Cape Town
Workshop
t 30 March 2004
Seminr
MINERALS 3I March - 2 April 2004
Site Visit
3 April 2004
Organised by:
Paterson & Cooke Consulting Engineers (Pty) Ltd
School of Civil & Environmental Engineering
University of Witwatersrand
Paste
2004
The organisers wish to thank the following sponsors for their support:
Anchor Sponsor
UTIRTH
Function Sponsors
INERALS
Seminar Sponsors
PATERSONaCOOKE
ER CO I{ 6 U LTI f{G EIIG IN EERS ECMP
FLOW PROCESS
RESEARCH CENTRE
Paste
2004
Programme Paste 31 March
2004 2 April 2OO4
Wednesday 31 March2004
Time Title
Topic / Authors Section
15.00-15.15 Discussion
Paste
2004
Page I
Programme Paste 31 March
/., 2004 2 April 2OO4
Friday 2 April2004
,+ut5
Time \ Topic/Title Authors / Speakers Section
8.30-9.15 \\ Keynote Address: Paste - as Backfill for Hartmut Ilgner 22
South African Mines?
9.r5-9.45 Boulby Mine Backfill System: ilkins, Chris
Commissioning and Operation Gilchrist, Michael
Fehrsen and Robert
9.45-10.15 Design Considerations for a 400 tph Paste Chris Lee, David 24
Backfill Plant at Kidd Creek Mine Landriault, Roy Durr
10.15-10.45 ExhibitionArea Tea / Coffee
10.45-11.15 Comparative Economic Frank Palkovits, R
Assessment of Mine 25
Backfill Systems Tenbergen and D
Landriault
n.-IL.45 Case Study from Orapa Mine. Co-thickening G J van Ryssen, A M 26
and Cycloned Dam Disposal System for Copeland and P van
Kimberlite Slurry der Westhuyzen
IL45-I2.I5 Development and Implementation of Gordon McPhail, 21
Thickened Tailings Discharge at Osborne Alasdair Noble,
Mine, Queensland, Australia George Papgeorgiou
and Daniel V/ilkinson
12.15-12.45 Discussion
Paste
2004o
Page 3
The Presenters Paste 31 March
2004 2 April 2OO4
Guillaume de Swardt, BSc (Civil) Eng, MSc Eng, Pr Eng
Director, ECMP (Pty) Ltd
Guillaume has been directly involved in the design, construction and operation of residue
disposal facilities since 1992. He is currently a director and partner in ECMP (fty) Ltd. His
main focus in ECMP is the operation and management of mine residue disposal facilities with
the line responsibility of Operations Director. His current interest is to develop all inclusive
risk, communication and information systems for mine residue disposal facilities.
Started Junior Engineer in Mining Industry 7974 tlIl 1983. Involved in various projects
ranging from shafts, metallurgical plants to shaft sinking. Completed 3rd and final year
Engineering full time at University Pretoria 1975 and1976.
Started own Mining Supply Company (1983 onwards). Developed alumina ceramics as high
impact chute liners and underground settlers for retrofitting. Introduced 3CR12high pressure
vertical and horisontal pump columns and thin wall compressed air columns into the mining
industry.
Developed, introduced and supplied the first of the now called Ultra Highrate Thickeners to
Vaal Reefs and Western Deep Levels. Patented the Ultrasep High Performance Thickener /
Clarifier in 1995. Joint venture with Bateman 1995 to 1998 to launch Ultrasep successfully.
Started Magra Process Engineering L999.
Patents: Flocsaver controller 1999, Magra Ultrasep 2000 in 2000, Polyprep in 2000,
Magrabrator 2003.
Andy Fourie graduated in civil engineering from the University of Witwatersrand in 1978 and
obtained and MSc@ng) from Wits in 1980. After working as a geotechnical engineer for
Steffen, Robertson and Kirsten in Johannesburg he obtained his PhD in geotechnical
engineering from the Imperial College, London. He lectured in geotechnical engineering for
four years at the University of Queensland in Australia before returning to South Africa in
1988. He is currently the Professor of Construction Materials in the Department of Civil
Engineering at V/its.
Page 5
The Presenters Paste 31 March
200,4 2 April 2OO4
Jahn Hohne has been involved in diamond exploration and mining since the mid-1970's
mostly in the Kimberley region of South Africa. His special field of technical expertise lies in
the design, construction and operation of a diamond recovery plant and equipment. He
currently operates and co-owns a million tonne per annum treatment plant for diamond
recovery from Kimberlite dumps in Kimberley as Super Stone Mining. He operates and owns
a fleet of mining and earth moving machines capable of handling a million tonnes of material
per month on contract to De Beers and several other small miners. He is also the managing
director of Ekapa Mining (Pty) Ltd a BEE mining company that inter alia owns and operates a
deep cone high rate paste thickener.
Bemard Krause graduated at the University of Pretoria, 1993 with a degree in metallurgical
engineering. His career started at ISCOR's Thabazimbi hon Ore Mine as process engineer
(1994 to 1998). He then accepted a position at ISCOR's Durnacol Coal Mine as snr process
engineer (1998 to 2000). He is currently employed at Ticor South Africa's Wet Heavy
mineral separation plant as snr process engineer (2000 to date).
His responsibilities included plant and process optimisation for iron, coal and heavy mineral
wet beneficiation plant as well as design and commissioning of a heavy mineral wet
separation plants.
PageT
The Presenters Paste 31 March
2004 2 April 2OO4
Frank Palkovits, P.Eng.
Senior Project Engineer, Golder Paste Technology Ltd.
Mr. Palkovits will be the Area Manager for the underground distribution system. He is a
Senior Project Engineer who graduated from Laurentian University with a B.Sc.Eng. (Mining
Engineering) in 1988, and from Cambrian College (Geological Technician) in 1981. Prior to
returning to university, Mr. Palkovits worked at INCO Limited in Sudbury and at Echo Bay
Mine's Lupin Mine. Upon completion of university, he worked at INCO Limited in Sudbury
for 12 years in the Mines Engineering and Operations departments in various capacities
including project and engineering supervisor and production supervisor. Since joining Golder
Paste Technology Ltd., he has been involved in conceptual through basic and detailed
engineering of paste backfill projects, including site visits and site reviews, estimations, flow
modelling, and U/G distribution system design, on-site flow loop testing and commissioning.
Mr. Palkovits is a registered Professional Engineer in the Province of Ontario.
Dr Angus Paterson graduated from the University of Cape Town in 1984 with a BSc@ng)
degree in Civil Engineering and completed his Masters degree in 1986. In 1987 he started
studies towards his doctoral degree under the supervision of Professor John Lazarus of the
Hydrotransport Research Unit at UCT. His thesis concerned the analysis and flow behaviour
of high concentration metallurgical full plant tailings used as backfrll in deep level mines.
While completing his PhD he worked as an independent consultant to several mining houses
and developed specialist software for the design of backfill distribution systems. During 1991
he worked for a firm of consulting engineers in Cape Town, and later left to form Paterson &
Cooke Consulting Engineers. Angus has worked on numerous slurry transport systems and
currently is working on several paste and thickened tailings projects.
Paul Slatter Pr.Eng. BSc(Eng) MSc(Eng) PhD NHD(PSE) FSAICE MRSSAf MWISA
Prof Paul Slatter founded the Flow Process Research Centre in 1993 at the Cape Technikon.
He has twenty-five years of teaching experience at the Cape Technikon and has received
fifteen Awards and Recognition at International Ivel for Achievement of Excellence. Prof
Slatter serves on several international committees and editorial boards and has produced over
200 educational and technical engineering publications. He has a strong commitment to
industrial research, and was a key member of the highly successful research team for the
AMIRA P599 project on high concentration suspension pumping. Prof Slatter presents
regularly to industry courses and workshops both in South Africa and abroad, and was
involved in the hrst Paste Workshop held in Melbourne in2003.
Paste
200,4 o
Page 9
The Presenters Paste 31 March
2004 2 April 2OO4
Andrew Vietti B.Sc. (Hons) M.Sc.
Senior Research Offrcer - De Beers Consolidate Mines Ltd.
Over the course of the next 13 years Andrew has been intimately involved in the development
of a water reclamation and tailings disposal system from laboratory through to pilot and
production scale within the De Beers Group which includes South Africa, Botswana;
Namibia; Tanzania and Canada.
Andrew is currently registered and studying towards a PhD in the held of Thickened Tailings
Deposition.
After graduating from Camborne School of Mines in UK in 1987, I was employeed by Impala
Platinum in South Africa in the Concentrator department, starting as a graduate metallurgist
and eventually becoming a production superintendent before moving to Australia in 1994 to
work for MIM at their Mt Isa Copper Concentrator. In 1997 I moved to Perth to accept a
postion as sales engineer with Eimco Process Equipment. Eimco vere acquired by GL&V
worldwide in2002 incorporating the Eimco, Wemco and Dorr-Oliver equioment brands under
the same company. I have recently relocated to our Gosford, NSW office to take up a positon
as Product Manager.
John Wates is a specialist in waste management and environmental geotechnics. He has been
responsible for design of many mine tailings storage facilities, and industrial and domestic
waste disposal sites. Projects involving chemical pre-treatment, filtration and thickening of
tailings and waste and cycloning, spigotting and open-ended deposition have been carried out
under his supervision. John has also been project leader for projects involving environmental
impact assessment, environmental and strategic risk assessment, closure planning, pumping
and pipelines and civil engineering. He has been approved as a category III water dam
engineer under South African law. John is currently the Business Development Director for
Golder Associates Africa. He is an active member of the South African Institution of Civil
Engineers and has served as chairman of the geotechnical and environmental divisions.
Paste
2004 o
Page 11
International Seminar on Paste and Thickened Tailings
Paste 2004
Paper I
Joe McAlinden
Keynote Address
Paste 2004
Cape Town
South Africa
Paper 2
Richard Jewell
Richard Jewell
Australian Centr
Abstract
The implementation of Paste and Thickened ffngs (P&TT) tehllogl on any mining
operation can only be justified if it is the mot cost effective solution availab.
avai The cost of
the plant required to tryke!_lqllgltransport a@not insignificant
and it is only *h.({ll5:9costs are taken into account that a true economic
comparison is possible. The imlementation of P&TT technology however can provide a
number of specific benefits to the extent that these become the reasons or "drivers" for
adopting the technique. This paper looks at these drivers in the context of above ground
tailings storages in Australia and the details of two specific operations are used as examples.
1. lntroduction
The thickening of tailings prior to deposition in above ground storages is a technique reported
by Williams (2000) as being first implemented in Australia in the early 1980's at the Elura
lead/zinc mine near Cobar in New South Wales. Cooling (2002) reports that after
considerable development work Alcoa adopted "dry stacking" for its Western Australian
Alumina refineries in 1985. Since then, thickened tailings deposition has been adopted by all
of the Alumina operations around the country as well as by a significant number of
metallurgical mining operations.
The reasons for implementing the thickening of tailings at these operations varies between
sites, but the common denominator in all operations is to save on the costs of storing the
tailings safely while ensuring minimum potential for pollution of the environment. Saving on
Paper: Thickened Tailings in Australia - Drivers Page 2
Author: R Jewell
costs satisfies the need of the operator to optimise profit, while the provision of a safe and
stable storage facility contributes to meeting the demands of the stakeholders and to the long
term "license to operate" of the mining company. U q,,1rP '' -
(
To fully evaluate the benefits of P&TT technology, an economic study based on "full life-
cycle costs" is needed. Capital costs, operating costs, timing and the time value of money
need to be applied across the full spectrum of the mine plan, including closure. Often,
complete mine and./or tailings plans need to be run to closure to truly understand the benefits
of this technology. Non-monetary benefits like improved public perception should also be
Savings are possible in water, energy and reagent conservation, reduced impoundment needs,
improvements in impoundment and embankment stability, more rapid closure and reduced
financial provisions. Additional capital and operating expenses for thickening equipment,
pumps and piping may pafally ofet these savings. In this paper, a brief introduction will be
given to these reasons for thickening and the relevance and impact of these "drivers" will be
discussed in the context of operations in which tailings are stored in above ground tailings
storage facilities (TSF's).
2. Drivers
would. come from both regulatory and operational areas. It appears however that in Australia
at least there are no formal pressures applied through legislation or by regulatory authorities.
The regulators prefer to leave the choice of how to manage tailings to the operators on the
understanding that a safe and stable structure will have to be achieved before an application to
relinquish a mining lease and its obligations will be considered. It is a case of telling
operators what the end product has to achieve and leaving it to them to decide how to get
there. In this environment, any decision to implement P&TT technology will be driven by
operational issues as will be discussed further below.
In Section 2 entitled Key Business Issues in the ACG publication by Jewell et al (2002), the
authors provide a very comprehensive review of the costs and benefits of the issues to be
considered in evaluating the proposed use of P&TT technology for any given project. These
include:
Tailings Containment:
. Since high-density tailings have less "free" water and can be discharged to form an
elevated cone, or stacked nearly horizontally, they do not need high specification
embankments comparable to those required for "conventional" wet tailings. Hence,
substantial savings can be realised through reduced embankment volumes for
comparable stack heights. Reductions in capital cost will result from the reduced need
for engineered embankments around Tailings Storage Facilities (TSF's), particularly
in the early years of project development. Furthermore, the reduced volume of
surpematant water reporting to the TSF minimises the need for underdrains, toedrains
and recovery bores to protect ground and surface waters from contaminated leachate.
ring main and spigots. However, transport of high-density slurries and pastes often
involves higher pipeline pressures and hence more expensive pressure rated pipes and
fittings. The pumping pressures may also require the installation of positive
centrifugal pumps.
Additional operating costs for thickened tailings stem mainly from flocculants, thickener /
'Where
cyclone I filter maintenance and piping and pumping operating costs. filters are used,
the higher cost of transportation of the dry material may be a factor. Of these costs, the
flocculation costs can be the most unpredictable and onerous. Savings and other benefits
include:
Water Conservation:
o In areas where water is expensive or scarce, retaining process water within the plant
can provide savings in both capital and operating costs.
Reagent Conservation:
o In some mining operations, companies may not want reagents exposed to or lost to the
environment for economic or environmental reasons. For example, savings can be
achieved in the alumina industry when caustic soda is recovered by prewashing or
from process water at the thickener, rather than being deposited with the tailings into
the tailings storage facility and importantl this reduces the potential threat from
caustic soda leaching into the groundwater. On the other hand, the use of cyanide in
the gold industry and the use of xanthates in base metals flotation, both of which break
down in sunlight can pose environmental concerns.
A tailings thickening operation allows recirculation of the process water within the
plant without having it first deposited in the TSF, thus considerably reducing reagent
loss to the environment.
The rehabilitation of tailings is frequently the most difficult, time consuming and expensive
component in the closure of a minesite. Tailings are often a mining company's single largest
liability during closure and have the potential to provide the largest cost savings. Govemment
agencies can require that expensive performance bonds be held for a long time for
conventional wet tailings facilities and any technique enabling rehabilitation to be expedited
has the potential to shorten that time frame.
phreatic surface resulting from free water in a decant pond or ongoing consolidation.
This can drive the transport of contaminants into ground or surface waters. Such
leakage can require the allocation of financial provisions to cover this contingency, as
-:ffi;,
W - tr,,t, .tru,',,
3. Examples
3.1.1 Background
The Peak Gold Mine (PGM) has been operating a Central Thickened Discharge (CTD)
tailings storage facility (TSF) since mining operations cormencedin 1992. During 11 years
of operations, the TSF has been operated to a large extent as first specified by the designers of
the facility and very few modifications to the TSF design and its operating strategy have been
necessary in that time. An oblique aenal view of the mine and the approximately 1200 m
The CTD facility at PGM is a very good example of the innovative use of the surface
topography to store the tailings in a shallow gulley adjacent to the plant. The principal
advantages of this approach compared to a conventional'addock" type TSF include:
Thickening at PGM is achieved with a High Rate Thickener located in the plant and the
underflow is currently pumped by a single'Warman 312 centnfttgal pump to the TSF through
a 150 mm dia. HDPE pipeline. The tailings from the CCD circuit feeding the thickener have
a solids content (wlw%) of the order of 35Yo and the thickener is currently producing
approximately 2,100 tonnes of tailings per day (700,000 tlyr) at an underflow solids content
of 60 to 61%. This throughput and density is probably close to the optimum for the plant and
both have been increased significantly with confidence gained in operating the plant over time
and the need for higher density underflow to provide steeper beach angles.
Figure 1: Aerial view of the CTD TSF at the Peak Gold Mine
The tailings were originally constrained solely by the low filter wall built in L992 that can be
seen at the toe of the tailings in Figure 1. This was sufficient to contain just under 2 Mt of
tailings deposited from the single discharge ramp near the head of the gulley. In 1996 the
filter wall was relocated just downstream of the original filter and increased in height and then
in 199912000 the filter wall and discharge ramp were raised again and 1 km of saddle
embankments were constructed in low points around the perimeter of the TSF. These
measures were designed to increase the storage capacity of the TSF to the original design
capacity of 5 Mt. The total cost of toe filters and saddle embankment construction to that
stage was less than AUD$IM.
The proving up of additional ore bodies has by now more than doubled the quantity of tailings
to be deposited in the TSF from that initially envisaged. By March 2001,3.96 Mt of tailings
had been deposited in the TSF, 0.532 Mt had been used in mine backfill and the forecast was
for a funher 6.6 Mt to be stored in the TSF. Measurements of the cone of tailings up till that
time showed that the beach was slightly concave, the maximum slope near the crest of the
cone was around 2o/o and the average slope of the order of L75%o. In order to store additional
tailings, the basic options available to the operators were to extend the footprint of the TSF or
alternatively to increase the slope angle of the "cone" of tailings and hence store more
material on the existing footprint. By modifyng the discharge from single to multiple point
discharge at the head of the ramp in 2002 and increasing the tailings density, the operators
have since increased the slope angle at the top of the slope to a reported 4o/o. By January
2004, some 6 Mt had been stored on the site and by 2008 the operators expect to have stored
8.5 Mt without having to extend the existing foorint.
. savings in the volume of make up water required, and in pumping costs due to the
recovery of significant volumes of water from the tailings in the plant.
The capital and operating costs associated with thickening additional to a conventional plant
include:
o the cost of the thickener installation and operations
o the cost of installing, maintaining and operating higher capactty pumps and pipelines
o the cost of flocculants used in the high rate thickener.
.
4
At the time of this survey, the TSF contained approximately 3 Mt of tailings and the toe of the
cone of tailings had reached virtually none of the perimeter saddle embankments. By
comparison, the depth of tailings in a paddock TSF on flat ground with a total foorint of
1200 m by 700 m would have to average more than 4 mto store 3 Mt of tailings assuming an
average density of say 1.2 tlm3 and the height of the perimeter embankments would be
The discharge pipeline extends about 1500 m from the plant to the end of the discharge ramp.
With a paddock TSF, this would have to be extended considerably to form a ring main around
the crest of the perimeter embankments. There would probably be no savings in a return
water pipeline as the water (rainfall and supematant) collecting in the catchment area below
the filter is returned to the process water pond adjacent to the plant.
One of the plant operators undertakes a daily inspection of the TSF to ensure that the
discharge at the end of the ramp is operating satisfactorily, that no water is ponding around
the perimeter of the cone of tailings and that the decant at the filter is free. This involves
considerably less time than would be required on a paddock TSF to rotate the spigots in use,
lift the ring main at each upstream embankment lift, maintain the operations of the decant
pond and return water pump, etc.
Using the phase relationships for tailings as for example provided by Vick (1990), and
assuming that the Specific Gravity of the solids (ore) is 2.7, it can be demonstrated that the
The initial cost of the thickener and the cost of flocculant used in the thickener are so
dependent upon the specifics of any operation that they will not be covered further here.
There may be additional costs involved in maintaining the thickener and transport system and
there will certainly be additional hassles for the plant operators with the thickener. The
properties of the thickener feed can vary considerably with the source of the ore. Added to
that, the limitations of instrumentation to monitor densities etc in-line mean that it can be very
difficult for operators to keep track of what is happening in a thickener. Hence, operating a
thickener to produce underflow densities as high as possible without bogging requires
committed and experienced plant operators.
3.2 Alcoa
3.2.1 Background
Alcoa currently produces six and a half million tonnes of alumina (and 14.1 Mt of tailings)
annually at its three Western Australian refineries located at Kwinana, Pinjarra and Wagerup.
The tailing storages for the Kwinana refinery are based upon the coastal sand plain and
despite constructing a thick clay seal under the original wet storage facilities, ground water
contamination quickly became an issue. This led to a number of development projects aimed
at lessening both the potential environmental and economic impacts of the residue storage
operations and ultimately to the introduction of what is termed "dry stacking" of thickened
tailings to all these Alcoa operations in 1985.
Author: R Jewell
An overview of the Kwinana operations are provided in Figure 3 showing the 160 ha "F"
Lake in its original wet storage configuration.
A comprehensive account of the development and operation of these residue storages has
been given by Cooling (2001), in which he notes that the initial costs of establishing dry
residue storage at Alcoa's three refineries exceeded $150 million. In this he lists the benefits
which have justified this expenditure as:
. A higher density deposit can be achieved reducing the overall volume of stored
residue.
o The progressive stacking allows the storage area to be taken to a height which would
not be economic with wet impoundments (it is currently planned to stack to a height of
40-50 m compared to an economic limit of 20 m for a wet impoundment).
o The higher density and increased deposit height means less land is used.
The exposure of less land area to residue and the drained condition of the deposit
significantly reduces the risk of ground water contamination.
o Safety hazards to people and wildlife are reduced
Improved surface stability and drainage mean that completed areas can be rehabilitated
quickly.
The conical shape of the final deposit surface provides a well drained surface for
rehabilitation compffed to the poorly drained convex surface shape which results from
wet storage.
high torque rating of 13 MNm. The thickeners have a capacity of 450-500 t/hr and cost
around AUD$30M each. They utilise synthetic flocculant at a rate of around 40 - 60 grams
per tonne of solids to aid settling and produce a slurry at a density o148-50o/o solids.
The Dry Stacking operations tpically receite a slurry at a density of between 700 - 800 gpl
(47-51% solids byweight). The yield stress of this material ranges between 25 and 50 Pa.
The slurry with these properties will spread across the drying bed at alayer depth of around
300 mm -500 mm on a bed slope of 1-L2 % (1:100 to 1:80 slopes). The target strength for
the dry stacking operation is a minimum of 25k Pa, the strength required to maintain a
minimum factor of safety against earthquake loading. This drying is enhanced by "mud
farming" activities that expedite evaporation. The time taken to achieve this minimum
strength varies according to the time of year, but is around 42 days at the height of summer.
Figure 4 presents an overview of another Alcoa residue operating area after being retrofitted
to dry stacking. The liquor ponds in the foreground serve as winter storage of alkaline run-off
water from the Refinery and the residue drying beds. The operational drying beds are the
areas to the left and rear of the photo. The areas in the centre of the photo are wet storage
areas used prior to the introduction of dry stacking. These areas will eventually be reclaimed
and converted to drying beds. The thickener can be seen between the operational drying beds.
The capital costs of the disposal areas for Alcoa vary between AIID$0.5M and AUD$llfha
and is highest at Kwinana. This provides a strong incentive to minimise the land arca
occupied by tailings storages by increasing the density of the tailings and the height to which
they are stacked. The dry stacked tailings achieve a solids content of around 70%o at the target
shear strength of 25 l<Pa and this is considerably higher than would be achieved in the same
time by wet residue. Similarly, this higher shear strength enables dry stacked tailings to be
built up to heights of 40 to 60 m on any given foorint compared to a maximum of around
20 mfor wet disposal.
Because of the decreasing surface area of the TSF as the stack height increases, the practical
height limit comes when the surface area is not large enough to provide the space required to
'WA
dispose of the drying tailings. The rates of storage achieved by the Alcoa locations are
all around the 14,500 tlyrlha range. For Kwinana, which produces 1.3 million tonnes of
residue mud annually, this translates to a minimum dtytttg area of 90 ha.
The initial costs of establishing dry residue storage at Alcoa's three refineries exceeded $150
million and the payback period for this expenditure has been estimated to be between 7 and 8
years. The operating costs ofthis dry stackng procedure are taken to average around 70% of
that involved for wet residue disposal.
There are savings in water consumption in that evaporation losses are considerably less with
dry stacking. There are savings in the recovery of caustic soda, but these are not as great as
might be imagined in that much of it cannot be recovered due to contamination in the plant
from organics in the primary ore, and pre-washing is undertaken prior to thickening to limit
the amount left in the tailings and potentially available for pollution.
The side slopes of the embankment are maintained at an overall 1 in 6, but there are savings
in embankment construction costs with dry stacking in that the embankments can be built of
the sand fraction of the tailings because of the low potential of seepage from the dry tailings,.
With the wet residue storages, the embankments were constructed of imported borrow
materials with clay liners to limit the potential for lateral seepage and pollution.
4. lntanqible drivers
A major driver in many operations, but unfortunately one that is almost impossible to quantify
is the impact the TSF has on the environment. This has a direct impact on the sustainability
of the operation, the company and indeed the industry. Even the perception of a potential to
degrade the environment (physically or visually) or pose a threat to the health of people of
wildlife can cause angst to those responsible for an operation where they are situated in a
sensitive environment and./or close to habitation and this is especially so for the Alcoa
operations.
At risk is the "license to operate" and this has very important implications for the ability to
gain permission for future mining operations. It is suggested that this means that mining
companies have to pay special attention to keeping community based liaison groups active
and well informed and to ensuring that tailings storage facilities are maintained and operated
such that they are safe, stable and have little potential for posing a threat to the environment.
The thickening of tailings to a paste or high density slurry prior to discharge is conceivably
the most positive technique one can apply towards achieving that state.
5. References
Cooling, D.J. (2001) "Thickened tailings storage as practised by Alcoa V/orld Aluminium"
Proc High Density & Paste Seminar 2001, Pilanesberg, South Afca, 10/11 May 2001.
Cooling, D.J. (2002) "Alcoa V/orld Alumina, Australia", Paste and Thickened Tailings - a
Guide, Published by Australian Centre for Geomechanics, eds Jewell, Fourie and Lord, pp
146 - r49
Jewell, R.J., Fourie, A.B. and Lord, E.R. (Eds) Q002) '?aste and Thickened Tailings - a
Vick, S.G. (1990) "Planning, Design, and Analysis of Tailings Dams", Second Printing, BiTech
Publishers, Pages I2-I5.
'Williams,
M.P. (2000) "Evolution of thickened tailings disposal in Australia", Proc Paste
Technology 2000 Seminar, Perth, L3ll4 4pri12000.
Paste 2004
Paper 3
J Stephen Slottee
/-z-
c? L**-< , eG-,-. ,:t*t
ABSTRACT ,V r-( ry- /4
rtb-.-;.1
Paste thickners without filters
"un "f"ltif
higher, dependi ing on the particle size distribution of the solids'
slumps
Slumps ap ches may be produced for certain applications. Early uz, /-', y'
integrated design
4
installations of paste thickeners have not effectively utilized an systems
for matching the thickener to the pumping system and the mine backfill plan. The destgn
of a paste backfrll process always starts with what is needed by the mine. Requirements
for strength and cycle time translate into binder cost and tailings paste slump, whether
from a thickener or hlter. Producing mine paste backfill using a paste thickener without
filtration is a technology that should be considered for the right application.
INTRODUCTION
Although used for thickened tailings disposal of red mud in Australia for many years, the
applicaiion of a thickener for making paste and thickened tailings outside of alumina is
rltivety new. Interest and installations in thickened tailings (also called paste) are
rapidly growing. Water shortages and environmental pressures are making consideration
of tfri.t tt"d tailings necessary in most new projects and in many existing plants. The
paste thickeners at the DeBeers CTP plant in Kjmberley, South Africa are the latest
example of thickened tailings disposal. Houman (2003)
The use of thickeners to produce paste for mine backfill is less developed than for surface
disposal but has attractive potential as an alternative to filtration for the right application.
This paper describes the recent use of paste thickeners for mine backfill, lessons leamed
fromtr early installations, and the integrated approach required for designing a mine
backfill process using a paste thickener.
A detailed description of a paste is outside the scope of this paper and can be found in
other sources such as Boger (1999) and Robinsky (1978)' In summary, paste a
suspension is characterizedby relatively high solids (75-90 wt%) content and non-
settling, non-segregating particles. The underground mining industry commonly uses a
tZ-inihslump o.r" -"u.utement as an indirect indication of the rheology and solids
content of a tailings paste. The slump and non-settling nature of paste is produced from
naturally occurring clays, silts, and fne sands. Tailings from milling or processing are
most frquently used for paste backfill. Moisture content for a given slump is determined
by the specific gravity and particle size distribution of the solids. Large particles can be
added to paste without changing rheological or transport characteristics. Pastes are
generally described as Bingham plastics with a yield stress, which is related to the
pr".ru.rt"quired to make a paste flow. Slump is an indirect measure of yield stress. The
high ,ri."ority of pastes makes pipeline design critical. Pastes are generally pseudoplastic
*h"r" the viscosity decreases with shearing such as in pumping and pipe line flow- The
slump of paste produced by a thickener, for example, may be higher than the slump at the
end of the pipeline delivering the paste to the mine stope'
Hydraulic backf,rll using sand-sized particles from cycloning for mine support has been
used for over fifty years. Paste backfill, which incorporates fines as well as coarse
particles, such as the complete spectrum of a mill tailings particle size distribution ("total
tails"), is a more recent development.
The design of a paste backfill system is dependent on the backfill materials available at
the mine site, such as mill tailings. Paste is produced by filtration or thickening. If
backf,rll strength is required, the paste is mixed with a binder, such as portland cement
and water is added to adjust the slump as necessary for pumping and pipeline flow. The
paste/binder mixture is pumped to the mine borehole for gravity transportation to the
rtop"r. The slump required in the mine is a function of the mining plan and cycle time
for accessing stopes. Typically a slump of 6-10 inches is required, and if the tailings are
too fine to produce the required strength with binder addition the tailings may be mixed
with alluvial sand or crushed rock to produce a wider particle size distribution and higher
strengths as described by Landriault (1995). Mine backfill with a paste is also a method
of tailings disposal, in which case a binder may not be used.
A paste backf,rll can be engineered to meet the mine requirements for strength by working
*ith th. parameters of paste solids content and the amount of binder addition. Some of
the advantages of paste backfill systems over hydraulic sandfill systems are identified by
Brackebusch (199a):
r Greater strengths are generated with less cement.
. Decanting water from stopes is unnecessary'
. Generally total tailings can be used, which may solve a mine mass balance problem.
Lower mining cycle time is possible because strength is achieved earlier.
. Barricades may be eliminated or simplified with low slump paste.
. Flexibility in mining methods is possible, including undercut-and-fill mining.
PREPARING PASTE WITH FILTERS
The conventional method of paste preparation is to dewater the tailings from a mill
operation in a high-rate slurry thickener followed by filtration. Filters include drum, disc,
and horizontal vacuum filters. Water may be mixed with the filter cake to achieve the
desired slump for transfer to the binder mixer. The thickener may be preceded by a
cyclone to remove excess water or fines to a surface disposal site. In some cases the
cyclone overflow can be dewatered in a conventional thickener and remixed with cyclone
underflow to form a paste (Brackebusch,1996).
The slump and solids content required for paste backfill is determined by the mining plan
which includes such parameters as mining method, cycle time for accessing stopes,
binder (such as cement) requirement, safety (such as controlling rock bursts), availability
of backhll material, the mine layout and so forth. The lowest slumps (high solids
content) of this range require filtration. The highest slumps (lower solids content) can be
produced by a paste thickener as an altemative to f,rltration. There are no hard criteria for
distinguishing where the paste thickener option becomes viable. Considering only the
process performance of a paste thickener, experience with pilot and full-scale operations
suggests for slumps higher than 9 inches, the paste thickener is an altemative for
producing paste without filtration. For slumps between 8 and 9 inches, the paste thickener
should be evaluated as a possibility. For slumps of 8 inches and lower, the paste
thickener is unlikely to be an option. These operating ranges are strongly determined by
solids particle size, throughput rate, and pumping/pipeline shear. For example, designing
to produce 9-inch slump paste for 800 t/hr solids may require an economically or
mechanically impractical thickener design, problems that may not apply for a 9-inch
slump, 80 t/hr solids design. Paste has a shearing properties affected by pumping and
pipeline flow. The delivered slump to the slope is affected by the transportation distance
and method of transporting (centrifugal, positive displacement, gravit etc.) Figure 1
shows a typical flowsheet for preparing paste with addition of a binder.
There are many examples of paste backfill prepared with filter. The use of paste
thickeners without a filter is less well established. In recent years mine paste backfill
installations have provided valuable information about the proper way to design a system
in which the paste is prepared by a thickener.
Example 1
The paste plant consists of a thickener, cement silo, building housing the flocculant
preparation system, electrical and control panel, and a positive displacement pump. The
thickener operates by making and storing paste produced from cycloned base metal
tailings (s.g. 3.2-3.4) from the mine. The during the backfill portion of the cycle the
stored paste is pumped at slumps of as low as 9 Y, inches (measured at discharge of the
thickener centrifugal underflow pump) and solids content of as high as 79-80 wt%o to a
mixer for cement addition.
Example 2
The paste plant consists of a thickener, cement silo, building housing the flocculant
preparation system, electrical and control panel, and a positive displacement pump. Using
base metal mill tailings (s.g. 2.7-2.9), the thickener has produced 8-9 inch slumps and76'
78 wt% solids, after underflow pumping. High cement consumption has been a problem,
which is related to the paste moisture content and the strengths required in the mine.
Example 3
The plant uses paste backfill, using a mixture of thickened tailings and cement to fill
excavated stopes to improve the structural integrity of fill, allow for underhand cut and
frll mining methods, and reduce dilution by reducing the amount of sand from fill that is
recycled back into the mill in the primary ore. The paste backf,rll system uses a paste
thickener and filter combination to produceT6 wt%o solids.
Example 4
OCI Chemical Corporation, located in'Wyoming in the U.S. produces soda ash (Na2CO3)
from the ore trona using a dissolution/recrystallization process. A tailings washing circuit
(countercurrent decantation) uses a paste thickener as the last stage. The thickener
produces paste underflow that is pumped underground without cement for tailings
disposal. OCI (2004)
DESIGNING A MINE PASTE BACKFILL SYSTEM WITH A PASTE
THICKENER
The installations described above show that for an appropriate application a paste
thickener can effectively produce the required paste for backfill. However, operation and
design of backfill systems using paste thickeners have not been without problems. The
problems resulted, not from the thickener, but because an integrated systems design
approach was not used to match the paste thickener with paste transportation, binder
(cement) addition, paste setup time, backfill strength, and other issues related to mining
methods.
An integrated systems design of thickened tailings and paste processes, whether for
surface disposal or mine backfill, is essential for the successful operation. Jewell (2002)
The systems approach recognizes that rheological properties of a paste are a common
design basis from the thickener to the pumps and pipeline design and to the application.
Designing one part of the process, such as the thickener, without integrating the rheology
of the thickener underflow with the transportation system, binder requirements, and
mining plan will usually lead to problems.
Cement Usage
In the project design phase def,rnitive determination of the strength requirements for the
stope backfill must be completed before the backfill preparation process is designed. For
one installation, not effectively using this approach resulted in unacceptable cement
costs. In hindsight it is possible that the mine required underflow solids content outside
the range of a paste thickener. Definitive determination of required strength and cement
content and the necessary paste solids content should be made before specifying the paste
production system performance and choosing the tailings dewatering technology.
The particle size distribution of the feed stream to a thickener is a primary influence on
paste moisture content and ultimately cemented backfill strength. The design of a paste
thickener must be matched with the expected particle size of the backfill solids. [n one
instance this particle size distribution was too fine to generate the required thickener
paste necessary for stope strength at the design cement requirement. The solution was to
install a cyclone to coarsen the feed to the thickener and, therefore, the underflow.
Cycle Time
Cemented stopes must reach the required strength in a time period that allows the mining
activity (blasting, hauling, etc) to proceed on schedule. In addition to specifying required
underflow solids content or slump, the mine backfill design must specify thickener cycle
time. Matching the paste thickener operation with the mine cycle time requires a design
for the necessary holding time in the thickener, and an operating procedure to produce the
required paste characteristics on a cycle time required by the mine.
Pipe Wear. Pump Design
Pipeline wear is always a problem for incorrectly designed tailings distribution systems.
Mine backfill paste tends to be a highly non-Newtonian suspension, which means there is
no generally available database for pump and pipe sizing. Early paste thickener
installations have demonstrated that paste produced from a thickener is no less subject to
these issues than is paste from a filter. Specialists in this field of paste transportation must
be part of the design team interacting with the designer of the paste thickener.
CONCLUSIONS
With the right particle size distribution, paste thickeners without filters can effectively
produce paste for mine backfill requiring slumps of 9 inches and higher, after shearing.
Depending on the application slumps approaching 8 inches may be produced. For
applicable tailings and backfill requirements, the paste thickener offers an alternative to
filtration for the production of paste.
The design of a paste backfrll process always starts with what is needed by the mine.
Requirements for strength and cycle time translate into binder cost and tailings paste
slump, whether from a thickener or filter. If a paste thickener can produce the required
slump at the discharge of the centrifugal underflow pump, then it becomes an option to
flrltation for paste production and the evaluation can proceed to capital cost and operating
expense analysis.
Boger, D.V. (1999) "Rheology and the Minerals Industry", Min. Proc. Ext. Met. Rev.,
Vol 20, Overseas Publishers Association, 1999.
Houman, J., Johnson, G. (2003) "Commissioning and Operation of the Paste Thickening
Farm at Kimberley Combined Treatment Plant", International Seminar on Paste and
Thickened Tailings, Melbourne, 20Q3.
Jewell, R.J., Fourie, A.B. and Lord, E.R. (2002) "Paste and thickened tailings - A guide".
The University of Western Australia,2002.
Landriault, D. (1995) "The Present State of Paste Fill in Canadian Underground Mining",
97th Annual General Meeting of C.I.M. Rock Mechanics and Strata Control Session,
Halifax, Nova Scota, May 14-18, 1995.
Robinsky, E.I. (1978) "Tailings Disposal by the Thickened Discharge Method for
Improved Economy and Environmental Control". Tailings Disposal Today, Proceedings
2nd International Tailings Symposium, 1978.
International Seminar on Paste and Thickened Tailings
Paste 2004
Cape Town
South Africa
Paper 4
Abstract
1. lntroduction
because poor sampling can give rise to significant variations in total solids content in the paste,
as well as in the particle size distribution. Both factors can have a very large influence on paste
viscosity, particularly at high solids contents. Thus a few percent increase in the solids content in
the paste can lead to a 50o/o or l00o/o increase in paste viscosity, at high solids concentrations.
Secondly, the specification of the appropriate and relevant shear rate (or shear stress) for
rheological measurements must be based on an understanding of the entire process application.
The identification of the relevant test shear rate window is one of several considerations for
selecting a suitable viscometric geometry. Other considerations include'typical" viscosity level,
maximum particle size, paste abrasivity, liquid evaporation from the sample before and during
testing, and whether the paste exhibits significant time-dependent property such as thixotropy.
Once an appropriate viscometer geometry has been selected, rheological tests need to be
undertaken to ensure that the resulting data are not significantly affected by the four main error
Paper: lmportant Considerations when Making Rheological Measurements on Pastes Page2
Authors: Dr and Dr NeilAlderman
sources, namely, secondary or turbulent flow, end effect, wall-slip and viscous heating. The flow
curve (plot of shear stress versus shear rate) free of the four main error sources will normally
then need characterising using one of several alternative flow models.
Thickened tailings and other examples of "high solids content" pastes typically exhibit shear-
thinning rheological properties at an intermediate shear rate range of approximately 0.1 to
1000 s-1. However, at both low shear rates down to about 10-6 s-l and high shear rates up to
106 s-lthe paste can exhibit Newtonian behaviour with a corresponding zero shear and infinite
shear viscosity resp ectively.
A consequence of this behaviour is that referring to slurries and pastes as either "Newtonian"
For any end-use of the viscosity data, the relevant shear rate (or shear stress) range must be
assessed and used in determining the measurement conditions for the viscometer. A cornmon
error is to allow the shear rate (or shear stress) range of the one and only available viscometer
to dictate the test conditions for measurement rather than to define the relevantrange upfront'
Another cotmon error is to use the full shear rate or shear stress range of an instrument when
a much narrower window is often all that is required
Methods exist for defining the shear rate range for many processing applications and
have been detailed in Aspen Technology's Process Manual (Alderman, 20044,
Alderman & Heywood, 2004). The minimum shear rate is often zero owing to the
presence of stationary equipment surfaces whereas the maximum shear rate can be
estimated from a critical evaluation of the application. An example is paste flow in a
circular pipe where the shear rate (local velocity gradient) values range from zerc at
the pipe axis to a maximum value at the pipe wall. To estimate the maximum shear
rate occurring at the wall, both the paste volumetric flowrate, Q, and the internal pipe
diameter, D are required. Here, the nominal wall shear rate is given by (Brown &
Heywood, 1991).
8V _ 32Q (1)
Y=
f nom
D nD3
which is the actual wall shear rate for a paste exhibiting Newtonian behaviour tn
laminar flow. The wall shear rate for paste exhibititg 494-N.*!qi* behaviour can be
obtained from
1+ 3n' 8V (2)
t*= 4n' D
dlnt* (3)
nt=
dln(8V/D)
DAP (4)
Lw -
4L
The correction factor (1+3n')/(4n') varies from 1 for a paste exhibiting Newtonian
behaviour to a maximum of approximately 2 for a paste exhibiting highly shear-
'When
evaluating several pipe sizes, it is unnecessary to estimate the wall shear rate for
each flow rate through each pipe size to obtain an overall shear rate range for pipeflow.
Instead, the wall shear rate range can be estimated using the lowest shear rate
assuming that the lowest flowrate passes through the largest pipe diameter, and the
largest wall shear rate can be evaluated assuming the highest flowrate passes through
the smallest pipe size. This will give alarger shear rate range than that which actually
occurs and therefore provides a conservative estimate for the relevant shear rate range.
Commercial viscometers and rheometers employ several geometries for flow curve
measurement. These can be grouped into two main types: rotational viscometers and tube
viscometers. The main rotational viscometric geometries are coaxial cylinder (bob-in-cup),
rotating disc, cone-and-plate and parallel plate, but it is recommended that flow curve
measurements for engineering design should be made using either a tube (Alderman &
Kruszewski,2004) or coaxial cylinder (Alderman,2004) if the paste rheological properties
are time-independent.
If the paste exhibits time-dependent properties (Boger, 2002), such as thixotropy, the tube
geometry is inappropriate and the cone-and-plate geometry can be useful because it provides a
near-constant shear rate throughout the paste sample, so allowing thixotropic characterisation.
Also, if the paste has a high viscosity level, either with or without a yield stress, the co-axial
cylinder geometry is inappropriate, as the inner bob will tend to "dig a hole for itself'. The
result is usually a progtessive decline in shear stress at constant applied shear rate, which is
sometimes mistakenly assumed to be evidence of the presence of thixotropic property.
Using syringes or pipettes for sampling pastes is discouraged as this will affect the shear
history of the sample prior to the flow curve measurement. Also, their use will increase the
variability of solids concentration and particle size distribution between samples' Sample
placement in the viscometer should be done carefully with a spatula, or pouring from a bottle.
Flow curve measurements should be made using at least two separate paste samples, allowing
assessment of viscosity variability with solids concentration and particle size distribution.
With controlled rate instruments, the flow curve is usually generated using at least three
cycles of ramping the speed, through either a sequence of intermittent steps or a continuous
ramp. This is done up to a maximum set speed in a given time period and down again over a
similar period. The resulting torque is measured. In controlled stress instruments, the torque is
ramped in a similar manner, and the resultant speed is measured. The time over which the
shear ramp up or down is to be carried out is decided by the operator, but often an appropriate
time that gives 30 s per step. Repeated shear cycles carried out on the test sample will show if
the sample exhibits time-dependent flow behaviour such as thixotropy. If the up and down
curves for the first and successive cycles coincide, the sample exhibits time-independent flow
behaviour. However, if hysteresis loops between the up and down curves are observed for
each successive cycle, then the sample is said to exhibit time-dependent flow behaviour, in
which case the experiment should be repeated with the speed (or torque) held constant until
the torque (or speed) attains a steady value before the speed (or torque) is ramped to the next
value. This will yield an equilibrium flow curve in which the up and down curves coincide.
There are four common effor sources for both tube and co-axial cylinder viscometers which
should normally be accounted for. Figure 1 summarises the ways these four errors affect the
flow curve. Higherthan-expected viscosity values will be obtained if the shear stress/shear
rate data are affected by end-effect in coaxial cylinder and tube viscometers. This effect
becomes most marked at progressively lower shear rates and at lower slurry viscosity values.
log t
L:
7"-r -T-:
log i
Data affected by secondary or turbulent flow will also give rise to apparently erroneously
large viscosity values. Lower-than-expected viscosity values will be obtained if the torque-
speed data are affected by the wall-slip effect and viscous heating. The importance of either
effect will often emerge at progressively lower shear rates for progressively higher paste
viscosity values.
Flow curve measurements are made for rimary) laminar flow conditions only. The
data collected must be checked to ensure they are not subject to secondary laminar
flow conditions in a co-axial cylinder viscometer, or transitional / turbulent flow
(5)
This inequality can be plotted as a double logarithmic plot of tr against 8V/D to give
the laminar limit line (Figure 2); data to the left of the line (which has a slope of 2)
correspond to laminar flow, while data to the right of the line must be rejected. Using a
value of n: I in the inequality (5) for a paste with power law non-Newtonian
behaviour leads to a conservative approach to data rejection'
rl)
SOt.IDS
)NCL,NTRATION
I 5.837
Q s.2q,.
f 4 01)4
l0 V 2.9(L
r t.fisl
\Vll shcar strcss,
t*,. (Pa)
0
r0 100 r0)
-l)
Wall shear rate. I V/D (s
Figure 2: Plots of t, versus 8V/D for Digested Sewage Sludge passing through a
18.85mm Tube with the Laminar Limit
(6)
number at the critical speed at which laminar flow breaks down, given by Alderman
(1996) as
Ta. = 2350 +# (r
Lrt
q-
P':''''
This inequality can be plotted as a double logarithmic plot of r against O to give the
laminar limit line; data to the left of the line (which has a slope of 2) correspond to
laminar flow, while data lying to the right of the line must be rejected. If the cup is
rotating and the bob stationary, the higher velocity at the cup radius has a stabilising
effect on fluid flow and secondary laminar flow occurs at a cup speed about 10 times
higher than that for the case where the bob is rotating (White, 1990).
5.2 End-effect
This error arises because the assumption that the tubes or cylinders are infinitely long
in the derivation of shear stress and shear rate equations cannot, of course, be met in
practice. End-effect effor can be dealt with by minimization (using tubes or bobs with
It is usual to treat the losses at the exit as negligible compared to those at the entrance.
Equations are available for estimating the entrance length to obtain 98Yo of fully-
developed flow of Newtonian, po\/er law and Bingham plastic fluids (Steffe, 1996).
However, it is preferable to correct for end-effect experimentally using u_""-bf
tubes of the same diameterbut of different lengths (Brown & Heywood, 1991).
53 Wall-slip
Wall-slip may occur when a paste is sheared. This effect gives a resultant wall shear
stress at a given wall shear rate lower than expected due to the formation of a thin
layer of fluid (caused by solid particle depletion at or near the shearing surface) having
a viscosity lower than the bulk of the paste. Conversely, for a given wall shear stress,
the measured shear rate is greater than the true shear rate. In tube viscometry, for
example, wall-slip is present when curves of wall shear stress versus nominal wall
shear rate obtained with different tube diameters do not superimpose after all other
corrections to the data have been made. This effect can be quantified through a wall
slip velocity, V.. Different methods are used to estimate V. values to correct flow
cr.lrve data for wall-slip error, which can often be substantial.
ewst-l
\-/ vs=Sr* -= ?U6 %
IJ" '"1
and this assumption is validated when the plots of 8V/D versus l/D for various
constant values of ty are straight lines. Otherwise, Vs is given by
Here an example demonstrates the method. Plots of wall shear stress versus nominal
wall shear rate (8V/D) were obtained for a paste using three tubes of diameter 0.408,
0.602 and 0.981 mm (Figure 3). The three curves did not superimpose, indicating
wall-slip effors. Wall-slip velocities were estimated using the procedure above and
ranged from 44 mm/s at a wall shear stress of 350 Pa to 68 mm/s at 600 Pa. The
corrected values of 8(V-Vs)/D were then used in place of SV/D for the three curves to
give a single wall-slip corrected curve (top curve in Figure 3). If the 0.408 mm tube
alone had been used, the paste viscosity would be underestimated by about 10%.
t
o-
tt,
th
tt
+oo
t
o
u)
Figure 3: Shear Stress versus Nominal Wall Shear Rate for a Paste
using Three CaPillarY Tube Sizes
It should be noted that the relationship between slip velocity and wall shear stress
using the tube geometry will not usually apply to other viscometric geometries. Thus,
if wall-slip is present and the application for the viscometric data is pipeline design,
the tube viscometric geometry should always be selected, if possible. Wall-slip
velocity data from a different geometry (e.g., co-axial cylinder) will not usually apply.
This can be identified when flow curve data obtained with a series of bobs of different
radii do not coincide after the data have been corrected for end-effect. V/all-slip is
normally dealt with either by using bobs and cups with walls roughened to eliminate
the wall slip or by experimental methods which allow the wall-slip velocity to be
determined and corrected for. The assumption that roughened wall surfaces eliminate
wall-slip should be verified using bobs and cups of varying degrees of roughness
(Cheng & Parker, 1976).
viscous heating. This is done by calculating the viscous heating limit given for the
The main classes of flow curve exhibited by pastes and slurries under steady-state shear is
given in Table 1. These are idealised representations as most pastes show more than one flow
curve classification over the measurable shear rate range of 10-6 to 106 s-1. As a result,
sometimes more complex models are required, such as the Cross or Sisko models. These
models can be particularly useful in product formulation, but for most process engineering
applications the simpler models involving just two or three model parameters will generally
suffice.
Having completed the calculation procedure for the cotected flow curve, the data may be
amenable to a single curve fit. Sometimes, because of considerable scatter in the data, it may
be more appropriate to construct at least two curves: a mean curve obtained from regression
analysis using all the data and a upper bound curve obtained from regression analysis using
(r,.!') data selected from the curve that was initially drawn by eye. The upper bound curve
would normally represent the worst case for many engineering applications. This curve would
normally account for any possible variations in solids concentration, particle size distribution,
particle shape and pH.
Table 1 Flow Curve Models
Further factors can cause difficulties in attempting to draw a single flow curve through the
data, including the use of two or more different viscometric geometries which may give
differing degrees of phase separation during shear, sample variability taken from the same
batch, and uncorrected errors associated with the use of any viscometric geometry.
7. Concludinq Remarks
Flow curve measurements can often be time-consuming in order to ensure sufficient accuracy
for process design applications. This may require additional laboratory experiments,
particularly if end-effect and/or wall-slip errors are present. However, ensuring that only those
viscometric data relevant to the applications of interest are undertaken will minimise the
effort involved. Figure 4 summarises the steps to take to ensure both accuracy and relevance.
lJ ,, ti , lcctQ-t-/'^ vvru)a 3
Determine shear rate / shear 1'
1'
stress over which flow curve tr
should be measured
-l
=
Consider the impact of slurry
z
properties on viscometric
geometry selection, including
1. Maximum particle size
o
z
2. Concentration
3. Temperature 0tr
ML
<t)
4. Pressure a7
>a
5. Corrosive? --l <
6. Volatile?
z(t
7. Abrasive?
3
g
j- -r\
:-- 2
Yes Undertake ad, iitional tests to a
---Are there sionifcant time:
!
beh rvtour
X:nteffects? =
No
----ir
---'.tn"A^t,
t
subject
error?
t---r\ Y-es
Minimise effects of error. or
correct flow curve data
No
Notation
D tube or pipe diameter m
L tubeorbob length m
R6 bob radius m
t
I nom
nominal wall shear rate in tube or pipe flow, defined by Eqn (l) s-l
References
Alderman, N.J. (1996) "Non-Newtonian fluids: obtaining viscometric data for frictional
pressure loss estimation for pipeflow", ESDU InternationalDataltem95}l2, London, [IK.
Alderman, N.J. (200a "Shear rate estimation for viscosity/flow curve measurement",
Industrial Rheology Manual, Volume 2: Measurement, Part9. Aspen Technology.
Alderman, N.J. & Heywood, N.I. (2004) "Make accurate and relevant measurements of slurry
flow curves", Cltemical Engineering Progress, April, in press.
Alderman, N.J. & Kruszewski, A.P. (2004) "Viscosity measurement : tube viscometry'',
Industrial Rheology Manual, Volume 2: Measurement, Part 8. Aspen Technology.
Boger, D.V. (2002) "Rheological concepts". Section 3 in "Paste and Thickened Tailings: A
Guide" eds. R J Jewell et al. Published by the Australian Centre for Geomechanics, Australia.
Brown N.P. & Heywood, N.L (1991), Slurry Handling: Desim of Solid-Liquid Systems. Pub.
Elsevier Appl Sci, Barking, UK. Now distributed by Kluwer Publ., Dordrecht, Netherlands.
Cheng, D.C-H, (1988) "High shear limitations in viscometers", Proc Xth Int. Conf. on
Cheng, D.C-H & Parker, B.R. (1976) "The determination of wall-slip velocity in the coaxial
cylinder viscometry''.In; Proc 7th Int Cong Rheol., Gothenberg, Sweden, p518.
Steffe, J.F. (1996) Rheological Methods in Food Process Engineering, Pub. Freeman, Michigan.
White, J.L. (1990) Polymer Eneineerine Rheolosy, John Wiley & Sons Inc, New York.
Paste 2004
Paper 5
Quesfions
Regions of Dewatering
colloidal particles floc network
* @ ^ * , ^E#f
-
fluid
111t1f1 failure
':!.............,.
.j,++.'t+++
Compresson Rheology and Thickening, Fiona Sofra', Slde 3
Paste 2004
Overview of Thickening
Glarification
zone
Settling
zone
Gompression
zone
Comparing Flocculants
9
e)
0a
q)
l+t
r J
Our Approach
- Gel Point, n
009
008
0.07
E
x 0.06
c oob
.c
.9
0)
oo4
I
0.03
0.02
0.01
20,4 'v
I lrv. wllFlvwlvlr I uvfrYJ sr]v rrrvr\vrrrrY rqJe \,
Author: F Sofra and P Scales
The Madel
consolidation zone
stress
$orao+aea
U GravitY
0.0001
0 0.1 0.2 0.3 0.4
Summary
. Operational thickening process is permeability limited (for
bed heights above a couple of metres)
0.4
0.35
c 0.3
o.ru
ct
o.z
c)
r 0.15
0.1
Ia
0.05
0
'100 '1000 I 0000
Time, f (s)
r(^
E
1oo
E=10
o
|t-
O1
c
vI 0.1
* a\
tr o.o1
pat
0.001
0 0001
o 0.1 0.2 0.3 0.4 0.5 0.6
0.4
0.35
0.3
E o.zs
I
q 0.2
E.
O)
' 0.15
I
0.'1
0.05
0
I 10 100 f000 't 0000
Time, f (s)
0.0001
o 01 0.2 0.3 0.4 05 0.6
Summary
?"1"-.. Acknowledgements
. MohV
. AMIRA
. United Utilities, Yorkshire Water, Brisbane CC
. PFPC
. People
. Ross deKretser, Janine Hulston, Shane Usher,
Anthony Stickland, Dan Lester, Adam Kilcullen,
Brendan Gladman
Paste 2004
Cape Town
South Africa
Paper 6
Phil Newman
Golder Associates (IlK) Ltd
Rens Verburg
Golder Associates Inc.
Mark Fordham
Somincor S.A.
Abstract
The Neves Corvo copper-tin mine in southern Portugal produces approximately 1.44 million
tonnes per year of highly-pyritic tailings (up to 80%o FeS2). These are currently disposed sub-
aqueously into the L34-hectare Cerro do Lobo tailings facility. Investigations are ongoing
into the application of tailings disposal in the form of paste to facilitate sub-aerial placement
and concurrent closure programme.
The acid generation potential of the tailings is very high and a phased approach to
understanding the geochemistry of the proposed paste has been adopted. After a successful
laboratory-screening program, six field cells were constructed and filled with different paste
recipes. Monitoring has been ongoing weekly since September 2002 and the results provide
useful insight to the acid generating process associated with the low-permeability paste.
The results to date indicate that the low permeability and high saturation associated with paste
tend to retard the oxidation process sufficiently to allow implementation of a practicable
operational solution for tailings disposal. Potential future hydrogeological and geochemical
impacts on groundwater quality resulting from paste placement have been simulated using
hydro geochemical modelling.
The investigation into the application of paste has been wide-ranging and thorough,
encompassing interaction with the mine paste backfill program and evaluation of
environmental impacts, closure and sustainability issues. Possible concepts for paste
Paper: Field Cell Testing of Sub-Aerial Paste Disposal of Pyritic Tailings Page 2
Author: Phil Newman, Rens Verburg, Mark Fordham
production have been evaluated, and engineering design for a pilot plant has been completed.
The pilot plant will be commissioned in May 2004 and includes a 8-m diameter paste
thickener expected to generate over 50,000 m3 of paste during a six month period.
1. lntroduction
The Neves Corvo Mine, located in southern Portugal and owned and operated by Somincor
S.4., opened in 1988 and currently extracts approximately 1.8 million tonnes of copper-tin
ore from its underground operation using a combination of drift and fill, bench and fill and
mini bench and fill mining methods. The tailings facility, the Barragem Cerro do Lobo,
(BCL), is a valley type impoundment located approximately 2llrrt east of the mine site. The
current areal extent of the facility is approximately 134 hectares and the maximum height of
the main embankment is 36 m. The facility is operated using sub-aqueous slurry disposal
with a minimum 1 m water cover maintained at all times, in accordance with industry
standards for acid generating tailings. The dam is currently being raised using the
downstream construction method. Given the anticipated mine life of 25 years (to 2029)
further raises will be required if sub-aqueous slurry disposal of the tailings is continued.
Somincor is familiar with paste, having done extensive flow loop testing in 1997 and
commissioned a paste backfill plant in 1999. The paste backfill plant uses cyclone underflow
as feed to disk filters, with the subsequent cake conditioned and mixed with cement and then
directed to the stopes by gravity.
The possible use of paste for surface disposal was first discussed in 1998 although a pre-
feasibility study was not completed until 2001. This study identified potential cost savings
with the introduction of paste, mainly associated with the elimination of further dam raises
(given the topography, these will grow significantly in size and cost) and a reduction in
closure costs through the adoption of concurrent rehabilitation. The study also indicated that
potential geochemical risks (associated with acid generation in paste) were significant and a
phased approach was adopted that would seek to quantify and mitigate such risks.
A series of laboratory screening tests were completed (Verburg et a|.,2003) followed by the
construction and regular monitoring of a number of field cells containing paste prepared to
different recipes. At the time of paper preparation, the field cells have been operational for
almost 500 days with regular monitoring of seepage and runoff. The cells were cored after 13
months and aged paste samples analysed. It is proposed that monitoring will continue until
The samples were prepared and subsequently monitored for 30 weeks, with temperature, paste
pH and paste conductivity measured. All samples were left undisturbed with the exception of
the agitated sample. The samples were misted with de-ionised water to maintain damp
conditions, and any excess water was removed from the samples during the testing period. A
detailed description of the testing procedure and the results is presented in Verburg et al.
(2003).
The results indicated the beneficial effects of higher moisture content on inhibiting oxidation
with the 250 mm slump paste having the 'best' perfonnance. The addition of lime and
cement provided some early buffering capacity, but once consumed, the behaviour of these
samples was almost identical to the non-amended mixtures. The bactericide provided some
benefits in terms of a slower rate of oxidation.
The semi-quantitative nature of the testing did not allow a rigorous scientific evaluation of
mechanisms and oxidation rates. However, the principal objective of the screening trials was
to identify a number of tailings mixtures for full-scale field trials. The following 3 recipes
were selected for field testing;250 mm slump with no additive, 250 mm slump with 0.5 wt%
cement, and250 mm slump with a surficial spray of bactericide.
sampling at depth, and surface runoff is collected in surrounding lined trenches. The paste
mixtures placed in each cell comprised:
runoff) commenced after cell completion and includes pH, conductivity, temperature, iron and
sulphate content. In addition, samples are collected each month for comprehensive chemical
analysis. On-site climatic conditions are recorded, including precipitation, evaporation and
temperature. Figure I shows the plan and cross section of the field cell and Figure 2 shows
two finished cells approximately 10 weeks after filling.
3m
LOnr
HDPE
eram 1 000
^
I
Needle-punched I
10.3m
il
4. Results
4.1 Field Cell Water Qualitv
As expected, seepage rates were highest in the irr'igated cells. Seepage rates were reduced in
the irrigated Promac@ cell relative to the other irrigated cells. This may be due to a reduction
in formation of surficial cracks in the Promac@ cell.
Seepage pH was near neutral (pH 6 -7) during the wet season, with a slight declining trend
over this period. Cement cells initially showed alkaline pH due to cement dissolution. With
the onset of the dry season (June 2003), acidic conditions were established in the irngated
cells, but ambient cells maintained a neutral pH. Runoff is generally more acidic than
seepage and all cells show a decreasing trend in pH.
Conductivity, sulphate and trace metal concentrations are gteater in seepage waters than
runoff, likely due to greater solid to solution ratio and increased travel time involved in
seepage processes. During the wet season, conductivity and sulphate concentrations were
greater in seepage from ambient cells than irrigated cells due to the reduced infiltration rate.
However, during the dry season irrigated cells generally show greater conductivity and
sulphate loadings. Conductivity and sulphate loading in seepage were much greater in the
paste and cement irrigated cells than the irrigated Promac@ cell. Iron concentrations in
seepage are largely coincident with reductions in pH. Trace metal concentrations in seepage
and runoff also generally increase with increasing acidity. Differences in behaviour are
generally more pronounced between irrigated and ambient cells than between the unamended
paste, cemented paste, and Promac@ cells within each of these two groups.
4.2 Field Cell Corino
All findings of the coring program are consistent with sulphide oxidation under near-surface
conditions. Paste pH values generally are lowest near the surface and increase with depth.
Conductivity shows an inverse relationship with pH, with highest values observed in shallow
samples. The solids content generally shows the same trend as conductivity. The total
sulphur content is directly proportional to the solids content, and displays a strong negative
correlation with paste pH. Neutralizationpotential is depleted in surficial samples.
The results from the solids analysis suggest that the ambient paste, ambient Promac@, and
irrigated Promac@ cells are the most reactive. At first glance, this is inconsistent with the
results obtained from the water quality analyses. However, it should be noted that it is
difficult to relate the solids results to sulphide reactivity as compositional trends are not solely
determined by sulphide reactivity, but also by transport of oxidation products. The chemical
compositions of the seepage and runoff are therefore considered a more reliable indicator of
sulphide reactivity.
months.
Seepage pH, sulphate load and iron load were evaluated to assess the relative degree of
sulphide oxidation occurring in the six cells. Table 1 shows individual and composite
rankings for each cell from I (most reactive) to 6 (least reactive).
The rankings in the table indicate that sulphide oxidation is occuring at a higher rate in the
irrigated cells in comparison to the ambient cells. Also, acidic conditions are established in
the irrigated cells before the ambient cells, which may in part be attributed to more rapid
flushing of oxidation products through these cells. Comparison of relative sulphate and iron
loads between the cells suggests that the irrigated paste cell shows the greatest reactivity,
whereas the lowest reactivity is observed in the ambient Promac@ cell. For the irrigated cells,
a clear difference in reactivity can be seen between the paste cell and the two other amended
cells, while all ambient cells have a similar reactivity.
Cell5-IrrigatedPaste 1 I I J
geochemical stability of the paste and the hydrogeochemical impacts associated with its
implementation. The valuable data from both the field cells and a detailed hydrogeological
site investigation were incorporated in a geochemical model to predict potential post-closure
impacts to groundwater quality downgradient from the facility. The model results provide
additional support for the conversion of the facility to sub-aerial paste disposal (including a
suitable impermeable final capping layer) as it was demonstrated that potential impacts from a
traditional, sub-aqueous setting are likely to be more severe than those from a paste stack.
The operational aspects of installation were considered for the pre-feasibility study, and it was
agreed that a pilot plant operation was required to more accurately determine the costs and
necessary protocols to ensure paste placement is efficient and effective.
The implementation of paste at the BCL will take advantage of the ability to place paste in a
modular fashion around the facility. A number of cells will be developed at one time,
probably at least 3, with paste being sequentially placed and allowed to develop sufficient
shear strength before the placement of the subsequent layer.
The use of stacking towers has not been discounted, although the placement and subsequent
stability of these remains an issue, given that up to 27 m of tailings have already been placed
in some places, presenting some foundation design challenges. However, paste stacks have
been developed without the need for stacking towers (e.g., Kidd Creek Mine in Ontario).
Gaining access across the paste cell upon the paste itself is feasible. The field trials have
indicated that initial desiccation is fast and facilitates both man and vehicular access in a
relatively short time. The impact of access on cell development will be carefully investigated
during the pilot plant operation, currently scheduled for May - November 2004.
The aim of the modular approach is to develop the cells to the proposed final height and
subsequently close them with either a temporary cap (to limit the penetration of oxidation
further) or the final cap.
Figure 3 presents a photo montage (sketches \ - 4) of the proposed approach. The base
photograph is some years old but is suitable for indicative purposes.
Sketch I (current state) shows the arrangement of the facility, indicating the positions of the
embankments and the aerial extent of the water cover at the time of the photograph.
Sketch 2 (enly stage of filling) shows an overlay of the overall proposed cell arrangement,
proposed location of the paste plant, proposed paste distribution lines and the location of the
existing spillway. Also included is the early cell filling sequence, showing cell 1 as being
closed and three active cells. The solid arrows represent the direction of cell closure whilst
the dashed affows represent the direction and sequencing of future cells. The reason for
commencing cell development in the northeast corner of the impoundment is to facilitate both
the release of water through the existing spillway in the event of unforeseen rainfall events,
and to minimise the potential for paste tailings spills on previously closed and rehabilitated
cells.
Sketch 3 rogressive paste placement) shows the progression of paste placement in the cells
and the closed area of the facility. As can be seen on this sketch, the open area of the facility
is progressively being reduced by paste placement, the result of which is decreasing
infiltration of contaminated seepage to the aquifer.
Sketch 4 (nearing final paste placement) shows paste deposition close to completion. The
spillway remains operational throughout paste placement.
Runoff generated during paste placement and progressive cell closure will need to be
managed to minimise the potential for cross contamination of tailings water and clean runoff.
On completion of paste placement, the paste plant will be removed from the site and the
spillway decommissioned.
(,
o
o
G
a.
It
aJ'
G
a-
o
Ut
at,
0)
l
o
o.
e)
o
(o
(t,
7. Gonclusion
The work completed at Neves Corvo provides valuable insight into the use of paste
technology for acid generating tailings. The testing program to date has demonstrated that
use of paste tailings represents a viable alternative that results in significant environmental
benefits relative to traditional tailings disposal. Based on the pre-feasibility study, use of
paste technology is also considered attractive from an economic standpoint. A pilot-scale
program will be implemented shortly to further evaluate and refine operational aspects.
References
Verburg, R., Johnsor, 8., Logsdon, M., Fordham, M. (2003) "A Rapid and Cost-Effective
Method for Bench Screening of Geochemical Performance and Disposal Options for High-
Sulfide Tailings" ICARD, dh International Conference on Acid Rock Drainage
Paste 2004
Paper 7
Paul Geraedts
Abstract
There is a crossover zone between the application fields of the different style positive
displacement pumps that are used for high-density slurry or paste disposal systems. In these
crossover zones, a number of specific design aspects will determine the appropriate pump
selection.
A crankshaft driven piston diaphragm pump generally is the most economic and the most
reliable solution for pumping fine paste tailings. On the other hand, hydraulically driven
piston pumps, are more forgiving to coarse particles and able to handle higher viscosities.
Coarse particles however will affect the wear life of pipeline system components, and it may
incidentally be better to screen them out.
The crankshaft driven piston diaphragm pump is capable of handling a yield stress up to
approx. 13 5 Pa at Bingham viscosity of 0.2 Pas, which corresponds with a slump value of
approximately 8-9". Higher viscosities may cause poor closing of the spring actuated pump
check valves.
Paste slurry transport systems often are designed for paste of around 8-9" slump. Though,
within the scope of a crankshaft driven piston diaphragm pump, many enquiries dictate the
use of more maintenance intensive hydraulic driven piston pumps, presumably to allow
excursions in viscosity. The paste consistency control in this slump range is critical because
small concentration changes can have a big impact on the paste viscosity.
Operators often run the paste system at lower than design consistency. As a result, the pump-
and pipeline operating pressure often is below rated values and the spreading characteristics at
the discharge point may differ significantly from the original design.
Modem thickeners are able to produce paste with little or no flow behaviour. Many tailings
systems use a centrifugal booster pump, that can change the paste by shear thinning to a
material that is easily pumpable over several kilometres distance with crankshaft driven piston
diaphragm pumps, and then levels equally over a disposal area.
Unlike conventional concrete pumps, industrial design hydraulic driven piston pumps are
designed for continuous trouble-free operation. The pumps are equipped with quick and
smooth piston reversal technology that reduces vibrations and pulsations. Until recent, large
1. lntroduction
There arc many reasons for an increased interest in high-density slurry or paste transport
resource is a key issue in many arid areas around the world. It would seem obvious that for
large mining projects, major water savings can be achieved by using high-density paste
systems. This is indeed the case and water savings are potentially very significant, however,
beyond a certain slurry density the returns are diminished as the system costs associated with
The effect of slurry on centrifugal pump performance can be significant and this increases the
power requirements. It is therefore not always clear whether centrifugal or positive
displacement pumps are more cost effective for high-density thickened tailings systems. The
higher capital cost of positive displacement pumps are offset by lower maintenance costs than
centrifugal pumps and aproper cost comparison of capital and running cost over the life of the
project needs to be done to confirm the most cost effective pumping system.
This paper will discuss some of the major displacement pump selection cntena for high
density and paste pumping systems and stresses the importance of selecting an appropriate
pump and slurry disposal system that best meets the project requirements.
Part of the detailed design of a tailings system is the verification of the slurry/paste parameters
and pumpability. GEHO PUMPS has in-house laboratory facilities where it conducts several
tests on slurry samples by which it can determine the pumpability, the rheolog the
settling/re-starting characteristics and the abrasiveness (Miller test). The lab test results are
firstly used to optimise the pump and pump component selections, but can also be used to
optimise the pipeline design. Where pipeline transport of high-density or paste tailings is
required, it is often best carried out by high-pressure reciprocating positive displacement
pumps. Each of the following three tlpes of displacement pumps avallable has distinct
The selection of the appropriate pump type depends on the viscosity and other characteristics
of the slurry such as the particle top size, as well as capital cost and maintenance
considerations.
The principal difference between a piston and piston diaphragm pump is the diaphragm that
protects the piston and liner from the sliding contact with the abrasive tailing. The tailing is
pumped by the combined mechanical hydraulic action of the crankshaft driven piston,
propelling fluid, diaphragm and suction and discharge valves. The diaphragm is of a pre-
moulded design to eliminate stresses caused by elongation of the diaphragm material and is
routinely replaced after 8,000 to 12,000 hours. It is guided during its movement by a central
monitoring rod and its stroke is controlled in either direction by adding or removing
propelling fluid as necessary. Roller bearings used within the pump afford the pump to
operate intermittently at avery low stroke rate against fulI discharge pressure. Each pump is
built to suit a particular service and meets proprietary criteria for e.g. power end rating,
diaphragm load, strokerate and valve velocity. The GEHO range of ZPlld double acting and
TZPM single acting crankshaft driven piston diaphragm pumps handle paste of thickener
consistency.
The only wear parts in the pump are the valves for which a number of options help to
optimize the life cycle and adapt to the slurry rheology, characteristics and behavior which
differ with each service. These spring loaded cone type pump check valves can handle small
percentage particles up to 8 mm in size.
At the Orapa Mine in Botswana a GEHO pump test was conducted in 1998 on thickened paste
tailings, a total of 3927 tons of solids was pumped to the test deposition site. The summanzed
results are shown in figure 1 (see attachments).
Figure 1 shows the expected pipeline system curve versus flow rate, and the measured pump
discharge pressure. The expected curves were estimated beforehand for arange of densities
and fines fraction. The results show that the pressure recorded during the test varied widely,
but all fell within expected discharge envelope. Tests were done for araruge of densities
between 1.50 and i^82 tlm3 . The average density during the test \/as approximately 1.59 tlm3
Using the data in Figure 1, an estimate of the yield stress can be obtained. As an example, the
maximum estimated yield stress would be expected from the upper data points that would
probably intercept the y-axis at 3000 kPa. If we assume this was recorded at the highest
density of 1.82 t/m3 , thenthe yield stress can be estimated by:
DAP
4
and pump dischatge pressure = L oo"LP + (H ,^,rgS
^)
3000 : 600x AP + (50x9.81 x1.82)
0.150x3512
LP =3.5l2kPa I m, so ro = =l32Pa
The ratio of coarse and fine particles changed continuously during mining, so we could not
determine a consistent relationship between density and increasing yield stress, as each
density would have a variable fine fraction. A photograph of the highest density pumped is
given below. As you can see the slurry is extremely thick and viscous.
Figure 2: Spigot discharge of high density kimberlite tailings (photograph courtesy Paterson &
Cooke, South Afca)
_l-
.,
Where the paste consistency and particle size are at the limit of what a diaphragm pump can
handle, consideration is given to hydraulic driven piston pumps (hereafter called "piston"
pumps). These are two cylinder, single acting pumps with continuous water flush on the
backside of the piston. These pumps are designed with a prerequisite low stroke rate, a low
piston velocity and large flow areas that also help to minimize \ryear. The piston is a metal
body with guide ring and replaceable cup seal, which slides in precision-machined hard
chrome, plated liner. The guide ring prevents the piston body from rubbing the liner, whilst
the 0,4 mm chrome plating doubles the wear life that is less susceptible to incidental scoring
and grooves. V/ithout a detailed review of the slurry consistency, performance indicators and
project requirements it is usually difficult to decide between the following two piston pump
models.
A model DHC piston pump with large diameter, high lift, actuator assisted suction and
discharge valves which handles rather viscous paste and particles tlpically not larger than 15
ntm.
A model DHT piston pump with transfer tube handling highly viscous paste and particles
sizes up to 80 mm maximum.
The model DHC is using the valve design from the diaphragm pump, except that the valve
cone bolts directly onto the actuator rod. The seating surfaces are conical shaped and provide
a metal to metal plus a metal to elastomer contact surface. The actuator load and activation
sequence is arranged such that avalve will not open against the pump discharge pressure. As a
result the system is inherently protected against back flow even if a valve is jammed open.
The model DHT pump water end comprises a hopper with transfer tube. This S-form pipe
pivots on the discharge side whilst the inlet is pending between the opening of the paired
cylinders. At the end of a suction stroke it covers the cylinder opening and isolates the filled
cylinder from the suction hopper. During the subsequent discharge stroke, the piston forces
the product through the tube into the discharge pipe. During switch over the tube partly
'When
overlaps and short-circuits the cylinders during a fraction of a second. pumping a paste
consistency, the back flow from discharge to hopper will be minimal and the resulting
pressure pulsation will not be insignificant but are quite acceptable in an adequately anchored
piping system.
'Where
The tube inlet opening is fitted with a nozzle, which slides over a wear plate. the
tailings are particularly abrasive, ceramics improve the wear and galling resistance.
Figure 5: GEHO model DHC pump
The GEHO sizes SR2 and SR3 hoppers are unique in that a hydraulic thrust load is generated
which pushes the tube against the wear plate. The tube nozzle is not required to move axially,
but is flexibly mounted to compensate for uneven wear and misalignment. The swivel
cylinders are positioned under a slight angle such that the nozzle to plate contact is maintained
when the pump is idle. The tube tuming lever is mounted on a tapered polygon, which
provides a long-term clearance free, and shock resistant fit. Positive tube thrust ends manual
tube end float adjustment and premature tube replacement due to washout and wear of the
splined shaft.
Cartridge mounted initiators are provided at either end of the hydraulic drive cylinders.
Signals from opposed initiators/pistons enable a plc controlled stroke reversal and either
piston to complete a full stroke. This method of leakage compensation enables the use of
standard hydraulic piston seals that afford a 5 to 100 percent capacity adjustment range. The
drive cylinder design, oil hydraulic flow sheet and plc programming aim to obtain a relatively
smooth piston reversal and reduce hydraulically and mechanically induced loads.
.When
cylinders become large, the piston bodies get excessively long and heavy. Large size
GEHO pumps are therefore fitted with a'linear transducer'in lieu of initiators - a slender bar
mounted onto the cylinder end cap and matching a borehole in the piston rod. The piston rod
itself is fitted with amagnet, which enables the transducer to define the piston position at
every moment in time. This system not only reduces length and weight of piston and cylinder,
it also enables control of the stroke length independent of the stroke rate.
Both piston pump designs do meet the demands of the Power, Mining and Minerals
Processing industries such as continuous operation, high availability, low life cycle cost and
an acceptable noise and pressure pulsation level. The capital cost for a larger size piston pump
including pulsation dampener will typically be equal or higher than for an equivalent size
o, which tentatively
diaphragm pump. A piston pump unit has efficiency in the rarige 50 to 60
South Deep Mine (South Africa) operates since 1993 a series of transfer tube type piston
pumps transporting backfill over 1250 meters from the underground crushing plant to the
stopes where the 85% Cw mixture of comminuted waste and aggregate helps prevent rock
bursts.
The backfill at OCI in the USA is less viscous and causes back flow, which generates pressure
spikes in the pump and piping. These (earlier design) DHT pumps are with a diverging
transfer tube bore, splined shaft and manual end float adjustment.
Other backfill projects e.g. Galmoy in Ireland are using DHC valved piston pumps. Kali+Salz
ensures the structural integrity of salt caverns with backfilling a paste that is a cemented high
One example of a large size piston pump which meets the 5o/o pulsations specified is at
Tarong North Power Station in Queensland handlingT5% Cw fly ash which is on the verge of
being pumpable with a diaphragm pump. The pump with actuator assisted valves must be
complemented by a level controlled air vessel, associated compressor and air receiver.
Figure 7: GEHO DHC valve piston pump at Tarong Power Plant
3. Conclusion
In slurries where selacting valves can operate, the crankshaft driven pump, in particular the
piston diaphragm pump is the best choice. This pump is the most economic and offers highest
availability.
For paste type materials where self-acting valves will not perform satisfactorily, GEHO offers
a complete rage of hydraulically driven piston pumps either with hydraulical actuated cone
valves or with a transfer tube, with numerous project specific design features to achieve
optimum performance and moderate operating costs. The most appropriate of the three types
of mainline pumps is best selected by a qualified pump supplier having relevant know how
and experience. Ordinary concrete pumps have obviously not performed to satisfaction and
It must also be ensured that the available thickener technology will produce the paste
consistency and quality required for disposal and pipeline transport.
GEHO PUMPS offers economical and reliable solutions with designs based on the company's
extensive experience in continuous operating pumps for the mining and processing industries.
Reference
Van Dijk, Harrie (2001) '?umping of high concentration tailings", Mining Environmental
Beekman, Ted (2001) "Different pumptypes for tailings - A selection guide", High Density &
Geraedts, Paul (2003) "High density residue transport and disposal", ALTA 2003 -
NickellCobalt - 9, Perth I9-20May
Paterson, Angus (2003) "The hydraulic design of paste transport systems", 2003 International
E
, d
v,
7
o
rrl
fi
cd
E t/l o
o o
r
J
H=
' t! I
() l E l x &-
o-h
t
N
t-* o
iE
u]
r.l
t: lot
t
n
to
o
il
l
tr
O
il lJ
Jf I ift (} o dt: zc}
l
-
tl
a5
1 l;f af\
e
q
tr
C}
o-
t
o-
c;
d-
EA
1
{ t*
$
{J
-) o rt {[
gq I r LI
r ,(} U) Y
7
: z u v
1
1 r a 1 ;g v> g
il-
fY.
l-
E
k
U
llrl e f fl
lJ
l
I
e. (,-
1
lJl Cl J 6 UI ( U' tl,
an
z-a E
c I
rrlw*rlfrl'.wrcl gE it t
t5
ctttr 5E f
dt
I c5 og t f
r
I
E
t, =H
F-
R
G >
r. ln
l c
t l- I
= ttl lrn I {
.tr
! .t-a
r.= 2fr
f=
6X t
3
\ ui<
(L= t I
I
I
I
l
I I
(J
r
tJ
t
vl
I
t 1,1
h
r I
r
!J
I [
i F
E
$ F
!
t
(
-\ ff
.\,. fl
*
I E
I
v
\x\
fi
o 8
t
OEt
u,.
E d rg
t(
o
o.
t4
E $ g f; H E H R
6++r)-sl
O
q
H
qlG
{ot) runud lgtlYHraQ dfid fr Ei q,
Paste 2004
Cape Town
South Africa
Paper 9
Nigel Heywood
Dr Nigel Heywood
Aspen Technology, Harwell, IIK
Abstract
The fctional pressure loss for the flow of thickened tailings in pipelines can be very high. By
injecting a small air flowrate into the pipe immediately downstream from the pump, frictional
pressure loss is reduced and in some cases there is also an overall power saving, provided the
tailings are rheologically shear-thinning and in laminar flow prior to air injection. This paper
sets out the main benefits to be achieved through air injection and outlines models which
allow prediction of the effect and the maximum reduction in fctional pressure loss.
1. lntroduction
V/ith the trend to produce higher concentration tailings from thickener underflows, frictional
pressure loss incurred across tailings pipelines will also rise, causing larger pump discharge
pressure and power consumption requirements. One way to reduce this fctional pressure loss
increase is to inject a relatively small flowrate of air (or other gas) into the pipeline immediately
downstream from the pump. The presen.r of tht air reduces the proportion of the inner wall of
the pipe that the tailings are in contact with, but at the same time, increases the mean tailings
velocity in the pipe, for a fixed tailings flowrate.
The overall result is a reduction in the frictional pressure loss across the pipeline, provided the
tailings are non-Newtonian (shear-thinning), and are initially flowing well within the laminar
flow regime. The more viscous and non-Newtonian the tailings, the greater will be this drag
reduction effect for given pipeline diameter, and tailings and air flowrates. The effect is most
Paper: Benefts and Modelling of Air lnjection into Pipelines carrying Thickened Tailings
Page2
Author: Dr Nigel
marked in horizontal pipeflow but is also present in vertical pipeflow. Depending on the relative
efficiencies of the tailings pump and air compressor, there may be an overall power saving.
Air injection systems can either be included at the design stage to reduce pump discharge
pressure and therefore pump porver consumption, or they can be retrofitted to existing
pump/pipeline systems if these were designed for lower concentration tailings and there is a need
to limit the existing maximum pump discharge pressure requirement. Altematively, for a given
tailings concentration and flow property, air injection can facilitate an increase in tailings
This paper provides design equations which allow the prediction of the air flow rate required to
achieve a specified reduction in frictional pressure loss, for known tailings flowrate, pipeline
diameter and tailings physical properties. The full potential of air injection has yet to be proper
exploited, and thickened tailings handling is an obvious application area.
Air (or other gas) is injected, either continuously or intermittently, into the discharge pipe of
the paste pump. The injection point is of the order of one to two metres downstream from the
pump and no special injection arrangement is usually required because tests have shown that
the use of specially-designed injection points have an effect on the distribution pattern of gas
and paste for only a few metres downstream from the injection point (see Figure 1). There is a
resulting reduction in the pressure gradient along the discharge pipe, provided that the paste is
shear-thinning and is initiallywell within the laminar flow regime prior to the gas injection'
gas
injection
point
(1) reduced discharge pressure requirement for a paste pump for a given paste flowrate
(3) extension of a pipe while maintaining the same discharge pressure and paste flowrate;
(4) application of an existing pump and pipeline combination to more viscous, shear-thinning
paste, while maintaining the same discharge pressure'
(5) reduced pump differential pressure, and therefore reduced slippage in some pump t1pes,
with a corresponding reduction in pump wear.
The resulting flow pattern is usually either the plug or slug flow pattem' The flow pattem
suggested in Figure 1 is therefore idealised but is useful for the basis of a simple plug flow model
If the paste flow is turbulent prior to gas injection the fctional presstfe loss always increases
progressively with an increasing air flowrate, for a fixed paste flowrate in the pipe. From a
knowledge of the paste flow curye, and the proposed paste flow conditions (flowrate and pipe
diameter), it is possible to calculate the paste Reynolds number and compare it with the critical
value for laminar flow breakdown for a chosen paste flow curve model in order to assess the
potential of air injection. Even if the flow is initially turbulent prior to air injection, there may
still be benefits to be gained. If there is a possibility that the pump may fail leaving a long plug of
viscous paste to start up, the presence of air in pockets along the pipe will assist in restarting the
paste. This is particularly useful if the paste exhibits a significant yield stress. The air pockets
ensure that the peak pressure gradient to be reached by the pump will be approached more
slowly, and so there is less likelihood of damage to the pump and its motor/gearbox.
Although the potential for pressure gradient reduction is immense, and much research has been
undertaken on the phenomenon to identify optimal conditions for maximising power reductions
(Dziubunski & Richardson, 1985, and Dziubinski, 2002), the take-up of these ideas in industry
has been relatively poor. This may, in part, be because of the lack of awareness of the potential
for air injection. In addition, there may often be concems over designing apractical system'
There is now enough information from many experimental and modelling studies to facilitate a
reliable system design. If the paste flow properties are characenzed through viscometry and the
paste flowrate and internal pipe diameter specified, the reduction in frictional pressure gradient to
be achieved for a given air injection rate can be predicted with reasonable accuracy. Also, the
maximum reduction in frictional pressure loss and the required air flowrate can be predicted.
Specification of the start-up and steady-state operating conditions also require some thought.
Because the discharge pressure of paste pumps is often quite high for tailings disposal, injecting
air at that pressure before the pump discharge pressure is reduced could be problematic and
create a safety hazard.Instead it is preferable to start the air injection into the pipe before starting
the pump if the discharge pipeline is empty, or, if the pipe is already full with stationary paste, to
start the pump and air injection concurrently and build up to the steady state pump discharge
pressure.
Owing to the reduction in absolute pressure in the pipe from the pump to pipe discharge, the
injected air will expand progressively as it flows with the paste down the pipeline thus
accelerating the paste along the pipe. If the paste were in laminar flow prior to air injection, two
pipe
situations arise. In the first case, the frictional pressure gradient falls progressively along the
just before pipe
as the air expands, and the minimum frictional pressure gradient is achieved
the
discharge if the paste is still nominally in laminar flow at that point. In the second situation
maximum reduction in frictional pressure gradient is achieved at some point along the pipeline,
and the frictional pressure gradient then starts to rise again until the pipe discharges.
A further consideration is to decide the most appropriate way of separating the air from the paste
on pipeline discharge. As most of the air will be flowing separately from the paste in the
pipe,
either a
the two phases should normally be relatively easy to separate on pipe discharge using
cyclonic device or a suitably adapted separator as used for gas/oil separations in the oil industry.
r order to be able to build models that can predict both frictional pressure gradient and
paste hold-up in the pipe, it is important to be able to identify the flow pattem that the
airaste mixture corresponds to. There are six flow patterns encountered generally in
airaste flows in horizontal pipes of which the plug and slug flow patterns are the most
important in the current context: Inplug (or elongated bubble) flow, bubbles coalesce at
increased gas rates giving rise to large bullet-shaped bubbles occupying more of the pipe
cross-section. In slug flow, lrothy slugs of paste carrying gas bubbles altemate with gas
slugs surrounded by liquid films. Although plug and slug flows are both well-defined, as
$
Evbtfr Flow b
oI
[:-:-ll, E
Plug Fld
Shollad-t{bvy Fd
A^A .Fbtt
=-...-..-l UPWARO VERftrAL FLOW
i".ii.1'.: : i : r';' :. i i. :: :. 2'.6
l.ll Flor
IRIZONIAL FLOW
Figure 3 shows a flow pattem map based on a data bank of flow patterns in two-phase
gas/non-Newtonian systems (Chhabra & Richardson, 1984). The influence of pipe
diameter has still, however, to be properly assessed.
t0
tt,rl51
E2 Ftff
,
9C5
En
.E
I ol
iE oo
E
a
A oz
01
@5
@t
001
{toz 05 0lo'2 05 l'o 20 50 10 20 I
Superficial Gas Velocity Vs m s'
A useful fctional pressure loss reduction is achievable when the power law, Metzner-
Reed Reynolds number, Re4, for slurry flow alone is less than 500, rather than the
critical value of Reynolds number of approximately 2000. Rertrn is defined as
e"v"- o" I
Rn'o
.wMR -
=
Kgn_r'P' [f + f"l
I (1)
For a given mean paste flow velocity, progressive increases in the air injection flowrate
will lead to a progressive increase in the reduction of friction pressure loss until a
maximum effect is reached. Figure 4 shows the drag ratio, $s2, defined as the ratio of the
pressure gradient for the airlpaste flow divided by the pressure gradient for paste flow
alone at the same paste flowrate, is plotted against the superficial atr velocity. Beyond the
.v'
t
D
CI
o 0.2+1
I 0.108
v 0,Bl
o 0.yl8
l.tl
^
paste move along the pipe in the form of alternate cylindrical plugs, as depicted in Figure
1. Using this plue flow model, the drag ratio can be predicted to l0o/o accutacy
(Heywood & Richardson, 1978) using
?t = )"'-" (2)
?'=#% (3)
This is the paste holdup (i.e., the fraction of the pipe occupied by paste) that would occur
if the two phases moved at the same velocity in the pipe. This equation is useful for an
initial scoping design and feasibility study and is more accurate at low paste and air
superficial velocities and high K-values in the power law model (Heywood &
Richardson, 1978).In particular, the mixture Reynolds number must not exceed 500 (or,
more usefully, the Fanning fction factor must be greater than 161500 : 0.032). The
mixture Reynolds number is based on the single phase power law Reynolds number:
Re!, =
It_t qn l" (4)
Lr+"j
In addition, the superficial gas velocity, Vg, must not exceed 1 m/s. Such conditions
correspond to the plug (elongated bubble) flow pattern. The model breaks down as the
condition for the maximum reduction in drag is approached, and a more detailed slug
flow model is required (Heywood, 1976, andFarooqi et al, 1980).
The maximum reduction in frictional pressure loss (minimum drag ratio) can be
correlated with the Metzner-Reed Reynolds number for paste flow alone and the flow
behaviour index, n, in the case of the power law model (Figure 5)' For a given paste
Reynolds number, the maximum fctional pressure loss reduction will increase
progressively as the flow behaviour index decreases. The minimum drag ratio was
found to be correlated (Farooqi & Richardson,lgS2) with a dimensionless parameter J
(Figure 6):
;k =t.er 0.05<J<0.35
where
(6)
and V* is the superficial slurry velocity at the point of laminar flow breakdown.
L
,F
ffi'
o .l K,
+.6 -3
O
x.5.f.
q
67
"1
ol5 7
E r J6
.J
t-,1
1.9 6
L7T .6t
f ? 5? r6t
.E
Y .26? -51
'E .tlfs .2J
t0.95
cl 26
t0 r00 1t0
Hetzner- Reed ReYnotds numbcr Ru,'oR
.
+
.7 l,N
t
.5 a
I
.5 .l
.Ft I
.
I
rJ
3L
a
+
- 4*
[Jl
,3
E
E
F-
Parameter, J
The optimisation of air incorporation into a shear-thinning paste in laminar flow in order to
maximise power savings has been studied (Dziubinski et al, 1985 & 1988). The superficial air
velocity, to obtain maximum po\Mer savings has been calculated and compared with
experimental data (Dziubinski & Richardson, 1985). In order to achieve any overall power
saving, the reduction in power by the pump as a result of the presence of air in the pipe must be
greater than the power required to compress the air to the line pressure for injection into the pipe'
To make the comparison, a coefficient of power saving, y, which normalises the power saving
(AN) with respect to the pump power required for slurry flow alone r is used:
AN (7)
'i,=No
They showed that ry could be expressed as
where qo is the efficiency of the pump, *d tl. is the efficiency of the air compressor. Values of
{ were determined for a range of conditions and maximum values identified by differentiating
Eqn (8) with respect to Vr. Maximum values were found to occur in the plug flow regime for all
experimental data used, i.e., when Vr. 1 m/s and Reun < 500. In this flow regime, we have
'When
already noted that the drag ratio may be expressed using Equation (2). Equations (2) ar
(8) are combined, differentiated to give dy/dVr, and this differential set to zero, this leads to the
maximum porer saving condition. First, it is necessary to select the desired values of the
following parameters:
V, superficialslurryvelocity
n flow behaviour index
r'p/I. the ratio of the efficiencies of the pump and air compressor
APsut n determined by the properties of the paste, its flowrate and the pipe dimensions.
V, may be determined by iteration. Maximum values of y can then be obtained from Eqn (8)
and are shown in Figure 7; it is applicable for 0.24 4 V, < 0.98 m/s. Figure 7 shows that:
(a) the power saving becomes progressively greater as qolt1. decreases; unless the efficiency of
the air compressor exceeds that of the pump there will be no power saving;
(b) for typical values of pump and compressor efficiencies (np : 0.6; q. : 0.85; \1\":0.7), an
overall power saving in gas/slurry flows is possible when the flow behaviour index is below 0.3.
For significant power savings, the flow must be laminar, the flowrate must be low, and the
compressor must have an appreciably greater efficiency than the pump.
6. Gonclusions
Technology for reducing the frictional pressure loss for the pipeflow of thickened tailings and
other viscous pastes using air injection is established. Simple models are avallable which
allow the prediction of the extent of the frictional pressure gradient, once the tailings
rheological properties have been measured under the right conditions (Heywood & Alderman,
2004). Now industry needs to apply this technology to tailings transport with the promise of
substantial benefits.
References
Chhabra, R.P. & Richardson, J.F. (1984) '?rediction of flow pattem for the co-current flow of
gas and non-Newtonian liquid in horizontal pipes", Can J Chem Eng,Yol62,Pages 449-454.
Dziubinski, M. & Richardson, J.F. (1985) "Two-phase flow of gas and non-Newtonian liquids in
horizontal pipes - superficial gas velocity for maximum power saving", J Pipelines, Vol 5, Pages
107-lll.
Dziubinski,M. (2002) "Hydrodynamics of non-Newtonian two-phase flow in pipes", Proc
6th European Conference on Rheology, Erlangen, Germany, September.
Farooqi, S.I. & Richardson, J.F. (1982) "Horizontal flow of air and liquid (Newtonian and non-
Newtonian) in a smooth pipe.2. Average pressure &op", Trans I Chem -8, Vol 60,323-333.
Farooqi, S.I., Heywood, N.I. & Richardson, J.F. (1980) "Drag reduction by air injection for
suspension flow in ahoizontalpipeline". Trans I Chem E,Yol58,Pagesl6-27 '
Heywood, N.I. (1976) Air Injection into Suspensions flowine in Horizontal Pipes. PhD Thesis,
'Wales,
University of UK.
Heywood, N.I. & Alderman, N.J. (2004) "Important considerations when making rheological
measurements on pastes". Proc Paste 2004 Seminar, Cape Town, South Africa.
Heywood, N.I. & Richardson, J.F. (1978) "Head loss reduction by gas injection for highly shear-
thinning suspensions in horizontal pipe flow". Proc Hydrotransport 5,Paper Cl. (BHR Group).
Notation
D pipeline bore (inside diameter)
Paste 2004
Cape Town
South Africa
Paper 10
Paul Slatter
Keynote Address:
Abstract
The operational flow rates of industrial paste pipelines often falls in an area in which it is
notoriously difficult to make reliable and repeatable viscosity measurements. The objective of
this paper is to highlight this problem and offer some practical solutions. After a review of the
relevant literature, the basis of the fundamental problems is illustrated by application to
practical example, and rheological measurement and characterisation are linked to the
pipelining problem. Discussion centres around the dilemma of the need for in-depth and
sophisticated rheological characterisation required for a full understanding of flow behaviour
juxtaposed against the need for robust on-site and on-line data for operational and control
pu{poses.
NOTATION
Symbol Description Unit
d vane diameter m
D internal pipe diameter m
K fluid consistency index, plastic viscosity Pa.sn
m Cross model exponent
n flow behaviour index
r radius m
T torque N.m
u point velocity m/s
V average velocity m/s
v shear rate, velocity gradient 1/s
q viscosity Pa.s
l. relaxation time constant S
Subscripts
0 at the pipewall, at zero
B for the Bingham plastic rheological model
at infinity
1. lntroduction
Paste pipelining is becoming increasingly desirable, driven by both economic and
environmental issues (Slatter, 2002). The rheology of the paste plays a critical role in
optimising the design and operational control of these pipelines. The range of the viscous
characteristics of paste makes their measurement particularly problematic, and the objective
of this paper is to illustrate this, and offer practical solutions.
The materials we will be considering will be assumed to be homogenous high concentration
non-Newtonian slurries, with particles typically -100pm. We also discuss the implications of
adding coarse material to this - as is done in co-disposal. Although rheology is a strong
function of several variables, one of the most important ones is concentration, and the main
characteristics of the pipelining problem are presented in Figure 1.
006
6
'd
E
oo
l
E
;c
.0)
!G
3 ooz
g
I
1 1.5
Flow Rate Q [m^3/sJ
Figure I : Pumping System curves for a typical mining tailings at Cu: 20o/o,30yo and 40%o n
a 600mm pipe
Figure 1 shows that while at low concentration, turbulent flow is the best option, but this is
simply not possible at higher concentrations where flows become laminar and the viscous
stresses - described by the paste rheology - dominate.
2. Paste Rheoloqv
Any material which is designated non-Newtonian means that it does not obey Newton's law
of viscosity, which can be stated algebraically as
For non-Newtonian materials, the viscosity q (as defined in Equation (1)) is not constant and
is a function of the shear rate '!' . There are three models (Slatter & Wasp, 2002) which have
been used consistently in the literature to describe the viscous characteristics of mineral
slurries viz: the Pseudoplastic model, where
The general form of these models is shown in Figure 2. The Newtonian model would appear
as a horizontal line in Figure 2.
.=
U'
o
o
CD
Yield Pseudoplastic
Shear Rate
to paste pipeline engineers because of their relative algebraic simplicity, they present
fundamental problems at high and low shear rates, where Newtonian plateaux (ie horizontal
asymptotes) are expected. The Bingham plastic model does achieve a high shear rate
Newtonian plateau, but still presents problems at low shear rates, where the yield stress causes
the viscosity to tend to the oblique asyrnptote in Figure 2 at a slope of -1.
These issues are more satisfactorily dealt with by the Cross model (Cross, 1965),
:10 - (s)
r+(l,i)- '
=U)
o
o
.2
Shear Rate
3. PASTE RHEOMETRY
Bob-In-Cup Rheometry. Nguyen &, Boger (1992) remark that concentric cylinder
rheometers seem to make their way into most laboratories, and the conveniences associated
with these instruments certainly makes them attractive (Slatter, 1997)' The basic
measurements are the torque T required to rotate the bob at a constant angular velocity fl.
Given that one of the primary objectives of rheometry is the establishment of the relationship
between the shear stress and the shear rate - often referred to as a rheogtam or a flow curve -
we start by casting this in its inverse form t = iG) ' Then
a=;1""'ft)f ,
(6)
T (t)
rbob = and r"uo
;"nrA
Since i:i(") is unknown prior to the analysis, this presents an inverse type problem. Data
reduction for non-Newtonian fluids in concentric cylinder instruments is therefore tedious and
represents a major disadvantage (Nguyen & Boger, 1992). This is further complicated by the
presence of a yield stress. Both these issues are relevant for paste rheometry.
Direct Measurement of Yield Stress. The value of the yield stress is required before
competent reductiori of concentric cylinder paste data can be performed. Direct measurement
of the yield stress is recommended by Nguyen & Boger (1992), and several methods are
available - two will be discussed here, the vane and slump tests.
The vane test involves immersing the vane in the fluid, slowly rotating it, and measuring the
time/torque response, from which the maximum torque T-u* is extracted. The special
geometry of the vane causes the material to yield within itself, rather than on the interface
between the sample and e.g. a bob. Wall slip is thus excluded. At yield, the sheared surface
comprises the cylindrical surface of the vane tip loci (Yan & James, 1997) and the yield stress
can be calculated from
r -- nd'
rmx [l-tl -Ty'
^ I r '^ I
Lorl
(8)
The slump test has been used by concrete engineers for many years to estimate the
workability of fresh concrete. Standard slump tests involve a conical frustum mould, but
Clayton et at (2003) recommend a cylindrical shape, and the reader is referred to this
reference for the different anallical approaches relating yield stress to slump height. These
authors point out that slump height is an empirical value which is a complex function of both
yield stress and density. Since yield stress is a unique material property, it is the more
appropriate to use for direct comparisons.
Vane-In-Cup Rheometrv. Because of its unique ability to exclude wall slip, as well as
several practical advantages, Bames & Nguyen (2001) state that rotating vane geometries
have become increasingly popular, and are particularly appropriate for paste materials.
Although further research is needed to establish the upper limit of rotational speed, the vane
geometry is very useful as a general means for measuring the viscosity of difficult liquids
(ibid). The reader is assured that paste material comfortably qualifies as a "difficult liquid".
Tube Rheometry. Because of the obvious geometric similarity, slurry pipeline engineers
often prefer the tube type of viscometer (Slatter & Lazarus, 1986, V/ilson et al, 1992).
Although tube rheometry usually requires the sacrifice of much of the convenience of
rotational rheometry (see e.g. Slatter, 1997), the development of the Balanced Beam Tube
Viscometer (BBTV) (Slatter et al,1998) addresses this, and the BBTV has been shown to be
a valuable and versatile research instrument. The basic measurements are the volumetric flow
rate Q and the pressure drop p over a known length of pipe L, which yields the wall shear
stress rs:DLpl4L. Then, for laminar tube flow the relationship between the bulk shear rate
8V/D, the wall shear stress re and the volumetric flow rate Q; and the rheogramis as follows:
8V 32Q --; 4 f.o ,r.i!)d"
;- = I (9)
--D nD' r; Jo
(Chhabra and Richardson, 1999). This means that in general the bulk shear rate
(8V/D) is a unique function of the rheogram i:i(t) and the wall shear stress (t6),
provided that there is no time dependency or slip at the wall and the flow is laminar.
lndeed, conversely this then provides a check for the presence of slip or turbulence.
o Being a definite integral, Equation (9) shows that the relationship between 8V/D and to
can be obtained by numerical integration using data directly from a rheometer, without
o It provides the link or pathway between the rheogram and the pseudo shear diagram, ie
given a rheogram,'we can use Equation (9) to construct a pseudo shear diagram (8V/D
vs re ), which can be used for design in laminar flow. Conversely, we can reverse the
process:- differentiating Equation (9) provides the link or pathway between the pseudo
shear diagram and the rheogram. This approach is used to derive the Rabinowitsch-
Mooney relationship, which enables us to construct a rheogram if we have a pseudo
shear diagram.
The main problem with tube viscometry is that 8V/D is not true shear rate but the wall shear
rate for a Newtonian fluid, therefore the bulk shear rate has to be transformed to the true shear
rate, (t ). For a flow curve where the form is unknown, Equation (9) after some
manipulation
o.:[-#),=i(#), (10)
(1 1)
If one plots a logJog pseudo shear diagram (to versus 8V/D) for the laminar flow region then
n is the slope of the tangent of this graph. The slope will only be approximately constant if
the fluid is a power-law fluid (Chhabra and Richardson, 1999)'
For a Newtonian fluid, the integrand of the Rabinowitsch-'Weissenberg integral is a cubic, and
becomes even steeper for shear thinning fluids. An important consequence of this is that the
integral value is dominated by the higher shear stress - and higher shear rate - values for
shear thinning materials. Low shear rate phenomena - such as the existence or otherwise of a
'Wasp,
"true" yield stress - are not relevant for laminar pipe flow (Slatter & 2}lz,Masalova et
Problem areas in rheometrv. The main problems which arise in rotational rheometry are
that the gap size should be greater that 10x the maximum particles size, together with the
influence of the yield stress which reduces the gap size. In tube viscometers, fully developed
flow must exist between the pressure tappings. Both tlpes suffer end and slip effects, which
must be detected and corrected
Perhaps the most iniquitous problem which arises with paste rheometry is that the
sophisticated software interface between user and instrument is capable of delivering such
credible but erroneous results, some of which are illustrated below.
4. PASTE PIPELINING
The fundamental pipeline problem is to predict the pressure gradient from the rheology. The
problem presents in three parts, laminar flow, the laminar/turbulent transition and turbulent
flow. The detail of these approaches is presented elsewhere (Slatter, 1999 and2001a) and was
used to produce Figure 1. However, it was previously shown (Slatter 2001b) that for yield
stresses above 1OPa, turbulent flow becomes progressively unachievable at realistic velocities
(<3mls) in industrial size pipes. Since paste materials typically exceed this yield stress value,
we shall restrict our discussion to laminar flow.
In particular, we shall consider some of the interesting issues which arise when a material
which behaves like a Bingham plastic with the properties shown in Table I is pumped at a
Tv K To SViD To
rheometer, without due diligence applied to the data reduction, then the deformed rheogram
shown in Figure 4 wilt result. The deformation of the rheogram is as a result of the yield
stress causing a partially sheared gap in the rheometer.
60
58
56
E*
llt Rt
"'
o
50
48
o
c46
U)
44
42
40
60
This deformed rheogram is an almost perfect fit for a Herschel-Bulkley fluid, and will glve
the results shown inTable 2.
Tv K n t
Pa Pa.sn
48.5 0.899 0.557 0.99986
Clearly Table 2 is in error (the correct values are as in Table 1), but the unwary or
inexperienced user may well attempt to use this information. Translating the information from
Table 1 and Table 2 into pipe flow velocity profiles and comparing them, we arive at the
result presented in Figure 5. Figure 5 shows; errors in velocity estimation of 600/o (without
even including experimental error), that this matter is not trivial, and that without a systematic
approach to rheometery, significant errors will result'
'1.8
_U) 1.6
.E. t.q
],,
'1
o
I oB
.c
o
0.6
fL o.4
o.2
0
0.1 0.15 0.2 0.25 0.3 035
Figure 5: Laminar pipe flow velocity profiles constructed from Table 1 and Table 2.
Tube rheometry. Although tube rheometers - and the BBTV in particular - are well suited to
rheometry for pipelining purposes, it is important to understand the limitations of this method.
One of the attractions of this approach is that much higher shear rates can usually be obtained
quite easily with a tube rheometer. Whilst this remains of great importance for turbulent flow
in pipes, it is quite irrelevant for us as we are interested in laminar flow, in true shear rates
less than 100 1/s and bulk shear rates less than 16lls. Therefore the hiehest velocities we are
interested in a 50mm tube would be 0.1 m/s, and most tube rheometers are smaller than this'
Velocities in this diminutive range are extremely difficult to measure accuratelY, and even if
this is possible, there are other problems. Figure 6 shows BBTV data for an 18 %(vlv) kaolin
clay suspension (Slatter, 1999). Now, the BBTV is designed to operate at extremes of
velocity, and can accurately measure those velocities of interest in this case. The problem
which comes to light is a diameter dependent low shear rate slip effect - which manifests
precisely in the rage we are interested in investigating! Nguyen & Boger (1992) show a
similar effect over the same bulk shear rate range and recommend the use of tubes with a
rough surface.
1 000 r
32mm
10
0.1 1 10 ,l00 1000 10000 100000
Pseudo-Sheor Rote 8V /D lt /"]
Pressure gradient prediction. Assuming that the problems associated with paste rheometry
are surmountable, we will then be able to integrate the rheogram (as per Equation (9)) and
obtain the pressure gradient. Using the information from Table 1, the integrand of the
Rabinowitsch-'Weissenberg integral takes the form shown in Figure 7 . The point of interest is
that 99Yo of the area under this function occurs above a shear stress of 55.7 Pa at a shear rate
o 14 lis. Rheometeric information at shear rates lower than 14 7ls (in this case) are not
relevant for pressure gradient prediction. This further reinforces the statement made earlier
that low shear rate phenomena - such as the existence or otherwise of a "true" yield stress -
are simply not relevant for laminar paste pipe flow pressure gradient prediction.
6
N
G
o-
C
.9
I
o 200000
l-
f
TL
150000
6-
o)
()
# 100000
c
50000
0
56
Coarse particles. It is now well known that coarse particles will settle in laminar paste flow
Line start-up. The problem of line start-up is quite different from operating point pressure
gradient prediction. The issue of static and dynamic yield stress is highly relevant here, and its
actual value will directly determine the pumping head required at line start-up' So while the
issues and measurement of yield stress are not directly relevant for pressure gradient
5. Discussion
The most appropriate rheometry strategy for in-line operational control purposes is probably
that adopted by Houman & Johnson (2002). A tube rheometer is incorporated into the normal
design of the thickener recirculation and transfer pumping arangements, and the signal can be
used to ensure that the flow properties of the paste remain within the pumping envelope
Although the problems described above apply to both rheometry and the pipelining problem
in a general way, it is clear that they are exacerbated by the particular range of values of
interest in paste pipe lining.
6. Gonclusion
The methods and significance of the different approaches to paste rheometry have been
presented. The difficulties associated with the measurement ranges required for paste
rheometry have been demonstrated and discussed. It is concluded that a broad and systematic
approach is essential if reliable and robust paste pipelining systems are to result.
Reference
Barnes, HA and Nguyen, Q D (2001), Rotating vane rheometry a review, J. Non-
-
Newtonian Fluid Mech.98 l-I4.
Clayton S., Grice T.G., Boger D.V. (2003) Analysis of the slump test for on-site yield stress
measurement of mineral suspensions, [rt. J. Miner. Process. 70,pp321
Chhabra, R.P. and Richardson, J.F. (1999) Non-Newtonian Flow in the Process Industry
Oxford : Butterworth-Heinemann
Cross M.M. (1965) "Rheology of non-Newtonian fluids: a new flow equation for pseudo-
plastic systems", J. Colloid Sci.20, Pages 417-437.
Houman, J. and Johnson, G. (2002) High density disposal of co-thickened Kimberlite slurry
Paste 2004
Cape Town
South Africa
Paper 11
Andrew Vietti
Andrew Vietti
De Beers Consolidated Mines Limited
Abstract
An attempted is made to provide an explanation for the observed suspension and compaction
behaviours of kimberlitic clay slurries based on three mechanisms affecting clay colloidal
properties. In addition, three models are proposed which allow for the prediction of both the
suspension and compaction behaviours based on an understanding of the slurry system
parameters.
1. lntroduction
Enhanced water recoveries are achieved by P&TTD systems, by concentrating the suspended
solids within so-called "low-density" slurries through novel high compression thickening
processes. These high-density tailings are then transported hydraulically to a surface disposal
site where they are deposited.
The operation of the P& TTD thickening and pumping processes are themselves critically
dependant on the settling and rheological behaviours of the clay slurries. For instance, the
thitkening process requires that the suspended clays be in a colloidally unstable state for
solid/liquid separation to take place and the subsequent compaction to a high-density state. It
is equaily important that the rheological behaviour of the high-density sluny be such that
hydraulic transport is possible.
The suspension and subsequent rheological behaviours of both low and high-density clay
slurries are dependant on a number of ore and water related parameters which affect the
colloidal properties of the suspended clays. This paper attempts to explain the mechanisms
affecting kimberlitic clay slurry suspension and compaction behaviour.
2. Saline Aqricultural Soils
o Saline soils
. Saline-alkali soils
o Non saline-alkali soils
1) The cond,uctivity of an extract taken from a saturated soil sample provides a measure
of the water-soluble cations within the soil (i.e. the salinity of the soil).
These soils have high conductivity; the ESP and pH values are low. Because of the
high soluble salt content, the soils are in a "flocculated" state and will form settling
slurries if suspended in water.
The ESP of these soils are high, however, these soils behave either as saline soils or
non-saline-alkali soils depending on the amount of soluble salts present (i.e' their
conductivity). If the conductivity of the soils is high, and the pH of the soil is low the
soils remaitt. in u "flocculated' state. If however, the soluble salts are leached out of
the soil, the properties of the soil change and they begin to behave as non-saline-alkali
soils.
The conductivity of these soils are low, the ESP and the pH values are usually high.
The ESP has a profound effect on the chemical characteristics of the soil. The higher
the ESP, the higher the pH and the more the soils tend to disperse. Typically,
problematic procssed kimberlite ores show behavioural characteristics similar to this
category of soil.
3. Alkalisation of Soils
Most importantly to the process of alkalisation, is the fact that it is the clay fraction in the soil,
which is able to adsorb nd exchange cations (notably sodium, calcium and magnesium) from
the surounding aqueous medium by ion exchange mechanisms. Normally, calcium and
magnesium ut" ttt. principle ions found in a saturated soil extract, however, under certain
.orritiorrr, the sodium ion can become the dominant ion if the salts become concentrated
through evaporation. ln this case, the saturation limit of various salts (e.g. calcium/magnesium
sulphates or carbonates) is exceeded and they are precipitated out of solution, increasing the
reltive proportion of sodium ion. Under such conditions, sodium replaces the original
exchangab calcium and magnesium cations on the clay surfaces, the ESP of the soil
increases and the soils become alkali.
Another factor, which plays a vital role in determining the ion exchanged nature of the clays,
is the chemical quality of the water, which contacts the soil.
The alkali hazard potential of a water used for irrigation is determined by the absolute and
relative concentrations of the cations in the water. If the proportion of sodium in the water is
high, the alkali hazard is high and conversely if calcium and magnesium predominate, the
hazard is low. An easy to remember rule of thumb is 'hard water makes soft land and soft
water makes hard land" (Richards 1969).
A unit which is used to determine whether a water quality is likely to create clays, which are
highly sodium ion exchanged is know as the Sodium Adsorption Ratio (SAR) value of the
*t"r. The SAR is the ratio of sodium ions to calcium and magnesium ions in solution and the
figures are derived from a normal chemical analysis of the water (in meq/l):
SAR:
(Co'* + Mg2*) l2
Since there is a fairly good correlation between the SAR value for a water and the ESP value
of a suspended clay, the ESP value of the clays in the irrigated soil can be estimated if the
SAR of the irrigating water is known (Richards 1969).
In order to understand the behaviour and interaction of clays in slurries (or soils), it is
essential to understand their structure. Smectites are classified as 2:1 type clays as a
consequence of the particular arcangement of the clay particle crystal lattice (Van Olphen
lg77\. The crystal lttice is composed of a single octahedral "gibbsite" layer (if the central
atom is Al3*) which is sandwiched between two silicon tetrahedral layers (Hurlbut & Klein
1977) (Figure 1).
Figure 1: Crystal Lattice Structure of a Tlpical 2:I Clay Mineral (Hurlbut & Klein 1977)
These crystal lattice layers are however, not uniform with respect to the chemical nature of
their central atoms. Often, isomorphus substitution of the existing atom by an atom with a
lower valance can take place (for example Mg2* for Al3* in the octahedral layer) that results in
an excess negative charge,which is distributed at the tetrahedral surfaces. The excess negative
charge is compensated for by the adsorption of cations onto the outer surfaces of the clay
crystal lattice structure. These cations are present even in the dehydrated forms of the clay,
however, in the presence of water, the compensating cations may be exchanged by other
cations in solution depending on how strongly they are bound to the clay surface. For this
reason, they are known as exchange cations and their concentrations can be used as a measure
of the amount of lattice charge or cation exchange capacty of the clay (Van Olphen 1977).
Smectite clay surface charge accounts for about 80% of the total cation exchange capacity
while the remainder is accounted for by the clay particle edges. It is generally concluded that
under aqueous conditions, hydroxyl (OH) groups will attach to the exposed silicon
tetrahedral and metal ion octahedral atoms at the clay edges (Svarovsky 1981). As a
consequence, the pH of the surrounding environment will have a profound effect on the
crystal lattice edge charge. Under acidic conditions, the OH- bearing gloups will become
protonated to carry an overall positive charge. As the pH is increased, the OH- groups will
become deprotonated until a point of overall edge neutrality is achieved. This pH is known as
the Point of Zero Charge (PZC) of the clay crystal edge. Further increases in pH will result in
total deprotonation of the OH- groups until an overall negative edge charge dominates (Figure
2).
E
(
.F
cq)
o
o_
o
o
N
'With
Figure 2: Surface Electrical Charge Characteristics of a 2:l Clay Respect to pH
Finally, all members of the smectite group of clays share one common feature, in that they
have ihe ability in the presence of moisture to absorb water and other polar molecules
between the particle unit layers and cause swelling of the matrix - hence the term "swelling
clays". Two forms of swelling mechanism are known which depend on the moisture content
to which the clays are exposed.
Firstly under low moisture content conditions a limited step-wise expansion of the unit layers
known as Interlayer (or Type I) swelling occurs. As more water molecules are drawn between
adjacent clay pltel"r oi to ihree layers of water molecules are covalently bonded to the
tetrahedral surfaces and in a semi-crystalline structure that resembles that of ice. This mode of
swelling leads to at most a doubling in the volume of the dry clay (Grimm 1968)'
The second form of swelling occurs at high moisture contents and can lead to an unlimited or
complete separation of individual layers and is known as Osmotic (or Type II) swelling.
Undr this cndition, the exchange cations dissociate from the clay surface and move to the
hydrated region between clay particles and as such they are regarded as being "in solution"
and hence lower the activity of the water between the particles. This allows more water from
the surrounding to move into the interlayer region by osmotic forces thereby increasing the
interlayer .*rlling. This form of swelling may continue indefinitely until normal electrical
double layers separate the individual clay particles (Figure 3)'
130
110
x-/o
Interplanar
Spacing o
(A) o
50
o
30
x+r#r
10
45
-1t2
c
Figure 3: Change in Layer Spacing of a2:1 Clay with lncreasing NaCl Concentration (Sequet
et al. 1975).
(Type I swelling is step-wise and limited while T1,pe II swelling is linear and unlimited)
The degree of osmotic swelling of a particular clay depends largely on the nature and
concentration of the cations in the contacting water and the degree of octahedral substitution
of the clay t1pe. Monovalent cations (such as Na*) in solution tend to cause unlimited
swelling since they are small and are able to "dissolve" more easily in the semi-crystalline
water layers which surround the clay particles thereby drawing-more wat^er between the
adjacent particles by osmotic action. bivalent cations (such as Ct* and Mg2*) on the other
hand tend not to cause unlimited swelling since they have a disruptive effect on the water
layer structure and as they are able to provide links between charged sites on adjacent silicate
sheets (Sequet et al. 1975; Mering 1946).
Conventional kimberlitic ore processing practices (particularly in the water recovery and
tailings disposal circuits) are undergoing rapid change. The concept of Paste and Thickened
Tailings Disposal (P&TTD) is gaining acceptance within the minerals industry, primarily as a
means of reducing water consumption, as well as improving disposal site stability and safety
(Robinsky 1999; Patercon et al. 1999). However, the technique is more sensitive to the
variability in the behavioural characteristics of clay suspensions than are the conventional
water recovery and tailings disposal circuits.
V/ithin a tlpical kimberlite processing circuit, ore is crushed and scrubbed before being
screened and processed further. Scrubbing represents the first contact between the ore and the
plant process water and may be critical in determining the behavioural characteristics of the
subsequent low-density tailings stream. Should a P&TTD circuit be installed, this stream
would typically be routed to a high compression thickener where the solid/liquid separation
would take place and the solids would be compacted to a high-density underflow product,
which would have certain unique non-Newtonian rheological characteristics. These
characteristics would determine the pump and pipeline design requirements and subsequently
the disposal site sizing requirements.
As stated, low-density kimberlitic tailings slurries (-1.5 mm) would t1pically constitute the
feed to a high compression thickening unit. The slurry is generally classified into two
fractions according to particle size, namely the +75 micron -1.5 mm "grits" fraction (which is
easily settlable) and the -75 micron "slimes" fraction containing the clay minerals. The most
important parameter affecting the thickener performance would be the suspension or settling
behaviour of the "slimes" fraction within the slurry.
1. Ion exchanged nature of the suspended clays:- increasing the sodium ion exchanged
nature of the clays will lead to increased dispersion behaviour of a clay slurry suspension.
2. pH of the suspension:- as alluded to earlier, the pH of the suspension greatly affects the
charge associated with the clay pafcle edges. Below the clay edge PZC, edge-to-face
particle interactions take place, resulting in particle aggregation and settling under gravity.
At suspension pH values above the PZC, clay slurries will tend to remain dispersed as a
consequence of negative particle repulsive forces (Figure 4).
4681012
Figure 4: Clanty Profiles of a Sodium Exchanged Clay Suspension (left) and a Calcium
Exchanged Clay Suspension (right) with pH.
a
J. Ionic concentration of the suspension'- A third and overriding mechanism affecting the
colloidal properties of clay suspensions is the absolute ionic concentration of the
suspension. The electrical double layer surrounding individual clay particles becomes
progressively compressed at high ionic concentrations, reducing interparticle distances
and allowing particles to interact. An ionic concentration is reached, known as the Critical
Coagulation Concentration (CCC) at which, the forces of attraction between adjacent
particles become greater than the repulsion forces and particle aggegation and settling
can occur.
Integration of the threemechanisms can provide visualisation models to explain the
suspension and compaction behavioural observations of kimberlitic clay slurries.
Typically, particle colloidal potential would increase with increasing clay ESP due to the
increasing ispersive nature of the clay, however, this trend is only truly expressed in a
naffow band between approximately pH 8 and 11. At lower pH values, clay particle edge-to-
face interactions take place and hence settling can occur even at high ESP values. Also at pH
values greater than 11, settling is once again observed at high ESP values simply due to the
CCC of the suspension being exceeded (Figure 5).
s
C
o
;.1
.F
(o
o'J
6)
()
C
o
O
tJ)
i:
I o.2
an 0.0
Figure 5: Model for Describing the Suspension Behaviour of Naturally Settling Low-Density
Kimberlitic ClaY Slurries
Observing and describing the suspended portion of a slurry provides one view of the
behaviour of clay particles within such a slurry. Another view, which would be more useful to
the hydraulic transport and deposition characteristics of the slurry, would be to describe the
behaviour of the settled or compacted portion of the slurry under the same conditions as
described in Figure 5.
In this case, it would appear that two compaction zones (or clay particle interaction zones)
exist - a zoe of relatively poor compaction surrounding a zoe of high compaction which
corresponds to the region of maximum clay dispersion (Figure 6). Indeed, the degree of solids
consolidation in the compaction zone is significant, up to 60% solids by mass (Figure 7).
Mud Bed Compaction After 90 hrs Mud Bed Compaction After 90 hrs
(Plan View) (lsometric View)
'12
11
10
-o- o
U) I
Figure 6: Model for Describing the Compaction Behaviour of Naturally Settled Low-Density
Kimberlitic Clay Slurries
'11
10
-9
.
:8
Figure 7: Solids Content of the Settled Bed of Naturally Settled Low-Density Kimberlitic
Clay Slurries
Initially, these observations may appear to be contradictory, i.e. a colloidal condition exists in
which maximum clay particle dispersion is allowed and in which maximum slurry
consolidation and compaction of any settled solids to take place.
In order to explain this seemingly contradictory behaviour, the orientation of the clay particles
in the settled bed was investigated and visualised using a cryogenic Scanning Electron
Microscope technique.
Figures 8 and 9 represent two micrographs describing similar clay mineral suspensions with
differing colloidal properties. Figure 8 describes the clay particle orientation within a slurry in
which the colloidal properties allow particle interaction to take place (i.e. within the particle
interaction zoe as described by the compaction model). It is noted that significant "edge to
face" interactions are allowed with the accompanied presence of a significantly high void
ratio and low compaction density.
,
.J
:'-: t-.
^" x:'
\riP
',,1
'
1" -.
,er *![..
U o.;i
l*ri*='ttc.
tritl.f, ,P'ltlfrUNIT
ELECTRON I{ICR0SCOPE UCT Detector= SE1 L-Nov-z3
Mag= 5. K X EHT= 5. kV I Probe= 18 pA tfD= 14 ll Photo No. =9
Figure 8. Scanning Electron Micrograph Describing Clay Particle Orientation Associated with
a Sluny in which the Colloidal Properties allow for Particle rteraction
Figure 9 describes the clay particle orientation of a slurry in which the colloidal conditions are
such that particle interaction is not allowed. In this case, a high degree of compaction of the
solids is achieved. It is noted that almost exclusively "face to face" interactions are allowed
with the accompanied presence of a significantly low void ratio and high compaction density'
EHT= 5 08 ku I Probo= 5 pA m= 13 nn Photo N0 =15
'l
tit,
:t
'}lil{t.U&g-.WlYe\t,!l\a'raE.-.&'-&3;.t',:--.d
UI{IT UCT Detector= SE1 l1'-Nov-23
ELECTRON I{ICROSCOPE
Mag= 5. K X EHT= 5. kV I Probe= 5 pA tdD= 14 nn Photo No.=18 !
Figure 9. Scanning Electron Micrograph Describing Clay Particle Orientation Associated with
a Slurry in which the Colloidal Properties do not allow for Particle Interaction
The effects which slurry colloidal properties have on clay particle association and hence paste
rheology are clearly demonstrated in Figure 10 in which a wide range of rheological
behaviours are expressed by a single thickened kimberlitic clay mineral paste. In this case,
under certain colloidal conditions, the slurry (dso of 2 micron) was shown to have no yield
strength at solids contents as high as 47o/o solids by mass, while in other conditions, it was
shown to develop significant strength at comparatively low solids contents (30% solids by
mass).
These rheological observations, when simply related to the solids content within the paste
provide no explanation of this apparently aberrant behaviour, however, if the same data is
xpressed as a function of slurry colloidal properties such as ESP and pH they can be
explained clearly (Figure 11).
E 200
o)
c
E
150
!
1oo
o)
-c.
U)
20 30 40
Figure 10: The Effect of Sluny Colloidal Properties on the Rheological Behaviour of a Single
Kimberlitic Clay Mineral Suspension and Expressed as a Function of Solids Content.
o
o,
q)
(r)
!
,
Figure 11: The Effect of Solids Content on the Rheological Behaviour of a Single Kimberlitic
Clay Mineral Suspension (at 41% solids content by mass) Expressed as a Function of Slurry
Colloidal Properties.
9. Gonclusion
However, as will all processes, pushing the boundaries of a technique requires gteater
attention to detail. V/ith regard to P&TTD systems, the detail appears to be at the level of the
surface chemistry characteristics of the suspended clay minerals.
As demonstrated, the colloidal properties of a slurry can have a profound effect on the
settling; compaction and rheological behaviour of many mineral tailings and particularly
those containing clay minerals. In these slurries, the effects of slurry density appear to play a
sub-ordinate role to the colloidal properties in determining the rheological behaviour of the
paste.
A complete understanding of the tailings suspension characteristics (i.e. both the mineral and
aqueous components) as well as the mechanisms which affect the clay surface charge
characteristics are required in order to manipulate process conditions to match the process
needs and to ultimately master the P&TTD system.
Based on the above and on visual observations, two models have been proposed in order to
predict the suspension and compaction behaviours for kimberlitic clay mineral suspensions.
Reference
Richards, L.A. (ed.) (1969) "Diagnosis and improvement of saline and alkali soils", US Dept
Agriculture Handbook No. 60.
Van Olphen, H. (1977) "An introduction to clay colloidal chemistry", John Wiley & Sons,
New York.
Klein, C. and Hurlbut, C. S. (1993) "Manual of Mineralogy (21't ed)", John Wiley & Sons,
New York.
Sequet, H.; De LaCalle, C. and Pezerat, H. (1975) "swelling and structural organisation of
saponite", Clays and Clay Minerals,Yol23, Pages 1-9.
Pste 2004
Paper 12,
Jim Wallace
Jim Wallace
GL&V Australia
Abstract
In 2001, Anaconda Nickel Ltd (now Minara Resources), identified a significant increase in
autoclave capacity and energy cost savings were achievable at their Murrin Murrin nickel
laterite plant if the average feed solids concentration could be increased. Anaconda
determined a 3%o increase in feed solids concentration from 36%o to 39o/o wlw was possible
GL&V Australia was commissioned to perform bench and site pilot scale paste thickening
testwork. Options were examined to paste thicken whole of autoclave feed and also to pre-
classify the feed, paste thicken the cyclone overflow only then reconstitute with the cyclone
underflow. The latter option was found to be the most cost efficient model.
Bench scale testwork was completed in April 2002, with these results being confirmed by
onsite Deep Cone pilot plant testwork conducted in August 2002. Based on the testwork
results, a 20m dia x 19m tall Eimco@ Deep ConerM Thickener was selected for the Feed
Preparation thickening duty. GL&V was awarded a contract to design, supply and site erect
the thickener cormencing in November 2002. The thickener was brought on line in October
2003 on schedule and budget, achieving targetperformance within two weeks of startup'
This paper discusses the testwork, sizing & design criteria, commissioning and scale up of
1. lntroduction
autoclave capacity and energy cost savings were achievable at their Murrin Murrin nickel
laterite plant if the average feed solids concentration could be increased. Anaconda
determined a 3Yo ncrease in feed solids concentration from 360/o to 39Yo wlw was possible
before pumping was constrained by increased mud rheology'
After initial bench scale testwork conducted in April 2002, GL&V Australia was
commissioned to undertake pilot plant paste thickener trials on site at Murrin Murrin. This
worked was conducted in August 2002, using a !.5m Eimco Deep Cone Thickener pilot
plant unit (as shown in figure I below)
Option 2 was to Pre classifY the feed in hydrocyclones and only paste thicken the
cyclone overflow comPonent then reconstitute the thickener underflow with cyclone
underflow.
The testwork was conducted on a 24 hour basis over a two week period and manned
continuously by GL&V personnel. Table I below provides a srmtmary of the results reported
by Vance (2002).
It is vital to dilute the feed down to the optimum solids concentration prior to flocculation.
This is even more so for paste thickening than conventional/ high rate duties. Failure to do so
can result in a significant increase in required polymer addition and a reduction in the
maximum underflow density achievable. Figures 3 & 4 show the respective optimum feed
solids for the whole of ore and cyclone overflow case studies.
,*o
ct)
Y tu*
x
f
Er*
ct)
.E
8u*
0
20% 30% 40 50% 60% 70% 80%
Weight % Solids
lf
The optimum feed solids concentration was well below the expected feed density to the
thickener circuit of 34-40o/o solids w/w depending on which flowsheet option was selected.
On the pilot plant thickener, feed dilution was performed by external dilution only. It is more
straightforward to conduct testwork this way than to install an internal dilution device due to
the range of dilution ratios and feed flows that are tested in the pilot plant. For full scale
operation, Eimco thickeners use the Eimco E-Duc@ feedwell dilution system.
3. Thickener Sizinq
Based on the results of the pilot plant testwork, paste thickeners were sized for both options.
Table II provides a sunmary of the sizing. (le"
Z I n/ 5"I
Table II: Summary of Thickener Sizing
i r rtt'taa)
QA.l O
* an additional 20o mud residence time was allowed for on top of that required in the pilot
plant as a scale upfactor (refer to section4.2)
20m was the largest diameter considered as there were already proven Eimco Deep Cone
Thickener installations at this size on Alumina Red Mud Disposal duty.
The maximum realistic dilution ratio for internal dilution systems is 5:1. External dilution of
the feed was therefore required prior to the thickener to achieve the feedwell solids
concentration for optimum flocculation and thickening performance. To maintain flexibility it
was decide to dilute the feed to l5o/o solids w/w for both cases prior to the thickener.
It was decided to proceed with option 2 aste thickening cyclone overflow only) as this
Effective internal dilution of the thickener feed should deliver intensive low shear
mixing of feed slurr diluant liquor and flocculant. Importantly, the E-Duc system
does this prior to feedwell entry and the commencement of settling. The design of the
E-Duc system has recently been refined to the new open channel E-Duc feedwell now
successfully operating at two large installations in addition to Murrin Murrin.
The thickener tank can represent 60-70%o of the overall cost of a paste thickener.
Determination of minimum required mud residence time to achieve required underflow
density is, therefore, an important factor impacting on thickener size and hence
thickener cost. Scale up of mud residence time required in the pilot plant to that
required at full scale is also an issue. Understanding of this parameter for paste
In any thickener, rake torque is proportional to the mud yield stress. As shown in
figure 6, the relationship between yield stress and solids concentration is exponential.
Paste thickeners tend to operate at the knee of the yield stress curve therefore can
experience relatively large increases in yield stress for minor increases in underflow
density.
150
120
90
60
30
0
30.0% 320% 34.0% 36 0% 38'O% 40 Oo/o
% Solids w/w
High rate/ conventional thickeners tend to operate with mud yield stresses <20Pa
compared to 100-300Pa for paste thickeners. Consequently the rake torque capacity of
the paste thickener mechanism and drive has to be in the order of ten times greater than
an equivalent diameter high rate thickener. GL&V utilise a proprietary torque model
which breaks down the thickener rake into segments and totalises the drag forces on
each segment as it moves through the mud at a given yield stress.
Based on this analysis, a thickener torque capacity of 1750kNm was specified for the
thickener and an Eimco 890P3 ring gear drive selected.
5. Proiect Execution
After the successful conclusion of the testwork phase, GL&V Australia were awarded a
contract to design, fabricate and site erect one 20m Eimco Deep Cone Thickener in
November 2002.
Due to need to conserve heat and a feed slurry temperature of 750, the thickener was designed
with a roof and insulation.
The contract was let on an open book basis to bring the thickener on line as soon as possible
by eliminating the time required for a tender evaluation process. Site erection began in April
2003 with the thickener ready for commissioning in October 2003. The project was completed
on time and on budget.
6. Gommissioninq
The thickener was brought on line on 24rh September 2003. After an initial minor electrical
fatlt, target performance was reached quickly. Less than two weeks after initial start up the
thickener stable operation was consistently achieved.
Table III provides a comparison of key design parameters between pilot plant testwork
and actual full scale results,
Table III: Comparison of Key Parameters
Actual flocculant consumption to achieve target underflow density was 30% less than
predicted by pilot plant testwork. This is probably attributable to a combination of the
open channel E-Duc system and increased mud compression in the fuIl scale unit.
Importantly the required mud residence time in the fulI scale unit was similar to that of
the pilot plant indicating the thickener rake design was able to deliver equivalent
raking intensity.
Actual rake torque was very close to the prediction from the GL&V torque model.
7. Gonclusions
Paste thickening of the autoclave feed slurry was able to deliver significant benefits to the
Murrin Murrin operation namely, increase autoclave capacity, reduced energy costs and
increased nickel PLS (pregnant liquor solution) tenor. There has also been a 6% saving in
total plant water consumption.
There was close parity of key parameters between the l.5m pilot plant and20m full scale
unit providing confidence for similar large diameter units in the future. This type of thickener
is now proven across a wide range of duties. Full scale performance can be predicted with
confidence from pilot plant and bench scale testwork.
8. Acknowledoments
GL&V Australia would like to thank Minara Resources( previously Anaconda Nickel Ltd) for
their kind permission to publish this paper.
The author thanks Ivan Hunter of SDF Pty Ltd and Kim Vance and Scott Burkett of GL&V
Australia for technical input on this paper.
9. References
Vance, K.V. (2002) "Thickening of Murrin Murrin leach feed slurry''GL&V testwork report.
Paste 2004
Cape Town
South Africa
Paper 1 3
Gase History of Cent ury inc and Sunrise Dam Gold Mine
Paul \ililliams
Keith Seddon
Australian Tailings Consultants
Abstract
This paper is a sequel to a paper presented at the Melbourne 2003 Paste and Thickened
Tailings Disposal seminar under the same title of "Delivering the Benefits" ('Williams et al,
2003). This paper presents case histories of two mature thickened tailings disposal schemes
and reviews their operational experience.
1. Genturv
1.1 lntroduction
Century is an open pit lead zinc mine in Queensland in what is called the Gulf country (Gulf
of Carpentaria). The location is shown on Figure 1. It is about 250 km north-north-west of
Mt. Isa and 300 km from the Gulf of Carpentaria coast. The mine is owned by Pasminco and
commenced operations at the end of 1999. Current ore throughput rate is around 4.5 Mtpa
producing about 4.1- 4.2 Mtpa of tailings. Mine life from start up was 20 years.
The mine is at an extremely remote location. The climate is sub-tropical with highly seasonal
rainfall. Average annual rainfall is 536 mm but nearly all of this can be expected to fall in the
January - March wet season. Road access is unreliable during the wet season. Concentrates
are pumped 300 km to Karumba close to the coast on the Gulf of Carpentaria, dried, and
barged out to bulk carriers for shipping to smelters.
Water supply for the mine covers from limestone dewatering at the mine pit, supplemented by
borefield water.
A preliminary options study quickly identified the Coglan Creek valley in the Termite Range,
around 6 km south-east of the concentrator as the optimum tailings storage site, Figure 2.
More detailed studies were then undertaken into up-valley discharge of unthickened tailings
versus down-valley discharge of thickened tailings.
When unthickened tailings slurry is discharged onto a beach hydraulic sorting occurs' This
produces a coarse sandy beach near the discharge point with finer and finer material being
deposited further and further down the beach. Discharge commonly occurs from an
embanlcnent crest and flows up-valley. A decant and runoff collection pond forms at some
distance down the beach. Slimes settle out from the slurry as it disperses into the pond.
The alternative is for tailings from the plant to go through a thickener before being pumped
out to the dispos al area. This has a number of advantages. Firstly, water and flotation
reagents are conserved by keeping as much as possible in closed circuit at the plant and
avoiding unnecessary evaporation losses. Secondly, since the volumes pumped out to and
returned from the disposal aea ae reduced, there are savings in pipeline and pumping costs.
Thickening the tailings also produces a non-segregating tailings slurry that will beach at a
relatively steep angle. Non-segregation means that the deposited slurry remains as a
homogeneous mixture of particle sizes from the top to the bottom of the beach. Hydraulic
sorting does not occur and slimes areas, which will be of low strength and slow to dry out and
therefore more difficult to rehabilitate, are avoided'
Thickened tailings are much more likely to produce a planar beach profile with little
difference in slope regardless of the distance from the discharge point. This is in contrast to
beaches formed by unthickened tailings which will be relatively steep near the discharge point
and then flatten out progressively further down the beach. The planar beach profile of
thickened tailings can be exploited to advantage by discharging the tailings down-valley onto
flat country until the beach runs out, or nearly runs out. The height of retaining embankrnents
can thus be minimised. A further advantage is that the decant runoff pond can be located
downstream of the stacked tailings, i.e. not on the tailings as would typically be the case for
unthickened tailings discharged up-valley. In most circumstances the pond size can be
minimised which in turn minimises seepage losses. Tailings leaching by the downwards
movement of water from a decant pond above is thus avoided.
The arrangement also ensures the maximum exposure of the beach to evaporative drying.
This results in high tailings density, thus minimising the tailings volume, and high strength'
High strength facilitates access for rehabilitation work. The final surface profile is self-
shedding and erosion resistant and is usually in keeping with the surrounding topography.
The need to stabilise the steep embankment faces of high earth or rockfill retaining structures
is avoided.
Thus, it can be seen that thickening the tailings leads to a number of environmental
advantages. However, the cost of the thickener, both the initial construction cost and the
operating costs in terms of power and flocculant dosage must be met. A cost comparison for
any given site and set of conditions is necessary to establish the optimum approach.
The unthickened tailings density was 18% solids. The owner decided to adopt a relatively
low figure for thickener underflow of 40% solids. Predictions for pit dewatering from
limestones overlying the orebody indicated that there would be surplus water. The Coglan
Creek site provide the opportunity to create a large evaporation pond and it was decided to
pump as much water out with the tailings as possible. No decant return from the evaporation
pond was proposed.
The cost comparison (1995 Australian dollars) between the unthickened up-valley scheme
and the thickened down-valley scheme is presented in Table 1.
It can be seen that the thickened down-valley scheme was the cheaper of the two and this led
to its adoption. The cost of the thickeners was offset by the greater pipeline distance and
pumping volume for the unthickened tailings. Total annual operating costs were also lower
despite the expense of flocculant, agan due to pumping costs. Note that operating costs and
on-going earthworks are much greater numbers overall than initial construction costs,
emphasising the need to focus on this aspect of scheme costing. Rehabilitation costs for the
down-valley scheme were lower due to the ease and speed of access onto the final tailings
surface. Note that in this case the long life of the storage results in the NPV costs of
rehabilitation being relatively insignificant.
The life-of-mine cost in NPV dollars of $0.05/t is very low by industry standards'
The layout of the down-valley scheme is shown on Figure 4, an aeial photograph after about
one year of operation, and on Figure 5, the final surface plan.
There is single point discharge at the head of the creek from the tailings pipeline. The
tailings are retained by a zoned earth and rockfill embankment which currently consists of a
main embankment approximately 1000 m long with a crest elevation of RL 170 m. kl
addition there are four smaller saddle dams. Process and stormwater is decanted from the
tailings storage area through two overflow weir structures. The weir structures consist of a
concrete channel formed from pre-fabricated concrete culverts. The decant level is
maintained and raised by placing concrete stop logs to ensure adequate freeboard for tailings.
The concrete channels discharge into 700 mm diameter polyetlene pipes. In addition,
underdrains are placed at 10 m intervals along the face of the main embankment. The purpose
of these drains is to de-water the tailings at depth and to promote densification and strength
increase. The decants each have a gate anangement which allows the water level in the dam
to be raised during severe storm events to minimize the loss of turbid water. The tailings dam
has an emergency spillway at RL t69 m. The embankment has been constructed full-width at
the base and will be raised centreline.
Decant overflow from the tailings storage and rainfall runoff is impounded behind Dams A, B
and asmall levee. These three dams form the Evaporation Pond. Dam A is about 30 m high,
220 min length and is a conventional earth and rockfill dam. Dam B and the Levee are of
similar construction but of lesser height being only l5 m and 4 m high respectively. The crest
of the various dams is RL 173.1 m and two spillways, each 50 m in width, are set at RL
170.9 m.
The dispos al areais designed as a zeo release scheme and it is expected that the water level
in the evaporation pond will increase with time due to the large bleed water inflows and
runoff from the 29 kn] catchment. Catchment diversion was studied but was not found to be
economically feasible.
The up-valley scheme would have had its embankment at locations A and B and at the Levee
site. They would have needed to be much higher than the water retaining structures.
Figure 6 shows the thickener underflow percent solids versus time for about the first
20 months. It can be seen to have been extremely erratic, especially for the first 5 -6 months
of this period. Although the average settles down to around the 40o/o design figure, the early
low densities and the on-going low-density "spikes" led to a fastet rale of rise at the
embankrnent than was predicted. This is illustrated on Figure 7. Overall beach slope was
only 0.3 - 0.4 %.
This was appreciated as being unsatisfactory, and a campaign was launched to increase the
thickener underflow density. This coincided with lower than expected yields from the pit
dewatering progam so that the need to dispose of excess water through evaporation was
reduced. Figure 8 shows the increase in monthly average underflow density commencing at
the beginning of 2002. The beneficial impact is clearly seen on Figure 9 with very little
increase in the tailings level at the embankment since the campaign began.
Slopes on the beach are shown on Figure 10. They vary from l.}Yo at the head of the beach
to O.5Yo at the toe. The concavity is due to the continuing variability in thickener performance
as evidenced by the daily percent solids plot shown on Figure 11.
A secondary benefit of the increased underflow density can be seen in the evaporation pond
water levels, The bleed water inflow to the pond has reduced by around t'2 Mm3lyt The
water level readings are shown on Figure 12. The rate of pond level fall in the dry season has
increased and after particularly wet years immediately after start-up, the overall trend is now
downwards.
2.1 lntroduction
Williams (2000), presented a generic study of comparative costs between a CTD scheme and
a conventional'addock" operation.
(Note that in the paper, the Australian usage of the term 'addock" applies, i.e. a 4-sided
storage embankment raised upstream).
This study showed that the all-up unit costs of tailings disposal by CTD could be expected to
be less than half of the cost of paddock style storages.
This case study of the Sunrise Dam Gold mine tailings storages provides a concrete example
of these cost benefits.
'W.4. Milling
Sunrise Dam Gold Mine is an AngloGold operation, located near Laverton,
operations commenced in 1997. Mining was initially by open-cut, but may progress to
include underground operations. The site is located in an arid, semi-desert area. Water is a
scarce resource, and much of the processing is done using hypersaline water'
Some of the costs associated with this project were included in outline in Williams et al,
(2003), at Paste 2003. They are presented here in significantly more detail'
The initial tailings storage was a conventional single cell paddock type dam, of 50 ha surface
area (nominal 700 m x 700 m), designed to contain 8 Mt of tailings, at atate of 1 Ma.
The
first raise of this dam was built by downstream construction, rather than the more usual
Year $M
Start-up construction (tee6 - e7) 2.16 (Actual)
TOTAL $s.63 M
1996 NPV @ r0% $4.31M
UNIT NPV COST $0.54ltonne
Notes: 1. Raises 2 and 3 to this storage were not built. The costs estimates are
The first cell provides a realistic check on the unit costs of tailings into a paddock type system
at this site.
By 1998 (the time of the first raise to the paddock system) mining reserves and the milling
rate had increased substantially. A detailed Options Study was undertaken to evaluate future
tailings disposal options. The most likely scenario was for 25 Mt of (remaining) reserves, at a
throughput of 2 Mtpa.
The principal comparison in the Options Study was between a CTD stack, and an extension of
the existing paddock tlpe storages.
The comparative costs of these two options are sum.marised in Table 3'
Table3: SunriseDam
Comparative Costs of 1998 Options Study
OPTION
CTI) Paddock (3 x 50 ha cells)
$M $M
Thickener 0.85
Notes: l. Day to day operating costs, including labour and power, are assumed to be
similar for both options and have been excluded.
2. Rehabilitation costs are based on an allowance of $15,000/ha' There is no
allowance for other rehabilitation works in this estimate.
J. The costing for the paddock system has not been reduced to include the existing
paddock cell. Allowance for this would reduce the paddock costs by around
$3.5 M.
The results of this study were clearly in favour of a CTD scheme, and this was implemented.
The original CTD scheme was designed for a total capacity of 28 Mt. This has since
increased, and the sunmary of costs presented below is based on a capacity of 30 Mt at an
ongoing rate of 3.5 Mtpa.
The construction of this scheme has been staged. The first stage of the CTD scheme was
constructed in 1999, together with installation of a thickener in conjunction with a plant
upgrade.
A small amount of additional work (principally progressive clearing and stripping of topsoil)
was undertaken in 2001. Additional work was completed in2003, and further work proposed
for 2004 will extend the capacity of the storage to the current design limit. The costs of this
scheme are summarised in Table 4. These are actual costs to 2003, with estimated costs
thereafter.
Table4: SunriseDam
Summary of CTD Costs
Notes l. In this case, rehabilitation costs have been increased to arate of $30,000/ha.
It is apparent from this table that the overall civil work costs (excluding rehabilitation) will be
about g4.1 M, spread over a period of 5 years. This is reasonably comparable to the 1998
estimate, which assumed 52.97 M for all civil works at initial construction'
The overall unit cost of disposal has increased by approximately 25o/o for that obtained
in the
Nevertheless, it is clear that the overall expectation that the CTD scheme would provide
cost component of any tailings storage, if a "whole of life" costing is carried out.
On a superficial basis, there is a tendency to say that (because of the greater surface area), the
rehabilitation costs of a CTD scheme must be prohibitively high compared to a paddock
scheme. This is not necessarily the case. In comparing options including very different
schemes (such as a CTD and a paddock scheme), it is important to ensure that realistic costs
For rehabilitation of a CTD stack, the most significant item is the surface cover. The
functions of an adequate caplcover are likely to include:-
- erosion resistance (to wind and water)
- support ofvegetation
- limitation of water/air infiltration
- "capillary break" to limit upward migration of dissolved species.
The costs of a capping layer are clearly site specific, and include considerations of the relative
importance of each of the above functions, the availability and cost of suitable materials, the
trafficability of the tailings surface, and the contractual a:rangements which apply.
- the need for additional fill on the top of paddocks if a "self shedding" profile is
required, and/or
- the need to cut spillways or place additional material around the outer rim to
ensure that overtopping of the embankment crest does not occur during
extreme rainfall events.
If these factors are not properly accounted for, the true rehabilitation cost may be significantly
under estimated. In fact, the costing of the paddock options listed above arc all guilty of this
sin of the omission - the rehabilitation costs for the paddock schemes are all based on the
surface area of caplcover required, without any inclusion of the other aspects. They may
therefore seriously under estimate rehabilitation costs.
3. Gonclusion
The cost estimates for the Century project (Table 1) indicate that the down-valley thickened
discharge scheme adopted for this site provides a remarkably cost-effective solution. The
operating history of the site confirms that to fully realise this benefit, continued attention to
thickener performance, and hence beach slope development, is required. With the progressive
improvement in slurry density the possibility of utilising the increasing beach slope to reduce
the height of future embankment raises is now being investigated.
The Sunrise Dam site provides an opportunity to compare the actual costs of conventional
paddock style storage with those of a CTD scheme. The overall costs of the CTD scheme are
typically less than half of the costs of a paddock style operation. These results confirm the
"generic" cost comparisons presented by V/illiams (2000).
References
Williams, P., Seddon, K. and Murphy S. (2003). "Paste and Thickened Tailings disposal
Delivering the Benefits". Paste and Thickened Tailings, Melbourne, Victoria.
Figure 1: Location
,i
:!-.-i
"-r-:r\ ,'
\5li
\'\
r'\ /il
j :. zo
t
.
U
50%
45%
40%
I
I
ts :s j---
z
U
I
ts I
z I
o 30% !
o I
q
o I
J
zsv" i
r
I
o I
,
20%
I
o
L
15%
10%
5%
o%l
612t1999
170
I
19 .l
i
les l--.-,
^
5t
roz l--
hi
$rosj
rr
oiG4l
Fl DECN] ?
<t
U! -
>i
u1 .62:___-
--- DECANT 1
o'6i i
ol . PREDCTD
E'60i
l
15s I
ii 156
f i.---
at r<:
ts
'., i
I
'' f-..-
I
'54 l----
Di.gg Nov-0
50%
|-
F
ou.
o
J roli
U
o
,
I
fi
o
zo";
10vo
9
o
u
16 .i
,u, I
'162
16'l
160
159
iss
i
157
i
,uu I
I
rs I
1s I'
r!N
@
Priic'
600
I
*"_l
,*[
,OO" ]
;l
{
3
Paste 2004
Cape Town
South Africa
Paper 14
By
ABSTRACT
This paper presents a case study of the EKAPA Mining (Pty) LTD's 15m Deep
Cone Paste Thickener supplied by Dorr-Oliver EIMCO' The plant located in
Kimberly South Africa treats the tails generated from the re-treatment of old
kimberlite tails dumps. The paste thickenef was designed to handle 140tlh (dry)
tails and produce a paste underflow product at a sg of 1.5 which is deposited in
the old De Beers Mine pit. A low moisture paste is required to prevent free water
penetration of the shale zones and the resultant destabilization of the pit. The
clean water from the thickener is retumed to the processing plants. The EKAPA
paste thickener plant started up in Octob et 2002 and has exceeded design
production rate by 55%. This case study will review the paste thickener
operation, and highlight the various modifications to the system that have resulted
in the dramatic increase in production throughput.
HISTORY OF EKAPA
Standing on the rim of the old De Beers Rhodes Pit it is easy to reflect back in
time to the beginning of this famous hole in the ground. This was one of the five
large kimberlite pipes in the Kimberly area. The town was originally named
filed. Mining of the De Beers pit differed from the Kimberly mine because of use
method. At the
conclusion of mining
this pipe in 1990
there remained a pit
600m in diameter by
250m deep. Photo 2
"section through De Beers Pit" shows the cut away view of the formations, the
kimberlite pipe, the open pit and undergtound mining areas.
Photo 2 - Mechanical equipment improved mining methods
Around the upper periphery of the pit were numerous coarse tailings dumps left
from the original processing of the ore. It was discovered that there were still
diamonds in those mountains of course tails that could be salvaged by modern
processing techniques and equipment. A number of small recovery operations
were set up and subcontracted by De Beers, who owns the properties, to process
these old tailings. The requirement was that these three subcontracted companies:
Super Stone Mining (Pty) Ltd., Dumpco Trust, and Sedibeng Mining (Pty) Ltd.
concentrate their tails slurry and recover the water (a very precious commodity
throughout Africa) for recycling. They turned to the newly-established
empowerrnent company EKAPA Mining Ltd. to provide the solution.
The tails from the EKAPA operation would go into the obvious receptacle, the De
Beers pit. That would be a perfect location with ample volume for the tails plus
providing the added advantage of environmental rehabilitation of the pit. There
was one concern however. The pit has a zone of permeable shale that could
weaken the pit walls if it became saturated with water. The management of the
railroad that passes near the upper edge of the pit was also concerned about this
risk. The requirement therefore was for the tails to be as concentrated as possible.
The search began to determine what equipment could produce a paste of the
quality that would be suitable for pit disposal. Bateman Engineering was
contracted to ovetsee the testing and sizing of the equipment for the EKAPA
project. The EKAPA objective was to produce a paste that was as dry as possible
but that would still flow by gravity or by pumping to the pit.
Ekapa test-work began in June - July of 2002. The feed to the proposed unit was
minus 1 mm kimberlite dump slimes, Some of the material turned out to require
both a coagulant and flocculant in order to obtain good solids separation and
overflow clarity. The selected coagulant was Floerger Zetafloc Ll00. The best
flocculant was determined, through test-work, to be Zetafloc 4017. It achieved
the best clarity in the overflow and gave the most efficient settling rate. Testing
found that there was a fairly wide range of solids settling rates, ranging from 20
meterslhr to as high as 70 meters/hr, depending on the ore source. Test work
involved the screening of the flocculants and running a tange of ore types to
establish what the operating parameters would need to be. The main test
equipment used was the Dorr-Oliver EIMCO "T" Floc test unit, which has the
capability to evaluate settling rates, retention time, effects of raking, and to
provide adequate sample to test for viscosity or slump of the material' It was
found that the material exhibited shear thinning characteristics, which would have
an impact on the method of transport of the underflow material to the pit's edge
and deposition.
Based upon the requirements shown in Table 1 a 15m Dorr-Oliver EIMCO Deep
Cone Thickener (DCT) was selected for the desired feed tonnage rate of 140 dry
dilution system would be required to reduce the incoming feed from 31% solids
down to 8o/o solids. This was necessary to reach the correct concentration of the
feed to achieve optimal settling rates at the lowest chemical consumption. The
internal E - DUC@ Feed Dilution System was selected.
During testing it became apparent that some of the old ore bodies were high in
clays and would require the addition of a coagulant to help them settle well. It
was decided to not include a coagulant addition system initially to see if the full
scale system could be adequately handled with only the use of flocculant. The
quantity of this poor settling material existing in the old tails dumps was an
unknown therefore the decision to wait and see concerning the use of coagulants
was made.
Typical
Total Production Rate, (Vhr) 140
FEED:
Feed pulp solids, (wt%) 31
UNDER FLOW:
Underflow solids, (wt%) 55
OVERFLOW:
Overflow flowrate, (m3/hr) 202
INSTALLATION
The EKAPA order was placed on June 18,2002 and the erection commenced in
August 2002. The unit was completed and commissioned in October 2002,
slightly ahead of the (5)15m Dorr-Oliver EIMCO Deep Cone Thickener units
down the road at the DeBeers CTP project. The 15m EKAPA unit was field
erected on a concrete foundation. It has a scalloped tank bottom which may be
clearly observed in Photo 4. The finished unit stands 16.2m high and has a
nominal retention time of 3 to 5 hours at design flow conditions. The tank is
raked by an EIMCO B60B3P drive rake assembly. The controls originally
consisted of: drive torque, tank level, nuclear density meters on the feed &
underflow, flow meter on the feed, and a turbidity meter on the clear water
overflow plus a nuclear density meter on the circulating pump loop.
The thickener is instrumented to allow the unit to perform in a safe and stable
manner. The minimum number of controls is offered to facilitate this. Feed from
the three processing plants is collected in the slimes collecting tank from where it
is pumped by means of a f,rxed speed pump to the thickener. This was the reason
for the tall, narrow feed collection tank. Incoming feed to the Deep Cone
Thickener from the feed collection tank is flocculated in proportion to the flow of
dry solids. The percent of dry solids is calculated by the Process Logic Controller
(PLC) using the feed density and the fixed feed pump curve flow rate. It is then
manually entered into the control system by the operator. The pumped feed is
automatically diluted to a suitable density by the E - DUC@ feed dilution system
which maximizes the flocculation efficiency by correct solids dilution and mixing
at approximately one part feed slime and two parts return water. The raked and
settled solids report to the bottom of the thickener to a special design-patented
hopper, which allows for shear conditioning of the underflow to occur for
improved pump flow conditions. This system was designed such that the
underflow material could be recycled back to the feed or sent to the pit depending
on its density and concentration. The discharge line from the hopper contains a
density gauge that is used to determine the density of the slurry. If it gets too
dense, it is diluted and if it gets too thin orbelow the 1.5 sg target, it is recycled
until the proper specific gravity is achieved. Once the correct parameters are
achieved by using the loop the operator actuates the discharge to the pit by
opening the appropriate valve.
The corect slurry density is primarily determined by the residence time in the
thickener. This is accomplished by signalling the raked bed depth by use of two
vibrating "Viga Probes" that indicate the lower or "Normal Operating level" and a
higher probe that actuates an alarm when the bed level reaches that point and is
about to go too high for good operation. The solids bed level is manipulated by
the use of the variable speed underflow pump that draws down the unit. The
pump is controlled by the underflow density meter in conjunction with the level
probes.
instructions which are managed through the PLC to control the thickener.
Photo 5 - Fully erected EKAPA 15m DCT
START UP EXPERIENCE
The start up of the 15m Deep Cone Thickener went fairly smoothly, with paste
being made on the first day. There have been a number of modifications and
changes to the original installation to improve the ease of operation - these will be
discussed under the next section titled 'Modifications'. The Deep Cone
Thickener has operated well and is capable of meeting the designed performance.
The data for the first six months has been very impressive with better than
expected results. The result from the last six months is equally good and is
presented in Table 3 'Actual EKAPA Production September 2003 to January
2OO4'. The underflow solids going to the pit are above the target of 1.5 averaging
Gombined Feed
Characteristics:
pH 8to10
AVE.18h/d (Estimated average with down time & mine shift variances)
42,362
Total 176.5d 484,630 m3 747,926 t 425,369 t kg 19,900 kg
Total 3177h
Average 152.5 m3lhr 235.4 Uhr 133.9 Vhr 99.6 g/t 46.8 g/t
Ave. sg 1.54
Table 4 - EKAPA's peak Deep Cone Thickener paste production & pumping rates
MODIFICATIONS
The initial results were exciting and the project was certainly a success. It was
determined over the next few months of operation that there were some modifications
that could be made in an effort to build and improve on the initial success of the plant's
operation. These modifications were for the most part undertaken by EKAPA based
upon their experience with the equipment. The following is a listing of the
modif,rcations:
o Flocculant Dilution Mixing Tanks added (1 part floc / 2 parts return water)
o Paste Gravity Vent Pipe discontinued
o Paste Pipeline was repositioned for best downhill gravity pumping advantage
The result of all of these changes has made the operation of the plant much smoother and
nearly operator free. The results for the last six months show the benefits of the overall
operation of the Dorr-Oliver EIMCO Deep Cone Thickener and its ability to consistently
produce qualitypaste underflow forpit deposition at a sg of 1.5 orhigher. The average
throughput of 120 to150 dry tlhr has been a positive benefit to EKAPA and has been able
to support the three process plant's variable production rates.
The decision to install a 15m Diameter Don-Oliver EIMCO Deep Cone Thickener to
dewater the EKAPA reprocessed tailings and deposit them into the old DeBeers Pit has
been a successful venture. This installation has helped to make diamond recovery from
the old dumps an economic success. The generation of tailings paste by the Deep Cone
Thickener has also made the pit reclamation feasible and practical. The plant has been
able to meet the design conditions of producing a 1.5 lm3 specific gravity discharge into
the pit at a production rate of 140 t/hr. The plant started up with minimal problems. A
number of modifications to streamline and simplify the process were made resulting in
improved operation of the plant. The plant has demonstrated the ability to produce paste
for disposal at rates up to 50+% over-design. Should the occasion arise in the future,
EKAPA would not hesitate a moment to install another Dorr-Oliver EIMCO Deep Cone
Thickener.
REFERX,NCES
Jewell, R.J., Fourie, 4.B., and Lord, E.R., "Paste and Thickened Tailings - A Guide",
2002 Australian Center for Geomechanics, Nedlands, Western Australia, Australia
Wheatcroft, Geoffrey, "The Randlords", 1986 Athenum, New York, NY, USA
International Seminar on Paste and Thickened Tailings
Paste 2004
Cape Town
South Africa
Paper I 5
G L de Swardt, J Houman
ru4!!vr
POLICY
AND
COMMITMENT o\
r
r
u
t>
IMPLEMENTATION
l=
lo
CONTINUOUS AND l=
OPERATION l>
t:
flt
MANAGEMENT REVIEW ,lr i
FOR CONTINUAL >T
IMPROVEMENT
- ------
7r4J I
tH l.a uvttvll-
The procedures for identification, maintenance and disposition of records should focus on those records needed
for the implementation and operation of the management system, as well as those records needed to record the
extent to which planned objectives and targets have been met.
The intent of the document control procedures is to ensure that organizations create and maintain documents in
a manner sufficient to implement the management system. However, the primary focus should be on the
effective implementation of the management system and on peormance, and not on a complex documentation
control system.
6
Procedures
Carefully written procedures are the basis for implementing the management system. Procedures should
specify *no is to carry out what tasks, give step-by-step instructions for how tasks are to be accomplished, and
include directions for dealing with departures from the procedures'
The emergency plan includes:
Paste 2004
Cape Town
South Africa
PaPer 16
Bernard Krause
Bernard Krause
Ticor South Africa (Pty) Ltd, South Africa
Abstract
After the successful commissioning of a similar plant 3 years ago, Ticor South Afca is
developing a process for the beneficiation of a second heavy mineral sand deposit on the
North coast of Kwazulu-Natal.
This deposit is again characterised by a high slimes content (> 30% -45rm material) with a
tendency to be shear sensitive - i.e. the rheological properties of a slurry containing slime
changes with shear rate and time.
Careful consideration needs to be given to this properly during the design of a processing
plant, since a number of physical manifestations have to be quantified and managed to
enable successful deposition of the ultra fine residue.
Lessons learnt at the current operation taught caution at assuming slurry rheological
properties and plant efficiencies.
This paper highlights the preparation and development of a process to enable the careful
*"urrement and harnessing f a variable slurry rheology to support a particular means of
residue disposal.
1. Introduction
A number of heavy mineral deposits along the North Coast of Kwazulu-Natal, South Africa
are characterised by a high in situ slime content (> 20o/o -45pm material).
Historically these deposits were labelled as being too difficult to process. Ticor South Africa
developed u pro..riing philosophy whereby the rheological properties of a high slimes
deposii are hrnessed to facilitate certain processes and carefully managed elsewhere in
order not to reduce plant perfofinance.
Figure 1 provides a graphical description of the concept flowsheet considered for treatment
of the orebody in question.
paper: Process Design Considerations for the Transformation of Shear Sensitive Slurry to Paste Page2
Author: B Krause
Hydraulic
monitoring
Sand Tailings
Mgn*tte
l-ieavy fVlinar
Concentrate
guns'
The dry ore at Ticor South Africa is mined hydraulically with high pressure monitor
1'he high viscosities associated with high slime bearing slurries enable transportation
of the
heavy inerals in trenches utilizing gravity and the elevation of the dune deposit.
The wet processing plant following, utilizes mainly gavity separation and magnetic
separationo r".orrrry lh. h"ur.y minerals from its lighter and more magnetic counterparts'
Above a certain slurry slimes content, the actual presence of ultra fine solids alters the
viscosity of the carrying medium (water) and ultimately reduces efficiency of the processing
plant. urther design onsiderations are required during pump and pipeline design as the
changing viscosity influences slurry flow properties too'
from
Hydro cyclones are used for size separation i.e. removing the ultra fine solids fraction
the feed to the processing Plant.
The slimes fraction (desliming cyclone overflow) is dewatered with appropriate thickener
produce a
technology. Careful ontrol of ttr solid-liquid separation efficiency is required to
thickener underflow with specific rheological properties required to support a specific
method of fines disposal.
Except where gravity can be used, wet slurry transportation is effected with centrifugal
pumps.
Author: B Krause
All of the above processes i.e. hydraulic mining, size separation, pumping and gravity
transportation contributes to various levels of shear and subsequent degradation of the ultra
fine solids size fraction.
Physically this leads to an increased ultra fine size fraction with subsequent greater surface
area, but on a more fundamental level this creates newly exposed mineral surfaces, which
may lead to changing surface chemistry and changing electrochemical properties as well as a
change in particle form i.e. spherical to plate-like.
The above properties are important considerations in the selection of solid-liquid separation
technology and supporting dewatering chemistry i.e. flocculants and coagulants used to
provide a dewatered slurry/paste for residue disposal.
2. Discussion
In order to quantiff the effects of shear on slurry properties and the subsequent management
of those effects, is it necessary to define the physical changes occurring within a slurry and
the means to measure those changes.
Slurry shear is defined as the energylagitation imparted to a slurry through the forces
prevailing in any of the processes in a minerals treatment plant, namely:
Shear sensitive is defined as having the potential to undergo a change in properties as a result
of exposure to shear. Of specific relevance are changes in rheological properties i'e' viscosity
A yftt stress and particle size distribution as a result of clay particle liberation and
breakdown.
2.1
2.1.1 Sizedistribution
Indications are that the slimes content increases with up Io 50o/o at the point in the process
where the slime is separated from the coarse sand for dewatering treatment. Of significance
is the fact that the -2tm content effectively doubles from pit to plant. Table I summarizes
the effect of process shear on particle size distribution.
Screen size ROM in situ ROM after minins ROM after feed prep
Fm %o Passins %o Passing %o Passins
425 90 90 90
355 81 8l 8l
250 53 58 59
2t2 42 48 50
150 28 34 36
106 23 28 32
75 22 27 31
45 20 25 30
10 t2 15 20
2 4 5 8
The data was obtained during pilot plant test work for thickener selection with a realistic
simulation of expected process & shear described elsewhere in this document.
Figure 2 provides a graphical description of size distribution change as the slurry progresses
along the process. The ultra fines solid content increases from an in-situ distribution with
20o/o free slime (bottom curve - "ROM in situ") to the size distribution at the point in the
process with 30% free slime where slimes removal for dewatering is about to occur (top
curve "afte feedprep" ).
100
90
80
70
Poo
c50
-" 40
30
20
l0
o
5o loo l5o 200 250 300 350 400
Scrn 5lze (Pm)
In effect the solids surface area virtually doubles as the number of -10rm particles
contribute to around 90% of the available surface area.
For the solid-liquid separation process this effect is even more pronounced as the number of
-1Qpm particles contribute to more thanggYo of the available surface area. This is significant
since by implication the flocculant consumption (surfactant used for dewatering) virtually
doubles on account of the shear imparted on the slimes fraction before presentation to the
thickener.
This degredation of fines does not stop, but carries on as processing continues. However, the
effect of shear degradation is less pronounced in dilute slurries'
Figures 3 and 4 show micrographs of the particular slimes (-45rm particles) and sand
samples (+45pm particles). The slime particles are angular with a large surface area and the
potential for degradation is clearly visible.
2.1.2 Rheology
,/udt22/
It is well known that the fraction of fines below 20Y/l controls rheology of thickened slurry or
paste. It therefore feasible to comment that rheology should increase as more fines are
liberated with shear.
The shear sensitive nature of the processing slurry is seen in the results of slurry pipe loop
tests. A typical test involves recordingthehoizontal pipeline pressure gradient over a range
of flow rates.
Pipe loop tests conducted by Paterson & Cooke indicated that the slurry characteristics were
chnging with time as the slurry was circulated through the pipe loop. The results obtained
are presented here in Figure 5 in the form of a pseudo shear diagram.
Author: B Krause
tr tr
trtr E tr
l,
a t' '
'
8o
^aa
ti
::"-'x
830
xxX
E
o
20
3
,ii"rJi-*"'o',
CUnSeaEd xshredfor2ominutes Xshredftr4o mnut
The data clearly shows a significant increase in slurry viscosity with increased residence
time in the pip loop indicating that the rheology of the slurry is time dependent and in all
probability a function of the liberation rate of ultra fine particles.
Tests conducted elsewhere also confirmed that the slurry shows definite time independent
shear thinning behavior and behaves as a yield pseudo-plastic fluid i.e. the apparent viscosity
decreases with increasing shear rate.
350
'.
lti
O
300 I
I :"'".:^:'J;^'^ a
^
E2
lia
..J;^"
P
^^^
U)
#,*
3
100
50
Figure 6: Process Viscometer test results for centrifugal pump vs peristaltic pump
(Courtesy Paterson & Cooke)
2.1.3 Mineralogy
The naturally occurring fines fraction of the orebody in question contains various minerals.
Especially dominant are quarts, kaolin & smectite and in lesser quantities are hematite,
goethite and mica.
Some of these minerals may contribute to slurry rheology, but results to date have been
inconclusive. The extent of mineral surface exposure with particle degradation and the
impact on rheology need to be quantified with further studies.
There are other minerals however, that has a pronounced influence on slurry rheology.
In the current operation gypsum is added to the ultra fine residue in the thickener underflow
collection sump as a passenger en rout to the fine residue disposal area. Deposition of the
gypsum at the residue disposal facility is a NNR requirement in order to dilute the radio-
active nature of the gypsum particles. The effect of glpsum on slurry rheology is shown in
figwe 7.
It is clear that the residue with gypsum has a signif,rcantly lower rheology than the residue
without Gypsum.
150
i = 1.21 t/m"
135
c (
...a
Thcke@UnderfldOnly ____.___.^c
-;:- L. "
120
,. , i :-
i!,psumAdded t
^ 105 -,c{"----r------Elii
ir., -u D
I
9 #ntr" :
I i
i
#ru irl
t i0-
$ r!
i6
o
s '
rt
30 id
i
15
i
2.2
The side effects of a varying rheology and it origins manifests physically right through
mining and primary beneficiation influencing slurry transportation, surge capacity, size
separation efficiency, mineral separation efficiency and solid-liquid separation effciency
and is especially prevalent at the residue disposal facility i.e. thickened slurry transport and
deposition on the tailings impoundment.
Surge capacity is supplied between the mining and beneficiation. Physically this means that
alarge tank or sump absorbs the differences in the slurry supply and demand rate between
the mine and the wet processing plant.
Surge capacity is increased under normal operating circumstances by settling of the heavier
sand particles with displacement of lighter sand and slime particles above and finally clear
water on top of the surge tank, causing in effect a densiffing of the feed slurry. A steep
density gradient is therefore expected along the vertical axis from tank bottom to top.
What transpired during commissioning of the current plant was that virtually no settling
occurred throughout the vertical dimension of the surge tank at the prevailing ROM feed
density and residence time. Figure 8 provides in indication of the actual density gradient
present in the surge tank.
11 12 l 1.4 1,5 16 17 l8
F.h. d..[Y
It is exactly this property that enables hydraulic mining and gravity trench transport of a
heavy mineral sand deposit used at the current mining operation and also to be used in the
next operation. Heavy mineral containing sand are transported along unlined trenches for a
number of kilometres from the point of mining to the wet processing plant without
deposition of the heavy mineral sand particles. Of course this is only possible because of the
nature of a dune deposit with the plant basically positioned at the lowest point of the deposit.
Size separation is effected by the varying resistance of particles to motion in a fluid based on
size and density i.e. Stokes law (note the viscosity term r1 below the line):
d2. g (o,-o)
ff
/ (settting velocity) -
ISrt
Goarse Underflow
From figure 10 it is clear that in excess of 40%o sand was carried over to the thickener and
more thn 20o/o slime was forwarded along with the coarse sand. Also the actual slimes cut
reporting to the degritting box contained in the order of 50 to 60% sand.
Due to the increased viscosity, size separation eff,rciency was so low that alternative means
of size separation had to be implemented i.e. hydrocyclones at the cost of increased shear
and particle degradation.
Heavy mineral separation is accomplished with various stages of spiral separators to effect
recovery and grde control. Separation efficiency is especially poor on the circuits
..rponribl. for recovery due to high prevailing slimes content. Figure 11 describes the
r6nltr of test work to determine the effect of shear on spiral performance. 4 Test samples
(similar slimes content f20%) and a 5th deslimed sample were subjected to various levels of
shear with 100% shear representing full-scale plant shear based on a shear model developed
by Paterson & Cooke.
MASS vs RECOVERY
1@
80
.70
fr
o
'
20
10
0
01020s40s7080s1
*TASS TO CONCENTMTE
Of significance is the fact that the sample with high slimes content and nil shear provides
similar separation efficiency to the deslimed sample. The samples subjected to various
levels of shear provides varying levels of poor separation efficiency indicating that there
may be more than just fines liberation contributing to rheology.
Figure 12 describes the relationship between Total Heavy Mineral Recovery $HM) and
mass pull from actual plant data (100%o shear). It is clear that a mass pull in excess of 90%o
is required for the same heavy mineral recovery compared to a 45%o mass pull on spiral feed
with little or no slime.
- 7
*
o
t/ 4
l,r I -ffi.rd1-2.*fu
//v.,'
///
r/l
dssMr
As indicated on the graph, double stage desliming is required to reduce slime levels to spiral
feed for efficient mineral separation at the cost of increased shear for further fine solids
processing.
Figure 13 describes the relationship between heavy mineral recovery and slime content as a
function of total dry solids.
Author: B Krause
l0 12 14 16 18 20 22 24 26 28 30
A more suitable relationship is the one between heavy mineral recovery and slime content as
a function of water, since an increased rheology can be countered to some extent by an
increased slurry dilution. Inside the current design density envelope however, the slime
content has to be contained below 6%by solids weight for effective mineral recovery.
Solid-liquid separation efficiency decreases with increased liberation of fines during shear
due to the creation of additional surface area. Rheology plays a lesser role during
dewatering on account of the very dilute condition of entry into the thickener, but becomes
very important again as a dewatered slurry converted to paste is required to be transported
large distances for deposition.
With increased shear and a subsequent greater surface area due to particle breakdown,
dewatering becomes more difficult and increased flocculant administering is required to
facilitate agglomeration of the ultra fine particles and subsequent settling.
A small component of coarser sand particles (+45pm) are required to assist with nucleation
of agglomerates to facilitate settling and dewatering. The specific minimum number needs
to be established.
As can be seen, due to the increased fines fraction, the settling rate of stream (la) is 40%
slower to that of stream 1 at the same flocculant dosage rate. This means that stream (1a)
may require up to 40Yo more flocculant for a similar settling rate, which illustrates the
significance of shear on settling characteristics with a single pump and hydro cyclone only.
0
\2040m1lm1Qlm1
-m
40
\
s
E
gr*
!
I 42o
-r40
160
lm
.N
Tlm(sd&)
@
Figure 14: Settling characteristics of various process streams with different shear rates
For conventional thickeners with its relatively flat bottoms, lack of natural movement of the
thickened mud due to increased viscosity necessitates a raking mechanism to transport
thickened mud from the remotest corners of the thickener to the usually centrally positioned
discharge facility.
Rake design is crucial in that just enough material needs to be moved to the discharge
facility to match the displacement rate of the underflow pumps.
This displacement rate is a function of rake geometry, rotational speed and thickener
underflow density.
In an oversupply situation the material would be churned like butter with new material
constantly being moved towards the exit, but not all of it actually finding it's way out. So
material would be raked and re-raked with some serious consequences to mud bed stability.
In an undersupply situation the pump would be sucking away at whatever material provides
least resistance if the rakes are unable to supply enough feed material. The net effect of this
is short-circuiting of more dilute material exactly above the suction of the pumps or ratholing
as it is commonly known.
The ultra high rate thickeners in operation currently, pictured in figure 15, were designed
with a bottom cone angle at 60o with supposedly enough motivation for natural migration of
settled mud to the central discharge.
During pilot plant trials in the days before existence of the current Primary Wet Plant,
chronic rat-hling and resultant slumping in the pilot thickener led to the installation of
scraper blades along the sides of the 60o cone as a preventative measure, still relying on
gravity to assist.
In the actual current scaled up operation the rheology of the thickened mud on the thickener
cone bottom causes resistance to downward flow and the installed scraper does not enough
to facilitate transport of mud to the central discharge. Resultant rat-holing causes erratic
discharge and low densities/yield stresses to the residue dam'
A density meter installed on each of the thickener underflow lines controls the respective
variable speed thickener underflow pump speeds, to keep the slimes density/rheology
constant.
As the density increases the underflow pump displacement rate increases to counter the
associated mud build-up leading to the initial increase in underflow density/rheology.
The rate of change of the underflow pumps and maximum displacement rate are both
variables able to influence thickener mud bed dynamics and underflow density/rheology.
The higher the mud density, the higher the viscosity would be and the more reluctant the
flow of material would be. In the event of the pump rate exceeding the densified fluid flow
rate along the surface of the 60" cone bottom towards the central discharge, it can only be
assumed ihat preferential suction of the more dilute material immediately above the suction
would occur or ratholing.
The residue disposal facility and the rheological properties of the material to be disposed of,
is the driving force behind the thickener underflow density specification.
Although thickener underflow properties are specified in terms of density mostly, slurry
viscosity and more importantly yield stress is the real driving force for residue dam
optimization and control. Density contributes mostly to rheology and is therefore a simple
parameter suitable to measure and control.
For the current operation it was established that the rheological requirements to the residue
dam is higher than the specified rheology values associated with the design density range
and higher too than the actual attainable density range.
Lower than expected viscosity of the thickener underflow result in low beach angles on the
residue disposal facility, which have the following negative effects:
o Slow bleed and runoff of bleed water to decant pumps, which means longer time
before evaporation can commence from the residue itself.
o Decanting of rainfall takes longer and even light rain causes long delays to
evaporative drying.
o The pool has a much wider influence (pool is very negative with respect to drying) as
the same volume of water results in larger area with flatter slopes.
The reduction in the drying performance of the residue results in lower in situ density and
hence less tonnage can be stored in the available volume.
The residue facility also requires a consistent product (in terms of viscosity). During periods
of low viscosity the material will not deposit on the beach in the required layer thickness.
Different areas of the residue dam also require different viscosities for optimum
performance. History to date has shown that there is a fair amount of variability in the
viscosity from the thickeners, which cannot be sufficiently controlled.
Regardless of the impact on rheology, the coarse sand component (+45rm material) to the
residue dam has to be minimised (<10% design) as the residue storage facility has limited
storage capacity and to date there is no alternative to residue disposal.
2.2.6 Slurrytransportation
In test work conducted on site where material of the new deposit was transported to and
processed with the current plant, the effects of process shear on higher density thickener
underflow was measured and quantified in a study by Paterson & Cooke.
The slurry shows definite time independent shear thinning behaviour and behaves as a yield
pseudo-plastic fluid.
The slurry rheology generally increases between thickener underflow pump suction and
discharge. This is attributed to physical changes in slurry properties when pumping and
during pipe flow.
Due to the time dependent properties of the slurry at a specific shear rate i.e. increased fines
liberation, rheology increases with every pumping stage. Also, the centrifugal pumps utllize
gland water for gland protection and rheology decreases at every centrifugal pump due to
dilution.
Figure 16 provides an indication of the change in particle size from the pump suction to
discharge as measured with Malvern particle size analyser.
100
90
80
e70
oo
e
uo
&o
30
Ezo
10
0
1 10 100 1000
**'"
--r tutrp suction-[,ra"." I"" lfl Discrprsetitvsn
.+ RJrp Sucton.-ydrter Rrp tlschargellydrmter
--
The ultra fine fraction of the size distribution changes with pump shear. It is possible that
the Malvern particle size analyser cannot distinguish between agglomerates and actual
particles, but as is clear from Figure 17 this change in size distribution has definite impact on
rheology i.e. yield stress with the finer size distribution providing the gteater yield stress.
Mass fraction of solds, Yo
-
r|1
.
!oo LI
6
A
A
*40
t A
b
15%
Vdqe Fractlon of Solids, C, (%)
Figure 17: Yield stress vs solids concentration by volume for suction & discharge of
-Smrodel a
, pump (CourtesyfPaterson & Cooke)
cenmrugil
For a residue disposal facility with specific yield stress requirements, the relationship
depicted is very important as far as management of the effects of shear and dilution is
concerned and subsequent design of the transportation system.
Figure 18 provides a more detailed description of the effects of dilution on slurry rheology.
The combined effect of dilution due to gland service water and gypsem could result in a
residue with a significantly lower rheology than the material produced at the thickener
underflow. The graph depicts actal measured parameters converted to a pseudo-shear
relation with a Paterson & Cooke developed process viscometer. An approximation of yield
stress has to be made based on an extrapolation of the test data due to the inability of the
instrument to measure yield stresses at very low shear rates. But the resultant rheogram
provides a good indication of expected trends with regards to yield stress.
160
8'*
'im
H.*
-P
Bo
$oo
40
o1.166 t/rl +117A Um' 1.193 U.rP 31210 tlttf tr'l 219 Um" x1.182 lmt
90% oyp6m
-Wh
80%
70%
g2 Ch.nge ol1.37. 3dkt3 @n@nbtld atl
g 60% gyspum additld cdld potdally redu
rlogy by.ppxlmately $',( ,
**
n
**
n*
/"
l0%
o%
Figure 19: Potential decrease_in rheology dge^tg dilution and glpsum addition
Courtesy ofPaterson & Cooke
Any increase in shear due to pumping is neutralized by the effect of slurry dilution and
gypsum on rheology.
The downside of course is that if those dilution constraints were not quantified properly
before hand, the residue disposal facility is bound to receive a disappointing rheology with
resultant endless operating problems where a greater rheology is required for optimum
deposition and drying performance.
Gp*m addition
o Due to the nature of a high slime sandy ore deposit, hydraulic mining is possible and
also desirable, since this is probably the second most cost efficient mining method
known today.
o Due to the high viscosity associated with slurries containing high slime and density,
surge capacity is reduced substantially, since no particle segregation i.e. densiffing
occurs and as such has to be increased by other means e.g. increased density or
bigger surge tanks.
o In order to maximise separation efficiencies on the beneficiation circuits, desliming
has to be done prior to beneficiation'
o In order to supply the thickeners with antltta fine solids component for dewatering,
effrcient size separation has to be maintained.
o Due to the shear sensitive nature of this slurry, particle degradation and subsequent
increase in the ultra fine solids size fraction with associated increased surface area
occurs.
o Particle settling and slurry dewatering efflrciency is adversely affected by slurry
shear.
o In order to minimise flocculant consumption, surface area has to be minimised.
o In order to maximise solid-liquid separation efficiency, slurry shear has to be
minimised.
o Due to the shear sensitive nature of this slurry, rheology changes dramatically.
o In order to maintain efficient residue dam operation, the coarse solids component
(+45rm particles) has to minimised (not allowed to exceed l0%by weight)
o In order to optimise residue dam operation, thickened slurry rheological properties
has to be maintained within a strict operating envelope.
o In order to maintain residue dam feed slurry rheology within a strict envelope, slurry
shear has to be quantified, measure and controlled.
o In order to maintain residue dam feed slurry rheology within a strict envelope,
thickener operation and mud discharge needs to be stable.
Although a big incentive exists to minimise slurry shear, it is clear that some processes will
induce shear, and therefore the onus is on the design engineer to:
. minimise shear
o quantiff this shear in terms of known properties and variables
. accommodate the side effects with proper design and a sound knowledge base i.e. no
surprises.
Apart from the profitability of the ore body in terms of the in situ heavy mineral content,
fines disposal or the ability to dispose of fines in an environmentally acceptable manner is
the big driver determining the feasibility of a mining venture'
The work described above is especially important for the design of the dewatering facility
and the dewatered slurry transport system in order to provide the residue storage facility with
the proper rheology.
The following process is followed to enable the design of a suitable solid-liquid separation
facility:
Ore characterisation includes subdivision of an ore body into distinct geographical areas
(zones) with similar mineralogical properties to enable metallurgical testing and subsequent
performance prediction.
The basis of selection of actual drill samples for ore testing and subsequent process design, is
the geological exploration data defining the ore body. Samples are drilled with narrow
diameter air core drilling and liberated in 3 meter intervals. A mineral analyses is made
available for every 3 meter interval and the sample property requirements are accommodated
by a weighted combination of samples.
Deposits are normally characterised in terms of the mineralogical constituency of the +45pm
particle fraction and the weighted distribution of +45rm and -45rm material (slime).
Further geographical classification (zones) is done at the hand of the Silica content of the
Carpco magnetic Ilmenite fraction between settings + 0.84 2.4A, important for heavy
-
mineral separation efficiency assessment. The slimes fraction of those distinct combinations
is the primary source of test substrate as this best describes the expected mineralogy
following resource utilization planning closely. Figure 2l provides a graphical description
of an ore body, with such zones superimposed in colour.
For the sake of the polymer screening exercise, a summarised mineralogical description of
the individual ore bodies is added and specified in terms of the minimum, maximum and
average assayed mineralogy values and slime content, also based on the mineral resource
utilization planning.
The mineralogical values are described in terms of the Total Heavy Mineral content (result
of a normal sink/float separation) and the ratio of Ilmenite to THM.
Finally, each zone is subdivided into the various present lithologies (Figure 22).
The individual samples totalled form the basis substrate on which polymer screening is
conducted.
In order to incorporate the preferred mining method and minimise shear, it is highly
desirable to minimise the number of pumping stage before presentation to the thickener and
to remove the ultra fine slimes fraction as early in the process as possible. Refer to figure 1
for a graphical description ofthe process.
In a study commissioned by Ticor South Africa and completed by Paterson & Cooke
(described elsewhere), it was established that it is possible to quantiff process shear in terms
of process parameters and to use this information for design purposes.
The basis of the shear model used to quantiff process shear is the hypothesis that the energy
dissipated per unit volume at any point in the process is proportional tot the process shear
experienced by the slurry.
The energy dissipated in the slurry through shear is calculated from the energy input, for
example pump power absorbed, and a utilization or efficiency term, since a certain amount
of energy will be dissipated through other means. The fraction of total energy input
dissipated by shear is expected to vary from one process to the next.
The modelling of the slurry transport and gravity separation process involves accounting for
each stage of the process where energy is input into the slurry. Energy input per unit volume
can be derived from the absorbed motor powers, slurry flow rate and the efficiency term'
This gives a value for the total energy dissipated per unit volume, which is considered to
contribute to the liberation of clay particles from the ore.
Calibration of the model is achieved by adjusting the efficiency terms for the various stages
of energy input. On actual process simulation scale calibration of the test facility is achieved
by adjusting the variable plant parameters. In both cased adjustment is made such that the
test thickener feed slurry has the same characteristics as the thickener feed of the current
operation, with a common parent sample as basis. The extent of shearing was determined by
cmparing the fines fraction of the thickener feeds using size distribution analyses & settling
tests i.e. dose response curves.
Process parameters that can be manipulated to achieve desired process shear within a mixing
tank are:
o Retention time
o Slurry volume
. Slurry solids content
Factors contributing to shear, but remain fixed according to process design are:
Due to the lack of fundamental knowledge of the changes in particle surface chemistry and
electro-chemistry due to shear, the tried and tested method of thorough empirical testing is
followed to identiff suitable, robust polymers. For valid polymer screening with realistic
performance results, properly modeled slurry shear is required, but on smaller bench top
scale since stock sample size is maintained below 5 liters.
Paterson &Cooke devised a method to simulate and validate sample shear as discussed
previously with appropriate apparatus and procedures. In this case process shear is
simulated with a single agitation unit at a constant agitation intensity and thus power input.
A constant process related slurry solids content is maintained with retention time being the
controlling variable. Figure 23 provides a graphical description of the shear vessel used for
bench top scale shear simulation.
The slurry sample is manually screened at 45tm after process shear and I0o/o +45rm grit is
added to the 45pm fraction by weight to simulate process conditions and for the sake of
repeatability. After careful, quantified shear the sample is considered ready for test work.
Figure 24 & 25: Flocculant screening with cylinder tests & dynamic benchtop thickener
o Polymer consumption
. Agglomerate settlingrate
o A parameter best described as robustness i.e. the ability to sustain settling within
dynamic thickener conditions
o Dewatering capacity and rheology.
?0
9@
:4
9
!
t
0
Figure 26: Polymer ranking by thickener underflow yield stress (actual data)
Selection cotmences with bench top thickener comparative test work, but ultimately
culminates in pilot plant comparutive test work.
As discussed previously, a proper simulation of slurry shear is required for realistic thickener
performance and dewatering capability assessment as a good representation of equipment
and process.
Figure 27 provides a graphical description of a pilot plant used for preparation of a feed
suitable to simulate thickener feed. Figure 28 provides a graphical description of a process
model used to calculate shear, hydration and retention time.
Figure 28: Schematic drawing of pilot plant used for shear simulation
During the sample preparation phase of the selection process an attempt is made to prepare
samples with as much ore variability as can be simulated, with the mine plan and geological
dataas basis and the test budget as constraint.
500
,l.5()
400
3s0
300
250
! 200
Hgh Ete
150
't00
uvnMat
50 '';F
0
1 r16 I 18 1.2 124 126 1.28
Thlck.ns underflow SG
A pre-sheared yield stress value is quoted indicating that it is measured at the thickener
discharge prior to pumping. With consent of the various role players, the yield stress is
measured and reported at a shear rate of 0.1 per second to correlate with the low,
determining flow rates expected at the residue discharge facility.
Thickener solids loading (tons per hour per square meter of thickener surface area)
determined with bench top testing is probably the more contentious parameter and is
associated with a greatt design risk. In order to reduce the cost of an installation, it is fairly
easy to increase the solids loading. Figure 30 provides a graphical description of a possible
relationship between solids loading, thickener size and associated cost for a certain
throughput rate.
R 11,000,0
10.m,0
37
a\ R
E35 R 9,m,000
!33 e s,m,o E
n z,o.ooo i
r\ R 6.m0.0
5
=
27+ -\ R 5.m,0
,.L R 4.0,m0
o.m 030 035 010 0m
Solld.lodlng (h/m2l
-tut
-
-stz6
Figure 30: Relationship between solids loading, thickener size and cost
Other variables however play a significant role in thickener selection and sizing with direct
influence on flocculation efficiency, settling performance and thickener underflow
density/rheology. This include natural emergencies i.e. winter temperatures and rain, slime
overload, ineffrciencies and malpractice for which a reduced solids loading is a requirement.
Therefore a safety factor has to be incorporated in the design.
A final criterion is possibly the scale up confidence associated with any fype of thickener
technology from pilot plant scale to proper sized thickeners. Past experience has shown that
dimensional similarity and associated surface effects are extremely difficult to scale up from.
This is probably more relevant to installations dealing with paste production and associated
higher slurry viscosities.
Even with steep cone angles associated with ultra high rate thickener and paste thickeners,
internal mud transport remains a challenge and more research has to be done on proper rake
design for these applications. Typical symptoms of some current installations are ratholing
and slumping contributing to reduced thickener underflow densities /rheology. These effects
are probably more pronounced for ultra high rate thickeners, due to shorter mud retention
times and subsequent lower buffering volumes.
Figure 3l provides a graphical description of paste production done right. This picture was
taken during thickener selection test work and the sample was produced with a pilot paste
thickener. This unit also provided the samples for the data used in figure 30.
The addition of chemicals to even out variability and produce a viscosity to suit the
deposition area is avery attractive option.
Steeper beach angles will mean that more material can be stored against the dune side of the
residue dam. For a one-degree increase in slope approximately one extrayear of production
can be accommodated.
Ifthe chemical addition results in faster bleed water release, the drying cycle would also
decrease, which would either result in greater throughput for the same achieved in situ
density or better density for the same throughput (both is beneficial). Higher density is,
however, the preferred method to increase viscosity in general as it results in less bleed on
the beaches due to less available water, and hence faster change from a bleed to an
evaporation drying mechanism.
In a simple world the slurry transport system should be the vessel only to bridge the physical
gap between the thickener and the residue disposal facility.
Quantiffing the impact of slurry shear on the actual thickener underflow slurry as a function
of transport system selection is therefore the remaining process variable n.."irury to ensure
constant delivery in terms of rheology.
From figure 32 for the pumping system in figure 20, an analyses of process shear may look
-17
as follows, considering a yield stress vs density relationship as described in Figure for
slurry with30%o solids by weight:
From the density vs rheology relationship in Figure 30 is it clear that a thickener underflow
density of I.2I - 1.23 is sufficient to piovide the residue dam with a yield stress of 60-80 pa.
Considering the case where no gland water is added to the centrifugal pumps (mechanical
seals), then the slurry may end up at the residue at a greater yielstrss i.e. ll0-130 pa
according to the example. Yield stress to the value of 50 Pa is added to the slurry for the
same density due to pumping related shear.
The thickener therefore is designed with excess rheology and according to the relationship in
Figure 29, thickener underflow density may reduce to accommodate the increased rheology
due to slurry shear, with benefits to installation costs.
Clearly it is not as simple as this, but the example illustrates that post-pumping shear needs
to be accommodated in transport system design.
3. Gonclusions
The work above describes the trials and challenges met during the development of a process
suitable for the profitable beneficiation of a dune sand deposit containing heavy minerals and
a high slimes component.
A similar operation was developed and commissioned as a green fields project 3 years ago.
This paper is concluded with a summary of lessons leamt during development of both
processes and commissioning & operation of the current wet processing plant with relevant
considerations for the design of a processing plant.
o Ore bodies with high slimes content need not be considered a threat. Rather, it poses
a challenge as well as an opporhmity in terms of harnessing the positive effecti and
managing of the negative effects.
o A thorough knowledge of the impact of slimes on process performance is required.
o This includes knowledge of process variables, which may be affected by slimes and
resultant rheology as well as a proper quantification of said effects.
o The processed slurry exhibits time dependant behaviour impacting on rheology,
particle size distribution and processing efficiency as well as time independent
behaviour i.e. a yield pseudo-plastic fluid
o Changes in slurry rheological properties and associated particle size distributions
with time are caused by the shear sensitive nature of the high slimes deposit in
question.
o This shear and the impact thereof on process efficiency can be measured, quantified
and utilized successfully for the simulation and design of a processing plant.
. Challenges as described above require the close co-operation of specialists and
industry to cross the barrier from mere challenge to operational success.
References
Abela, R., "Test work to determine the effect of energy input into the ROM material on
rougher spiral performance" MULTOTEC report SPR01/218/RA, Issued to Mr B Krause,
January 2001.
De Lange, J. "IHM wet plant settling test work carried out at Hillendale plant", Thickener
optimisation nitiative, Ticor South Africa Report (Sept 2001).
Farrow, J.B. et al. "Improving thickener technology", AMIRA P266D Project,4th Progress
Report
Goosen, P.E., Krause, 8., Paterson, A.J.C, Paulsen, E. "Variable rheology of heavy mineral
tailings", HMC 2003 Heavy Minerals Conference, Cape Town, South Africa, October 2003.
Grobbelaar, W.P., Krause, 8., Goosen, P.E. "A case study for quantiffing process shear
action on a shear sensitive slurrr", Idh International conference on hydrotransport,
Santiago, Chile, April 2004.
Krause, B.J. "Optimization of Hillendale ultra high rate E-CAT thickeners", Thickener
optimisation initiative, Ticor South Africa Report (Jan 2003).
Paulsen, E., Goosen, P., Paterson, A. "Commissioning test report for Fairbreeze Mine
Simulation Apparatus", PCCE Report THM-451.R02 Rev 0, Issued to Mr B Krause,
November 2002.
Pste 2004
Cape Town
South Africa
PaPer 17
H A C Meintjes
Meintjes HAC
SRK Corsulting, Johannesburg, South Africa
Point of dePosition
Surface of hYdraulic
filled beach
Clav ll 11
beaching perspective, at low shear rates, an estimate
clay 900 t1 11 11
of the yield stress in the rarr1e of 20Pa to 40 Pa with
a best estimate of about 30 Pa (at a shear rate of
clay I 800 11 16 11
3.3 Flume beaching tests Figure 6. Flume test with beaching slope of 1: 53
Beaching tests undertaken in a flume have to be
planned, taking effects of scale modeling into ac-
count. In most instances for non-segregating fine
grained slurries, the grading of the slurry is fine
enough not to require scaling. The rate of testing
will affect the beach slope developed in the flume'
From a practical point of view, the geometry of the
flume has to be considered when interpreting the
flume test results. In some instances the materials
used for flume construction may also affect the in-
terpretation of the test results.
Flume test results were undertaken for the fines
disposal of a mineral sands project, with some of the
test results shown in Figures 5 and 6.
Figrue 7, Flow of slurry within flume
180
160
y= -0.0363x+ 166.86
4 INTERPRETATION OF FLUME TESTS
e140
Etzo R2 = 0.9932
For the purposes of this paper, two dimensional
c 100 analyses of the flume tests were undertaken. The
.E ao reason why this is justified, is based on the relatively
9oo small contribution of shear stresses on the side walls
f;oo (i.e. geometry of the slurry within the flume)'
20
Both limit equilibrium and finite difference
0
methods of calculation were used to assess the shear
o looo
o=tt"i""t(t#h)
4ooo sooo
stresses within the flume. In the case of limit equi-
librium, the slurry strength parameters were varied
Figure 5. Flume test with beaching slope of 1: 28 until a factor of safety of 1 was obtained. In the case
of finite difference calculations, the slurry strength
Tpically the flume beach slope test results varied parameters were varied until there was a significant
between 7 in25 and 1 in 75. increase in displacements within the body of the
These flume test results (Figures 5 and 6) show slurry within the flume, indicating that failure of the
very steep beach slopes, which if direct applied to slurry was eminent. Similar results were obtained
field conditions would provide for very steep beach from both sets of calculations, which are supported
slopes, which are not readily observed in the field. by the observation that the criteria selected, will pro-
Therefore interpretation of the flume tests is re- duce similar outcomes.
quired. Figure 7 shows the nature of the flow of the
The numerical model was constructed as a two- oped within the impoundment will be approximately
dimensional model, with the same slurry geometry 1 in 100 under similar operating conditions.
as shown in the flume test result (Figure 8).
tffi
23*S 14:16
LreE 5@t di a@
-r-Gqq< 5Gq
1eF 14117
.l+ 10@
-27tr41q< 5775E+ s.ffi{< .ffi@
-29H2<r 4BE{1 i G q< smq
D 1:0 T
ttl
I
E
rM+6
z.w' :r
@@r d<rYt- 5!(.cS
O IE
Figure 8. Geometry for flume test analyses Once the flume test has been interpreted to de-
termine the undrained shear strength of the slurry
A numerical model was developed which allowed (assuming that the rate of slurry delivery corresponds
sufficient flexibility so that for the same slope con- with the roposed rate of slurry delivery in the field),
figuration, a flume test, a 100 mm layer and a 200 the expected field slurry slopes can be estimated if
mm layer could be modeled, with minor changes in the thickness of the layers to be placed is known.
input parameters.
The slope stability calculations were undertaken
a slurry
was ob- l@
ts of the &s la:45
!4 212
5.@q .x< 4.G
various numerical analyses can be summarized as -treq <y< 5G4
I in25 21 36 63
1in50 t2 20 38
I tn75 l4 27
Figure 10. Slope stability for 200mm layer at I in 25 slope
1 in 100 11 21
1 in 200 11
A number of calculation methods can be used.
The simplest is to use an infinite slope stability cal-
Additional slope stability calculations were un- culation (Table 3).
dertaken to assess the shear strength implications for These calculations can be checked using more so-
100 and 200 mm layers of placed slurry flable 2). phisticated limited equilibrium calculations or finite
Using the results presented in Table 2, the impli- ifference numerical methods. For example if a
cations of undrained shear strength requirements for 200mm layer is placed at a slope of 1 in 200, then
be assessed.
the infinite slope stability calculation requires an
zontal base)
undrained shear strength of 11.8 Pa, whilst the finite
in 25 is de- difference method shows that 11 Pa is required.
veloped, of 21 Pa is
This finite difference calculation was checked using
required the flume'
a finite difference model where two different lengths
sl
If this 00mm laY-
of slope were modeled: 14m (Figure 11) and 100m'
ers within an impoundment, with an undrained shear The two finite difference calculations agreed to
strength of 2l Pa, then the beaching slopes devel- within 1 Pa, which confirms the validity of applying
infinite slope stability calculations to the slurry lated tests to determine slurry properties at the cor-
placement problems. rect rate of deposition.
There are a number of dimensions associated with
Table 3: Summary of undrained shear strength c, (Pa) required the correct rate of deposition. The first relates to
to support slurry layer on slope viscosity testing at low shear rates similar to the
Slope Layet Layer Layer Layer Layet deposition rate say 4m to 10m from the delivery
50mm 100mm 150mm 200mm 250mm point. In terms of conventional viscosity testing, the
1in20 29.4 58.8 88.2 117.6 t47.0 shearing rate should not be more than say 30 1/s.
I in50 11.8 23.5 35.3 47.1 58.8 The second relates to the rate of volume flow per
1in70 8.4 16.8 25.2 33.6 42.0 metre width of the flume, which should approxi-
I in 100 5.9 1 1.8 t7.7 23.5 29.4 mately correspond with the proposed rate of deposi-
I in 150 3.9 7.8 118 t5.7 19.6 tion per metre of the tailings circumference, if the
I in 200 2.9 5.9 8.8 r 1.8 14.7 slurry will be deposited from the edge of the im-
I in 250 2,4 4.7 7.1 9.4 11.8 poundment.
The viscosity tests have to be re-interpreted to as-
The numerical modeling showed that the devel- sess in detail the related yield stress which will ap-
opment of failure is related to a very thin failure proximately correspond with the undrained shear
zone where the shear strength of the slurry is ex- strength obtained from an interpretation of the flume
ceeded and consequently high shear strain rates de- tests. More development work is required in this
velop. The development of high shear strain rates is area to clarify definitions. As stated above, the pre-
also an indicator of the inception of failure within liminary indications are that there are viscometer test
the numerical model. facilities available which will be able to provide
yield stress data which can be applied to the prelimi-
nary assessment of undrained shear strength values
related to assessment of the beaching slopes.
tr
23k3 23S 6 CONCLUSIONS
,1E.md< 1lG{1
I @q n< 5oG-d
L..1...4
beaching tests which should be applied to prediction
| .all
of slurry slope perfonnance under field conditions.
A framework has been presented how small scale
beaching tests for non-segregated slurry can be suc-
cessfully interpreted and applied to field conditions,
Figure I l. Slope stabil for 200mm layet at 1 in 200 slope if enough is understood about the differences be-
tween the rheology at small scale and field condi-
If the flume tests are being undertaken at remote tions, including shear thinning effects.
sites, where computers are not available to interpret Lessons from a number of recent projects have
the flume tests, then it is also possible to evaluate the shown that the rheology of the slurry could also
flume tests using infinite slope stability calculations, change during pumping from the plant to the im-
by making a relative approximation of the depth of poundment site, which also has to be taken account
the flume slope: in most instances it is good enough of in the assessment of the field beaching character-
to use the average slope depth for such preliminary istics.
calculation purposes.
7 ACKNOWLEDGEMENTS
5 DISCUSSION
The valuable contribution of the various testing
When pilot plant test work is undertaken together laboratories and field testing set-ups are acknowl-
with any test work related to the pumping properties edged which have made this interpretation possible'
of the slurry, there are many opporfunities to evalu-
ate the beaching characteristics of the slurry, using
viscosity tests, slump tests, vane shear tests, and
flume tests. There is a need in all the beaching re-
8 REFERENCES
Melent'ev, VA, Kolpashnikov NP and Volnin BA 1973' Hy-
draulic fill struciures. Energy, Moscow (translation from
Russian), 241pp.
Blight, GE Ig94. The master profile for hydraulic fill tail'1y51
Teaches. Proceedings Institution of Civil Engineers (UK)
107 :27-40.
Blight, GE 1998. Tailings beaches formgd in air and water'
iuitittgt and Mine waste 1998, Fort Collins, Colorado, USA
McPhail, bI and Blight, GE 1998. Predicting tailings beach
profiles using energt and entropy. Tailings and Mine waste
1998, Fort Collins, Colorado, USA
Sofra, F and Boger, DV 2001. Slope predictionfor thicken-ed
toiltngt and pastes. Tailings and Mine waste 2001,
Balkema, Rotterdarr PP' 1 69- 1 80.
Fourie, eg 1OO:. Impoinnt material characteristics of thick-
toilingt Intrnational Seminar. Paste and Thickened
"rd
Tailings, Melbourne, 14 to 16 Ma Section 27.
International Seminar on Paste and Thickened Tailings
Paste 2004
Cape Town
South Africa
PaPer 18
TheThinFilmApproximationforLaminarFlowinopen
Ghannels for non'Newtonian Fluids i
Rainer Haldenwang
Flow Process Research Centre, Cape Technikon, South Africa
Paul Slatter
Flow Process Research Centre, Cape Technikon, South Africa
Raj Chhabra
Department of Chemical Engineering, IIT, Kanpur, India
Abstract
Developing a flow analysis which is fundamentally conect and mathematically rigorous for
laminar non-Newtonian flow in rectangular open channels is only possible if one uses the
long smooth tilting flume was used to test various concentrations of kaolin and bentonite
well as carboxrmethylcellulose (CMC) solutions. By using the appropriate
suspensions as
Reynolds number, Haldenwang et al (2002), showed that all the laminar data could be
collapsed onto the standard Moody diagram used for water. In laminar flow the friction factor
can be predicted for all the materials tested. The actual velocities vrere compared with the
velocities predicted by De Kee et al (1990). To be able to compare the data only the laminar
data points with an aspect ratio less than 0.1 were used.
Paper: The Thin Film Approximation for Laminar Flow in Open Channels for non-Newtonian Fluids
Authors: Haldenwanq, Slatter & Chhabra
For the CMC, the predictions were very good and within the expected experimental error. For
the yield stress fluids the predictions were poor, with effors of up to 10 times.
It is not yet clear why the mathematically and physically rigorous approach of de Kee et al is
not able to predict the thin film flow for yield stress fluids. This has implications for the use
of test flurnes with side walls to predict sheet flow of pastes on tailings disposal sites.
KEY \ilORDS: open channel, flume, non-Newtonian, rheology, laminar flow, sheet flow,
Reynolds number.
NOTATION
2
A - aea m
D - diameter m
Re - Reynolds number
R - hydraulic radius m
x - coordinate
v - shear rate s
p - density kd^t
Tg - wall shear stress Pa
1. lntroduction
The flow of non-Newtonian open channel flow is more complex than pipe flow due to the fact
'When
that there is a free surface open to atmosphere. estimating depth of flow and flow rates
there is always one more unknown than in pipe flow to take into consideration.
If non-Newtonian flume flow formulae that are fundamentally mathematically and physically
rigorous are to be developed, it can only be done for sheet flow. This is what De Kee et al.
(1990) did when they developed Reynolds numbers for a number of rheological models for
sheet flow.
One would have suspected that these models would be able to accurately predict the non-
Newtonian flow encountered in the test work done at the FPRC.
To test this the laminar flow data was extracted from the database and the actual velocities
were compared with the predicted velocities as presented by De Kee et al. (1990).
2. Literature
De Kee et al. (1990), developed flow prediction formulae for viscoplastic fluids including one
for yield-pseudoplastic or Herschel-Bulkley fluids like the kaolin used in these tests.
They considered the flow of a purely viscous time-independent fluid down an inclined plane
and assumed the following:
o Laminar and one dimensional flow
o Incompressible fluid
o Steady state
. Fully developed flow, i.e. no end or side effects
o No slip at the wall and no waves at the surface
From the equations for continuity and momentum the shear stress at arty level in the fluid can
be shown to be:
ru = ro + K(-i-)' (2)
to : pghsino (3)
In their paper De Kee et al. (1990) then derive from these a formula of the average velocity.
The average velocity that they propose is as follows:
v- [*)(*)('-*)'''['.(*)*)
(4)
This formula is mathematically and physically rigorous within the assumed conditions.
Coussot (1997) suggests that in a rectangular flume the flow can be taken to be sheet flow
when the depth to width ratio is less than 0,1.
To give some statistical meaning to the results the log standard error (LSE) was used to
compare the data sets.
The log standard elror, LSE, was found to be useful for comparing models by Lazarus and
Nielson (6), and is defined as follows:
LSE=@ N-l
(5)
3. Experimental Procedure
The database used was obtained from experimental work conducted in the experimental
The in-line tube viscometer is fitted with a high and a low range differential pressure sensor to
measure the pressure drop in the tubes over a set distance. Each line is also fitted with a
magnetic flow meter to measure the flow rate. In addition the relative density and temperature
are measured with a mass flow meter.
The fluid heights in the flume are measured at two positions with digital depth gauges which
are operated manually.
All the data are sent electronically to a data acquisition system linked to a PC where all the
One of the assumptions that De Kee et al. (1990), made is that the flow must be laminar,
steady and fully developed. The flow section in the 10 m flume where the observations were
made was 6 m from the entrance. At that position the flow especially in laminar flow at low
flow rates was fully developed. This was established by measuring the flow depth at two
positions one meter apart and ensuring that they were within limits the same.
Because most of the fluids were viscous the surface was smooth.
The flume which was 300 mm, obviously was not infinitely wide therefore only the data
points in laminar flow which had an aspect ration of less than 0,1 were selected.
3.2 Rheoloqicalcharacterisation
From the tube viscometer the pressure drops and flow rates are obtained. The pressure drop is
used to calculate the wall shear stress and the flow rate the wall shear stress. The wall shear
stress is plotted versus the pseudo shear rate a pseudo shear diagram.
All the laminar flow data obtained from the three tubes will be co-linear if there is no wall
slip. This data is then transformed with the Rabinowitsch -Mooney transformation method to
true shear rate using (equation 6).
(-r).=[#)iandn'=# (6)
The rheological parameters can be obtained using the yield pseudoplastic or Herschel-Bulkley
rheological model. This is shown in (equation 7).
ro=y+K(-t); (7)
This model was used to characterise kaolin. For the Bentonite tested, which is a Bingham
fluid, the value of n becomes 1 and for the CMC which is a power-law fluid, the value of the
yield stress, tr, becomes 0.
4. Experimental Results
Some of the data points taken in the 300 mm flume for the 8% kaolin with depth to width
ratios of less than 0.1 (Coussot, 1997) were collated. The Reynolds numbers were kept to less
than 2000 as the model is only for laminar flow. The actual velocity is compared with the
velocity calculated by using Equation 4.
The results are depicted on Figure l.
0.01 0.1 1 10
-
Figure 1: Actual versus predicted velocity (De Kee et al.) for 8 o/o
kaolin in 300 mm flume.
The results shown in figure 1 indicate that the velocity is significantly over-predicted. The
LSE of the velocities of the 8% kaolin data set in the 300 mm flume is 0.1336.
Another set of data, this time 6Yo kaolin in the 300 mm wide flume, was used and the results
are presented in Figure2.2. The LSE of the velocities of the 60/okaolin data set in the 300 mm
flume is 0.0974 which is slightly better than for the 8o/o Kaolin.
0.1 'l
Figure 2. Predicted velocity (De Kee et al.) versus actual velocity for 6Yo kaolin in 300 mm
flume.
The same trend as seen for the laminar flow of kaolin can be observed for bentonite as shown
The LSE for the 4.So/obentonite velocities in the 300 mm flume is 0.1684 and the LSE for the
6o/obentonite velocities in the 300 mm flume is 0.2246. This is worse than for the kaolin data
sets presented.
0.01 0.1 1 10
Velocity (actual) m/s
Perfect prediction Experimentaldata
- - - - Error factor +5 times - - - Error factor -10 times
-
Figure 3 Predicted velocity (De Kee et al.) versus actual velocity for 4.5%o bentonite in
300 mm flume.
10
- E-
q EEH
I
tt
.uo sEE
E
E.'
-'tt '
1 s
o
,
()
Y
(,
c
'6 0.1
-9
o
0.01 0.1 1 10
Velocity (actual) m/s
prediction E Expermental data
- - - - Error factor +20 times
-Perfect
Figure 4. Predicted velocity (De Kee et al.) versus actual velocity for 6%o bentonite in
300 mm flume.
The following are laminar flow data sets for different concentrations CMC and these are
E
l
o
o
o
Y 0.1
I
o
o
o
.9
o
0.01 0.1 1
Figure 5 Predicted velocity (De Kee et al.) versus actual velocity for L8o/o CMC in
300 mm flume.
,
E
G1
o
o
o
Y
o
o
' 0.'l
-9
o
Figure 6. Predicted velocity (De Kee et al.) versus actual velocity for 2.8%o CMC in
300 mm flume.
0.01 0.1 1
Figure 7. Predicted velocity (De Kee et al.) versus actual velocity for 3.8Yo CMC in
300 mm flume.
U'
E
=G
o
o
o
Y 0.1
o
o
'
-9
o
0.01
0.01 0.1
Velocity (actual) m/s
predicton r (perimental data
- - - -+20%devation - - --20%deviation
-Perfect
Figure 8. Predicted velocity (De Kee et al.) versus actual velocity for 5.4o/o CMC in
300 mm flume.
The LSE for the 1.8% CMC velocities in the 300 mm flume is 0.0389, for the 2.8% CMC it is
0.024, for the 3.8% CMC it is 0.0139 and for the 5.4o/o CMC it is 0.0099, which is
significantly better than for the kaolin or the bentonite. There is also an improvement in the fit
with increase in concentration.
5. Discussion
The results shown from Figure 1 to Figure 4 for kaolin and bentonite, which are both yield
stress fluids, indicate that the velocity predicted is significantly different to the actual
velocity. This indicates that for the data presented in this thesis, the model cannot be used for
the yield stress fluids, even when the aspect ratio is limited to less than 0.1.
For the CMC the predictions are very good as shown in Figure 5 to Figure 8. The predictions
are within 30% of the actual, which is within experimental error range.
It is not clear why this model (Equation 4.2) is not able to predict the velocity for the yield
stress fluids, as this is a mathematically and physically rigorous prediction.
Perhaps the effect of the yield stress requires a smaller aspect ratio to approximate to sheet
flow. This was tested in Figure 9, where the aspect ratio was plotted on the secondary axis.
The aspect ratios are all less than 0.1 in this figure. It can be seen that the aspect ratio does not
seem to be linked to the deviation. There are effors over the range of velocities regardless of
whether the aspect ratio is high or low. Points A and B both are well predicted with A having
a relatively high aspect ratio and B a low aspect ratio. Points C and D are poorly predicted,
and again the two points have very different aspect ratios. One can therefore not link the error
to the aspect ratios.
1.5 0.1
0.09
0.08
tt c
L 0.07 E
:'l
5 o 0.06 o
A
o
(E I A
0.05 fE
6a
ata o
' ^a a a 0.04 o
CL
a
-9
()
0.5 AA a
0.03
A
. '
i
.r.1 6A
^y'.
o 0.02
0.01
0
0.5 1.5
Figure 9. Actual versus predicted velocity (De Kee et al.) for 4.5%o kaolin in 300 mm flume.
For the database used this model, which was developed for thin film flow, cannot be used for
the yield stress fluids kaolin and bentonite, but it predicts the flow of CMC adequately as long
as the aspect ratio is less than 0.1. This indicates that for yield stress fluids the sidewall has a
In summary, test work on viscoplastic fluids - even though the aspect ratios are less than 0.1
- does not seem to be analogous to the true sheet flow expected at paste disposal sites.
6. Gonclusions
For the data presented, the sheet flow model presented by De Kee et al. (1990) was shown to
predict velocities well for a pseudoplastic fluid like CMC, but it was not useful for predicting
the flow of yield stress fluids such as kaolin and bentonite slurry. This was a disappointment
as the model is mathematically and physically rigorous. It is not clear why this is so except
that it could be that the sidewall effect for the yield stress fluids is more prominent. It could
also be that the so-called yield stress, which was derived from the curve fitting of the pipe-
flow data, is very sensitive for this model. The indications are also there that the sidewall
effect could be significant for these yield stress fluids even if the aspect ratios are small.
The evidence presented in this paper shows that sheet flow test work performed on yield
stress fluids in flumes with side walls maybe fundamentally flawed.
7. References
DE KEE, D., CHHABRA, R. P., PO\ryLEY, M.B. & ROY, S. 1990. Flow of viscoplastic
fluids on an inclined plane: evaluation of yield stress, Chem. Eng. Comm. (96): 229-239.
HALDENV/ANG, R. SLATTER, P.T. & CHHABRA, R.P. (2000). The flow of Non-
Newtonian fluids h Open Channels. 9'h International Conference on Transport and
Sedimentation of Solid Particles, Cracow, Poland. 269-280.
LAZARUS, J.H. & NIELSON, I.D.A generalised correlation for friction head losses
1978,
of settling mixtures in horizontal smooth pipelines, 5th Int. Conf. On the hydraulic transport of
solids in pipes, Hydrotransport 5, Paper 81.
Paste 2004
Cape Town
South Africa
Paper 19
Rob Williamson
Knight Pisold Consulting
Angus Paterson
Paterson & Cooke Consulting Engineers (Pty) Ltd
Abstract
Southern peru Copper Corporation appointed Ifuight Pisold in a joint venture with Paterson and Cooke
Consulting Engineers to undertake a feasibility study into the disposal of 150 000 tonnes per day of copper
tailings in near-paste consistency.
Southern peru Copper Corporation currently disposes of tailings in their existing Quebrada Honda tailings dam,
with the impoundment formed by compacted cyclone underflow product. The proposed paste disposal system
was conceived as an alternative to the raising of the Quebrada Honda dam to cater for tailings disposal over the
next 30 years of mine life. The paste tailings would be disposed onto a large flat plain (Pampa Purgatorio)
upstream of the existing Quebrada Honda dam pool. The pampa has a general floor slope of about 5%o, and is
bounded by a semi-circle ofhills on three sides, thereby providing a platform ofhigh ground for the deposition
of the paste product. A beach slope of 6% is required to develop the required volume to store tailings over the
next 30 years. A viscous, high yield stress near-paste tailings is required to form a beach at this slope'
afea.
Paper: Paste Tailings Disposal on a Major Scale at Southern Peru Copper: A Case Study Page2
Authors: U O Salas, RWilliamson, A Paterson
1. lntroduction
The Southern Peru Copper Corporation (SPCC), owned by Groupo Mexico, operates two
open cast copper mines in the mountain desert region near the town of Tacna, in the south of
Peru. The Cuajone mine concentrator at elevation 3 500 mamsl has a current capacity of
87 000 tonnes/day, while that of the Toquepala mine, at elevation 3 100 mamsl is
60 000 tonnes/day. The tailings from the two mines is currently combined into a single stream
and is deposited onto the existing Quebrada Honda tailings facility at approximate elevation
1 200 mamsl, some 2 000 m below the plant elevations.
The estimated useful life of the Quebrada Honda facility is until approximately 201l. As the
ore reserves of the two mines will extend to about 2036, firrther storage capacify for the
disposal of tailings is required.
In November 2002, SPCC issued an enquiry for the study of tailings disposal onto the Pampa
Purgatorio site, immediately upstream of the Quebrada Honda site. This site was identified in
previous pre-feasibility studies as having the potential capacity to store the tailings to the end
of mine life.
The study contract was awarded to the Lima, Peru, off,tce of Knight Piesold Consulting (KP),
in association with Knight Piesold, South Afica and Paterson and Cooke Consulting
Engineers (PCCE), Cape Town, South Africa.
The tailings from each mine is thickened to concentrations of 54 to 63% by mass using
thickeners, and then flows by gravity in a series of concrete canals and natural stream beds
(quebradas) to the Quebrada Honda tailings facility, as illustrated in Figure 1. The specific
tailings data from each mine is given in Table I.
The distances from the Cuajone and Toquepala mines to the Quebrada Honda dam are
approximately 60 km and 32 krrr respectively. The route from the Cuajone mine includes
26krn of tunnels with slurry canals at flat gradients. The static head differences are 2 300m
and 1 900 m respectively. The concrete canals in the tunnels and elsewhere are formed at 1.0
to 1.5 o/o gradient, with the natural quebradas being much steeper in places.
The mixed tailings stream is transfered from the open canal to an 1 100 mm diameter HDPE
pressure pipe upstream of the crest of the Quebrada Honda tailings dam to provide sufficient
driving head for the cyclone clusters situated on the crest of the dam wall.
The tailings dam impounding wall is formed by downstream deposition of the coarse cyclone
cluster underflow product. The deposited coarse tailings are deposited into paddocks along
the downstream toe of the dam and are compacted at optimum moisture content before
placing the next layer in the paddock.
The cyclone overflow product is discharged into the interior beaches of the dam where it is
stored in a saturated state at very flat beach angles. A supernatant pool is formed at the beach
extremity, and water is recovered from the pool by means of a floating pump station equipped
with a series of vertical spindle pumps. Some of the water is deliveed to the canal just ahead
of the pressure pipe section, for dilution of the tailings slurry to a concentration (+- 34%) that
achieves optimal split efficiency at the cyclones. The balance of supernatant water is
discharged to stream, as the static head and distance for return to the plant are too great for
viable operation.
There are eleven cyclone cluster structures positioned at intervals along the crest of the dam
wall. Each cluster station is equipped with 4 x 500 mm diameter cyclones. Eight cyclone
clusters operate at any one time. Figure 2 illustrates one of the eleven existing tailings dam
cyclone stations.
OUEERADA INCAPIJQUIO
In
I
-l *o
I
l+- OUEERADA SMARRNA
CNI DE DESVO
QURADA
SPUMR EOX
DILUON WAE
SUPERNATANT PML
FUBRAOA HN DATI
The Quebrada Honda dam is cuffently 93 m high, and has a crest length of 3 000 m at an
elevation of 1188 mamsl. The current planned final height of the dam is I28m at
1 188 mamsl, requiring significant extension of the wall length, and the formation of new
earthfill saddle dams to contain the beaches. The tailings dam is situated in an area with high
levels of seismicity.
The declared ore reserves at 31 December 2000 of the Cuajone and Toquepala mines total
l2ll.0 and 675.6 million tonnes respectively. The tailings generated are abor,tt 98%o,
requiring future storage of 1 849.6 million tonnes. The conservatively estimated useful life of
the existing Quebrada Honda tailings dam is until20ll, with a remaining storage capacity as
at 1 January 2001 of 483 million m3, or 579.6 million tonnes at a field dry density of 1.2 t/m3 .
The balance of tailings for storage in a new facility is I 270 million tonnes, or possibly more,
depending on whether the existing dam is developed to it's full potential capacity or not.
Earlier studies commissioned by SPCC investigated numerous sites and options for the future
disposal of tailings from the two mines. These studies identified the most favourable site as
the Pampa Purgatorio, alarge, gently sloping plains area immediately upstream of the existing
Quebrada Honda supernatant pool. The site is bounded at its upper end by a semi-circle of
hills. The site was identified as having good potential for disposal of tailings in paste
consistency, by discharging the tailings from a series of locations around the hills at the upper
extremity of the site, thereby creating a wedge of tailings thinning out as it approaches the
existing supernatant pool. By depositing the paste tailings in this way, very little pre-
deposition earthworks would be required, and the existing tailings dam and pool would serve
to collect any process water or stormwater discharge form the new facility. The required
storage capacity for tailings could be generated providing the paste tailings can be deposited
aT a 6Yo slope over the existing site floor at approximately 5o/o natural slope. The concept is
illustrated in Figures 3 and 4. SPCC's primary objectives in commissioning the initial studies
were to:
The 2003 study was done to confirm the concept and to investigate the technical and financial
feasibility of achieving deposition as envisaged. The study was undertaken in two phases;
Phase l: Bench scale testing and desk top study to validate the concept and to determine
physical properties of the tailings including its rheology and further to investigate
the potential deposition characteristics of the tailings at high concentrations.
Phase 2: Pilot plant scale testing of a thickener unit, including pilot scale flume flow testing
and tailings deposition testing at various concentrations. Various options for
increasing the tailings slurry concentrations from existing levels of about 57% by
mass to the required concentration for paste or near paste deposition at about 65 to
rrf.n lI t I tf Or:lfrt
trt'lH
Figure 4: Typical cross-section through the Pampa Purgatorio Paste Disposal Site (Section D)
The following investigations and tests were undertaken during Phase One:
1. The site has the potential capacity for tailings storage to the planned end of mine life
provided that beach slopes of at least 60/o can be achieved
2. The site is suitable for tailings deposition and storage, with very few detracting features
3. Cuajone tailings is more viscous than Toquepala tailings
4. The rheological properties of the Cuajone tailings increases rapidly above 65% solids
concentration by mass, and of the mixed tailings aboveT0%o
5. The rheology of the tailings is sensitive to variations in pH, flocculant dosage and clay
content
6. Small scale flume deposition tests illustrated that beaches can be formed at sufficiently
steep slopes at tailings concentrations of between 65 and 67Yo concentration, providing
that deposition energy is low, deposited layers are thin, and a drying cycle is incorporated
between each deposited layer. However results must be treated with some caution because
of the effects of :-
a. Viscosity and scale
b. Layuthickness
c. Discharge flow rate and energy of deposition
Phase One provided sufficient confidence to proceed with Phase Two pilot plant tests on site
o Installation and testing of a pilot scale paste thickener, which produced material for
launder flow and beaching tests.
o Launder flow tests in a 4 m long, variable width flume.
Further rheology tests using freshly prepared material'
o Beaching tests in I m wide by 15 m long flumes, and in a cone shaped deposition facility.
r A spray bar test facility for mixed tailings at existing concentrations of about 54%.
The pilot plant testing resulted in the following conclusions being drawn:-
l. The pilot scale thickener was able to produce an underflow at concentrations tp to 67Yo,
but not always on a consistent basis.
2. The particle size distribution of the tailings, particularly from the Cuajone plant, can vary
considerably.
3. Flume flow tests validated the theoretical canal flow calculations undertaken in Phase
One.
4. Flow in some of the existing concrete canals will not be possible at high concentrations
due to existing flat canal gradients.
5. The pilot scale beaching tests again demonstrated that effective beaching angles can be
achieved providing that the deposition energy is low, the deposited layers are thin, and
that a drying cycle is incorporated between each successive layer of deposition. The beach
profiles were slightly concave, to convex, depending on the slurry concentration.
6. Because of the small scale of the pilot plant tests in comparison with the envisaged full
scale operation, it could not be concluded with confidence that the required beaching
angle of 6Yo cotild be achieved at concentrations as low as 650/o by mass. The technical
and financial analysis was therefore undertaken at an assumed concentration of 67% by
mass, thereby requiring the use of specific paste thickeners as opposed to conventional
high rate thickeners.
7. Tailings deposited at low concentrations using the spray bar technique did form effective
beaches by gravitational sorting, and with a concave profile. This would however not
satisff the specific demands of deposition onto the Pampa Purgatorio site.
The Phase Two report recommended further studies with near to full scale testing of thickener
performance and the deposition technique before embarking on construction of a full scale
facility
6. Thickeninq Aspects
The combined thickener underflow concentration fom the four existing thickeners at the
Cuajone concentrator is 55%o by mass. This is limited by the thickener capacities to produce
higher concentrations with the high percentage of clay fines present in the Cuajone tailings. A
further limitation is the carrying capacity limitation at higher slurry concentrations of the
26km long concrete canal at l%o gradient in the tunnels leading fom the Cuajone plant
The combined underflow concentration of the four thickeners at the Toquepala plant is in the
region of 640/o, but this is diluted to about 54%by the introduction of services water to the
tailings stream.
Early studies indicated that it may be possible to deposit the thickened tailings stream onto the
Pampa Purgatorio site at a mass concentrationof 650/o to form the required 6%obeach slope.
Under this scenario it would only be necessary to further thicken the Cuajone tailings stream,
and to divert the services water away from the Toquepala stream.
Early studies concentrated on adding a new thickening plant at the following locations:-
r At the Cuajone concentrator.
o At the tunnel R4 concrete canal exit, namely 26km from Cuajone'
At the Pampa Purgatorio disposal site.
Early investigations were also made into the possibility of using vacuum belt filters for
treating a portion of the tailings stream to very high concentrations, but this concept was
abandoned because ofhigh capital and operating cost'
The Phase 2 pilot plant tests indicated that a final mass concentration of at least 670/o wod
be required to conf,rdently form the deposited tailings beach slopes of 6%o. This had significant
additional technical and cost complications due to the high yield stress and viscosity.
In order to achieve a final mixed tailings stream mass concentration of 67%o, it is necessary to
thicken both Cuajone and Toquepala tailings streams to this concentration. Thickening to this
concentration could not be confidently undertaken with conventional high rate thickeners, so
it was essential to use paste type thickeners. To assess the thickening options, proposals for
thickening equipment were invited from several vendors.
For the case of adding additional thickeners at the existing Cuajone thickener complex, a new
thickener installation would be required that would produce a very high underflow
The proven ability of paste thickeners to consistently produce the high required
concentrations
The limitation of the 26 km long, l%o gradient canal to carry the tailings stream at a
concentration of 66Yo
The option of adding additional thickener plant at Cuajone was therefore not considered
viable.
To achieve a 67%o mass concentration at Tunnel R4 exit (at the end of the 26 km tunnel and
concrete canal) required that the entire tailings stream of 87 000 tonnes/day at 55o mass
concentration be re-thickened. This required twelve 25 m diameter paste thickeners. While it
was found to be feasible to prepare suff,rcient space for this installation, other complications
arose. The tailings underflow stream from the thickeners at 67Yo concentration was too
viscous to flow down the natural quebrada to the disposal site. Therefore pumping through a
new pipeline would be required.
Furthermore, a new tunnel would be required to transfer the thickened stream from the natural
quebrada to the Pampa Purgatorio disposal site.
As a sub-option of this scheme, a further case was studied where partial thickening of the
Cuajone stream was undertaken at Cuajone, to the concentration carrying limit of the 26kn.
Canal. This achieved partial and economical water recovery for re-use in the Cuajone plant.
The option however has the same complication as the main Tunnel R4 option.
Thickening of the Toquepala stream from 64 to 67Yo mass concentration required the
treatment of the entire 60000 tonneslday of tailings using eight 25m diameter paste
thickeners. The resulting tailings stream at 67Yo concentration was found to flow satisfactorily
down the existing natural quebradas to the Pampa Purgatorio disposal site. Toquepala services
water would again require to be diverted away from the tailings stream to prevent unwanted
dilution.
The option of installing new thickeners at the Pampa Purgatorio disposal site was also
investigated. In order to achieve water recovery at the Cuajone plant, limited thickening to the
concentration carrying limit of the 26 km concrete canal was first undertaken. Twenty 25 m
diameter paste thickeners are needed to treat the combined Cuajone/Toquepala stream of
147 000 tonnes/day .In this case it is not necessary to divert the Toquepala services water.
The pipeline transport requirements were determined using the slurry rheological properties
obtained during the pilot plant test work in Phase Two. The data shows that there is a steep
increase in viscosity and yield stress above 65Yo solids concentration by mass, as shown in
Figures 5 and 6.
l. The 26 km concrete canal from Cuajone to Tunnel R4 exit has insufficient gradient to
transport the Cuajone tailings at solids concentrations greater than 60%om
2. The Cimanona Canal, transporting the combined tailings, has a maximum carrying
capacity of 64%;o.
The canals cannot be used to transport 67% solids concentration by mass tailings.
200
180
160
110
d lzo
P OO
.9 80
60
40
20
0
5s% m% 65%
55% 6070
50% 6s%
Solds @nnbal by weghl (%hl
l: Cuejore Solds co6nbelion by reghl (%m)
Phase l: Mixed
- Phass 2: Mxed r r Phse2:Cu4one l. Mxed l: Cuejone
x WITS ve yleld skess - - r Phase 2: Mxed r - r Phse 2i Cuajone
- -Phase
-Phese -Ph
Figure 5: Yield Stress versus Solids Figure 6: Plastic viscosity versus solids
It was desirable to utilise the available gravity head between Cuajone and Toquepala mine
sites and the Pampa Purgatorio to eliminate or reduce the pumping requirements. However, a
key component of the deposition strategy requires that slurry is placed in a controlled manner
and that the deposition point is rotated on a regular basis discussed in Section 8. In order to
meet the deposition strategy requirements, the flow streams from Toquepala and Cuajone
need to be combined in a pressure pipeline.
Analysis of the requirements at 650/o and 670/o solids concentration show that at 670/o there is
not enough gravity head between the Cimarrona Canal and the Pampa Purgatorio and a high
pressure pump station would be needed. It is not possible to combine the Cuajone and
Toquepala flow streams at a higher elevation as they are in different valleys upstream of the
Cimarrona Control point.
l. It is not possible to use gravity flow to transport the combined flow aI 67%o solids
concentration from the Cimarrona canal.
2. A positive displacement pump station is needed to pump 67% solids concentration,
regardless of whether the flows are combined at Cimarrona or the Pampa Purgatorio.
a
J. The limits of centrifugal pumping are approximately 65% solids concentration for the
mixed tailings, which is too low for the required 6%obeach slope on the Pampa Purgatorio.
8. Deposition Aspects
The surface flow of thickened tailings is complex process, incorporating the following
factors:
Absorption of moisture into the previously deposited dried layer, together with
evaporation, causes an increase in the slurry yield stress and viscosity, thereby increasing
the deposited beach angle.
The layer thickness at the point of discharge always tends to increase. This leads to a
decrease in the natural beach slope, and the material thus tends to flow further. This
mechanism explains how the deposited material will continue flowing for long distances.
Therefore as long as there is a supply of fresh tailings at the deposition point, flow down
the slope will continue.
Conversely, as deposition stops, the layer thickness decreases, the beaching angle
steepens, flow stops and a beach is formed.
The results of the Phase one and phase two tests, combined with Knight Piesold's extensive
tailings disposal experience, indicated that the following deposition principles must be applied
in order to achieve the required 6%obeach slope across the Pampa Purgatorio deposition site:
The above factors imply that a carefully designed and managed deposition system will be
required to achieve the desired result.
The length of deposition available around the semi-circle of hills at the upper end of the site is
approximately 8.4 km. The deposition design divided this into four deposition zones, each
fitted with two sets of eight 250 mm diameter off takes, one from each of the 550 mm
diameter supply pipelines. Each 550 mm diameter pipeline requires a dedicated set of positive
displacement pumps to pump the 670/o mass solids concentration thickened tailings.
Each of the 250 mm diameter off-take points is fitted with a diaphragm valve and discharges
into a 50 to 75 m long perforated drop pipe of 550 mm diameter. The drop pipe leads the
tailings discharge to floor of the deposition site, and dissipates the discharge energy through
the perforations in the pipe. The drop pipe becomes buried as the beach rises, with flow being
This system has been successfully implemented on numerous other tailings dam installations,
and was demonstrated during the Phase two pilot trials. Figures 7 and 8 illustrate the
deposition concepts.
C 55omm
|
PIPE
i'.* 1'o',8"
/-
ELEVATION
NT
The extensive pilot plant testing and conceptual pumping, deposition and plant layout design
work, followed by cost estimation has enabled the study team to reach firm conclusions about
the proposed Pampa Purgatorio project. The conclusions can be summarised as follows.
l. The pilot scale tests demonstrated that the required 60/, beach slope can be achieved at
tailings slurry concentrations of 65% by mass.
2. However, even though thorough pilot scale test work was undertaken, the scale of this was
too small in relation to the prototype project to have full conceptual stage confidence in
the tailings concentration required to achieve a 6Yo beach slope. The conceptual design
and costing was therefore undertaken assuming a required 670/o concentration.
J. There are significant deposition flow energy effects that can influence the beach formation
angle, and there still require to be demonstrated.
4. The pilot scale thickener trials demonstrated the ability of a paste type thickener to
produce underflow concentrations of 67%o by mass.
7 . There is no cost effective solution to recover all of the water from the thickening process
for return to the plant. Some solutions for partial recovery for re-use were however
identified.
8. The existing low gradient concrete canals are a restriction to tailings slurry flow at high
concentrations.
g. Tailings slurry at 67%o concentration cannot be transported via gravity and require positive
Finally, it is concluded that the disposal of high concentration tailings onto the Pampa
Purgatorio site is technically feasible, and is very attractive in terms of the low cost of
developing the new disposal site. However these advantages are off-set by the high capital
and operating cost in producing and transporting the high concentration tailings slurry.
ACKNOWLEDGEMENTS
The authors would like to thank the Southern Peru Copper Corporation project management
team for their assistance during the study and their permission to publish this paper.
Paste 2004
Cape Town
South Africa
Paper 20
Topics to be discussed
. ls it as simple as it seems?
A laboratory-sized fl ume
200
180
160
^,140
Eno
.! roo
ao
oo
-40
20
0
0 200 400 00 800 1000 1200
200
E tso
a
=
E'*
.9"
b50
. 25
1.0
c
o.a
o.o
q)
.e.0.4
(5
c
Eoz
z.
0.0
0.0 0.1 0.2 0.3 0.4 0.5 0,6 0,7 0.8
250
E
ZOO
;!)
1so
=
.9 100
G
-50
ct
0
0 500 1 000 1 500 2000
DstqnceoffloY/ long hech lmml
*
s.o
$
g
40
I eo
20
500 1000 1500 2000 2s00
Flume width (mm]
Then what?
0
2
4
g6
E8
o-
3ro
12
14
16
o.2
'l
1.2
140
120
s 100
c
o 80
c
o
o 60
L
()
(It 40
20
=
0
20 40 60 80
1 400 70
1200 60
1 000 50
800 40
600 30
400 20
200 l0
0
1.3 1.4 1.5
200 600
0
0.25
E
-c 0.5
o
(l)
o
0.75
1r.,
800
700
; 600
! soo
400
b 300
$ zoo
100
0
1 1 100 1200 1300 1400 1500
-tr-
300
! zso
$
o.
zoo
g 150
o
100
.g
Eso
c
50
()
Trr (days)
{s4
,g
,gs
L
s->. 4.
>
I
t1
s240st8$
Tir {d}ys}
c0
o
-20
E
8_ -40
9a
o -60
9g
=
-80
io -100
=E -120
o= -140
Time (days)
Time (days)
Conclusions
Paste 2004
Cape Town
South,Africa
Paper 21
Santiago Luna
Southern Peru Copper Corporation
Ronald Scheurenberg
Knight Pisold Consulting
Andy Fourie
University of the Witwatersrand
Abstract
As part of the feasibility study into the disposal of 147 000 metric tons per day of highly
thickened copper tailings for Southem Peru Copper Corporation (SPCC), pilot scale tests
were carried out on the beaching characteristics of the tailings. The principal objectives of the
tests were to determine the beach slope that could be achieved with the tailings at
predetermined solids concentrations, the effectiveness for beach formation of selected
deposition procedures, the properties of the tailings deposits and an indication of the scale
effects. The test program included the deposition of prepared samples of tailings into a 140
mm wide laboratory test flume; deposition of tailings abstracted from the discharge channel at
the mine, thickened in a pilot thickener to solids concentrations of 65Yo and 67%by mass and
deposited into 1.0 m wide test flumes 15 m long as well as into a cone shaped basin with an
internal angle of 45o at the SPCC site in Peru; and deposition of the tailings currently
produced by the mine, at a solids concentration of 52o/o to 54% by mass, through spray bars
Paper: Pilot testing for disposal of highly thickened tailings at Southern Peru Copper Corporation Page2
Author: U Oliveros S, S Luna, R
onto natural ground. In all cases the floor slope of the flumes, cone or ground was 5olo to
The pilot scale tests showed that the required beach slope could be achieved at the scale of
testing done, provided that the deposition procedure was properly controlled. The properties
of the tailings in situ - dry density, degree of saturation, amount of consolidation, strength
developed, etc. - are dependent on the deposition procedures and the scale, as well as on the
initial solids concentration of the tailings.
1. lntroduction
Southern Peru Copper Corporation (SPCC) operates two copper mines - Cuajone and
Toquepala - on the western side of the Andes mountain range in the south of Peru. The
following quantities of tailings solids are currently produced:
The two mines and concentrator plants are located about 34 km apart at an altitude of
approximately 3100 to 3500 mamsl. The tailings from both mines are currently
conveyed in natural gullies (quebradas) and concrete lined canals over a distance of
about 42 km to a common deposition facility at Quebrada Honda at an elevation of
approximately 1200 mamsl. New tailings deposition facilities must be identified and
commissioned before the Quebrada Honda facilit in operation since 1994, reaches
SPCC have identified a possible area for future tailings disposal on the Pampa
Purgatorio, aplain at the foot of the mountain range and upstream of the Quebrada
Honda deposit. The area under consideration is fairly flat with a ground slope of
around 5%o andcovering approximately 25 km2. In order to utilize this area effectively
it has been proposed to thicken the tailings to such a consistency that it could be
The principal objectives of the beaching tests were to determine the following:
the beach slope that could be achieved with the tailings at predetermined solids
concentrations in the range of 60Yo to 70%o by mass
the effectiveness for beach formation of selected deposition procedures, including
intermittent deposition with various cycle times
the variation with time of the dry density, moisture content and degree of saturation of
the tailings beach
o the strength, consolidation and permeability properties of the tailings deposits
a an indication of the scale effects.
The variables that were applied to the formation of the test beaches included
solids concentration, layer thickness, drying time, rate of flow (related to size of test
equipment) and scale effects.
The test program, performed between March and July 2003, included the following
test arrangements:
deposited into test flumes and into a 45o cone shaped basin at the SPCC site in Peru.
In all cases the floor slope of the flumes, cone or ground was 5Yo to match the natural
ground slope at the proposed deposition aea.
2. Tailinqs Properties
Typical properties of the tailings that are currently produced, as measured at various times
during the test program, are summarizedin Table I. These properties do vary depending on
the mineral being processed at any particular time.
Relative density (SG) of solids 2.719 MT/m' 2.755 MT/m' 2.727 MT/m'
Sedimentation tests carried out in Ifuight Pisold's laboratory in Lima, Peru, indicated that
the dry density achieved by the tailings after settling depended on the drainage conditions as
well as on the initial solids concentration:
The permeability of the tailings after drained settling averaged 1x10-5 cm/sec.
Samples of mixed tailings were tested in the laboratories of the University of the
'Witwatersrand preliminary indication of the
in Johannesburg with the purpose of obtaining a
beaching angles that could be achieved at solids concentrations of 60, 65 and 70%by mass.
The tailings were pumped from the pan mixer to the test flume, 1800 mm long by 140 mm
wide, set at a slope of 5%o. The material was placed at the beachhead through a slotted pipe
and allowed to flow down the length of the flume until it reached the end. The tailings profile
was measured immediately following deposition and again after drying under a hear fan
until the surface started to crack. A second tailings layer was deposited and the beach profile
measured immediately after deposition and again after standing for a day.
The beach profiles achieved for the three solids concentrations, for both the first and second
layers, are presented in Figure 1. The axes are defined as follows:
o Dimensionless Elevation :Wy
where h: height above beach toe at a point, y: total beach height.
o Dimensionless Distance :}llx
where H: distance from the beachhead at a point, x: total beach length.
1.0
0.9
0.8
c
.9 o7
6
o
ul 06
o 05
c
' o.4
co
.E 0,3
0.2
0.1
00
0.0
The pilot plant was installed downstream of the point where the Cuajone and
Toquepala tailings streams join. It consisted of a thickener and an agitator system,
supplied and operated by the company Outokumpu. It was fed in turn from the
Cuajone and the mixed tailings streams. Thickened tailings at the required
concentration were obtained from the thickener underflow.
A general view of the pilot plant and the associated test flumes and cone is shown in
Figure 2. Figures 3 and 4 show the test flumes and cone respectively, after the
deposition of several layers of tailings and a drying period.
During the test period, the average daily minimum temperature was 10oC and the
average daily maximum was 23"C, while the evaporation averaged 4.6 mm/day.
Ten similar flumes were constructed in order that the tailings at selected solids
concentrations (65% and 67%o by mass) could be deposited on successive days with
varying layer thicknesses (betweenT0 and 150 mm) and drying periods between layers
(5,7 and 10 days). The rate of flow that the pilot thickener underflow could produce
consistently was between 15 and 18 Vmin. So as to achieve the required layer
thickness, the period of deposition was generally between 6 and 12 hours.
The tailings were deposited at the upper end of the test flume through a vertical drop
pipe with horizontal holes at various levels in order to minimize the flow energy. The
slurry was sampled at the start and end of each deposition so as to check the tailings
properties. The tailings levels were measured on stakes fixed near the head, the middle
and the toe of the flumes at various times before, during and after deposition in order
to determine the variation in layer thickness with time. At the end of each drying
period, samples were taken in 75 mm diameter PVC tubes to enable the density,
moisture content and degree of saturation to be determined. Additional undisturbed
Shelby tube samples were taken from the penultimate layer for laboratory permeability
and shear strength tests.
Tailings at 67%o solids concentration from the pilot thickener was deposited on the
cone, shaped basin with an internal angle of 45o and a radius of 25m, from a drop pipe
at the apex, at a nominal layer thickness of 15 cm at intervals of 6 days. Because of the
low rate of delivery of the tailings, it required some 15 hours to deposit each layer.
With the low energy of deposition the tailings tended to form meandering streams of
width about 50 cm.
The tailings levels were measured on stakes fixed at intervals over the area of the cone,
at various times before, during and after deposition. Samples were taken at various
points in order to determine the tailings properties. The beach profiles measured the
end of the drying period of each layer are given in dimensionless form in Figure 5.
Other test results are summarizedin Tables III, IV and V.
1.0
c
o 0.9 \
(E 0.8
\
-9
ll
t,
0.7
0.6
t.>t
\-1.\
tt 0.5
-c
c
.9
0.4 -\]
o 0.3
o 0.2
.E 0.1
o
0.0
o.o 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 10
Dimensionless Distance
Table III shows the average properties of the tailings in the channels and cone after the
drying period:
41
Two sets of spray bars were set up, each approximately 40 m long with 30 mm diameter holes
drilled at 1.5 m centres in the 150 mm PVC pipe. The slurry was abstracted from the canal
delivering the current tailings to the operating Quebrada Honda tailings facility. Each of the
spray bars deposited the tailings onto natural ground (dry silty sand) at a slope of
approximat ely 5Yo, as shown in Figure 6. One set was meant to provide a 3 day dryrng period
and the other 5 days between depositions. The rate of delivery of tailings to the spray bars
varied between 260 and 570 l/min and deposition was continued for up to 72 hours. The
planned durations could not always be achieved because the 200 mm supply pipeline became
The rate of flow was measured at the start and near the end of oach deposition and samples
were taken to enable the tailings properties to be determined. The tailings levels were
measured on stakes fixed in the ground at intervals over the area. The results of the
6. Discussion
Table fV shows the average properties of the tailings slurry placed in the channels,
cone and spray bars:
Solids
7o Fines 7o Sand Pulp density
Concentracin
(%) (%) (%) (MT/m3)
Table V shows the average beach slopes obtained in the tests and the degree of
consolidation of the tailings during the drying period. Almost all the consolidation was
observed to take place during the first 48 hours after deposition, after which significant
cracks would appear in the deposit.
Solids
Slope (%) Consolidation
Concentration
Scale effects are important and in practice the beach slopes could be affected by the
flow energy, the amount of moisture lost by evaporation and to the underlying layet,
viscosity, yield stress and the thickness of the layer. Conversely, the thickness of the
layer is governed by the rate of discharge of tailings from the deposition point, the
slope of the underlyrng layer, the consistency of the tailings and the amount of
moisture lost to the underlying layer.
Additional tests are recom.mended at a larger scale with solids concentrations of 65%
and 670/o with flow rates that resemble that from at least a single discharge point to be
used on Pampa Purgatorio in practice. These tests would be required to confirm the
following points:
- Verify if it is possible to obtain beach slopes of 6%o with tailings of 65% solids
concentration.
- Evaluate the spacing between deposition points for the effective formation of the
beaches.
- Evaluate the minimum flow energy required to be able to transport the tailings for
the full length of Pampa Purgatorio.
- Confirm the drying cycles needed to ensure the required degree of saturation is
achieved.
7. Gonclusions
The results of the beaching tests in the flumes and the cone, as well as in the spray bars, led to
the following conclusions:
- Intermittent deposition with low flows and defined drying periods is effective for beach
formation even with tailings of relatively low solids concentration.
- The average beach slopes achieved for solids concentrations of 67%o and 65%o wete 6'40/o
and 5.7o/o respectively. Additional tests at alarger scale are required to establish whether
a beach slope of no less than 6Yo canbe achieved with concentrations of less thart 6T%o.
- The structure of the deposit is neither monolithic nor isotropic. The contact surfaces
between layers are preserved.
- The consolidation (without any applied load) of the tailings layers is about 22o/o, neatly
all of which occurs in the first 48 hours.
- After the drying period, the average dry density measured in the beach was 1'64 MT/m3,
while the moisture content varied between 25 and30Yo.
- The effective fction angle of the thickened tailings was 35o. The permeability of the
tailings was 1.2x10-6 cm/s.
Paste 2004
Cape Town
South Africa
Paper 22
Hartmut Ilgner
Auhor:H llgner
Peste 2004
Outline
Current Situation in SA
I
I
p to.oo
E
Cemented
ED
Backfll
E
o t.oo
I
.l I
ut
o I
at I
oE
o.ro
t Gohe sion less
Backfill
0.01
0.01 1.00
Strain Log [o/ol
Well-Placed
Hydraulic Fill at Depth, 15 Degree dip
Underground
Platinum Mining
. 1000 to2000 m
. High thermal gradient'
. Traditionally mining only Merensky reef.
. lncreased mining of UG2, below mined out Merensky.
. Potholes to be included in support and mine layout.
. Drive to mechanise mining.
Significant Savings
in Heat Transfer due to Fill
0m
Backfill
compressron
and load
taking.
1 70 days 300 days 400 days
Paste ,Y,
Paste Backfll, ilgner, Slde 11
Paste 2001 2Aa4 v
Coal Mining
. 40 m to 400 m depth.
. Bord and pillar method.
. Various coal grades.
. Huge remaining primary mined out areas.
Government Paradox:
Ashfilling
underground
AMD or
process
water
Gontamination
V
Paste
20a4 'v
Pasle Backfill, tlgne| Slde 16
Paste 2004
gr-.:;:,:;
Ashfilled Coal,
60 % Filling Height
Non-violent Failure
of upper, unfilled coal sqecimen
Behaviour and
Difference of Post-Peak-Fail ures
25
daneter= 100mm
20
.E
o.
leght = 200 m
=!,
tt, 15
o
an
G
t)
l0
L
o
5
0
0.00 0.01 0.02 0.03
Vertical Strain
Pasle Backill, ilgner, Slide 22
Pasae 2001
Overburden loading
onto backfilled pillar
Displacement
Paste Backfll, llgner, Slide 23
Paste 2004
80
T'
70
F60
550
filo l *ElaseCase:
*BaseCase:
60m
100m
rj,
-,-Elase case: 180 m
* {160 mwith Ashrfill
n + 100 mwith Ashfill -lstop cf trring l--
l0
<:*180 mwith Ashfill
---1 \
0 -
-14 -13 -12 -11 -10 -9 4 -76-54-3-2-1 01234
Tne [Yeas] Ashfi ll lnvesrnt Period Extended tife of Mrie
Summary
Paste
200,4 'v
Pasle Backfilt, llgnet, Slde 33
Pasle 2004
It lasts....
Don't take
Shortcuts !
Paste 2004
Cape Town
South Africa
Paper 23
Matt Wilkins and Chris Gilchrist, Michael Fehrsen and Robert Cooke
Tailings
1.8 mtpa coarse salt (soluble waste) - Centrifuge cake
0.2 mtpa fine clay (insoluble waste), salt and calcium sulphate - Filter cake
Backfill Project
. 1996 GPL initiated a study to investigate the feasibility
of disposing filter cake in the worked out areas of the
mne. lssues investigated were:
. Backfilling a fine tailings (clay and salt)
. Binder requirements
. Distribution system challenges (1 100 m deep shaft and l1 km
horizontal transport distances)
. lmplications of backfill in dry, low humidity environment
particularly regarding potential dissolution of pillars
. Transfer of overburden stress onto the backfill
. Surface preparation Plant.
. Study indicated that the proposed system was feasible.
Backfill Project
. ln 1998 CPL committed to a four year project involving
extensive laboratory test work and pilot scale plant
operation.
. The project received financial support from the
European Gommission's LIFE-Environment programme.
. Managed by GPL with assistance from the UK's Mineral
lndustry Research Organisation.
' PGGE appointed as principal design consultants.
Backfill Properties
. Binder addition rates o 10o/o to 15Yo are requred. After
carefut consderation, mine decided not to use binder:
. Prohibitive cost
. Backfill would be stiffer than containing rock mass.
. The basic slurry properties are:
. Filter cake solid density: 2'6 Um3 (variable)
. Slurry densitY: 1.5 Um3 to 1.6 Um3
. Bingham yield stress: 32 to 70 Pa
. Bingham viscosity: 0.006 to 0.039 Pa.s.
. Viscosity control
. Minimise water in backfill
. Wide variation in filter cake properties
. Horizontal transport distance varies from 5.5 km to 11 km.
. Batch operaton
. Filter plant operation
. Accurate control.
SEAWATER
Mixing Tank
Header Tank
Valve Station
Storage Tank
Charge Pump
Energy Dissipaters
Valve Station
Piping in Roadway
Segmental Bend
Panel Discharge
Underground Placement
. No maor ventitation concerns related to backfilling'
. Pillar dissolution shown not to be an issue.
. Placement panels selected on the following basis:
. Form a natural sump
. During panel closure, the backfill should not be forced into
areas where future access is required
. Panel conditions must permit entry to install pipe work and
observe placed backfill.
. Need to allow for pillar consoldation and lateral
deformation:
. 360/0reduction in area after four years
. Additional260/o over the next ten years.
31
Boulby Backftt syslem, wit'iinsea at, stide Pa Ste S-
Pdste2oo4 2OO46-'Y_**
Backfill
Sysfem Commissioning
V
Boulby Backfiil Syslem, Wlklns et el, Slide 33 Paste r{1
Paste 2OO4 2DO4 v
Leaking Gland
Sysfem Commissioning
. The backfill distribution system is operating as
specified. Measured flow rates are slightly higher than
design due to cautious approach in staing with low
viscosity slurry.
. Main problems encountered during commissioning
were:
. Poor peormance of storage tank agitator
. Valve sleeve adaptor failure
. lntermittent fibre optic cable termination fault
. Leaking and failure of low pressure victaulic couplings.
Operating Experience
. The new-repulping plant has proved to be significantly
more reliable than the old sYstem.
. The backfill system distribution system has proved
simple to operate. The flow rates have been consistent
over a six month operating period indicating the
reliability of the viscosity control loop.
. The backfill system has been shut down with the shaft
column of slurry for a two week period. The system re-
started without any Problems.
. A pipeline wear monitoring programme has been
implemented.
Conclusion
Paste 2004
Paper 24
Abstract
Falconbridge Limited is currently constructing a 400 tph paste backfill plant at its Kidd Creek
copper-zinc mine in Timmins, Canada and is scheduled to begin production in August of
2004. The paste backfill system is unusual in a number of ways including:
I The excavation of old, dewatered tailings and transportation to the paste plant by
roadJrall.
2. The flexibility of the plant to handle alarge number of different tailings, aggregate and
binder materials in a large range of recipe possibilities.
3. The underground distribution system is designed to run full to surface during the
majority of its operating hours by substantially increasing the capacity of the plant so
that it can produce paste faster than the pipeline can receive it.
The paste plant is currently under construction and on schedule to begin producing paste in
August 2004.
Design Considerations for a 400 tph Paste Backfill Plant at Kidd Creek Mine Page2
Chris Lee, David Landriault, Roy Durr
1. Svstem Description
1.1 Overview
The paste system's primary function is to provide a cemented backfrll material to satisfy the
mining requirements for Falconbridge Limited's Kidd Creek Mine D. Mine D is the
extension to depth of the existing Kidd orebody and includes mining depths from 6,800 ft
below surface to 10,000 ft below surface. The current mining operations use cemented
rockfill as their primary method of backfill, however, the capital and operating costs required
to use rockfill at greater depths was substantially higher than paste and, in fact, made the
Mine D project uneconomical. Through a number of feasibility studies, paste was identified
as being an economic option that could provide the high strength materials and short cycle
The implementation of a paste backfill system at Kidd Creek is unfortunately not easy or
typical. This is primarily due to the cost and technical issues associated with the delivery of a
suitable material to the Kidd Mine site. The Kidd Met-Site (Mi11) is located a distance of
32kr away from the mine by rail or 50 km away by road, which makes delivery of a dry
(less than 23o/o moisture), fine tailings material expensive and technically difficult. A large
number of options were evaluated regarding the sourcing, dewatering, transportation and
processing of backfill materials. In the end, it was decided that the best option would be to
reclaim tailings from an old gold mine to supply the fines required to make a paste and to
supply sand for the remaining bulk of the materials. The tailings will be excavated from an
old dewatered tailings impoundment and either trucked or transported by rail to the mine site
where they will be stored in a storage dome. Sand will be delivered by a contractor and also
be stored in a dedicated storage dome.
The tailings and sand will be loaded from the storage domes into the batching plant by front
end loaders. The batching plant is very similar to a concrete batching plant and discharges a
tailings/sand paste to one of two holding hoppers which feed the undergtound distribution
system. The process flowsheet for the system is shown on Figure 1.
LOADIR
A6GRECA.lt
S-IORAGE OOMI
'AlLl'lG
S
S]ORAGI DOMI
latLu{Gs
SURGE 'IANK
ACGRTGAE
SCALE
'lo
lo
U NDERGROUND UNDIRGROUND
2004
lnternational seminar on Paste and Thickened Tailings: Paste
Page 4
Design considerations for a 400 tph Paste Backfill Plant at Kidd creek Mine
Chris Lee, David Landriault, Roy Dun
During the initial years of paste production when the tonnages required are relatively small,
the tailings will be delivered to the mine site by truck. When the tonnages become more
significant, however, tailings will be transported on the back-haul of the ore train which
delivers ore from the Kidd Mine site to the Kidd Met-Site. Large tonnages can be effectively
transported in this way with only a minimal cost at the loading and unloading facilities.
content (despite the fact that cyanide was used during processing). Levels of heavy metals
and other contaminants are also quite low. This is important because the transport of the
tailings will inevitably result in some dusting. To further combat the effects of dusting, trucks
and rail cars will be covered during transport.
The reaction of the tailings with the binder is dependent on the tailings chemistry and
mineralogy and has a huge impact on the amount of binder required and, therefore, the cost of
the fill. The tailings showed excellent strength properties when combined with a 90/10 mix of
blast furnace slag and cement. Although the chemical reaction is complex, it can be
simplified by describing it as a reaction between the blast furnace slag and the iron in the
tailings which is catalysed by the small proportion of cement. In addition, the iron is present
in a non-sulphur bearing mineral which eliminates the post-curing production and growth of
gypsum crystals (which can have a detrimental effect on the long-term strength of the
backfill). The strengths of the tailings with the blast furnace slag/cement mixture vs. the
strengths with cement only are shown on Figure 2.
35
3.0
aCL
- 2.5
/
o
ztJ
.
z.o
J
v)
,/
D
lJ
,
,/ ,/
1
O- 4E
r'v
o
=
o ,/
c I
J
z.
tr 1.0
/ -/
z
o
C)
z
05
0.0
14 21 28 35
Tailings will be transported by rail to an unloading facility where they will be unloaded by a
backhoe that is mounted on a platforn over the rail cars. This method is currently used at
Kidd Creek for unloading of concentrate and is shown on Figure 3. The backhoe will
discharge into a dump truck which will move the tailings a few hundred metres to the tailings
storage dome.
Tailings and Sand will be stored in two large storage domes. The size of the two domes is
sufficient to hold enough material to filI an entire stope. Although this is far more than is
generally required, it allows the mine to start a pour and be 100% confident that the pour will
not be intemrpted because a lack of availability of materials. In addition, the large surge
capacity allows the materials supply contractor to work relatively consistent hours with a
small crew to fill the domes over a long period of time which is independent of the peak
materials consumption rate during pouring. Although this seems to be a relatively minor
point, there is considerable savings when comparing this method with 'just in time' delivery
due to the large size of the crew which is required to deliver 400 tph of tailings and sand'
2. Batch Plant
The batch plant resembles a concrete batch plant with the exception of the tailings circuit
which allows the moist fine tailings recovered from the tailings atea to be handled, weighed
and batched into the process (see Figure 1)'
The tailings are rehandled out of the storage dome and into a live bottom feeder which is
essentially a hopper with a bottom that is composed of 12 screw conveyors that conveys the
material towards the discharge onto a belt conveyor. This tlpe of equipment is required to
move the fine sticky tailings material onto a conveyor. The tailings are then conveyed into a
continuous mixer where they are diluted with water and mixed to a paste consistency before
being discharged into a surge bin. From this surge bin, batches are withdrawn into a weigh
The sand is rehandled out of the storage dome and into a hopper before being conveyed
through a vibrating screen and then into one of two large surge bins in the plant. From these
surge bins, sand is discharged through a clam gate and into a weigh hopper.
Binder is stored in two silos which ae an integral part of the paste plant building. Binder is
discharged from one or both binder silos through the action of a screw conveyor which files a
weigh hopper. A weigh hopper is also provided for the water that will be required in the
batch.
Once all the weigh hoppers are fuIl, they will dump into the batch mixer (shown on Figure 4).
Additional water will be added, as required, to bring the paste to the desired consistency
before it is discharged through one of two discharge doors on the bottom of the mixer and into
one of two paste hoppers.
Figure 4 - 9 mi Mixer Front View (Left) and No.1 Dump Door (Right)
Each paste hopper is mounted directly over a borehole which extends to the 1600 Level
(approximately 1,600 ft below surface).
One of the items that makes this plant different from the vast majority of paste systems is the
flexibility of the recipe that can be handled by the plant to suit the materials that ate available
now and in the future. The design production rate of the plant is 400 tonnes of dry solids per
hour and can be composed of any variation on the following:
o 28o/o-I00% tailings
o 0o/o-72o/o primary aggregate (sand)
. 0o/o-72Yo secondary aggregate
o 0o/o-8%oprimarybinder
o 0o/o-8o/o secondary binder
Essentially, the plant is equipped with a system for handling a single wet and sticky material,
two different materials (which are free flowing and can be handled in bins) and two types of
binder. This includes tailings, sand and./or aggregate from any source which is available. The
flexibility of the recipe allows Kidd to use whatever materials are the most economical at the
time that paste is being poured. For example, at lower tonnages when all tailings are being
trucked to the Kidd mine site, it may be more economical to use a low percentage of tailings.
'When
the rail haulage of the tailings has been installed, the delivery of tailings may be more
inexpensive than the delivery of sand and a higher percentage of tailings may be beneficial.
Another advantage to the flexibility of the recipe is that waste materials such as crushed waste
rock from the old open pit at the Kidd mine site can be co-disposed of before closure of the
Kidd mine site to reduce clean-up costs.
It is generally agreed by (most) experts in the field of paste technology that running a
distribution pipeline under fuIl flow or choked flow conditions is preferable to running under
slack flow. This reduces the velocity of the paste, the production of vacuum and hence the
wear on the system components. The degree to which this general rule should be followed is
much more contentious, however, and runs the gambit from INCO's Garson mine where the
paste is actually floated down the borehole on top of a column of water during initial filling so
that the system is never under slack flow conditions all the way to a gold mine in Red Lake
which produces a significantly higher slump paste than design and fills less than 10o/o of the
borehole (due to the mine's preference for a lower angle of repose in their cut and fill stopes)'
In this case, Falconbridge clearly stated their preference for a full system. Although this
seems simply a matter of adjusting the rheology of the paste to balance the head with the
system losses, the implementation of any sort of rheological feedback system is made
impractical by the long response time from when a batch is prepared and when the effects of
that batch on the system are observed by underground instrumentation.
Instead, the solution which has been adopted at Kidd Creek is to simply overdesign the
capacity of the plant so that the underground distribution system is 'force fed until it chokes''
Although this doesn't sound like the most elegant of solutions, it is an effective method of
instantaneous compensation for changes in the balance of the system. Of course, the rheology
must be within a reasonable range to balance the system at the design tonnage; however, the
rarLge is expected to be quite large since the effect of changing flowrate on rheology (i.e.,
friction losses) is quite substantial at the expected typical conditions. Although this solution
will not completely eliminate slack flow during start-up and upset conditions, it is expected
that it will substantially reduce the occunence of slack flow to the point where wear on the
underground distribution system is minimized'
This solution can be implemented at the Kidd Creek mine with little cost impact since this
ty,pe of plant can be scaled up from a capacity of 250 tph (which is required to meet their
filling schedule) to 400 tph (which allows us a large excess capacity to keep the borehole
choked) with only a small increase in capital. This type of solution would be much more
costly for a plant where dewatering of the tailings and pumping is required.
Blpass boreholes are installed between main levels below 6000 Level so that the paste does
not have to hit each 100 ft between 6000 Level and7400 Level. Bypass boreholes also act as
a backup system so that there are essentially two separate lines below 6000 Level. Blpasses
extend from 6000 Level to 6400 Level to 6800 Level and to 7400 Level.
PASTE PUNT
SURFACE
LEGEND
MAIN LI NE BOREHOLE/P I PI N G
AUXILIARY BOREHOLE//PIPING
LEVEL PIPING
@ CAMERA
38 LEVEL
46 LEVEL
KIRUNA RAl,,lP
52 LEVEL
56 LEVEL
60 LEVEL
64 LEVEL
68 LEVEL
75 LEVEL
as paste flows through the system faster than it enters the system. Until the pipeline is
full (in particular, the longer horizontal piping at the discharge into the stope), the
system will experience areas of free fall or slack flow. This perfect vacuum combined
with the velocity of the paste results in pipe vibration and high wear at the transition
from uphole to downhole. Extending piping out onto each level ensures that the paste
is backed up into the borehole and the wear is not concentrated on a single transition
point. In addition, since the rheology of the paste will be changed by the operators
(depending on the location of the pour), the height of the paste in the borehole will
vary which will distribute the wear along a greater length of the borehole.
2. Friction losses were predicted during a number of flow loop tests. However, flow loop
tests can only be relied upon to give a good estimate of the friction losses. This, and
the fact that the recipe used in the plant can be changed substantiall5 means that the
system must be flexible enough to handle a wide range of rheological properties. To
accommodate this, the design of the underground distribution system will be fine
tuned during the operation of the plant by extending the length of the horizontal piping
so that the boreholes are kept 'reasonably' full even under start-up conditions. Of
course, the level in the boreholes will change depending on the rheology of the paste
which will, in turn, depend on the location of the pour. However, in general,
boreholes will be kept as fuIl as practical during the filling of the system.
3. Placement of instrumentation, pressure relief spools and cleanouts is much more
practical when they are placed in low wear areas. The cascade system allows us to
place these essential items in a low wear environment (the horizontal level piping) as
4. Gonclusions
The paste system at Kidd Creek Mine is vital to the success of the Mine D operation. The
system which is being constructed at Kidd Creek takes advantage of the availability of
dewatered tailings and an existing rail link to overcome some of the challenges presented by
the large distance between mine and tailings source. In addition, the paste system is designed
to give the operators maximum flexibility to use whatever materials and process routes are
With commissioning due to be completed in August of 2004, it is hoped that the results of the
start-up and operation of this system will be of interest at future paste conferences.
Paste 2004
Cape Town
South Africa
PaPer 25
R. Tenbergen, P.Eng.
Senior Process Engineer, Golder Paste Technology Ltd.
D. Landriault, P.Eng.
President, Golder Paste Technology Ltd.
Abstract
The process for selecting the type of a mine backfill system often begins and ends with the
higher capital cost of a paste backfill plant, and the selection of a conventional slurry system,
even though, in many cases, it can be shown that the Net Present Value (Ni-PV) for a paste
In terms of shareholder retum, mine personnel must look simply beyond the initial capital
cost and towards a value-added approach that is based on basic proven improvements to the
overall mining operation provided by paste backfill. Paste backfill ty,pically provides an
filling rate and reduced dilution handling. Lastly, higher quality fill means less backfill
(
dilutiorland more ore to be hoisted, which positively impacts on the net smelter return.
Also overlooked is the global effect that a paste backfill plant has on the overall operation
when also utilised for surface disposal. Cost reduction in dam construction, operational
Comparative Economic Assessment of Mine Backfill Systems Page2
F. Palkovits, R. T and D. Landriault
improvement in tailings management facility, higher placement density in backfill and surface
disposal, and improved water recovery are all possible with the use of paste technology.
This paper illustrates a simplistic approach for evaluating the NPV of cemented and un-
cemented hydraulic slurry backfill, and cemented paste backfill. The approach is based on a
typical bulk mining operation and incorporates mining and backfill costs, cycle times,
dilution, production rates and net smelter return (NSR).
1. lntroduction
Often the payback for a capital project must be shown to occur within 12 to 24 months before
will be considered by management, and some projects with high perceived risk
that project
(with new or unfamiliar technology) have little chance of getting approval, even when the
economics appear to be overwhelming. While a cement paste backfill system is considerably
more capital intensive, there is now good evidence to show a positive effect on the NPV.
Historically, the main drivers for paste have been the high cost of binder for slurry backfill
systems, and potential reduction in dilution provided by an "engineered backfill". Considering
that binder alone is up to 80o/o of the cost of backfill, a reduction of up to 50Yo in binder
consumption and operating cost is significant, though pales in comparison to the payback
resulting from reduced backfill dilution alone. Other major improvements, as noted
elsewhere, include overall mine productivity through reduced dilution, stope cycle time, and
worlace conditions. Significant amount of water and slimes from hydraulic fill systems can
More recently, paste technology has successfully proven to dramatically improve the
economics and environmental aspects of surface disposal of tailings. The elimination or
significant reduction in large tailings dams, cyclone stations, water return systems, water
treatment systems and all associated services, roadways and power systems can cut tens of
millions from project development costs.
In this paper, we present a holistic approach to the following case study to demonstrate how
paste technology for underground backfill and surface disposal can dramatically impact the
NPV. The calculation of the NPV uses current operational and cost data gathered from
projects in Latin America.
The case study includes an underground polymetallic mine and milling complex in
Latin America, schematically shown in Figure 1. It is developed in and alongside the
slope of a mountain, with several access adits situated up the slope. The uppermost
level is 3900 m, with mine offices on surface at 4000 m. Mining extends below the
valley floor to 3400 m. The mill is located on the valley floor at 3500 m altitude and
about 5 km from the main mine offices.
PROPOSED DEWA'-ERING
PROCESS PLANT
Proven and probable polymetallic ore reserves are stated at 15M tonnes, containing
zinc-lead-copper sulphides with gold and silver values and bismuth. Similar tonnages
are suggested as inferred resources. Design criteria is given in Table 1.
Current mining methods include mechanized cut and fill, post-pillar cut and fill and
short-hole up-hole mining. Ore zones vary from 2-3 metres wide, sub-vertical zones to
t
large 80-100 metres wide, steeply dipping massive zones. Ore grades are variable, but
an average of 5.0Yo zinc,3.0o/o lead and 0.28% copper will be assumed. Production
stopes are spread out laterally over 2 kr and over 500 metres vertically, with over
50% of the production coming from the post-pillar cut and fill mining in the Main Ore
Zone.
with piston-diaphragm (PD) pumps. A booster PD pump is situated part way up the
mountain slope to provide fill to the upper areas of the mine. Approximately 45o/o of
the total tailings by weight end up in the stopes as backfill. Although cement is not
currently used in the classified tailings backfill, potential changes to a bulk mining
method would require the use of cement. Pillars would be in rock or low grade
The mill is operated 345 days per year, with a nominal throughput of 3,600 tpd. The
process includes crushing, grinding, flotation, tailings discharge and vacuum filtering
Fines from the cyclone station are transported by gravity pipeline to the TMF. In this
example, less than 18 months of capacity remains so that timing to complete
permitting and approvals and construction for a new TMF is critical' A possible new
TMF site has been identified in an adjacent valley system and this, combined with the
underground voids, will have sufficient capacity for the 15 Mt ore reserve, equating to
9Mm3 volume. The selected tailings transport pipeline route will cross over the
underground operations at 4000 masl, and will be suitably positioned to allow the
tailings to be used for underground backfill. The relative elevations and mountainous
terrain limits the storage capacity of a dilute slurry TMF and requires considerable
investment in high-pressure pumps/pipeline systems. Slurry TMF requires major
containment dams, storm water diversion systems, and pond water decant systems.
3.
Three common backfill systems and surface disposal systems are presented in this case study,
as follows:
The design criteria sunmary table below is used as the basis for the assessment.
5,o0o,o0o m'capacity
Un-cemented backfill systems necessitated the use of ore pillars to provide support and
contain unconsolidated backfill, resulting in lower ore recovery. Cemented fills opened
the doors to changes in mining methods; in this case replacing post-pillar cut and fill
mining method with bulk stoping. SubJevel stoping may replace short-hole up-hole
mining.
4. Options Evaluation
CGREGAIE EIWR
WOERGROUrc
Hydraulic fill requires relatively larye quantities of water for pipeline transport
An input of 80% is used for ore recovery in this model. This assumes that the main ore
zone contains a significant portion of the mineable reserves and will continue to be
mined by post-pillar cut and fill mining. Though backfill dilution would continue from
the floor during ore removal, post-pillar cut and fill mining permits waste and low
grade sorting in the stope to reduce internal sources of dilution.
Cement or other binders (i.e., fly ash, blast fumace slag) are added to hydraulic fill to
permit a greater recovery of ore reserves by providing stable, free-standing fill for
regional ground support. Up to 50Yo more binder is required than with paste due to the
poor water:cement ratio. The percolation water drained from the filI carries cement and
other fines affecting safety and mine dewatering facilities. Beaching and segregation
occurs within the stope, leaving bands of tailings with high and low binder content'
and a poorer bulk fill quality. For this example, 15 wto/o fill dilution is used; tlpical
used in
experience suggests l0o/o and 25% by weight. Ore recovery value of 98o/, is
this economic model.
presently there are over 30 underground paste backfrll plants in operation globally
ranging from 20 tonnes per hour up to over 360 tonnes per hour' Though we see a
recent increase in interest, the slow response by industry is due to perceived risk and
to
higher capital costs. A quality engineered backfill means less backfill dilution to muck,
crush, hoist and process, thus substantially lowering the operating cost. As an example,
dilution at one operation using paste backfill was reported by staff to be 0o/o (Zeto)by
weight! Mine staff suggested that substantial savings could be achieved through a
reduction in binder content. It should be investigated, but consider that 1 wt% binder
costs $1.00/t of backfill or $0.50/t of ore (assuming $100it cement). Should binder
reduction lead to 1% dilution then, for 3,400 tpd, the mine would have to muck, tram'
hoist and process 34 torures of backfill. Table 2 illustrates the effect on ore recovery'
annual ore production and life of mine by (a) adding cement to hydraulic backfill,
and
2,890 3,332
Actual Ore Production Rate (d) 2,720
Table 3 illustrates stope cycle time with cemented hydraulic slurry and cemented paste
for a mechanized 25,000 tonne bulk stope operation.
One major savings in cycle time using paste backfill is that the little amount of
contained water is used in cement hydration, and does not require decant systems
within stopes and drain time. Development waste rock can be dumped directly into the
stope at controlled rates, with corresponding reduction in cost and increase in
productivity.
Evaluation of the pumping system to transport mine tailings to the TMF indicates that
tailings thickening prior to pumping would be the most cost-effective and operational
simple solution, even though large piston-diaphragm pumps and high-pressure lines
would. be needed. Typical mass balances for the various streams are shown in Table 4.
4.2.1
Should management decide to continue with hydraulic fill, no change in that system is
required. Thickening of the cyclone overflow is recommended with PD pumps to
transport the tailings to the TMF. A PD pump and pipeline system (installed) is
estimated at US$8.6M.
Thickening of the full plant tailings would require paste backfill for underground
disposal due to the underground fill mass balance. PD pumps would be selected again
and the higher flow rate would require alarger PD pump and pipeline. The estimated
It wilt be shown that the paste option for backfill and surface disposal has the lowest
overall capital cost and, due to related benefits in the mining operation, generates the
4100 m to distribute paste to the TMF and underground via gravity with pump assist.
Deep tank paste thickeners generate a high slump paste underflow that is more
amenable to long-distance pipeline transport, though would require more binder than a
low slump paste. A plant may also be designed around thickened tailings feed and
vacuum filters. At 4000 masl, downgrading of the vacuum system will be required.
Pressure filter systems are considerably more expensive, often prescribed where water
balance is extremely important. Filtered tailings can be mixed to varying shear strength
(slump) to optimise pipeline transport to the stopes or the TMF. The capital cost of the
tailings thickening and transport system, and paste plant is estimated US$21.6M.
A conventional slurry TMF to contain thickened cyclone overflow will require alarge
dam to store the fine tailings or slimes. With a water cover, consolidation time will be
long and the mass will remain liquefiable for a considerable period. A 70 m high dam
would be required and constructed from bonowed materials since the coarse tailings
are required for backfill. Dam construction volume 3.0 Mm3 would be required at
US$10 per m' placed and compacted. With water decant and diversion systems, the
capital cost are estimated at US$33.0M.
Paste may be used to significantly reduce risk and major capital costs by eliminating
large dams in conventional slurry TMF. The primary objective of the tailings
deposition plan is to maximise the desiccation of the tailings by thin layer deposition
and exposure to the environment prior to subsequent deposition. Stacking angles may
be increas ed (5% to l0o/o or more) gaining additional capacity. More expedient
reclamation and closure of the site is expected. A paste or filter cake stacking system
would require a simple 3 m to 10 m high toe berm at the foot of the disposal site to
hold back the flow of paste since there would be no hydraulic minimal loading. Toe
berm and minor water collection facility are estimated at US$2.0M.
STORM WATER DIVERSION
The capital costs are summarizedin Table 5, which show that it is not appropriate to look at
each aspect of the project in isolation. Unforeseen benefits from project specific synergies
may come to light, such as in this example.
The operating costs as shown in Table 6 have been collected from previous projects
and compiled to provide a reasonably accurate assessment of the options.
The NSR shown in Table 7 is the value of the concentrate after deductions for
concentrate transportation, smelting, refining, insurance, marketing, penalties and price
participation. The capital expenditures are assumed to be completed in the first year.
The evaluation can simply be shown as:
LME
Miil Head Gr. x Recovery
Item Grade Recovery
x LME ($/t)
($/kg or oz) (%l
Znc s30% 0.93 90 49.08
The NSR and capital and operating costs for the three options were incorporated into the NPV
calculation in Table 8. The un-cemented hydraulic filVcyclone sluny TMF is the most
expensive option and provides the lowest NPV. A factor to reflect the risk of a conventional
TMF within a highly active seismic location coupled with extensive historical knowledge of
dam failures may be applied by investors for such projects.
Surface Disposal 5 5 8
Underground Backfill 10 41 2l
NP Cost (Capital and Operating Cost) 321 347 257
8. Gonclusions
The intent of this paper was to provide a realistic assessment of paste technologies compared
to conventional slurry system for both underground backhll and surface disposal. We have
hoped to have dispelled the notion that paste is a high capital cost system for questionable
benefits and economic retums. Each project or operation is unique and an evaluation that
investigates the costs, process and downstream implications should be conducted to confirm
the best approach.
References
Graham, C.B. and Morrison, D.M. (2003) "Technology - Is it Working For You?", 2003 CIM
Tenbergen, R. (2000) "Paste Dewatering Techniques and Paste Plant Circuit Design", Minefill
2001, Fort Collins, USA.
Landriault, D.A. (1995) "Paste Backfill Mix Design for Canadian Hard Rock Mining" in
Proceedings of the 97th Annual Meeting of the CIM Rock Mechanics and Strata Control
Session, Halifax, Nova Scotia.
Landriault, D.4., Verburg R., et al. (1998) "Short Course Notes on Paste Technology for
Underground Backfrll and Surface Disposal Applications", Golder Paste Technology
Worlrshop, Tailings and Mine Waste '99,Fort Collins, USA.
Henderson, 4., Newman, P. et al. (7997) "The Cost Advantages of Using Paste as a Backfill".
Hassani, F.P., Bois, D., et al. (1992) "Economic and Technical Feasibility for Backfill Design
in Quebec Underground Mines", Quebec Mines Survey, CRM Report 71226002.
Goldie, R. and Tredger, P. (1992) "Net Smelter Retum Models and Their Use in the
Exploration, Evaluation and Exploitation of Polymetalic Deposits", Geoscience Canada,Yol
18, No.4, p 150-171
Golder Associates (ttK) Ltd. (1999) "Optimisation of Backfill Operations at Madem Lakkos
and Mavres Petres Mines (Stratoni Operations)" Ref: 99522653'
Ley, G.M.M., Steed, C.M., Bronl*rorst, D, Gustas, R. (1998) "Mining under Backfrll", CIM
Bulletin, Vol9l, No. 1020, p 65-71.
Robinsky, E.I. (1999) "Thickened Tailings Disposal in the Mining Industry" Toronto,
Ontario, Canada: E.I. Robinsky Associates Ltd.
Verburg, R.B.M. (1997) "Environmental Benefits Associated with the Use of Paste for
Canada.
Paste 2004
Cape Town
South Africa
PaPer 26
2004
lnternational seminar on Paste and Thickened Tailings: Paste
International Seminar on Paste and Thickened Tailings
Paste and Thickened Tailings - Paste 2004
G. J. van Ryssen
Abstract
to
There has been a major thrust within DeBeers and Debswana diamond companies
reduce water consumption associated with their processing of kimberlite.
In many of the
residues with
newer plants, finer crushing to liberate more diamonds has resulted in finer
streams generated
an increase in water consumption. Co-thickening of the grit and slime
high rate
from this process has been pursued to improve water recovery in the plant using
thickeners. Such a system was implemented at Orapa Mine' The disposal system
used
walls with
the presence of grits in the slurry as an opportunity to replace impoundment
cycloned walls. This case study looks at some fundamental issues associated
with the
and their impacts
design of a new plant with co-thickening and cycloned disposal facility,
1. lntroduction
following
The decision to combine the slimes and grits for disposal was based on the
argument;
o the traditional
better overall water consumption would be achieved compare to
separate disPosal sYstems,
one residue disposal facility with reduced land use would be cheaper
o to
These questions will be dealt with in the paper, for this particular case study. The results
3. Design lnformation
For the expansion project at Orapa, an existing, fully operational No'l Plant provided
good information on the residues (compared to a new mine) on which a new No'2 Plant
could be designed. The following details are relevant:
70 6 >1,6mm <25mm
Coarse Tailings
77 20 >0,3mm <1,6mm
Grits
13 178 (RD:l,28) <0,3mm
Slimes
In the design of the new No.2 Plant, production was to be doubled, but the new plant
would incorporate new technology not operating in the No.1 Plant' This new technology
high rate
included amongst other equipment and changes; a recrush circuit, scrubbers and
thickeners. Both the re-crush circuit and scrubbers were expected to increase the ratio
of
fines to coarse. Asister mine had a similar plant and therefore the prediction of the
increase was based on operating experience. A high rate thickener had been
installed at
the No.l Plant some years previously and a number of trials had been run with
various
piston
ore types, grit and slime ratios. This had extended to pumping trials, including
thickener
diaphragm pumps. For the expansion project, a second (different) high rate
was installed. once again, a number of trials were run to determine the thickener
performance, generate more rheol ogy data, and cyclone performance' This information
pumping
was used to generate the design parameters for the No.2 Plant slurry thickening,
and disposal system.
When the No.2 Plant was commissioned it was soon apparent that the slurry composition
was quite different from that projected'
<1,6mm
Slimes 45 to 50 (RD:1,30 - 1,40)
The table shows clearly that the amount of slimes generated was far greatet than
expected. The expected grit:slime ratio of 50:50 was in fact in the range of 25:75 to
type. This impacts directly on the thickener performance and
30:70 depending on the ore
The re-crushing and scrubbers generated much more slime than expected. The crushed
ore enters the scrubber with alarge volume of water, and by inter-particle abrasion, the
gravel is ground down to finer particles. The scrubbing action generates more ultra-fines
than crushing. Kimberlite is well known to contain smectite clays which are exposed in
the process. These are active clays and readily swell, absorbing water which they do not
easily release.
'When
These clays continue on to the thickener. the thickening trials were run, the slimes
fed into them were not generated from a scrubber, and therefore did not have the same
composition as those to be produced in the No.2 Plant. In the trials the effect of the clays
would have been less, and the higher grit content would have minimised their effect.
For No.2 Plant, the lower grit content and high ultra-fine component, resulted in the
thickener achieving underflow densities less than l,30tlm3 and having a reduced through-
put. Through modifications, improved controls and an additional thickener, densities of
l,35tlm3 can be consistently produced. Another type of high density thickener is being
considered to further improve control and increase density'
This impact of the clays on thickening continues to the downstream deposition behaviour
at the dam. These include;
The slurry pumping design work was based on a non-settling slurry behaviour since it
was expected that a slurry with a 50:50 grit:slime mixture at an RD >I,45 would not
segregate or settle outin the pipe and would exhibit slug flow. On this basis low flow
velocities were accepted and pipe blockages after periods of no flow were not expected.
At the new grit:slime of 25:75 and lower RD's of 1,301/m3, the slurry rheology changed
to a settling, segregating slurry, requiring higher flow velocities. ha addition the holding
tank and pumps would now have to cope with much larget volumes of slurry'
.- /
ta
'.
Instead of one pump set and pipeline in operation at a time, up to three were required to
cope with the volumes and limited tank capacity with an associated increase in energy
consumption. The impact on dam construction and management becomes difficult under
such circumstances.
the centrifugal pumps required to pump the slurry were close to the limit of centrifugal
pumping. Therefore their cost and energy demand are akeady high for centrifugal
pumping as are the costs for the high pressure pipelines and gland water system.
From the cyclone testwork, a mass recovery to cyclone underflow in excess of 55% was
expected. A mobile cluster cyclone arrangement was installed made up of five 250mm
cyclones each. Twelve of these were spaced around the dam wall, each capable of
receiving the full plant slurry production. It was intended that this would provide great
flexibility in operation. The walls were to follow the centre-line/downstream
construction method up to a design height of 30m with 18 degree slopes with a 9m high
underdrained starter wall. Due to the high mass recovery, wide walls were planned,
rising at arate of 5m/yr while the basin rose at 3m/yr. It was recognised at design stage
that cycloning a high density slurry would result in some short-circuiting of fines to the
underflow, that would inhibit consolidation. This was to be monitored and managed
accordingly.
While cyclones prefer lower density slurries for solids separation and dewatering, the
lower RD alone did not improve cyclone performance. The 250mm cyclones were
exposed to a high range of variables during the first 3 years of operation, often having to
operate outside of their optimum ranges;
a water hammer due to frequent stop-starting of the pumps which damaged the
vortex finders.
The result was that the cluster units were scrapped and replaced with one hundred
350mm diameter single cyclones.
operation, the feed pressure could be controlled. The larger cyclones were selected since
they are able to recover 40% by mass to underflow (the minimum required for stable
walls) with some ultra-fines over a greater range of feed (density and composition). With
smaller cyclones the recovery would have been lower and they would not have been able
to cope with large feed ranges. Therefore the quality of the underflow may have been
worse than for the larger cyclones.
The larger cyclones proved to cope well with the variations and tlpical gradings
shown in Figure 2.
Figure 2: TypicalOrapa Slurry/Slime Gradings
100
90
80
70
o
.=
6 60
o
o 50
o
o-
o
. 40
I5
E
o
+cyclone Underflow
+Cyclone overflow
+No.1 Plent - Slime
slurry Feed
The cyclone overflow was finer and more dilute than anticipated, with a latget volume
reporting to the basin. This resulted in a significantly higher basin rate of rise (6 to 7mlyr
versus 3m/W in the design), largely attributed to a lower basin dry density (0,681/m3
versus l,2t/m3 in the design). To maintain freeboard, it was only possible to construct the
cycloned walls in an upstream manner. This would have contributed to the high rate of
rise. The basin remained as a thickened slurry or gel, with little to no consolidation
taking place, and a huge volume of water being locked-up and lost.
The wall dry density was also lower than anticipated, l,40tlm3 versus l,60lm3 in the
design, mainly due to inadequate dryrtrg time and consolidation.
8. Wall Stability
Clearly upstream construction of the wall over an unconsolidated overflow material with
an underflow material having compromised consolidation properties of its own was not
going to result in a stable wall to the design height of 30m. It was only throug good
operational management of the cyclones that freeboard was maintained. However the
inevitable signs of stress in the walls appeared, with seepage at elevated levels on the
walls. Following a series of large storm events (330mm within 3 weeks), seepage
became widespread around all walls and the stability of the dam was clearly under threat.
' ':
9. Water Gonsumption
Despite the high lock-up of water in the basin, the dewatering of the cyclone underflow
and concentration of water into the basin, 40Yo or more of the water deposited on the dam
was returned for re-use. However water consumption increased because of the lower
slurry densities or poor water recovery in the plant.
Comparative water consumption figures for four plant residue systems are as follows:
Table 4: Water Consumption Comparison
Comparative Factors No.l Plant - No.2 Plant - No.2 Plant - No.2 Plant -
Actual Conventional Design Actual
Average water
consumption 0,31 0,79 0,39 0,66 to 0,90
Dewatering of residues
and destination
impoundment impoundment
deposited)
Obviously the No.l and No.2 Plant water consumptions cannot be directly compared
because the processes and residue products are different, but clearly the No.2 Plant
design needed to consider new methods to reduce water consumption from 0,79m3ltto a
lower value. This provided the motivation for co-thickening of the grit and slime. Water
consumption as low as 0,31m3lt for the No.2 Plant would be difficult to achieve, but
0,39m3ltwas acceptable at design stage for the co-thickening system.
However, the actual lower grit:slime ratio and increase in ultra-fines in the No'2 Plant
meant that the water consumptiontarget had to be revised to 0,66m31t. The impact was
that the wellfield capacity had to be increased to sustain demand and that the life of the
finite aquifer may be shortened. Alternative reliable water sources have to be sought to
guarantee supply for the mine's long life.
10. Gosts
When a system does not operate according to design, there are costs to modify it in
addition to difficult operating conditions. The system was re-examined and original
decisions questioned to see if it is the most economic system, and trade-off studies of
alternatives were undertaken. The outcome of some of these studies was as follows:
o the ori$nal cycloned slurry dam capital cost was split equally between a) the
pump station and pipelines and b) the entire dam infrastructure.
o the new cyclone slurry dam capital cost was made up of a) two thirds for the
pumping and pipelines due to a booster pump station and the dam being
further away from the plants and b) one third for the dam infrastructure.
o a 30 year trade-off study between the current No.2 Plant system and a
traditional residue disposal system (de-grit plant to separate grits dump, and
thickeners to impoundment slimes dams), showed that the costs were similar,
but did not include increased water supply or closure costs.
o the new dam was required 6 to 7 years premature of its planned date,
Projects have been initiated to research alternative disposal systems for the future. The
upside for the mine was that diamond liberation at the No.2 Plant as a result of the re-
crushing and scrubbing increased significantly compared to that achieved in the No.l
Plant.
'41. Lessons
This case study is a clear example of where an increase in fines in the residue and in
particular the ultra-fines, had significant impacts on the entire disposal system and on
the water supply system.
The extremely high cost of slurry pumping and distribution for the new slurry dam
compared to the original dam illustrates that disposal systems need to be evaluated for
the life of mine and as complete systems. Thickening, transportation, disposal and
closure should not be looked at in isolation.
Even with so much knowledge of the ore; all the operational experience of existing
plants, processes and equipment; and large scale tnal data for the new plant design;
the composition and properties of the slurry were not correctly projected or
understood. In particular the effect of scrubbers on ultra-fines generation and water
absorption was not foreseen.
Large scale trials can lead to incorrect results if they are not run on the corect ore or
feed. For the high rate thickener trials, the ore type was harder than normal; the feed
contained a higher grit content than the No.2 Plant would have produced; and the
slimes had not been generated from a re-crushing or scrubbing process.
',2. Gonclusion
The two objectives of the No.2 Plant slurry disposal system were to minimise water
consumption and reduce disposal costs. Clearly the water consumption tatget was not
achieved and this was directly related to re-crushing and scrubbing generating more slime
and smectite clays with significant downstream impacts, as well as unexpected wellfield
impacts. The mine has not only spent money modifying the residue handling system, but
has had to build a new dam 6 years earlier than planned, at great cost. Once again this
objective was not achieved.
On the positive side, 40%o or more of the water deposited on the dam has been returned
for re-use, as planned. Now that No.2 Plant operations have stabilised and once the
second slurry dam is fully operational, it should be possible to correctly assess whether
the co-thickening system was the correct disposal solution for the mine.
International Seminar on Paste and Thickened Tailings
Paste 2004
Cape Town
South Afric
PaPer 27
1.1 INTRODUCTION
Osborne Mine
Osborne has been operating since 1996 and generates approximately 112,000
tonnes of tailings per month. The majority of water loss in the process water circuit
occurs in the tailings system as a result of evaporation. Thickened discharge
provides a means for significantly reducing water losses since, by putting less water
on to the tailings facility, there is less opportunty for evaporation.
Against the background of the above potential benefits Osborne elected to carry out
trials to assess the efficacy of the implementation of thickened discharge.
Speciflcally the following issues required detailed evaluation:
. The method of generating thickened tails and the reliability of this method
. Pumping implications
. The beaching profile of the thickened tails, specifically whether it would stack at a
steep enough angle to enable a reduction in lift requirements'
Gordon McPhail, Alasdair Noble, George 2004 tnternational Seminar on Paste and Thickened Tailings
P apageorg iou, Dan W ilkin son Development and lmplementation of Thickened Discharge at
Osbome Mine, Queensland, Australia
o The practicality of management options for dealing with periods when the process
plant is unable to generate thickened tailings
. Geotechnicalstability
. Erosional stability
1.2.1 Mineralogy
The tailings comprises mostly magnetite but contains a number of sulphide minerals
notably pyrite with traces of chalcopyrite together with gypsum. Sulfides make up 3
to 4% of mass of the tailings. The sulfides are moderately reactive and to counter
their acid generating capacity lime is added in the course of mineral processing. The
gypsum is a result of the neutralisation reactions.
Over the long term the tailings has significant potential to generate acid although
measurements to date both in the lab as well as in the field indicate that the reaction
rates are slow.
Gordon McPhail, Alasdair Noble, George 2004 lnternational Seminar on Paste and Thickened Tailings
P ap ageorgiou, D an W il kin son Development and lmplementation of Thickened Discharge at
Osborne Mine, Queensland, Australia
As a result of the high iron content of the tailings the particle specifc gravity is 3.4 to
3.6.
100
s
z ?L-
&)
70
Ktr4,t
F
60
s
10
30
l0
0
0mf ool
Plrdch st{nm)
l+wg3+wM - wgn
1.2.3 Slurrycharacteristics
Slurry testing was conducted by Paterson and Cooke Consulting Engineers in the
pipe loop located at Alrode in Johannesburg. Tests were conducted a range of water
contents from 50% solds to 74% solids (slurry relative densities of 1.56 to 2.12
respectively) to determine settling rates, friction characteristics and pump de-rating
factors. Figure 1.2-2 shows some of the results of the testing.
Page 4
Gordon McPhail, Alasdair Noble, George 2004 lnternational Seminar on Paste and Thickened Tailings
P a pageorgiou, Dan Wilki n son Development and lmplementation of Thickened Dscharge at
Osborne Mine, Queensland, Australia
,i
t
[.i
rr
I
f :r
*t
a
,,
fFo
-o ti
qt
r;l1 n
;'fr tl Ur ft tr,
'-hE!
t lE tH + f Q.lV f .: ' r t td cr^* ri.'i't1i
pre June 2000Tailings was pumped to old tailings storage facility in purely
June 2000 The concept of high density tailings was frst discussed and a program
decided upon to investigate its potential.
July 2000 Control loops were tuned through entire circuit from grinding, through
flotation to tailings to allow more stable operation and allow targeting
of higher densitY.
Sept 2000 JKSimmet models were developed for tailings cyclones and smaller
rubber spigots trialled to increase cyclone underflow density and
increase the amount of solids feeding the thickener. Rubber spigots of
first 57mm then 51mm trialled. Cyclone underflow densities with the
various spigots were as follows:
Page 5
Gordon McPhail, Alasdair Noble, George 2004 lnternational Seminar on Paste and Thickened Tailings
Oct 2000 Feedwell deflector cone mounts were modifled to allow for a larger
spacing. The attaching rods were extended allowing a gap of 280mm
compared to 140mm. This allowed more material to be fed to the
thickener by preventing the feedwell from boiling over.
Nov 2000 Steadier operation of the thickener and operating with higher
flocculant dosage and a higher bed level enabled the thickener
underflow density to be increased from around 55% solids to around
67% solids.
Dec 2000 A Bredel pump was installed in the thickener underflow line to prevent
blocking of the U/F line and allow operation with higher density
undefflow stream.
Jan 2001 Pipe loop test work was conducted in Johannesburg with Patterson
and Cooke to determine characteristics of pumping Osborne tailings.
Specifically to quantify the pressure drop per unit metre at a variety of
densities, to measure the settling and deposition velocities and to
measure the beach angle formed by deposition at various densities
and the segregaton of solids during deposition. Pipe loop tests were
conducted at slurry densities of 620/0, 690/o,73o/o,74o/o and 76% solids.
Deposition velocities were determined to be in the range of 1.5-
Feb 2001 Modifications were made to the thickener underflow cone to remove
obstructions and smooth the flow allow the Bredel pump to be
removed and allowed a return to gravity flow of the thickener
undelow material.
Feb 2001 Tailings was being pumped at a density of around 68% solids which
allowed pumping to the far corner of the old tailings storage facility
after the last wall lift, something that had been difficult to achieve
previously. Surveys of the slope deposited at 68% solids showed that
the beach angle was approximately 1 in 40, compared to slopes of
around 1 in 90 with 55% solids deposition'
Gordon McPhail, Alasdair Noble, George 2004 lnternational Seminar on Paste and Thickened Tailings
Papageorgiou, Dan Wilkinson Development and lmplementation of Thickened Discharge at
Osbome Mine, Queensland, Australia
Apr 2001 lnstalled higher ranging bed mass gauge in thickener to allow more
stable and reliable operation at higher bed masses. lnstallation of
density gauges in thickener underflow line and in tailings pipeline to
allow better monitoring of operation.
improvements, the pumping and piping duties for the new tailings dam
were specified and the design of the new system commenced' The
design of a new dam to incorporate high density deposition
commenced.
Sep 2001 Design criteria for pumping and piping system were fnalized with pipe
diameter selected to suit high density deposition. Valving, flushing
systems and instrumentation were installed to suit high density
pumping.
Oct 2001 A change of flocculant supplier and type to better suit duty allowed
significant increase in thickener underflow density.
Figure 1.3-l: Isometric view of the old tailings facility showing the thickened mounds in
the corners.
Page 7
Gordon McPhail, Alasdair Noble, George 2004 tnternational Seminar on Paste and Thickened Tailings
June 2002 The new tailings dam was commissioned, initially pumpng to
conventional spigots at around 68% solids' The reduced elevation of
the dam, the reduced pumping distance, the higher density and more
approprately sized pipeline all lead to a reduction in pumping duty
from in excess of 1800kPa with three pumps in series to one pump
only with pressures of around 400kPa.
Aug 2002 Rubber cyclone spigots of first 47 then 41mm were trialled to further
increase cyclone underflow density and overall tailings density'
Following trials a mix of 41mm and 45mm ceramic spgots has been
installed to allow operatonal changes to suit changing ore types.
Sep 2002 High density trial planned and location selected. Trial beach fitted with
pipe work, monitoring instruments etc.
Oct 2003 High density deposition trial commences at spigot "SD1". Densities of
around 74-760/o solids targeted.
Oct 2003 lnitial trial phase concluded with measurement of rainfall erosion
potential, liquefaction potential and final slope.
post Oct 2003 High density continues at SD1 and the new hgh density spigot ND2 to
maximize tonnage of tailings placed at high density while analysis of
trial continues and submission to the regulators is drawn up for
approval of high density tailings as a long term strategy'
Feb 2004 Plan to install a flow meter in thickener underflow pipe, as almost
invariably any problems with tails line beginning to block and causing
problems originate from a loss of flow from the thickener underflow'
Details of the thickened trial on the new dam are described below.
Page
Gordon McPhail, Alasdair Noble, George 2004 lnternational Seminar on Paste and Thickened Tailings
Papageorgiou, Da n W il kinson Development and tmplementation of Thickened Discharge at
Osbome Mine, Queensland, Australia
Figure 1.4-1 below show a series of isometric views of the thickened trial on the new
tailings facility. The isometrics are produced from detailed topo surveys'
Use was made of a natural ridge within the new tailings facility to form a launch point
from which to advance a single discharge pipe. As tailings beached and filled to the
pipe end additional lengths of pipe were added and this discharge point advanced
forwards. ln this way an advancing cone was formed. ln addition, with each
advance, the pipe discharge elevation was raised so that as the cone advanced the
advancing face was at a rising elevation. This, together with the natural fall in the
topography enabled the formation of a cone face of approximately 10 m in vertical
height from the end of the beached tailings to the discharge head. To control the
deposition direction a 60 degree "Y" section was introduced in one pipe length before
the discharge point and discharge alternated between the two branches.
Figure 1.4-1 lsometric views of the thickened trial on the new tailings facility
=
Gordon McPhail, Alasdair Noble, George 2004 tnternational Seminar on Paste and Thickened Tailings
Figure 1.4-2 below shows a cross section through the mound drawn at natural scale.
The slopes of the mound are 1 in 20 for the upper half of the beach length and 1 in
30 for the lower half giving an average beach slope of 1 in 25'
280
270
2
100 200 300
Figure 1.4-2: Gross section at naturalscale through the thickened trial mound
Figure 1.4-3 below shows the measured beach profile calibrated against the beaching
prediction methodology by McPhail [1] and
Gordon McPhail, Alasdair No.ble, George 2004 tntemational Seminar on Paste and Thickened Tailings
100.00%
90.00%
80.00%
o
E
o 70.00%
o
o 60.00%
o 50.00%
o
E 4A.00to
o
o 30.00o/o
o
o. 20.0Oo1"
'10.00o/o
0.00%
0.001 0.1 1 l0
Partcle slze (mm)
274
^o 272
270
co 268
E zoo
fr, z
262
260
+Calibrated
-Mound
Figure 1.4-3: Galibrated and measured beach profes
Page 11
Gordon McPhail, Alasdair Noble, George 2004 lnternational Seminar on Paste and Thickened Tailings
P a pageorgiou, D a n Wilki nson Development and lmplementation of Thickened Discharge at
Osborn e M in e. Queen sl and, Austra I i a
100.00%
90.00%
80.00%
Eq 70.00%
o 60.00%
o
o
o
o
50.00o/o
40.00%
o
30.00%
o
20.00v.
10.00o/o
0.00%
0.001 0.1 10
Figure 1.4-5 and Figure 1.4-5 below show two cross sections through the thickened
mound. The locations of piezocone soundings are also indicated on the sections.
+
'fF
4
U
Figure 1.4-5: Section through the thickened mound from the outer confnng
embankment on the left to the pool wall on the right (Vertical scale exaggerated)
D
T
E
o
i
II
Figure 1.4-6: Section through the thickened mound along the delivery line route
(Vertical scale exaggerated)
Page 12
Gordon McPhail, Alasdair Noble, George 2004 lnlernational Seminar on Paste and Thickened Tailings
Approximately 270,000 tonnes of tailings had been placed in the mound at the time
of the survey. At an average measured in situ density of 2.2 Vm3 this represents
approximately 127,000 m3 of tailings fill.
Figure 1.4-5 below shows the stage-capacity curve for the thickened tailings mound
based on actual deposition tonnage records. The stage-capacity curve is developed
for an advancing cone arrangement and focuses on advance length to calculate
slope length, marginal rate of vertical rise and total height.
FwrMHsl F-v"r,**l
250 250
zoo 2@
150 5 rso
100 ;t
tm 90 80 70 60 50 40 30
loo.ooo 150,000 200,000 250.000 300,000
29Jen{3
28-Feb{3
30-Mar{3
29-Apr{3
29-May{3
28-Jun{3
28-Jul{3
27-Aug-O3
26-Sep{3
F*Pr"d*tb;l
Figure 1.4-7: Stage'capacity curve for the thickened mound
It is particularly noteworthy that the vertical marginal rate of rise is generally of the
order of 20 m per year. This has been achieved with no evdence of slumping,
cracking or excessive seepage at the advancing toe.
To achieve the above special care was taken to ensure that only thickened tailings
was discharged on the trial mound. A by-pass arrangement that was triggered as
soon as the density dropped below a slurry relative density of 2 (70% solids) was
incorporated into the flow control system. Flushing was kept to the minimum
Page 13
Gordon McPhail, Alasdair Noble, George 2004 tnternational Seminar on Paste and Thickened Tailings
required to ensure that the delivery line would be clear enough to erode free on re-
direction of the slurry to the trial area.
On average, over the duration of the trial, the slurry density has been maintained at
an average of 2.06 (72% solids) with regular excursions to 76% solids.
1.4.5 Piezometricmeasurements
The mound has been instrumented using both standpipe piezometers as well as pore
pressure transducers. The latter have experienced zero drift or failed scorng yet
another victory for simplicity. The piezometers comprise PVC pipe slotted over the
bottom 1m and covered with a geofabric sock. Figure 1.4-8 shows the locations of
the piezometers and Figure 1.4-9 shows typical standpipe piezometer readings'
c
/
Figure 1.4-8: ptan of thickened mound showing locations of piezometers (crosses) and
piezometer cone soundings (triangles)
274 0
27t O
273 0
212,O
212 0
211,O
210 0
21t O
270 0
269.0
269 0
2E 0
268 0
261.O
The rise and dissipation of the water levels is clearly evident from the piezometers;
during deposition particular attention was paid to the extent to which piezometer
levels recovered between deposition epsodes as well as to the rising trend in the
dissipated water level. lt is also evident that the tailings coped well with the rates of
rise in excess of 20 m per year.
1.5 TAILINGSGEOTECHNICALASSESSMENTS
ln the course of the trial samples were collected from the tailings beach and
submitted for to
Pretoria University where particle size
geotechncal testing
distributions were determined stress path testing and consolidation testing was
carried out. ln addition field density tests were conducted by local laboratory
personnel. The sections below summarise pertinent results.
A vital issue in respect of the method of placement of the thickened tailings is that of
liquefaction induced by slope failure. lt is common to assess liquefaction potential by
determining whether the material is contractive. Stress path testing in the method
described by Papageorgiou [2] was conducted.
The stress path testing confirmed that the tailings could be contractive if at
sufficiently high void ratio. The tests were conducted on loose handtamped
samples, saturated, consolidated and then tested undrained with pore pressure
measurements. Figure 1.5-1 shows the stress paths for tests at a range of initial
densities and Figure 1.5-2 shows the initial and final void ratios as well as the derived
5116 Paths (p'{ plot)
critical state line.
123
l,*
a
I
Io oB
06
Tailings at stress and void ratio states that plot below the envelope indicated in
Figure 1.5-2 are considered dilatant while those that plot above the envelope are
considered contractive. Tailings that plot within the envelope are considered semi-
contractive.
To assess the likely stress state of an initially slurried sample of tailings undergoing
drained consolidation testing was conducted in a Rowe Cell. This approach allows
the determination of consolidation coefficients and permeabilities at a range of stress
states and also allows the determination of the consolidated void ratio at each stress.
Figure 1.5-3 shows void ratio vs square root of time plots for an initially slurried
sample of tailings from which the coefficient of consolidation and the permeability at
each stress state were determined as indicated in Table 1.5-1
Figure 1.5-4 below shows the consolidation curve for the initially slurried sample.
Page 16
Gordon McPhail, Alasdalr Noble, George 2OO4 lntemational Seminar an Paste and Thlckened Tallings
Papageorgiou, Dan Wilkinson Development and lmplementation of Thickened Discharge at
Osbome Mine, Queensland, Australia
0.(x)
0.20
s
; o.4o
I
!E 0.60
-g
! 0.80
1.00
120
Ro<rtTlm m05
Figure 1.5-3: Void ratio vs Root time curyes from the Rowe Gell
Table 1.5-1: Goefficents of consolidation and permeablties from the Rowe Gell
tests
0.77
50 1.32 7736
100 1.42
200
400 7347 1.0 x10''
Page 17
Gordon McPhail, Alasdair Noble, George 2004 lnternational Seminar on Paste and Thickened Tailings
Papage org iou, Dan Wil ki n so n Development and lmplementation of Thickened Discharge at
Osbome Mine, Queensland, Australia
The following points are noteworthy given the flne-grained nature of the tailings:
. The coeflcients of consolidation are high, indicating that the tailings drain well.
This is in line with insitu observations.
. The permeablities are relatively high - approaching that for a fine graned sand
at low stress levels.
p'=lou'. (+2Ko)ll3
where u' is the vertical effective stress in the odometer and Ke the coefficient of
earth pressure at rest which is equal to 0.425 for an effective angle of friction of 32
degrees, a value representative of the tailings based on past testing..
14
't2
p'(kPa)
Page l8
Gordon McPhail, Alasdair Noble, George 2004 lnternational Seminar on Paste and Thickened Tailings
Pap ageorgiou, Dan Wilkin son Development and lmplementation of Thickened Discharge at
Osborn e Mi ne, Q ueensland, Au stral i a
o The slurried sample is in the dilatant range stress state from the outset of the
consolidaton test.
. The slurried sample enters the semi-contractive range at a p' of 80 kPa (u' = 130
kpa which at a density of 2.2 Um3 is approximately 6m of tailings)
. Since the maximum tailings slope height will be less than 37.5m the tailings is
unlikely to cross the extrapolated intersection point and will therefore, at worst,
exist in a semi-contractive (but nonetheless dilatant) state.
There is scope for considerable discussion in regard to the above since it is arguable
whether simple extrapolation of the consolidation and critical state lines is valid'
Unfortunately it is not possible to extend the stress path testing to beyond the stress
levels indicated due to equipment limitations. These limitations are common to
almost all commercial and university labs and relate to the maximum pressure in the
triaxial cell apparatus.
I.6 PIEZOGONEMEASUREMENTS
t_ n ;o u
'" ROOT-IHE GRAPH
rq)--"
I
rci
I
I
-i I
t_
\
P
n- +*
a
tr
t\
t+
t\
t3
t\
I
l-----)
b
j
Figure 1.6-1: Piezometer cone sounding results at a location 40m behind the crest
pont
R(X'T-XE GRAPH
rc ,c :D .o tu 0 rr a0
})'
-
;I
t-
o:
l
'l I
I
ot
0 td o
*F lC.t ..lh l
Figure 1.6-2= Piezorneter cone sounding results for a test at the crest pont
Gordon McPhail, Alasdair Noble, George 2004 lnternational Seminar on Paste and Thickened Tailngs
The following points are noteworthy from an assessment of the piezometer cone
results:
o Dissipation rates are similar to those measured in the Rowe Cell which indicates
that the lab test is reasonably representative of the field situation
Water pressures in the slope are in reasonable agreement with the standpipe
piezometers.
o The cone resistance is greater than 0.5 MPa and, in the case of the sounding at
the crest point, is generally 2 MPa. lt is interesting to note the soft zones in the
sounding back from the crest. This is attributed to variations in the slurry in the
course of the trial.
1.7 SEEPAGEASSESSMENTS
The average placed tailings relative density has been 2.06 (72% solids)' ln
geotechnical terms this implies a moisture content of the tailings immediately after
deposition of a maximum of 28%. There is some run off/bleeding but the majority of
the water is initially locked up as interstitial water. The key question is how much of
this water seeps down to the phreatic surface and how much is drawn back out of the
mass through evaporation. To obtain a macro estimate of the recharge rate to the
phreatic surface a seepage model of the mound was generated' The program
SEEpy was used with axi-symmetric settings. Figure 1.7-1 shows the calibrated
\s
section.
&ldrah tx = 0 017 us
tu dd=09b006U5
To achieve calibration influx rates from the beach face were varied and seepage
pressures compared with the piezometer and piezocone measurements. lt was
found that while foundation conditions were central to the predicted profile of the
phreatic surface it was the infiltration rate that determined its location.
o The infiltration rate is less than 1% which indicates that most of the water
movement that takes place is through evaporation from the deposited mass.
The seepage modelling provides confrmation that with thickened discharge rates the
seepage is reduced by between a half and one order of magnitude.
Application of the consolidation and seepage data to slope stability analyses has
shown that factors of safety are above 2 and probabilities of failure below 1 in 10,000
even for very conservative assumptions on pore pressures, drainage conditions and
shear strength. This is in keeping with the flat slope angles that are generated by the
thickened tailings mound.
I.9 EROSIONASSESSMENTS
Erosion of the beaches of tailings placed at50% solids is known to be very low. The
thickened mound, on the other hand, is considerably steeper and the question of
erosion of these slopes both during operation as well as after decommissioning
needs to be assessed. To this end erosion testing and modelling has been carried
out. The erosion testing has been carried out by Landloch Pty Ltd and Australian
consultancy specialising in field measurements of erosion rates. Field
measurements are carred out using a rainfall simulator which has been designed to
ensure that the kinetic energy transmitted by the simulated rainfall is similar to natural
rainfall. Gulleying is assessed by over-land flow tests. Figure 1.9-1 below shows the
testing on both tailings as well as material that would be sued as topsoil or "growth
medium". ln both test types sedimentation samples are collected at short time
intervals and flow rates accurately measured.
Gordon McPhail, Alasdair Noble, George 2004 lnternational Seminar on Paste and Thickened Tailings
P a p ageorg iou, Da n Wil kin son Development and lmplementation of Thickened Discharge at
Osbome Mine, Queensland, Australia
Figure 1.9-1: Rainfall and gulley eroson simulations (top is tailings and bottom s
growth medium)
The results of the field measurements are used to derive erosion parameters that
have been applied in the dynamic erosion modelling program SIBERIA. SIBERIA
models long term landform evolution and works with a digital terrain model (DTM) of
the surface. The DTM is adjusted with each iteration in the simulation which means
that the model is able to simulate gulley formation.
Figure 1.9-2shows an isometric view from the DTM of the potential thickened tailings
facility before erosion simulation. Figure 1.9-3 shows an isometric view of the DTM
after 500 years of erosion on the bare tailings surface. lt is evident from the isometric
that there has been erosion of the confining embankment and spillage of eroded tails
over the crest of the eroded confining embankment. The pool wall has also been
swamped. Not withstanding these observations the cross-section in Figure 1.9-4
shows that erosion depths on the tailings surface are less than 500 mm over a 1,000
year smulation. This low erosion is attributed to the fact that the thickened tailings
mound largely mirrors the slopes found in the surrounding country.
Page 23
Gordon McPhail, Alasdair Noble, George 2004 lnternational Seminar on Paste and Thickened Tailings
Figure 1.9-3: lsometric showing eroson of the tailings after 500 years of simulation
2
rE
w
N
2
!n
E^-
2 to
Qn1
a2n
,.
um
26
R
22
L__t I
01 msms6t
Page 24
Gordon McPhail, Alasdair Noble, George 2004 tntemational Seminar on Paste and Thickened Tailings
Figure 1.9-5 shows an isometric of the DTM after 500 years of simulation for a
situation where the tailings mound is covered with growth medium. lt is evident from
the isometrc as well as the section shown in Figure 1.9-6 that erosion depths are
considerably reduced. The growth medium is representative of the cover materials in
the surrounding country. lt is only the embankment slopes that show excessive
erosion. The high eroson above the pool wall is induced by erosion of the pool wall
itself.
Figure 1.9-5: lsometric showing eroded mound with a cover of growth medium
*:r
'
4,F
6;;
r"
d.
u
il
Figure 1.g-6: Section through tailings mound with growth medium cover after erosion
simulation
To place the Erosional performance in context Figure 1.9-7 shows an isometric of the
old tailings dam after 500 years of erosion simulation based on the same parameters
as used for the thickened tailings mound with growth medium cover. The influence of
the steepness of the slopes of the confning embankments is clearly evident.
Gordon McPhail, Alasdair Noble, George 2004 lnternational Seminar on Paste and Thickened Tailings
Papageorgiou, Dan Wilkinson Development and lmplementation of Thickened Discharge at
Osbome Mine, Queensland, Australia
Figure 1.9-7: lsometric showing eroson after 500 years simulation on old tailings dam
It is concluded from the above that the thickened tailings mound is likely to perform
considerably better than a conventional tailings dam with 1 in 3 outer slopes of
similar height.
A summary of the modifications to the plant which were made to allow production of
high density tailings were as follows:
o Tuning of all control loops from grinding through floatation and tailings to
allow stable operation at desired density.
. Reduction in tailings cyclone spigot sizes from 70mm to 57, 51, 47, 45 and
41mm to increase underflow density and send more solids to the thickener.
Page 26
George
Gordon McPhail, Alasdair Noble, 2004 International Seminar on Paste and Thickened Tailings
Papageorgiou, Dan Wilkinson Development and lmplementation of Thickened Discharge at
Osbome Mine, Queensland, Australia
. Change of type of flocculant to better suit duty and allow higher density
underflow.
Total cost of plant changes approximately $16 000. lnstrumentation accounts for
approximately $13 000 of this. This cost is for equipment only and doesn't include
temporary trials.
Several aspects of the new dam pumping and piping system were designed
specifically to suit the pumping and deposition of high density tailings. These
included:
. Pipeline diameter and pressure rating designed specifically to suit both high
density tailings and "normal" density.
o lnstallation of flow meters at plant and at valving station on dam wall to give
indication of line blockages and ruptures.
1.10.3 OperationalGhanges
Page 27
Gordon McPhail, Alasdair Noble, George 2004 lnternational Seminar on Paste and Thickened Tailings
P a pageorg iou, Da n Wilkin son Development and lmplementation of Thickened Discharge at
Osbome Mine, Queensland, Australia
Strategies are in place for stopping deposition at a high density spigot should
the density fall. This occurs immediately if the plant is shut down
unexpectedly or if some major change occurs and also happens if the density
drifts low for around half an hour and efforts to increase the density are
unsuccessful.
1.10.4 Performance
The performance of the plant is reasonably steady with densities of betv'teen 72o/o
and 76% solids able to be obtained for the majority of the time. Occasionally there is
a difficulty in maintaining density which is attributed to a different SG or ore due to
ore type changes. While the density drops, the performance at the dam may not
necessarily be impaired. Test work is continuing to assess how the density impacts
the deposition behaviour with different ore types.
The pipe loop test work allowed the pumping pressures required to be predicted.
When this exercise was conducted the pressures calculated were obviously too high
based on operational data from pumping to the old dam. A number of iterations of
calculations were conducted after which a scale up factor from the pipe loop test
work was arrived at. Use of this scale up factor effectively calibrated the theoretical
Page 28
Gordon McPhail, Alasdair Noble, George 2004 lnternational Seminar on Paste and Thickened Tailings
The predicted pressure drops after applying the scaling factor were in the range of
O.2B- 0.46 kPa/m in the density range of 68% to75% solids. These correspond to
pressures of around 400-680kPa at the furthest spigot, which are right in the range of
the operating data. The test work indicated that at densities exceeding 760/o solids
the pumping performance would drop markedly with pressure drop per metre
increasing rapidly. This is validated by operational experience. lf the density
increases above around 76%-78% solids for any length of time then the flow begins
to drop in the pipeline and the pipe starts to sand up. lf this is not noticed and
remedied quickly the flow drops completely and the emergency line must be switched
to while the duty line is flushed. To date this has occurred several times, so far the
line has always been able to be cleared by flushing water and the line has not had to
be split along the length to allow clearing.
lmplementation of thickened discharge over the trial period has shown that significant
reductions in water lass are achievable. The simplified water balances shown in
Figure 1.10-1 and Figure 1.10-2, which are prior to and after implementation of the
thickened trial respectively, indicates a reduction in borefield abstraction from 81
m3/hr to 51 m3/hr, a reduction of 37o/o based on an achieved average percent solids
in February 2004 of 75o/o. This reduction is will increase further in the event of a
decision to proceed with thickened discharge beyond the trial since it will be possible
to direct a proportion of the process water underground to drive hydraulic equipment
and, in so doing, displace borefield water.
Page 29
Gordon McPhail, Alasdair Noble, George 2004 lnternational Seminar on Paste and Thickened Tailings
Figure 1.10-2: Simplified water balance during the thickened trial (which is on-going)
Storage of tailings in elevated beaches above the height of the dam wall
means that the wall needs fewer lifts to store the same volume of solids,
leading to considerable savings in wall construction costs. lt is estimated that
these savings may be in the order of $2.5 million over the 6 year life of the
dam.
Page 30
Gordon McPhail, Alasdair Noble, George 2004 lnternational Seminar on Paste and Thickened Tailings
Papageorgiou, Dan Wilkinson Development and lmplementation of Thickened Discharge at
Osbome M i ne, Quee n sl a nd, Au stra lia
. Reduction in water losses. The quantity of water deposited in the dam at75%
solids is around 500,000 cubic metres per year, compared to around
1,200,000 cubic metres at 55% solids. The reduced quantity of water sent to
the dam has to result in a reduction of losses through entrainment in the dam
and evaporation. lt is difficult to quantify this gain but based on flows from the
borefields it is likely that the reduction in losses is around 30,000 cubic metres
per year. Each cubic metre of water from the borefield costs around $0.36 to
pump to site so this translates to a saving of some $1 1,000 per annum.
. The reduction in water losses has a direct cost saving, but the reduction in
water drawn from the borefields also has signifcant benefit in terms of
responsible environmental behaviour and a reduction in post closure costs
until the level in the artesian aquifer is restored.
The sum of the smaller benefits is around $56,000 per year, but clearly the main
benefit is the potential to save $2.5 million in wall raise costs.
Page 31
Gordon McPhail, Alasdair Noble, George 2004 lnternational Seminar on Paste and Thickened Talings
Papageorgiou, Dan Wilkinson Development and lmplementation of Thickened Discharge at
Osborne Mine, Queensland, Australa
:. .tF +,-.
Figure 1.11-1: Potential long term thickened tailings deposition geometry at Osborne
The scenario above formed the base DTM for the erosion assessments described
previously.
Page 32
Gordon McPhail, Alasdair Noble, George 2004 lnternational Seminar on Paste and Thickened Tailings
Papageorgiou, Dan Wilkinson Development and lmplementation of Thickened Discharge at
Osborne Mine, Queensland, Australia
1.12 CONGLUSIONS
From the thickened discharge trial and assessments summarised in this paper it is
evident that there are considerable merts in implementing thickened discharge. Not
only are there significant reductions in capital expenditure and water use but there
are mproved operating conditions. The tailings mound presents less of an issue with
regard to erosion management both during operation as well as after closure. lt has
been shown by the geotechnical assessments that there will be less potential for
seepage into the foundation materials and geotechnical risks will be manageable.
Thickened discharge has shown potential to bring about the above benefits since
technology began to make it possible to thicken to appropriate densities reliably as
well as to pump the thickened slurry. The assessments documented in this paper
provide tangible evidence that this potential is realisable.
1.13 ACKNOWLEDGEMENTS
The authors would like to extend their gratitude and appreciation to Osborne Mines
for allowing permission to publish this paper.
References
Page 33
International Seminar on Paste and Thickened Tailings
Paste 2004
Paper 28
Wallie du Toit
Author: W du Toit
ittedtoMagra-UltrasepHP
t,*
j,,u
. 10:35 Stop Feed i,*
. 10:40 Flush i*u
ril
. 10:52 Gontrolled SG 1.44 r,r
r,fl
E t6 s g !g lfl lg lll lt
underflow at SG of 1.35.
. Viscometer meter read off
135 000 cP scale.
. Slump Sample Decreased
in Height by only 30 mm.
' Sample from 50 cents
Rheometer Would not Release
. This Sample Dried in the
Sun for Two Days and lt Still
Shrinks Daily. See at our Display Stand.
Thckenng by vibrclion, du Toit, Slide I Paste
Paste 2001 2004
t:::
Density to Maintan i;:i;
Pumpability and to ;i
Avoid Blockages. ',::::
Gorrect Pump to be iri
Selected. :;
2004
lnternational seminaron Pte and Thickened Tailings: Paste
Paper: Another Approach to Exploit the Rheology of Mineral Tailings
Page 6
Author:Wdu Toit
Paste 2004
Paper 29
R fVeHf, J \ilates
The feasibility of surface paste disposal at South African mines is set to improve with time as the
importance of water saving is emphasised through environmental awareness and as confidence in
thickening, conveyance and placement technology grows. The capital cost associated with
surface paste disposal usually makes this type of facility expensive especially when compared
with the cost of a paddock or spigot deposition facility that is raised by upstream construction
(that is a conventional South African disposal facility). The lifecycle costs for a paste disposal
facility has however been shown in a number of feasibility studies to be significantly less than
for the normal paddock or spigot facility. Other advantages associated with paste include, no
pool control requirement, limited seepage and phreatic surface development, overall reduction in
the risk of structural slope failure and less monitoring and control. These advantages are
beginning to play a stronger role in new tailings facility design and site selection. In particular,
where paste is involved, there is a further advantage to be gained by construction of an
impoundment wall using the waste rock (overburden) arising either from the mining operation or
the waste rock dump. This paper will show the outcome of investigations into a new disposal
facility at Venetia Diamond Mine and demonstrate the role that topography and site shape plays
in the site selection process and how this might change the criteria normally used to identify
suitable sites if one is looking to dispose of paste.
KEYWORDS
The viability of surface paste disposal in the South African mining industry is often assessed on
capital cost alone. There are however other considerations that need to be taken into account to
determine the viability of paste disposal. Water saving, reduced environmental risk and reduced
operational costs can improve the feasibility of disposing of fine residue in the form of paste.
Venetia Diamond Mine are planning to increase their plant throughput and will require a new
fine residue deposit (FRD). The investigation into a suitable site and disposal method examined
the possibility of paste disposal at the preferred FRD site. This paper will detail the aspects that
made this an athactive oPtion.
WALL CONSTRUCTION
2.1 TopographY
To make surface paste disposal possible an impoundment needs to be built, the height of which
will vary depending on the available area for disposal. For the same storage capacity, a large
foorint will reduce the height and therefore the cost of wall construction. A smaller foorint
thus increases the height and cost of the wall construction'
volume to be reduced. These features are shown in Figure 1 below which is a 3D rendering of
the site as viewed from the north.
In comparison a hypothetical site on flat terrain in the same location would have had a wall
volume or foorint area approaching four times larger than for this site.
FIGURE 1 - Proposed site for the FRD
2.2 Wall building material
A cost advantage can be gained by constructing the impoundment wall out of arising
waste rock (where pit mining is still in progtess). The cost of wall construction is the
extra over for shifting the waste rock deposition centroid. The obvious implication is that
there is an advantage to considering this type of wall construction at the early mining
development stage. Alternatively waste rock could be borrowed from the waste rock
dump located nearby. There are several considerations for this type of construction:
The rate of waste rock removal should be enough to at least match the rate of rise of
the new FRD.
The extra haulage distance may necessitate the purchase of additional haul trucks and
spreading equipment.
The topography of the wall site must be suitable for waste rock placement by haul
trucks.
The size distribution of waste rock generally negates the possibility of building a
water retaining structure, a water-tight upstream lining is thus required to ensure that
fine tailings does not migrate through the rock wall'
For the trade off study the wall was to be lined with a layer of compacted soil borrowed
from the FRD basin.
Figure 2 below is a 3D rendering of the site as viewed from the north showing the
proposed waste rock wall in place. The wall dimensions are schematic, as if the waste
rock wall were to be an extension of the original waste rock stocile.
FIGURE 2 - Proposed site for the FRD showing the impoundment wall in position
WATER REDUCTION
The deposition rate for the new FRD is 200 000 dry tonnes/month and the relative slurry
density for Venetia slurry is typically 1.45 tonnes/m3 1t.O tonne of water per tonne of dry
fines). Experience with paste thickening has shown that the upper end relative density
(RD) forpaste thickening of kimberlite slurryis around 1.60 tonnes/m3 10.6S tonnes of
water per tonne of dry fines). The water saving could thus be 70 000 m3lmonth out of a
possible 200 000 m3lmonth. Although any water saving is significant a 35%o saving in
water may not be enough, on its own, to make paste thickening economically viable.
The quintessential def,rnition of paste is that when placed it does not bleed water, thus
reducing a rage of environmental and structural risks. The removal of the supernatant
pool typical of a slurry storage facility reduces the risk of overtopping and the risk of
liquefaction. There is a reduced source of water (only rainfall) available to drive seepage,
which is a comfort to legislators and a relief for mines that may be rendered unfeasible
because of the need for expensive lining systems for their disposal facilities.
Kimberlite fine tailings are highly variable from a rheological point of view, thus offering
a negative effect on fwo (in particular) important aspects of paste development. Complex
control measures are required to maintain a particular RD target and/or the paste pumping
system needs to be robust enough to cope with changes in rheology brought on by
inconsistent particle size distribution and RD.
Locating the paste plant close to the disposal area and on a high point serves two
purposes:
o The expensive pumping system can be reduced or even eliminated by allowing paste
to flow under gravity to the FRD basin. This may even require occasionally diluting
the paste to improve its flow properties and;
o The control of RD can be based on a system that achieves the most water reduction
possible as apposed to achieving a specific target density or rheological property.
7
For Venetia the paste plant would be located on the southern end of the waste rock
stocile. Slurry would be pumped to the paste plant using the existing plant pumping
system where it is diluted to optimise flocculent efficiency. Reclaimed water is returned
to the existing plant. Paste underflow from the high compression thickeners is discharged
onto an inclined launder and flows under gravity to the FRD basin.
COSTS
The cost of a paste plant to process 200 000 dry tonnes/month of Venetia's fine residue
was estimated at R35 million. This cost is additional to a conventional spigot disposal
facility, but is set off against savings in the slurry delivery system and retum water
system. The cost for construction of the waste rock impoundment wall was calculated by
using arate of R0.96 tonne-km at an additional 1 lan haul. When compared to a starter
wall conskuction consisting of compacted soil the two walls were similar in cost despite
the difference in size. The order of magnitude capital cost of the paste disposal facility
compared favourably with that of a conventional spigot disposal facility:
Significant savings in the operational costs were found in slurry delivery and facility
management.
CONCLUSION
Capital costs for paste disposal facilities are high, but a well-chosen disposal site and
paste plant location can reduce capital and result in significantly lower lifecycle costs.
The saving in water consumption on its own is currently not enough to make paste
thickening economically viable.
Fine residue deposited in a dewatered state makes the facility structurally safer and
greatly reduces the quantity and therefore the impact of seepage. In future where lining
systems are required to protect the groundwater regimg the viability of paste disposal
will improve further.
REFERENCES
Venetia Diamond Mine Fine Residue Deposit Trade Off Study - Golder Associates -
May 2003.
lnternational Seminar on Paste and Thickened Tailings
ABSTR,AGT:
The role of raking is still not well understood. - Statement from Paste 2003,
Melbourne.
This vibration is applied in such a manner that the contents and not the tank or
structure is vibrated.
The theory, the means, the product and the results of the operational systems
will be discussed in the presentation.
l.lntroduction
Since the inception of the modern thickener, the primary roll of the rake
mechanism was to move the slurry in the thickener toward the centre discharge
cone. A few innovations such as pickets have seen the light, and have in certain
instances contributed to thickener operation'
ln general, the effect of raking on the actual underflow density, and thus the
rhology of the underflow is still a very grey area, and not well understood.
The need to clean increasingly higher volumes of dirty water in ever shorter times
has led to several innovations, both chemically and mechanically, hence the
development of coagulants, flocculants, feedwells, highrate thickeners and high
performance thickeners, each with its own pros and cons'
The introduction of High Performance Deep Vee Thickeners allows the industry
to exploit the free settling rate rather than the hindered settling rate as the
determining factor in thickener sizing'
Even though it allows the particles to settle more readily, the matrix formed
between the solids, flocculant and water, is entrained into the thickener
underflow. When pumped to the tailings disposal this leads to major liquid losses
due to hold-uP and evaPoration'
1 .2 Maqrabrator ConcePt
The concept of the Magrabrators is to utilise the force of vibration and apply it in
an oscillating movement to the consolidated slurry (compaction zone) in a deep
cone thickener,
ln order to facilitate the discharge of the thickener the Magrabrators are conve-
nieny placed just above the underflow discharge. This is the region where the
highet density and viscosity occurs. The high density or viscosity can lead to
"bridging" and subsequent blockage of the outlet'
A series of plates, curved to conform to the shape of the cone are typically
installed in the "Deep Vee" section of a deep cone thickener.
They are installed in such a way that the plate is surrounded by slurry and the
osciilating plate vibration will be transferred into the slurry, and not the tank
structure. The amount of plates are determined by the size of the thickener and
the type of slurrY thickened.
These plates are individually vibrated by means of small, typically 0.2 kW 4 - pole
oscillating electrical motors mounted on frames on the outside of the tank. Each
plate is mounted to its frame by means of two bolts passing through rubber
vibration dampers and seals. The frame on the outside is suspended by means
of an adjustable spring supported on a bracket welded to the cone. This spring
allows adjustment for centralising the frame and also isolates the vibration from
being carried over to the structure.
The motor speed and therefore the oscillating frequency can be changed by
means of a frequency invertor. By changing the out of balance by means of a
percentage scale the amplitude is also adjustable.
1.2.3 lnstallations
The Mineral Sands installation boasts the first in-line flow, density and viscosity
(promass 83 Coriolis Principle) measurement instrument in the world. Being
available up to 50 NB only we used a by-pass line for characterising the slurry.
This instrument was provided to us kind courtesy Endress and Hauser'
lnitial results indicate that the viscosity and thus shear, exceeds the 200 Pa limit
for slurry versus paste by far, and all underflows generated are well within the
paste regime.
lncreasing the density from 260/o solids to 35% releases 887 litres of water per
tonne of solids. Annual savings of 1,9 million m3 is attained. How much space
does this occupy ?. Savings in pumping alone amounts to 30,4% with 1,5
Megawatt drive motor.
We have a slump test result for you to see - dried for two days. Density 1,35
mineral sands slimes, viscosity off the scale of 135 000 cP'
2. Pumps
The only serious drawback to the physical modification of the slurry in the
compaction zone is the fact that the density and viscosity becomes such that
conventional means of pumping becomes impractical if not impossible'
3. Gonclusion