.
The development of synthetic polymer flocculants in the 1960’s revolutionised thickener designs.
Utilising the higher settling rates associated with flocculated feed, this new generation led to a new
thickener type that are collectively referred to as High Rate thickeners. These achieve a much higher
solids loading than conventional thickeners, owing to higher flocculant consumption and improved
control, which produce a fast-settling agglomerated solid. Various configurations exist for this type of
thickener, depending on the feedwell design, flocculant rate, and dilution control. Compared to the
earlier generation of thickeners, the High Rate thickener has a higher solids loading, higher flocculant
consumption and requires more attention to feed slurry dilution, flocculant dosage and bed profile.
While the high rate thickener’s smaller diameter reduces the site footprint, it does so at the cost of
lower residence time and a more “twitchy” operation. On the positive side, the high rate thickener has
options for recovery from an upset situation that the earlier generation did not.
Investigations into the new high rate technology have led to an improved industrial understanding of
High Rate thickener operations. These have benefited from advances in computational flow design
(CFD) modelling and effective industry/research co-operation. (Farrow, 2000; Rudman, 2008.)
One advantage of the smaller unit area is that above-ground thickener construction has become
viable. As a result, there has been a trend for thickeners to be constructed above ground for ease of
maintenance and for safety access, which simplifies issues such as confined space entry and
maintenance limitations for caisson and tunnel access. However, there are still issues with
maintenance access for underflow pumps, particularly where high viscosity underflow streams
demand the shortest distance between the thickener and its underflow pump.
An extrapolation of the High Rate thickener is the Paste thickener, which is often applied for tailings
management and other cases where a high underflow slurry density is required. (Gollaher et al 2010)
Like the High Rate thickener, the Paste thickener uses high flocculant dosage and optimised feed
slurry dilution to achieve acceptable solids settling rates. The increased underflow slurry density
requires a higher rake torque for a given thickener diameter. Floor slope is more acute to both assist
raking and to improve underflow bed compaction. Sidewall height is typically greater to provide
sufficient underflow residence time for compaction.
Underflow slurry viscosity yield stress increases with increased underflow density, particularly when
the density exceeds a given threshold. This may affect the underflow handling systems, especially in
achieving net positive suction head (NPSH) requirements for the underflow pumps. Various options
have been developed for overcoming this obstacle, which will be discussed separately. However, it
will remain a challenge for thickener designers and operators as long as the push for increased
density continues and until thickener feeds actually comply with their design criteria on a reliable
basis.
Rakeless thickeners were initially developed for high-scaling applications such as alumina processing,
but have been extended beyond this industry. Examples are the E-Cat (Harris, 2002) and Ultrasep
thickeners, which use internal dewatering cones to reduce the thickener diameter. The side wall and
floor angle for these thickeners are generally more severe than that found in paste thickener design.
For further information, Klepper (2009) has provided a comprehensive review of thickener design
evolution.
Thickener Selection
Within a typical metallurgical circuit, there are various applications for thickeners that will drive the
design towards one or another of the thickener configurations discussed above. Examples are given
below.
Leach Feed Preparation circuits typically aim to achieve a dense underflow slurry, but are not as
reliant on overflow clarity, provided that it does not impact on flocculant dilution, gland seal water
quality or other process requirements. For a pressure acid leach autoclave, for instance, a 1%
(absolute) increase in underflow slurry density increases the autoclave output by about 3%. The
figure below describes the variation in annualised metal extraction rate for a “typical” autoclave as a
function of the slurry density that is fed to the heating circuit.
21000
16000
11000
6000
30% 35% 40% 45% 50% 55%
Feed Thickener % Solids
As Wallace (2008) suggests, capital “savings” on feed thickening should be treated with caution; they
may prove to be a pyrrhic victory if they are achieved at the expense of reduced autoclave capacity.
At many hydrometallurgical sites, the process economics are dominated by fixed costs, making
attainment of design production fundamental to profitability. In these cases, a conservative approach
to thickener sizing is justified.
Thickener design configurations usually consider the high rate or paste thickener options, with a
tradeoff between capital cost and process performance. Designs that assume a lower settling flux will
usually achieve higher leach feed slurry density on a reliable basis, but at increased capital cost.
Countercurrent Decant (CCD) applications also seek higher underflow density, in order to maximise
washing efficiency from the available wash volumes. Like the leach feed thickener, a high density
thickener design with the appropriate settling flux is desirable.
The theoretical washing efficiency of the CCD train can be calculated from an equation provided and
proven by Page (1976) and Barnea (1978) respectively.
Option 1 involves capital installation that may be impractical for the operation. Option 2 is a good
short term fix for improving wash efficiency, but it comes at the cost of increased downstream
volumes (hence more capital), increased water consumption, and potential for more metal losses to
final raffinate or barren liquor streams. Efficient solids-liquid separation, evidenced by high
underflow density, will usually maximise the circuit’s economic return.
As long as flocculant dilution quality is satisfactory, overflow clarity is a lesser priority in CCD circuits,
except where CCD overflow is fed to a solution purification circuit (e.g. Solvent Extraction or solids
precipitation). In these cases, the entrained solids in the CCD overflow either foul the process or
contaminate the product respectively. Designers can avoid these problems by either installing a
clarifier to process the dirty CCD overflow, or by over-designing the CCD1 thickener to reduce the
solids entrainment. (e.g. enlarged CCD1 diameter or parallel CCD1 thickeners). These options
usually reduce the load on downstream polishing filters ahead of the sensitive process area.
Solution Clarification applications are typically required where the downstream circuit is sensitive to
entrained solids and the feed slurry has a low solids loading. (e.g. PLS cleaning ahead of SX, gland
water polishing.) Clarification thickeners are installed for this duty where the solids loading would
overload a conventional polishing or cartridge filter. The Pin Bed, Dynamic Bed and Floating Media
clarifiers are variations on the standard clarifier, and can obtain overflow TSS levels below 50 mg/L.
The Ishigaki Polishing Filter can be seen as a high-intensity extension of the floating media clarifier.
Concentrate recovery thickeners are often located ahead of the concentrate filtration process to
reduce the hydraulic load on the filters. The typical objective is to obtain a reasonably high density
underflow, consistent with delivery to the filter elements. Concentrate solids are usually simple to
settle, resulting in a higher solids flux rate.
Thickeners for tailings densification need to produce a dense thickener underflow if the tailings are to
be deposited in a tailings storage facility or processed by filtration. Paste-style thickeners are often
used in this capacity, particularly where water balance considerations are significant.
In summary – form follows function. Design of a given thickener reflects its workload and the specific
process priorities for the given application. When a clear overflow is required from a dilute, slow
settling slurry, the thickener geometry will be shallow and wide to provide the settling area required for
clarification. On the other hand, where a dense underflow is required, additional residence time in
the slurry compaction zone is provided, with attendant “beefing up” of the rakes and rake drive
systems.
Thickeners may operate in parallel duty, especially when the volumes to be processed exceed the
capacity of a single stand-alone unit. Having thickeners in parallel improves the process availability,
but at the cost of additional capital and (to a lesser extent) operating costs.
TESTWORK
Most thickener design is based on a combination of batch and continuous bench-top testing. The first
step is typically the small scale “jar test”, where a slurry is allowed to settle over time in a 1-2 litre
measuring cylinder; the height of the interface between the supernatant and the top of the hindered
settling zone is plotted over time to give an indication of settling rate. This test is a simplification of
the full-scale process, but is adequate for initial sizing and for evaluating the effects of ore type, water
chemistry, flocculant type, flocculant addition rate and slurry feed dilution. This information can be
used to decide the setpoints for subsequent continuous testing.
It is critical, of course, to ensure that the sample being presented to the jar test is “typical” of the full-
scale feed. This involves discussions with mining and geological personnel to ensure that the
sample is relevant and clearly understood in the context of the mining process. This understanding
should be based on the sample’s location in the orebody and to the mine plan. The understanding
should include detailed sample “provenance”, its geochemistry, its particle size distribution and its
mineralogy. The sample itself should be presented in much the same way as it would be delivered to
the full-scale thickener, especially if there are upstream processes such as beneficiation, grinding or
leaching / precipitation stages involved. The sample should not be dried unless this is done in the full-
scale process. Aging of the sample, especially when it consists of a slurry in a saturated (e.g.
gypsum) solution is a potential risk, which can be evaluated by performing a standardised jar test on
the fresh sample so that it can be repeated on the sample (after storage) to indicate the extent of
aging.
Water chemistry, particularly the pH, salinity and scaling potential should be consistent with the
envisaged process conditions, as should the operating temperature. Site water, or a comparable
synthetic, should be used. Side by side tests on the site and synthetic water should be used to
demonstrate the validity of the synthetic solution.
Though quite basic, the jar test has the virtue of simplicity, which permits it to be used for multiple
variability tests on a range of feed samples and conditions. Fast settling, low flocculant requirements
and clear overflows are obviously desirable outcomes from the testwork, but it is also wise to
investigate the “Bad Actors”, where the settling performance was poor; this can lead to an
understanding of mineralogical or other factors that need to be controlled in the full scale process.
Continuous small-scale thickening tests can be performed using 100-200 mm diameter cylinder units,
which are fitted with rakes for underflow compaction. Process conditions are usually varied to
demonstrate the likely effects of feed rate, flocculant dosage and flux rate on thickener underflow and
overflow results. (Pearce 1980). Continuous raked cylinder tests consume more time and resources
than a simple jar test and are used to demonstrate the process conditions indicated by the jar tests,
particularly the effects of thickener loading on overflow and underflow quality. However, while the
continuous cylinder test can achieve an underflow density approaching that of the full-scale thickener,
the scale-up factor from bench to full scale operations still relies on operator experience. Ruhanen et
al (2018) provide an example where vendor – client co-operation can enhance the process design
and plant performance.
A further output from the continuous raked cylinder test can be an underflow sample for rheological
testing. This sample has the advantage of being taken from a known process under clearly
understood conditions with a relatively known shear history, which is usually an imponderable
variable. Testwork output usually the relationship between slurry yield stress and slurry density. This
rheological information is critical for design of the thickener rake, rake drive and the underflow
pumping systems. Most yield stress data is collected using a vane-style viscometer to avoid the “slip
zone” interface experienced by the conventional cup-and-bob viscometer. During testwork, the
sample may be densified in a centrifuge or diluted with site water to provide a relationship between
slurry density and its rheological properties. For reference, it may be useful to correlate this data
with alternate viscosity measures, such as slump ring or Marsh funnel data. Rheological testing can
also determine how the underflow slurry responds to shear once it has been extracted from the
thickener. This can be usefully applied to establish whether underflow slurry shear thinning systems
are justified.
Where applicable, discard slurry from the rheological testwork may be used for determining the
slurry’s filtration characteristics. (Bickert, 2011)
Pilot-scale continuous thickener testwork can be conducted using larger raked thickeners; these are
typically about 1 metre diameter with a sidewall of about 4 metres. These can confirm the small-scale
continuous test results, but are more commonly used for generating slurry samples for continuous
pilot testing of the downstream circuit. Larger slurry samples may also be required for generation of
tailings characterisation data as an input to tailings storage facility design.
A practical issue with all thickening testwork is a lack of industry standards, which makes it difficult to
apply or transfer testwork data into the design and procurement phases. Testwork is often conducted
by thickener vendors, who may consider the results to be proprietary; even in cases where this is not
the case, those vendors who do not partake in the testwork are reluctant to trust their competitor’s
data. Project ownership of the test program, using valid feed samples, representative process
conditions and “typical” test procedures for third party testwork may be warranted on this basis.
THICKENER COMPONENTS
This section will discuss the various elements in a thickener, their function and their desirable
features. It should be remembered that not all of these components are necessarily provided by the
thickener vendor, and that the project manager is responsible for dealing with the interface issues that
arise.
Feed Tank
The purpose of the feed tank is to deaerate the slurry and to deliver it as a consistent slurry to the
inlet to the thickener feedwell inlet. Dilution water may be added (as necessary) to provide a target
slurry density ahead of flocculant addition. Coagulant, if used, can be added at this tank or in the
slurry line feeding the feedtank, if additional residence time is necessary.
It is essential that the slurry line to the thickener feedwell remains flooded to prevent intermittent air
entrainment in the line. If permitted to occur, aeration of the feedwell occurs, resulting in floc
breakage, poor settling and dirty thickener overflow. Large gravity head differentials between the
feed tank outlet and the feedwell should be avoided to reduce the risks of surging flow and of
unnecessary turbulence in the feedwell. Sizing should discourage solids settling in the tank, but tank
and tank supports should be designed to handle dense underflow slurry.
If stage-wise flocculant additions are made, a flocculant sparge may be located on the feedline. The
potential for solids settling in the line downstream of the flocculant addition point should be
considered, especially with respect to the mechanical loads on the feed line and its supporting
brackets. Flat or quiescent zones where slurry solids may settle should be excluded from design.
Feedwell
The feedwell has a number of critical but conflicting purposes – namely to provide a controlled
turbulent zone for flocculant distribution, inter-particle collision and agglomeration, followed by a
quiescent zone where the resulting flocs may settle without excessive breakage due to shear. To
quote Triglavcanin, 2009:
“Delivering dilution water and mixing this water with incoming feed slurry, addition of
flocculant, maximising flocculant-slurry interaction to build aggregate size, unabsorbed
flocculant percentage, volume average of solids volume fraction, momentum dissipation ratio,
dilution recycle ratio, volume average shear, maximum shear rate at exit, aggregate size on
exit, uniform discharge profile and more — all of these need to occur in the right order and in
the right proportions within the thickener feedwell, the secret to thickener performance.”
Clearly, there are several operations taking place within the feedwell. From a practical perspective,
the feedwell also serves as a diagnostic of thickener health: if flocculation and settling is not clearly
visible in the feedwell, then the operation is headed for unstable conditions.
Some feedwells incorporate provision for slurry dilution, using supernatant liquor or upwelling liquor
currents. Others have forced slurry dilution systems attached as “bolt-on” mechanisms to the
feedwell. Finally, the feedwell should deliver the solids slurry evenly across the feedwell discharge to
prevent dense slurry streams sloughing to specific regions in the thickener.
Feedwell dimensions are designed to provide the required turbulence and residence time for
flocculation, and to deliver the flocculated slurry to a suitable level in the thickener profile. The
feedwell internals may include flocculant spargers, baffles, ledges and other fittings to modify the flow
dynamics.
Feedwell design is greatly assisted by CFD analysis of the mixing and shear patterns in the feedwell,
especially as a function of thickener loading. This can be used for feedwell design and
understanding of its flow dynamics, as well as help with location of slurry inlets and floc dip tubes,
Figure 2 – CFD model of the discharge profile – underside of an Outotec Vane Feedwell ®.
More information on how optimisation of feedwell internals such as shelves, floc inlets and slurry
discharge arrangements can improve thickener performance is provided by Heath. (2014)
Slurry Dilution
Flocculant consumption efficiency is maximised at an optimum solids density in the slurry prior to
flocculant addition. (Gregory, 1989). This effect seems to occur when the combined effects of
efficient flocculant dispersion and particle collision probability are optimised. In general, flocculant
consumption efficiency falls off more rapidly in cases where the slurry density is higher than optimum,
compared to the case where the slurry density is below the optimum level. (i.e. it is better to be
slightly on the dilute side of the optimum solids concentration.)
It should be noted that actual feedwell solids concentration will not be the same as the weighted
average solids concentration of the feed and dilution flows; the residence time differences between
the feedwell liquor and the settling solids means that the overall average feedwell % solids is actually
lower than the mathematically predicted value. This factor provides a safety margin to the feed
dilution process, but should not be used as a front-line control because it is a “Catch-22” situation –
the additional dilution due to these residence time differences only becomes significant when there is
effective flocculation in the feedwell, which in turn relies on optimisation of the slurry dilution ahead of
flocculant addition.
Dilution water is often added at the feed tank, but, (especially where large dilution volume is required),
the dilution can be provided at the feedwell using supernatant liquor from the thickener itself. Various
arrangements for doing this are available, using either passive or active (or forced) dilution flow
streams. An example of a forced dilution system is shown in figure 3.
Passive dilution methods rely on indirect factors such as the lower level of the feedwell slurry (relative
to the thickener overflow launder) or on venturi-style induction of supernatant liquor into the mixing
area, which is induced by the velocity of the incoming slurry past one or more angled nozzles.
Feedwells can also be designed to promote upwelling of supernatant dilution liquor at the core of the
feedwell. Active slurry dilution systems, on the other hand, physically pump the supernatant into the
feedwell or the feed tank. The active dilution systems can include variable speed drives to provide a
direct control over feed dilution, whereas the passive methods lack this “dial up” capability.
Most modern thickeners rely on polymer flocculant additions to achieve the target settling rates and
overflow clarities. Efficient flocculant utilisation requires flocculant adsorption on only a fraction of the
active sites on the particle surface; additional flocculant adsorption on a given particle does not
improve its settling characteristics, but wastes reagent that would be better used on another,
flocculant-starved particle.
For this reason, flocculant solution is typically diluted before dispersion into the slurry. Flocculant
solution is normally formulated and distributed at a nominal concentration of ~2.5 g/L, then diluted by
a factor of about 10 ahead of dispersion into the slurry. (Which also brings the flocculant solution
viscosity to a manageable level.) If dilution water is being added to the feedwell, it is often convenient
to further disperse the flocculant solution in this stream to improve its distribution, provided, of course,
that the dilution water is reasonably solids-free. If not, then the flocculant will be selectively and
wastefully adsorbed onto the supernatant solids, the remaining feedwell solids will be flocculant
starved, and the thickener is at risk of sliming. Prudent thickener designs include a back-up
flocculant dilution system for cases when the thickener overflow is unsuitable.
Flocculant mixing with the slurry solids requires high shear conditions for flocculant dispersion within
the slurry, followed by evenly-distributed flocculant adsorption, particle collision and agglomeration.
Multiple flocculant additions may be beneficial if this assists the flocculant dispersion or if the
secondary flocculant addition repairs broken or poorly flocculated solids. This second motivation is
not desirable as it represents a “band-aid” solution to an underlying issue in the feedwell.
Thickener Body
The thickener body, or tank, is a thickener’s most obvious element and its shape indicates its design
priorities. For a given flowrate, thickeners in a clarification duty will have large diameters with low
sidewall height. Where the intention is to produce a dense underflow, the conical section is steeper,
the sidewall is higher and diameter is often smaller to accommodate the rake torque capacity
limitations. Increased slurry bed depth has been correlated with increased compaction rates. (Hogg,
1993).
Thickener diameter is based on testwork data to achieve the desired overflow and underflow
conditions at the lowest life-cycle cost. If the thickener is too small, then either the supernatant “rise
rate” will be too high (which results in entrained solids in overflow) or there is insufficient residence
time in the compaction zone to achieve the design thickener underflow density. Testwork should be
used to determine the appropriate rise rate (m/h) and solids settling flux (t/m2/h) to match the design
requirements. An interesting optimisation option has been reported by Stoilov et al (2016), with the
retrofitting of lamella modules on an existing thickener to enhance its fines capture capability.
Thickener sizing should account for variability in both the quantity and quality of feed that is presented
to it. Testwork data should indicate the frequency and extent of slow-settling solids, such as clay
minerals. Where the thickener is located indoors (e.g. to prevent freezing), then design will favour
high rate type thickeners.
Thickeners can be mounted underground, with access tunnels or centrewell caissons to provide
maintenance access to the underflow pumps. These arrangements may present safety or
maintenance issues, and above-ground configurations are often considered preferable for these
reasons. However, even where the thickener tank is mounted on supports, the number and spacing
of the supports need to be considered if crane access is needed for pump maintenance. Cantilever
support of the tank floor may be required for pump access; this may be available if it is specified in the
initial stages of design, but the option becomes more impractical and expensive as design proceeds.
(e.g. Footings design, floor plate thickness, floor support structure.)
Above-ground thickener tanks sometimes include a central cylindrical discharge boot, which both
helps to support the thickener and provides the location for discharge nozzles (and some
instrumentation). Discharge nozzle sizes should include allowance for head entry losses in the
transition between the thickener boot and the underflow pump suction line.
Floor plates for above ground thickeners often use a catenary-profile “petal” to combine strength and
lightweight construction.
Materials of construction for the thickener tank become a significant issue where it represents the bulk
of the thickener mass and is exposed to the process fluid. Corrosion resistance is a significant
concern, especially in acidic, saline or high-temperature applications. The decision to use protective
coatings or complete alloy components for these situations needs to be made on a case-by-case
basis. The limitations of each option need to be considered. (e.g. Natural rubber in a high
temperature application.) Where epoxy coatings are used, their brittle nature should be considered
throughout the life cycle, including the hydrostatic testing phase. Coatings require non-destructive
testing and inspection throughout their life cycle, which must be factored into the ownership cost.
Thickener assembly, especially for the larger units, can be labour intensive and a burden on site
infrastructure during the construction phase. The thickener vendors have adopted various strategies
for reducing this cost and the cost of delivery logistics. To reduce site man-hours, assembly
techniques such as flanged joints between thickener panels have been proposed.
Contents of the larger thickeners represent a significant volume of process fluid, which needs to be
considered for containment and water balance considerations. Fortunately, sudden loss of
containment from a thickener is a rare event, but it can be provided for with a bund draining to an
emergency storage pond. A more mundane event is the draining of a thickener for maintenance
purposes. The sudden surges in the water balance (both on draining and refilling the thickener) are
an operational headache, especially where the application has a constrained water balance. A
suitably sized storage pond with provision for liquor recovery should be considered for these
conditions.
Thickener Rakes
The purpose of the thickener rake is twofold: to sweep the settled slurry from the outer perimeter of
the thickener floor to the central discharge point, and to increase the underflow solids density by
mechanically compressing the slurry against the leading faces of the rake.
Thickener rakes are usually centrally-driven using modest electrical or hydraulic drives to deliver high
torque, low speed outputs. Rake torque is measured and controlled to prevent damage to the rake
during high load events, and rakes are often fitted with rake lift mechanisms to deal with transient
high-load conditions. A secondary feature of the rake lift is that it enables the rake to be “driven
down” manually to break up donuts and other solids accumulations on the thickener floor.
Torque requirement is proportional to the yield stress of the given slurry, (Rudman et al 2008). As a
result, thickener rake assemblies have needed to become more robust as slurry density has
increased in recent years. Rake torque capacity is a critical factor in thickener design, especially
when handling viscous underflow slurry. Torque requirement increases with the slurry yield stress,
rake diameter and solids delivery rate.
The rake’s slurry delivery rate is calculated by the thickener vendor, usually with an allowance for
maldistribution of solids settling across the thickener, and assisted by the floor slope. Unless the
characteristics of the thickener feed are well understood, it is wise to make allowance for feed
variability and the expectations of subsequent process users in specifying the rake torque limits.
Dewatering rods (aka pickets) are often attached to the rake to improve the slurry compaction.
These consist of a series of rods extending vertically from the rake arms into the hindered settling or
supernatant liquor zones. They operate on the principle of compressing the slurry immediately
ahead of the rod; the clear liquor produced in this process collects in the low-pressure zone on the
trailing face, where buoyancy forces drive it out of the compact slurry zone. Hence it is important that
the dewatering rods extend well beyond the compaction zone to ensure that this avenue for liquor
release remains open.
The dewatering rod concept has been extended to the hindered settling zone to improve the rate of
compaction in this area. Industrial trials of this option are reported by Loan et al (2012).
Static rods, while not technically part of the thickener rake, are sometimes fitted inside the tank to
assist the raking mechanism by retarding the tendency of the slurry bed to rotate with the rake.
Thickener “donutting” is a common problem, often associated with over-flocculation or high-viscosity
slurry.
When a “donut” forms, the underflow slurry rotates as a massive bed with the thickener rake. In the
absence of the rake-bed speed differential, the rake performs minimal compaction of the slurry. The
thickener underflow discharges through a “rat-hole” above the thickener underflow nozzle, this
discharge is usually about the same slurry density as the feed. Slurry underflow pumps are not
designed to handle high volumes of low density slurry; as a result extra solids accumulate in the
thickener (on top of the rotating donut) until either the thickener overflows or the rakes bog. An
example is given in figure 4.
Figure 4 – “Donutting” of a 30 metre gypsum – metal hydroxide thickener.
The thickener rake is normally suspended from the bridge and drive mechanism. The base of the
rake shaft may be free-floating in the lateral dimension, or constrained by a concentric “steady
bearing”. In this latter case, the bottom of the rake drive shaft is located within a loose-fitting cylinder,
which prevents shaft deflection that would otherwise cause the outer arms of the thickener rake to
“dig in” when a rake strikes a solid bank of solids on the opposing side of the thickener. The steady
bearing saves damage to linings and to the rake itself, but still requires maintenance to deal with wear
and abrasion. It is also very important that the rake remains within the steady bearing – if this is not
the case, then the steady bearing can divert the thickener rake to adopt a planetary motion around the
outside of the steady bearing, which may damage the thickener rake and / or the sidewall.
Under exceptional conditions, the steady bearing can be subjected to heavy lateral loads; it needs to
be firmly secured.
Construction of rakes may justify the use of alloy metal instead of mild steel with rubber or other
coatings. This is because the wetted parts on the rake do not weigh as much as the thickener tank,
(hence lower materials cost) and the acute angles within a rake’s construction provide more
opportunities for imperfections in the coating surface, which then initiate the corrosion process.
Underflow Pumps
Thickener underflow pumps manifest a number of challenges, especially when handling high density,
high viscosity slurry. In these cases, the main issues arise from delivering the slurry to the pump
suction; these are further exacerbated when the suction line length is extended to meet maintenance
access imperatives. Usually, the pump itself is not the issue – it is a combination of inadequate
suction line design in the face of challenging slurry transport properties.
Slurry pump design for high viscosity slurry may include opting for “wide throat” inlet designs, which
were originally invented for froth pumping applications. These pumps are fitted with scoops
extending into the suction line discharge to promote free flow of slurry into the pump inlet.
Control over underflow slurry viscosity can be achieved by either controlling the slurry density or the
solids structure. Slurry density can be controlled by maintaining a thin compaction zone in the
thickener or by adding dilution water at the thickener underflow discharge nozzle. Both of these
options usually negate the purpose of the thickener and would not be considered as long term
solutions.
Slurry solids structure can be improved by altering the flocculant dosage or flocculant type, (Spitzer,
1992), especially in cases where the flocculant is over-dosed. Beyond this, where the slurry displays
shear thinning characteristics, the slurry structure can be modified to reduce its yield stress and
improve its transport properties. Proprietary shear thinning arrangements are available from the
thickener vendors; these involve withdrawing slurry from the thickener underflow, passing it through a
high-shear pumping system and injecting the modified slurry either into the thickener boot or the
underflow pump suction line.
A further option has been invented and patented by the CSIRO. (Rajisingam et al, 2016) Known as
DRT (Drag Reduction Technology), it operates by injecting a small volume of liquor into the suction
line to form an annular slip surface between the suction line wall and the bulk slurry.
It is suggested that many of the underflow pumping issues can be controlled by adopting a more
conservative approach to underflow slurry line design. Where thickener underflow slurry is
characterised by high density, high viscosity and a fine particle size distribution, the risk of solids
settling in the line is not relevant, and larger line diameters should be installed to enable efficient
delivery of the slurry to the underflow pump inlet. Also, as has been discussed above, nozzle entry
losses at the interface between the thickener boot and the pump suction line should be considered;
the thickener underflow nozzle should be oversized to manage the head loss at the nozzle entry, then
fitted with a reducer if necessary to match the pump suction line.
THICKENER CONTROL
Instrumentation
Responsibility for instrumentation and control is often shared between the thickener vendor and end-
user. Instrumentation duties for thickener controls may include:
Rake torque
Rake lift and level
Rake drive system controls (e.g. lubrication, hydraulics and cooling)
Bed pressure
Bed interface level
Slurry feed volume
Slurry solids feed mass
Thickener feed dilution flow
Flocculant dosage rate
Flocculant flowrate and distribution
Overflow clarity
Underflow density
Underflow slurry volume
Underflow viscosity
Shear thinning control
It should be noted, however, that most thickeners are not instrumented to this extent.
It can also be seen that the instrumentation and controls can be collected into the following duties:
Solids inventory
Solids inventory can be monitored directly using the bed pressure if the overflow stream density is
known. The bed pressure is related to the average density of the slurry column above the pressure
sensor by the relationship:
Where pressure is reported in kPa(g), density in t/m3, height is in metres, and gravitational
acceleration is 9.81 m/s2.
Once the geometry of the thickener is known (overflow launder and pressure sensor levels), then the
measured pressure during operations can be compared with the hypothetical “base case”, in which
the thickener would be filled with supernatant liquor. The “excess pressure” represents the solids
inventory.
Solids inventory can also be tracked – at least in theory - if solids flows into and out of the thickener
are measured. This requires measurement of feed and underflow volumetric rates and their
respective slurry densities. Ideally, this should be used for underflow pump rate controls, but
instrumental errors usually cause deviations from this. In these cases, the bed pressure can be used
as a trim value on the underflow pump output to account for flowrate and density value errors.
The level of the notional slurry interface within the thickener can be crudely estimated by assuming
that the thickener contents comprise a dense underflow (as measured at the underflow nozzle) and a
clear overflow. The “average” column density gives the ratio of the dense and supernatant heights
within this column. (Correction is necessary to account for non-linearity in the height-volume
relationship in the thickener cone.) This determination is a crude approximation, representing the
most optimistic estimate of supernatant volume in the thickener, because it does not allow for any
hindered settling zone between the compacted and supernatant phases.
The gradual nature of the slurry-supernatant interface makes direct measurement of the interface
location difficult. Ultrasonic measurements have had varied success, especially for low-powered
units. Physical measurement is more reliable but needs more operational and maintenance
resources. Physical measurements range from simple manual “dipping” of the thickener (e.g. air
sparge), sampling of the supernatant (e.g. “Sludge Judge”) to automated bed profiling (e.g. “Smart
Diver”). The information that can be determined from these systems is used to forewarn operators of
impending loss of bed level control.
Flocculant control
For very dilute (e.g. clarifier) feeds, flocculant dosage is normally in direct ratio with the feed slurry
volumetric flowrate. In other cases, flocculant delivery is controlled as a ratio of floc dosage (g floc/t
solids) multiplied by the solids mass flow. The solids flow rate is calculated from slurry flowrate and
the known (or assumed) slurry density. Flocculant dosage (grams of flocculant per tonne of solids
feed) can be a simple “plug in” process value but can also be determined by instrumental
determination of the feedwell solids settling rate. This measurement can be taken automatically using
proprietary instrumentation, (e.g. “Clarometer”), which is available from the flocculant vendor.
Flocculant efficiency is significantly affected by the slurry density at the point of flocculant addition.
Slurry density is controlled where necessary by dilution with process or supernatant liquor, to variable
degrees of accuracy. Where “passive” dilution methods are used, the dilution water flow is an indirect
consequence of another system property (e.g. differences in feedwell and general thickener surface
level, induced liquor flowrate into a flowing slurry stream), and control can be limited - and probably
non-linear with respect to thickener requirements. An “active” dilution system, either at the feed tank
or the feedwell, can be controlled in direct response to the thickener requirements as long as the feed
solids concentration, slurry flowrate and the dilution water flow are known.
Overflow clarity
Overflow clarity is normally determined by visual inspection and by manual semi-quantitative
measurements using a clarity wedge. The disadvantage of this method is that readings are infrequent
and tend to report a problem only after it has occurred.
Underflow density
Thickener underflow density is normally measured manually (using a Marcy balance or similar) or
continuously by using a nucleonic density gauge. Twinned with volumetric slurry flows, density
values are used to calculate slurry solids content for mass balance purposes, and also for thickener
diagnostic purposes.
Underflow viscosity
Underflow viscosity is normally inferred from pump output diagnostics (e.g. pump power) and is also
measured manually on laboratory samples. (e.g. Marsh Funnels, slump tests, laboratory viscometry.)
Deviations from the “normal” density-viscosity relationship can be used to initiate investigations into
the underlying issues (e.g. flocculant dosage, solids mineralogy or sizing).
In some cases, particularly in tailings pumping applications, an in-line slurry viscometer may be used.
Output from this instrument is used to protect the downstream line from bogging, and also to ensure
that underflow density is optimised to achieve a given slurry rheology.
THE FUTURE?
The thickener will remain with us, although its form may adapt to improvements in our understanding
of geometallurgy, materials, instrumentation and rheology. While the thickener may be partially or
fully replaced in some applications, such as CCD processing, by alternatives such as ion exchange or
resin in pulp (RIP), it is difficult to see a viable alternative to the thickener in most of its current
applications. Even though there is a trend to augment the thickener with a filter for dry stacked
tailings disposal, the thickener is always a practical and essential part of the flowsheet before the
filter.
Thickener developments have been driven by industry and academic initiatives from the last 20-30
years (e.g. CFD modelling, AMIRA Research) but feedback from these implementation steps have
been slow. Potential reasons for this are inadequate market awareness, conservative design, capital
constraints, variations introduced by the changes themselves and poor communications between
vendors, designers and operators.
At the risk of going out on a limb, here are a few ideas for future development:
Feed characterisation
Feed solids characterisation – specifically the particle size distribution and mineralogy, and their
effects on flocculation requirement and underflow rheology, are areas where the tools for
improvement are being developed but have not yet been applied in mainstream operations. Solids
mineralogy is especially important for understanding the nature of a resource and its potential for
meeting process design criteria. However, this is rarely done during process testwork, often
compounded by an imperfect understanding of the orebody beyond its geochemical value.
In the future, it is hoped that far-sighted project managers will include mineralogical variability as an
input variable in their resource investigations, and develop an understanding of this factor on
operational issues such as ore handling, slurry thickening and rheology. As mentioned above, the
more perceptive of the process investigators will identify the root causes of the unfavourable settling
and rheological testwork data so that long term solutions can be developed.
In an ideal world, the geological and process personnel would use the same vocabulary and testwork
methods for identifying problem mineralisation. This requires improvements in the understanding
from both disciplines but would be greatly facilitated if common “fingerprint” elements or parameters
could be used.
Flocculant control
The tools for improvements to the flocculation process have been identified in the AMIRA (and other)
research, but industry is still catching up with their implications. The importance of slurry dilution and
flocculant dispersion within the diluted pulp is recognised and is generally considered in design and
operation. However, the quantitative details of dilution control and flocculant addition are not pursued
rigorously, which is surprising, given the reagent cost and the costs of inefficient flocculation on
thickener performance. During design, flocculant dosage control falls in the gap between the vendor
and the owner; an integrated flocculant dosage strategy is rarely considered. For instance, where
the owner may specify a given overall flocculant dosage and flocculant dilution regime, the individual
flocculant solution flows to each of the floc dip tubes is not measured; where flow regulation is
installed, its “control” consists of the number of turns on a manual valve. This does not make for
consistent process control, especially when the controls are subjected to frequent tinkering by the
field operators.
It is often suspected that the location of the flocculant sparge locations are determined on a
reasonably ad-hoc basis, with no connection to the conditions used in the testwork. With increased
uptake of CFD feedwell design, this is expected to improve, but, notwithstanding the difficulties in
scaling up such parameters, it remains an area for further investigation.
Feedwell design
Improvements in computing capacity have opened the practical opportunity to use CFD modelling for
feedwell design. Besides regulating the degree of shear and the zones for quiescent settling in the
feedwell, these improvements can help avoid short-circuiting and unbalanced discharge from the
feedwell. However, this technology is not universally applied. Over time, this is expected to become
more widespread as newer thickeners are commissioned. In some cases, retrofitting of existing
feedwells (e.g. installation of shelves, relocation of flocculant sparges, baffles etc) will be justified.
Every time there is a change in thickener feed characteristics, there is the risk that the flocculant
dosage will prove inadequate, causing the hindered settling zone to rise, which, unless controlled,
results in a dirty thickener overflow. Especially in cases where feed and flocculant dilution relies on
clear supernatant liquor to be effective, this results in a short term shutdown (1-4 hours) to enable the
system to stabilise. The issue compounds itself if the thickener overflow is used for both flocculant
and feed slurry dilution, because the entrained solids add to the burden on the flocculant.
An obvious solution is to continuously report and trend the slurry interface level. This would enable
preventative action to be taken before a “recovery” operation becomes necessary. So far, no single
method for reading the interface level has achieved widespread adoption; perhaps none of the current
offerings combine effectiveness, simplicity and economy. (Or publicity.) Over time, the more
technically effective methods will be simplified to the point where they will obtain greater market
acceptance. Alternately, the ultrasonic instrumentation will become more sophisticated and capable
of interpreting the bed profile on a reliable basis (some claim this is already happening).
As an interim measure, installation of optical turbidity sensors on the thickener overflow and (perhaps)
1-2 metres below the surface would be a viable option. On the other hand, some would propose that
existing technologies could already perform the same function. In any case, there is clearly a need
for a reliable monitor of the overflow clarity to act as a contingency control.
At the qualitative level, endoscope technology will be applied to provide the operator with a remote
image of the feedwell and supernatant liquor conditions.
Underflow density
Obviously, the physical design of the thickener needs to be capable of achieving the target underflow
density. Besides providing sufficient surface and volume for settling and compaction in the thickener
body, the thickener design needs to include adequate rake torque capacity to handle a high density,
high viscosity slurry. Rake design will include properly designed dewatering rods to assist
compaction, but structural improvements in this area are still required to prevent the rods from
breaking off and fouling either the rake or the underflow pump.
Underflow pump and pump suction line systems will need to accommodate the increased rheological
problems associated with higher slurry densities. Part of this may be solved with adoption of shear-
thinning systems – where slurry properties permit – and future thickeners should be designed with
provision for retrofitting a shear thinning system if deemed necessary. Underflow nozzles will be
oversized to account for line entry head losses and to permit large bore pump suction lines to be
installed. Suction lines will be shortened, with provision for pump maintenance access provided by
changes to pump design, built-in monorail cranes, thickener support structures etc.
Alternately, axial booster pumps immediately downstream of the thickener discharge isolation valve
could be considered as a means of both partially shearing the slurry, and, more importantly, delivering
slurry to the underflow pump with the required suction head. An axial-flow pump may present an
elegant solution for high yield stress slurries, but careful design is necessary to manage the erosion
and maintenance issues it introduces.
From a controls perspective, stable operations are essential for obtaining optimum underflow density
on a reliable basis. This, in turn, requires that the thickener is fed with a known quantity of consistent
feed, and that it maintains a steady solids inventory. Taking efficient and effective flocculation as a
given, these steps need the following information:
Package thickener control systems are being offered to meet this gap, but uptake has not been as
enthusiastic as would be expected. The reasons for this are unclear.
If the thickener control system is designed “in-house”, then the control needs to be taken into account
from the earliest phase of design. On the other hand, there is an opportunity for the thickener vendor
to provide both the instrumentation and the logic for process optimisation.
CONCLUSION
Over the years, an impressive volume of high-quality development work has been conducted to gain a
better understanding of thickener design and operation. However, the continuing problems appear to
be due not so much to a lack of knowledge, but a failure to apply it. It is felt that most of the future
improvements that can be made will be made using existing knowledge. Its application will require
improved understanding and communications between the disciplines and organisations involved in
the design, supply and operation of the thickener.
ACKNOWLEDGEMENTS
I would like to thank SNC-Lavalin for their permission to publish this paper, and my colleagues at
SNC-Lavalin for their constructive reviews of the various drafts. I also wish to thank the owners of the
images used in this paper for their permission to use them. Finally, I would like to thank the many
investigators over the years, whose work has made recent and future thickener technology
improvements possible.
REFERENCES
Barnea, P.M., 1978. “Mathematical Unification of an Equation for Solute Recoveries in
Countercurrent Decantation.” J. South African Inst. Mining and Metallurgy, Jan 1978, pp 143-5.
Bickert, G. Langer, B. 2011. “Filtration Test Work – Extracting the Whole Story for Studies and
Designs.” MetPlant 2011, AusIMM.
Fawell PD, Owen, AT, Grabsch, AF, Benn FA, Labbett, DM, and Swift JD. 2009. “Factors Affecting
Flocculation Within Gravity Thickeners”. AusIMM
Farrow, JB. Johnston, RRM. Simic, K. Swift, JD. 2000. “Consolidation and Aggregate Densification
During Gravity Thickening.” Chem. Eng. Journal 80 (2000) pp 141-8.
Gollaher,T, Johnson, JL, Biesinger, MT, Accioly, AHL, 2010. “Paste Thickening Tailings – Recent
Examples of a Rapidly Emerging Technology”. SME Annual Meeting, 2010
Gregory, J., O'Melia, CR. (1989) “Fundamentals of Flocculation”, Critical Reviews in Environmental
Control, 19:3, pp185-230 DOI: 10.1080/10643388909388365
http://dx.doi.org/10.1080/10643388909388365
Harris, GR. 2002. “Rakeless Thickener Operation at St. Ives Gold Plant.” MetPlant 2002.
Heath, AR., Arbuthnot, I. 2013. “Advances in Thickener Flocculant Delivery”. MetPlant 2013,
AusIMM.
Heath, A., Barycki, A., Baladran, J., 2014. “Vane Feedwell Flocculant Sparger Optimisation Trial”.
International Minerals Processing Conference, IMPC. 2014.
Hendriksson, B. 2006. “Thickening Rake Mechanisms” International Mining (Supplement) April 2006,
pp 4-7.
Langton, M. 2013. “Development of the Net Generation Thickener Featuring the Radlow Feedwell.”
ALTA 2013.
Loan, C, Lawler, N and Arburthnot, I. “Industrial Scale Results from Shear Enhanced Thickening
Technology”. AusIMM, 11th Mill Operators’ Conference, 2012.
McCarty, MF. Olson, R. 1959. “Polyacrylamides for the Mining Industry”. Mining Engineering, Jan
1959, pp 61-5.
Page. P.M., 1976. “A Simple Equation for CCD Equations.” EMJ, Oct 19756, pp 86-7. Reprinted in
EMJ Operating Handbook of Mineral Processing, pp 165-6.
Pearse, MJ, 1980. “Factors Affecting the Laboratory Sizing of Thickeners.” AIMMP, 1980.
Rajisingam, R, Pawelek, M, Wiltshire, PG, Wu, J., Robinson, Dave. 2016. “CSIRO Drag Reduction
Technology for Improved Slurry Pumping – An Industrial Trial at Glencore’s Murrin Murrin Operations
Pty. Ltd.” ALTA 2016.
Rudman, M. Paterson, DA, Simic, K 2010. “Efficiencies of Raking in Gravity Thickeners” Int. J. Min.
Proc. 95 (2010) pp30-9.
Rudman, M. Simic, K. Paterson, D. Strode, P. Brent, A. Sutalo, JD. 2008. “Raking in Gravity
Thickeners” Int. Journ. Min Process (86) 2008 pp 114-30.
Spitzer, D, and Avotins, PV. 1992. “The Effect of Flocculants on Rheological Properties of Thickener
Underflow” SMME, Chapter 33.
Triglavcanin, RA. 2008. “Designing Feedwells for the 21st Century. MetPlant 2008, AusIMM.
Wallace, J. 2005. “Increasing Leach Capacity through Paste Thickening”. ALTA 2008.