Tailings Transport

[Link]
Keynote Address

Tailings Transport – Back to Basics!

P.T. Slatter
Cape Peninsula University of Technology, Cape Town, South Africa

ABSTRACT
The underlying principles of tailings transportation are strongly influenced by
the paradigms of conventional Newtonian engineering hydrodynamics. However,
fundamental differences exist, and the objective of this presentation is to high-
light these, and their practical implications. The Bingham plastic rheological model
is accepted as an appropriate point of departure, and the fundamentals of the
development of the engineering hydrodynamics of tailings transport are presented.
In particular, the meaning, measurement, importance and implications of the yield
stress are explored. The exploitation of tailings rheology for pressure gradient
prediction is then presented, emphasising the key role that the yield stress plays in
turbulent, transitional and laminar pipe flow. Discussion centres around the prob-
lems associated with the application of each flow regime, and the case for opera-
tion in the transition zone is argued.

Paste 2005, Santiago, Chile 165

Tailings Transport - Back to Basics! Slatter, P.T.

1. INTRODUCTION
Tailings disposal has undergone fundamental changes in recent years (Jewell
et al., 2002), and environmental and economic drivers have pushed tailings trans-
port designers and operators to consider operating at considerably higher con-
centrations than before (Slatter, 2002). Slatter (ibid) makes the point that real
engineering is just as much an art as a science, and is by definition a rather grey
exercise, with many varying degrees of what constitutes a good design in any
particular application. Higher concentration designs will not necessarily reduce
environmental contamination, or always ensure significant cost and water sav-
ings. It therefore behoves us to re-visit the basics of tailings transport, lest we fall
into the trap of believing that the “new” will always be “better”!
Fluid mechanics is arguably the most empirically based engineering science.
The main reason for this is that the last unsolved problem of Classical Physics is
Fluid Turbulence, and we are probably still a decade away from a solution which
is widely acceptable to physicists (McComb, 1990, 2005). It comes as no surprise
then that the paradigms of classical Newtonian engineering hydrodynamics edu-
cation and practice are firmly rooted in dimensional analysis.
The hydraulic transport of tailings material fits broadly under the heading of
the flow of non-Newtonian suspensions, and is closely related to, and profoundly
influenced by, classical Newtonian engineering hydrodynamics. The essential dif-
ference between the two is the presence and influence of solid particles. There are
two major consequences of the presence of particles. The smaller particles influ-
ence the viscous behaviour (rheology) of the mixture, and the larger particles will
tend to settle. Both of these will influence the flow patterns and velocity distribu-
tions fundamentally. The objective of this presentation is to highlight these fun-
damental influences, and their practical implications for tailings transport.

2. THEORY AND LITERATURE

2.1. Classical Newtonian Engineering Hydrodynamics
Since the work of Osborne Reynolds in the late 19th century, the paradigms of
classical Newtonian engineering hydrodynamics have been based on dimensional
analysis of the major variables. For pressure gradient and flow behaviour predic-
tion, the relevant dimensionless groups are the Reynolds number, the friction
factor and the relative roughness. Flows present as laminar, transitional or turbu-
lent, and the interrelationship between the dimensionless groups in each regime
has been summarised classically using the well-known Moody diagram (Moody,
1944). This approach has strongly influenced the development of techniques for
pressure gradient and flow behaviour prediction for non-Newtonian flows, and a
similar, extended set of dimensionless groups has been developed.

166 Paste 2005, Santiago, Chile

 Tailings Transport

2.2. The Bingham Plastic Rheological Model

Rheology is often best understood as a dynamic property of microstructure.
At rest, the attractive forces between particles or agglomerates form a three-di-
mensional structure, which extends to the walls of the container. The stress re-
quired to break this structure and initiate shear, is called the yield stress. Below
this stress, the material is envisaged to behave as an elastic, Hookean solid. As
shear stresses and shear rates increase, the agglomerates gradually re-orientate
and disintegrate, resulting in a decrease in the viscosity of the material. This pro-
cess is known as shear thinning. At very high shear stresses and shear rates, the re-
orientation and disintegration process reaches equilibrium, and the viscosity be-
comes constant.
Although this portrayal of the relationship between viscosity and microstruc-
ture is idealised, it is useful for engineering purposes, and the simplest rheological
model which can accommodate this behaviour is the Bingham plastic model, which
can be formulated in terms of shear stress ô;

τ = τ y + K γ , [1]
or viscosity η;

τy
η= + K. [2]
γ
The two terms on the right-hand side will be equal when

τy
γ b = . [3]
K
The importance of the boundary shear rate γ b is that it marks the boundary
between yield stress and plastic viscosity domination of viscosity. This is shown
graphically in Figure 1.
100

Yield Stress Plastic Viscosity

10 y Dominated Dominated
Viscosity

0.1

0.01

0.001
1 10 100 1000 10000 100000 1000000

Shear Rate

Figure 1: Graphical presentation of Equation [2] showing the boundary shear. rate γ b .

Paste 2005, Santiago, Chile 167

Tailings Transport - Back to Basics! Slatter, P.T.

It is significant to note from Figure 1 that the ordinate intercept of the oblique
asymptote is in fact the yield stress value ôy. This illustrates the fact that – for
practical purposes – the viscosity values for shear rates less than the boundary
shear rate γ b are directly proportional to the yield stress. This serves to emphasise
the importance of the yield stress value when operating in the region to the left of
the boundary shear rate. The yield stress ôy is a strong function of particle proper-
ties, concentration and solution chemistry, with the plastic viscosity K usually
being a weaker function of these.

2.3. Dimensional Analysis

Dimensional analysis of the Bingham plastic model as applied to circular pipe
flow will yield three dimensionless groups. Besides the standard friction factor,
the other two dimensionless groups are of prime concern here. They are the
Reynolds number and the Hedström number:-

ρV D
Re B = , [4]
K

D2 ρ τy
He B = . [5]
K2
One of the problems with dimensional analysis as shown above is that it sepa-
rates the rheological parameters ôy and K, and is unable to accommodate the
additive relationship of Equation [1]. An alternative approach (Slatter & Lazarus,
1993) is to analyse the viscous stress as a unit, yielding

8ρ V2
Re 2 = .
 8 V  [6]
τ y + K 
 D 

2.4. Laminar Flow

Laminar flow is one of the very few areas where a rigorous analytical solution
is available, known as the Rabinowitsch-Weissenberg integral;

ô0 2
ô . γ (ô )dô . [7]
8V 32 Q 4
= =
D ð D 3 ô 30 ∫0
Application to Equation [1] will yield the well known Buckingham equation;

8V τ o  4τ  1  τy 
4
= 1 −  y +   ,
D K  3  τ o  3 τ
  o

  [8]
 

that can be used for design in laminar flow.

168 Paste 2005, Santiago, Chile

 Tailings Transport

2.5. The Laminar/Turbulent Transition

The dimensional analysis presented above can be used to develop an approach
for determining the critical velocity Vc for large pipes (Slatter & Wasp, 2000);

0.5 5
Re = 26 He for He > 1.5x10 . [9]
Bc B

This results in a critical velocity which can be calculated directly from

τy
V = 26 . [10]
c ρ

2.6. Turbulent Flow

A turbulent flow model appropriate for non-Newtonian slurries is based on a
particle roughness turbulence effect (Slatter, 1999). A roughness Reynolds num-
ber was formulated to accommodate the larger of either the particle roughness
size (usually taken as the d85 size) or the standard pipe roughness k (Slatter & van
Sittert, 1997) as,

2
8ρ V
Re = *
r  8 V* 
 
( )
τ +K [11]
y
 max : d85 , k 
where V* is the classical shear velocity.
This roughness Reynolds number was used to correlate the classical rough-
ness function B in the same way as for Newtonian fluids.
If Rer d»#ð 3.32 then B = 2.5 ln Rer + 5.5. This is analogous with smooth wall
turbulent flow for which the flow behaviour can be predicted from

V R
= 2.5 ln   + 2.5 ln Re r + 1.75 . [12]
V* k
If Rer > 3.32 then B = 8.5. This is analogous with fully developed or rough
wall turbulent flow for which the flow behaviour can be predicted from

V R
= 2.5 ln   + 4.75 . [13]
V* k

Paste 2005, Santiago, Chile 169

Tailings Transport - Back to Basics! Slatter, P.T.

3. YIELD STRESS DOMINATION

While it is clear from cursory examination of Equation [10] that the yield
stress has a dominant effect on the laminar/turbulent transition for large pipes, its
dominant effect in other areas is more subtle and will be highlighted below.

3.1. Operational Area

We can formulate the critical bulk shear rate Γc in a pipe at the laminar/turbu-
lent transition as

8 Vc
Γc = . [14]
D
Combining Equations [3], [5] [10] and [14], we derive that

γ b He
= . [15]
Γc 208
If we substitute the lower limit value for the large pipe regime ie He = 1.5x105,
then we get

γ b
= 1.86 ≈ 2 . [16]
Γc
The important consequence here is that the boundary shear rate is always at
least twice the critical bulk shear rate, for large pipes. This situation is shown
graphically in Figure 2.
100

Yield Stress Plastic Viscosity

10 y Dominated Dominated
Viscosity

0.1
8 Vc
c
0.01 D
0.001
1 10 100 1000 10000 100000 100000

Shear Rate
Figure 2: Bingham plastic viscosity diagram showing the maximum transition point for
large pipes.

170 Paste 2005, Santiago, Chile

 Tailings Transport

This means that in the laminar, transition and early turbulent region, the
behaviour of these flows will be dominated by the yield stress value.

3.2. Yield Stress Measurement (Slatter, 2004)

The determination of yield stress has traditionally been performed using at
least three different approaches.

3.2.1. Flow curve extrapolation

Flow curve or rheogram measurements of shear stress vs shear rate are ob-
tained from either tube or rotary rheometry. After competent data reduction,
Equation [1] can be fitted to the data points. This results in the yield stress effec-
tively becoming a fitting parameter, rather than a true measure of structural fail-
ure. Furthermore, Nguyen & Boger (1992) point out that competent data reduc-
tion cannot proceed until the value of the yield stress is independently deter-
mined.

3.2.2. The slump test

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 al. (2003) recommend a cylindrical shape, and the reader
is referred to this reference for the different analytical 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.

3.2.3. The vane test

The vane test involves immersing the vane in the fluid, slowly rotating it, and
measuring the time/torque response, from which the maximum torque Tmax 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.

4. PRESSURE GRADIENT PREDICTION

The predictive models presented above can now be exploited to utilise the
slurry rheology for predicting pipe flow behaviour and pressure gradients. Pres-
sure gradient prediction usually begins with rheological characterisation of the
material over a range of concentrations. This information can then be used to
predict the pressure gradient

Paste 2005, Santiago, Chile 171

Tailings Transport - Back to Basics! Slatter, P.T.

4.1. System Curve

The system curves for water and a typical tailings at three concentrations is
presented in Figure 3.
0 .3 5

Turbulent
0 .3
Hydraulic Gradient

0 .2 5

0 .2
Laminar

0 .1 5

0 .1

0 .0 5

Water
0

0 0 .1 0 .2 0 .3 0 .4 0 .5 0 .6

Flow Rate
Figure 3: System curves for water and tailings at three concentrations.

Figure 3 shows the general form of the laminar and turbulent regions, and the
effect of increasing concentration.

4.2. The Importance of the Yield Stress

In order to reinforce the comments above regarding the importance and influ-
ence of the yield stress value, the effect of yield stress on the friction head loss can
be investigated. The results of such an exercise are presented in Figure 4. For this
exercise, all variables were held constant – only the yield stress has been varied.

1 4 0

1 2 0

Turbulent Laminar

1 0 0
Friction Head Loss

8 0

6 0

4 0

2 0

0 5 1 0 1 5 2 0 2 5 3 0 3 5 4 0

Yield Stress Ty

Figure 4: Variation of friction head loss with yield stress for typical tailings slurry.

172 Paste 2005, Santiago, Chile

 Tailings Transport

Figure 4 shows a direct proportionality between friction head loss and yield
stress in the laminar regime. A much weaker, but monotonically increasing, rela-
tionship is seen in the turbulent regime.

5. PARTICLE SETTLING
So far, we have dealt with the effect of small, essentially non-settling particles,
which modify the viscous characteristics of the carrier fluid. Larger particles,
because of their tendency to settle, have quite a different effect on the behaviour
of the mixture.

5.1. Settling in Turbulent Flow

In turbulent flow, turbulent eddies are used to suspend and transport settle-
able solids through the pipe. If there is insufficient turbulence, particles will form
a stationary bed, and pipe blockage can occur. Although early research on the
deposition velocity was largely empirical, a quantum leap in this area was achieved
with the work of Wilson who applied a mechanistic approach based on the two-
layer model (Wilson, 1979). This work is presented in the form of nomograms,
which are still very much in use. In practical terms, the deposition velocity is the
velocity at which, as average pipe velocity is decreased, a stationary bed will
begin to form on the pipe invert. It is desirable to operate above this velocity to
avoid unstable operation and pipe blockage.

5.2. Settling in Laminar Flow

In laminar flow, there are no mechanisms such as turbulent eddies to suspend
settleable solids, and they will accumulate in a bed on the pipe invert (Pullum et
al. 2004). The only method of transporting these materials is to ensure that the
forces acting on the settled bed are sufficient to maintain the sliding motion of the
bed. The same mechanistic two layer model approach has been used successfully
to model this behaviour (ibid).
Although the deposition velocity in laminar flow is a complex function of
particle characteristics and carrier fluid rheology, it has been found that a pres-
sure gradient above 1 kPa/m is usually sufficient to ensure a sliding bed (Cooke,
2002).

6. OPERATION IN THE TRANSITION REGION

There are at least two practical difficulties emanating from the above discus-
sions, as we attempt to design and operate tailings transport pipelines at higher

Paste 2005, Santiago, Chile 173

Tailings Transport - Back to Basics! Slatter, P.T.

concentrations. The first is – considering turbulent flow operation – that due to

the increased viscous stresses at higher concentration, it may not be possible to
achieve enough turbulence to operate above the deposition velocity at realistic
velocities. The second is that if we operate in laminar flow, the required minimum
operating pressure gradient of at least 1 kPa/m may result in excessive energy and
pumping plant costs.
Furthermore, a principal omission from the above topics is any consideration
of operating in the transition region.
Figure 5 shows the results of a standard pipe flow test curve for a 6%v/v
kaolin suspension in a 5 mm pipe. Superimposed on this is a plot of the standard
deviations of the pressure transducer readings taken at the same time (Huq &
Slatter, 1999).

500 2000
Wall Shear Stress
L/T Transition
400 1500
Wall Shear Stress [Pa]

Press Std Dev [Pa]

Press Std Dev

300 1000

200 500

100 0

0 -500
0 5000 10000 15000
8V/D [1/s]

Figure 5: Pipe flow test data showing pressure readings standard deviation.

Figure 5 shows clearly that there is a dramatic spike in the pressure deviations
in the transition region. This is a reflection of the regime chaos at transition,
which has been reported by several other investigators (eg Park et al., 1989 and
Graham et al. 1999).
For our purposes, there are two advantages arising from this. Firstly, the re-
gime chaos indicates that there is a high transport potential for settleable solids.
Secondly the extreme nature of the spike shown in Figure 5 indicates that identi-
fying and tracking this region would be possible with an intelligent control sys-
tem. A tailings transport system operating in this region may well have advan-
tages, and is a region – perhaps exotic, but – well worth investigating.

7. CONCLUSION
The fundamentals of tailings transport have been developed from the para-
digms of conventional Newtonian engineering hydrodynamics, using the Bingham

174 Paste 2005, Santiago, Chile

 Tailings Transport

plastic rheological model as the appropriate point of departure. Emphasis has

been given to the meaning, measurement, importance and implications of the
yield stress, illustrating the key role it plays in turbulent, transitional and laminar
pipe flow. The case for operation in the transition zone has been argued against
the problems associated with the application of each flow regime.

ACKNOWLEDGEMENTS
The author wishes to acknowledge the contributions of his colleagues and
students at the Flow Process Research Centre of the Cape Peninsula University of
Technology. Special thanks are extended to the organising committee for the honour
which the Keynote Address Invitation embodies.

REFERENCES
Cooke, R. (2002) Laminar flow settling: The potential for unexpected problems,
Proc. 15th Int. Conf. on Hydrotransport, Banff, Canada, 3-5 June.
Clayton, S., Grice, T.G. and Boger, D.V. (2003) Analysis of the slump test for on-
site yield stress measurement of mineral suspensions, Int. J. Miner. Process. 70,
pp. 3–21.
Graham, L.J.W., Pullum, L., Chryss, A. and Rudman, M. (2001) Transition to
Turbulence in Non–Newtonian Pipe Flow, 6th World Congress of Chemical En-
gineering, Melbourne, Australia, pp. 23-27.
Huq Md, N. and Slatter, P.T (1998) Real-time identification of the laminar/turbu-
lent transition, 6th Cape Technikon Conference in Civil Engineering, Cape
Technikon, 8th October, 1998.
Jewell, R.J., Fourie, A.B. and Lord, E.R. (Eds) (2002) Paste and Thickened Tail-
ings – A Guide. Published by the Australian Centre for Geomechanics, University
of Western Australia, UniPrint, 2002, pp. 173.
McComb, W.D. (1990) The physics of fluid turbulence, Oxford University Press,
Oxford, ISBN 0-19-856160-1.
McComb, W.D. (2005) Private Communication.
Moody, L.F. (1944) Friction factors for pipe flow, Trans ASME November, 1944.
Nguyen, Q.D. and Boger, D.V. (1992) Measuring the flow properties of yield
stress fluids, Annu. Rev. Fluid Mech. 24, pp. 47-88.
Park, J.T., Mannheimer, R.J., Grimley, T.A. and Morrow, T.B. (1989) Pipe flow
measurements of a transparent non-Newtonian slurry, Journal of fluids engineer-
ing, Vol. 111, pp. 331-336.

Paste 2005, Santiago, Chile 175

Tailings Transport - Back to Basics! Slatter, P.T.

Pullum, L., Graham, L.J.W. and Slatter, P.T. (2004) A non-Newtonian two-layer
model and its application to high density hydrotransport, British Hydromechan-
ics Research Group 16th International Conference on Hydrotransport; Santiago,
April, 2004. ISBN 1 85598 054 1, pp. 579-594.
Slatter, P.T. (1999) The role of rheology in the pipelining of mineral slurries, Min.
Pro. Ext. Met. Rev., Vol 20, pp. 281-300.
Slatter, P.T. (2002) Non-Newtonian Laminar Pipe Flow – A Place In The Sun At
Last!; Invited Keynote Address - 11th International Conference on Transport and
Sedimentation of Solid Particles - Ghent, September, 2002, ISBN 83-87866-34-2,
pp. 33-40.
Slatter, P.T. (2004) The Measurement And Role Of Rheology In Paste Pipelining;
Invited Keynote Address, Paste 2004, International Seminar on Paste and Thick-
ened Tailings, Cape Town 31 March-2 April, 2004.
Slatter, P.T. and Lazarus, J.H. (1993) Critical flow in slurry pipelines; British
Hydromechanics Research Group 12th International Conference on Slurry Han-
dling and Pipeline Transport HYDROTRANSPORT 12; Brugge, Belgium: 28-30
September, 1993, ISBN 0 85298 874, 5 pp. 639 – 654.
Slatter, P.T. and van Sittert, F.P. (1997) The effect of pipe roughness on non-
Newtonian turbulent flow; 9th International Conference on Transport and Sedi-
mentation of Solid Particles - Cracow: 2-5 September, 1997, ISBN 8385582232
pp. 621 – 635.
Slatter, P.T. and Wasp, E.J. (2000) The laminar/turbulent transition in large pipes;
10th International Conference on Transport and Sedimentation of Solid Particles
- WrocBaw: 4-7 September, 2000, ISBN 83 - 87866 - 12 - 1, pp. 389 – 399.
Wilson, K.C. (1979) Deposition Limit Nomograms for Particles of Various Den-
sities in Pipeline Flow, 6th International Conference on the Hydraulic Transport
of Solids in Pipelines, BHRA, Canterbury, UK, September 1979.

176 Paste 2005, Santiago, Chile

 Session X Título sección

Determining the Optimum Location of a

High Rate Thickener for a Thickened Tailings System

A.J.C. Paterson
Paterson & Cooke Consulting Engineers Pty Ltd, South Africa

ABSTRACT
This paper presents a cost comparison between pumping large volumes of
dilute thickener feed to a remote thickener located at the tailings storage facility
(TSF), and using high pressure positive displacement pumps pumping thickened
tailings from a thickener located at the process plant to the tailings residue facil-
ity. The study considers only the pump and pipeline system capital and operating
costs, and specifically excludes all costs associated with preparation and place-
ment.
This study shows that for a 2.5 km distance both options have a similar net
present cost. As the distance increases to 5 km it is better to locate the thickener
at the process plant and pump high density thickened tailings using piston dia-
phragm pumps. This is mainly due to the differences in annual operating costs
over the life of mine as it is not energy efficient to pump large volumes of dilute
slurry long distances.
Each site assessment is obviously project specific and this trade-off study shows
that it is not immediately obvious as to which pumping solution is the most cost
effective.

Paste 2005, Santiago, Chile 177