You are on page 1of 34

Pipeline Flow of Settling

Slurries

Presentation to Institution of Engineers Australia (Mechanical Branch)

Jeff Bremer - 23 rd April 2008

Overview and Aims


1. Explain physical laws underlying the behaviour of settling
solids in slurry pipeline flow.
2. Compare theories associated with pipeline flow. Why are
there so many?
3. Show where and how the theories disagree.
4. Present some preliminary results from recent work
(J. Bremer, V.Lim & R.Gandhi )

??

QUESTIONS
1. Where and why are slurry pipelines used?
2. What is a settling slurry?
3. What are the main features in pipeline flow?
4. Engineers are good at using theoretical and empirical best
fit theories. Whats the problem?
5. What are the underlying equations and physical phenomena?
6. What are the theories of pipeline flow?
7. What do we know that is right, and can we easilly confirm that
we have the right answer?
8. Whats the latest, and where to in future?

Slurry Pipelines
Slurry pipelines are used mostly for short haul duties, e.g.
dredging (~300m ), process plants (~300m) and tailings
(~3 km) In some long haul duties, minerals are pumped
many hundreds of kilometres.

Alumbrera copper concentrate


pipeline (316 km), Argentina

ENGINEERED BY PSI
Photos with permission of PSI Australia Pty. Ltd., 66 Kings Park Rd.,West Perth, WA 6005,Tel. no. (08) 9463-6606.

Slurry Pipelines
Each type of duty has its own best operation point, where
the size of the particles and the tendency to settle has a
strong impact on capital and operating cost.

ENGINEERED BY PSI
Photos with permission of PSI Australia Pty. Ltd., 66 Kings Park Rd.,West Perth, WA 6005,Tel. no. (08) 9463-6606.

Settling Slurries

Non Settling Slurries


contain particles that
remain in suspension
for a long time

NON-SETTLING

Settling Slurries
contain particles that
will fall and settle at
the bottom of a
container

SETTLING

Particles < 40 m

Particles > 40 m

Viscosity modified by
particles

Wide range of sizes from

Increasingly non-Newtonian
as concentration increases

Small (suspensions) 40 m
Medium (transition)
200 m
Large (heterogeneous) 2 mm
Very Large (hetero ) 5 mm

Transport velocity must increase as size increases

~ 200 m
~ 2 mm
~ 5 mm
~ >200 mm?

Settling Slurries

SETTLING
Particles > 40 m
Wide range of sizes from
Small (suspensions) 40 m
Medium (transition)
200 m
Large (heterogeneous) 2 mm
Very Large (hetero ) 5 mm

~ 200 m
~ 2 mm
~ 5 mm
~ >200 mm?

Transport velocity must increase as size increases

Settling Slurries

SETTLING
Particles > 40 m
Wide range of sizes from
Small (suspensions) 40 m
Medium (transition)
200 m
Large (heterogeneous) 2 mm
Very Large (hetero ) 5 mm

Dead Donkeys?

~ 200 m
~ 2 mm
~ 5 mm
~ >200 mm?

Pipeline Flow of Newtonian Liquids


P
HW =
g

L V2
= f
D 2g

Darcy-Weisbach equation
f

L
D

HW

head loss due to friction

(m)

friction factor

(dimensionless)

length of pipe

(m)

internal diameter of pipe

(m)

accelaration due to gravity

(m /s)

mean Flow velocity

(m/s)

Moody Diagram

HeadLoss
HW
P

2
v

2g

H1= 1 +

+z1

2
v

2g

H2= 2 +

+z1

PipeFlow

C.Y. OConnor Pipeline c.a. 1899

Features of Settling Slurry Pipeline Flow

Fixed Bed

Fluidised
Fluidised
Bed

Homogeneous
Homogeneous
Flow

Heterogeneous
Heterogeneous
Flow

1. Size does matter.

Hydraulic gradient, i (m/m )

V1

V2

V4

V3 =Vdep

Settling Slurry

Larger particles require


increased transport velocity

Smaller particles (particularly


fines <40 m) can modify
viscosity. Helps to suspend
larger particles.

2. Flow velocity generates


turbulence which keeps
particles suspended.

Water
Carrier

Mean Velocity , V (m/s)

3. The system curve has a


minimum that bounds different
flow / friction processes

Newitts Classification of Slurry Pipeline Flow

Solids

Concentration

Newitt et al (1955) described a range of flow flow/deposition


phenomena after observing sand and coal particles in 25mm Perspex
pipes. His classifications are still used today.
Newitt, D. M., J. F. Richardson, M. Abbott, and R. B. Turtle. 1955. Hydraulic Conveying of Solids in
Horizontal Pipes. Trans. Institution of Chemical Engineers 33: 94-113.

Frictional Head loss Mechanisms


HeadLoss,5mmgravel,Cv=10%,DN400Pipe
500

Since we
understand the
behaviour of water
(the carrier) we can
calculate the
frictional head
losses caused by
wall friction - HW

The remainder must


be friction losses
between

450

400

350

H M = HW + H S

Frictional Head
Loss due to
solids - Hs

)r
300
e
ta
W

m
(s 250
o
L
d
a
e 200
H

Water
Settling
Slurry
Deposition
Point

150

Frictional Head Loss due to


wall friction of carrier fluid
with pipe- HW

100

(a) particles and fluid

50

0
0.00

2.00

4.00

6.00

8.00

10.00

12.00

FlowVelocity(m/s)

14.00

16.00

18.00

20.00

(b) particles and pipe


wall
(c) particle-particle
collisions.

Durand Theory -1952

= 82.1.5

V 2

iM iW
CD
= 82.
CV .iW
gD S

Durand, R. 1952. The Hydraulic Transportation of Coal and Other Materials in Pipes. Colloq. of National Coal Board,
London.

1.5

Durand Theory (contd)


HeadLoss,5mmgravel,Cv=10%,DN400Pipe
500

1. Durands Theory is purely correlative.

H M = HW + H S

450
400
350

2. The curve fit was for 305 points, for sand


and coal running between 200 m and 25
mm.

Frictional Head
Loss due to
solids - Hs

r) 300
te
a
W
m
( 250
s
o
L
d
a
e
H 200

Water
Settling
Slurry
Deposition
Point

150

4. As transport velocity becomes large, the


slurry curve converges to water head loss
from above.

Frictional Head Loss due to


wall friction of carrier fluid
with pipe- HW

100

3. The results are in Head of Carrier Fluid


usually water.

50

0
0.00

2.00

4.00

6.00

8.00

10.00

12.00

14.00

16.00

18.00

20.00

FlowVelocity(m/s)

i M iW
C V .iW

V 2

= 8 2.
gD S

= 8 2 . 1 .5

= H W (1 + C V . 8 2 . 1 .5 )

CD

1 .5

Nothing proves that such a formula is


rigorously exact. Doubtless exists a
more accurate and more complex
means of notation, but the one given
above groups quite favourably

More Theories
(To name a Few)
Correlation

1. Durand 1952
2. Homogeneous Mixture Theory
3. Newitt et. Al - 1955
4. Rose and Duckworth 1969

Correlation

5. Heyden and Stelson - 1971

Correlation

6. Volcado and Charles 1972

Correlation

7. Wasp et al - 1977

Part theory part


correlation
Correlation

8. Lazarus Neilson 1978


9. Wilson - 1992
10. Wilson Addie & Clift 1997
In Current Use
Not in Use

No Problem Ive got a Computer


HeadLossat6.6m/s,5mmgravel,Cv=10%DN400Pipex1000m
800
700
600
) 500
m
(
s
o
L 400
d
a
e
H 300

Lazarus Neilson
WilsonAddieClift
Durand

200

Water

100
0
0

FlowVelocity(m/s)

10

Answers Using
commonly accepted
theories can vary by
several hundred
percent AND
MORE!

Settling and Drag Forces on Particles

Depends on density
, particle diameter,
shape, Reynolds
number and
surface effects

Settling and Drag Forces on Particles

Particles > 150 m

Drag coefficient as a function of Reynolds number for smooth spheres


and cylinders (Munson et al. 2002, 582)

Known correlations
to correction CD
based on shape
effect
Slip Velocity to Produce drag force FD

Settling and Drag Forces on Particles

Turbulent fluctuation of particle velocity in the direction of flow

Settling and Drag Forces on Particles


HeadLoss,5mmgravel,Cv=10%,DN400Pipe
500

Solids concentration
approaches input
concentration
Hs=constant

450

400

350
)r
300
e
ta
W
(m
s 250
o
L
d
a
e 200
H

Hs

Frictional Head
Loss due to
solids - Hs

HW
Frictional Head Loss due to
wall friction of carrier fluid
with pipe- HW

50

0
6.00

+ H

Settling
Slurry

100

4.00

Water

150

2.00

= H

H M = HW (1 + CV .82.1.5 )

Deposition
Point

0.00

8.00

10.00

12.00

14.00

16.00

18.00

P
HW =
g

L V2
= f
D 2g

20.00

FlowVelocity(m/s)

In the limit the slip velocity is roughly constant as the average velocity of
particles in direction of flow equals approaches the velocity of the liquid
i.e.Vsolid = Vliquid the homogeneous limit . In other words Hs << Hw

In Durand Theory in the limit Hs

zero

Comparison of Theories
H e a d Lo s s ,5 m m gra ve l,C v= 1 0 % ,D N 4 0 0 P ip e x 1 0 0 0 m
800

700

600

500
)
m
(
s
Lo
400
d
a
e
H

L azar u sN e ilso n
W ilso n A d d ie
C lift
D u r an d

300

200

100

0
0

10

12

14

16

18

20

F lo w V e lo cit y(m / s)

Location of The Deposition Velocity and Head Loss at Deposition is


the Key to having an accurate Theory.
Clearly the state of the art is not good

Comparison of Theories
Head Loss, 100m particle, Cv=10%, DN100 pipe x 1000m
500
450
400
) 350
m
( 300
s
s
o
l 250
d
a 200
e
H
150

Wilson Addie Clift


Durand
Lazarus Neilson
Water

100
50
0
0.00

2.00

4.00

6.00

8.00

10.00

Velocity m/s

Agreement is less critical at 100 m

Wilson Addie and Clift Theory

Slope M

Determined in tests on 400 m sand. Pressure gradient = 0.5 x sliding fr friction factor

Lazarus Nielsen Theory (1978)


Lazarus Neilsen Theory is a correlation theory that claims to be
more accurate than Durand and Newitts theories.
They proposed that the mass flow rate ratio (M*), defined as
the ratio of mass flow of solids to carrier fluid, should be used
instead of the volumetric concentration (Cv)

Lazarus Nielsen Theory (contd)


They plotted friction factor fM for the mixture against the base
friction factor fB to develop their final correlation.

Current Work Particle Drag & Deposition Head and Velocity


Collaborators : J. Bremer (SKM) , Vincent Lim (K.J. Beer),
Ramesh Gandhi (PSI California)

Began by describing the equations of


drag and pressure loss due to solids at
the deposition point.
Assumes : All particles fluidised at the
minimum in the pressure gradient curve

Fixed Bed

Fluidised
Fluidise
d Bed

Homogeneous
Homogeneou
s Flow

Heterogeneous
Heterogeneou
s
Flow

Hydraulic gradient, i (m/m )

V1

V2

V4

V3 =Vdep

Settling Slurry

Water
Carrier

Mean Velocity , V (m/s)

Particle Drag and Deposition Velocity and Head Loss(contd)

Particle Drag and Deposition Velocity and Head Loss(contd)

Pesky mean path length constan

Particle Drag and Deposition Velocity and Head


Loss(contd)
All terms in the final equation are rearranged to solve for the
Slip velocity V

This is Measurable from experiment!

Particle Drag a Virtual Experiment Based on Durand


Points

4 10 5

Particle Drag a Virtual Experiment Based on Durand


Points
System Parameter

Value Range

Unit

Lower

Upper

Carrier density ()

1,000

1,250

kg/m3

Carrier viscosity ()

0.0008

0.001

Pa.s

0.1

0.9

2,160

4,000

kg/m3

(40 m)

0.02 (20 mm)

0.05

0.4

Pipe diameter (D)


Particle density (p)
Particle size (d)
Concentration

by

volume (Cv)
Pipe length (L)
Pipe roughness

1,000

m
Smooth

200 Virtual data points (deposition velocity, and pressure


at the deposition point) obtained using Durand equation to

4 10 5

Virtual Experiment Results


Deposition Velocity

Deposition Velocity Average Error 0.05 %


-- Maximum Error 0.42 %

4 10 5

Virtual Experiment Results


Head Loss at The Deposition Point

Head Loss

Average Error 0.55 %


-- Maximum Error 1.8 %

4 10 5

Conclusions
1. Not all is well with the theory of slurry transport.
2. There is considerable disagreement amongst theories
regarding
1. Deposition velocity
2. Head Loss at Deposition

3. There is no clear agreement on the forces and friction


associated with various mechanisms, (e.g. fluidised bed,
heterogeneous flow, homogeneous flow etc) or the velocities
at which they occur.
4. Many of the theories blow up when large particles are
involved. Say > 2mm. Comparison between calculations at
these sizes indicates a need for model studies in future
developments.
5. Where possible dont pump at sizes > 150 m.

You might also like