Вы находитесь на странице: 1из 13

Concentration at the pipe bottom at deposition velocity for transportation

of commercial slurries through pipeline


D.R. Kaushal
a,
*
, Yuji Tomita
a
, R.R. Dighade
b
a
Mechanical Engineering Department, Kyushu Institute of Technology, 1-1 Sensui-cho, Tobata-ku, Kitakyushu, Fukuoka 804-8550, Japan
b
Civil Engineering Department, YCCE Nagpur, Nagpur, Maharashtra, India
Received 23 August 2001; received in revised form 27 November 2001; accepted 25 December 2001
Abstract
In commercial slurries, the solids being transported are multisized and their size may span three orders of magnitude. The minimum
operating velocity is usually kept as 0.5 m/s more than the deposition velocity. Deposition velocity is the flow velocity at which the
deposition of solids occurs. In the present study, Kaushal [D.R. Kaushal, Prediction of particle distribution in the flow of multisized
particulate slurries through closed ducts and open channels, PhD Thesis, Department of Applied Mechanics, I.I.T. Delhi, 1995], Mukhtar [A.
Mukhtar, Investigations of the flow of multisized heterogeneous slurries in straight pipe and pipe Bends, PhD Thesis, Department of Applied
Mechanics, I.I.T. Delhi, 1991] and Seshadri et al. [V. Seshadri, R.C. Malhotra, S. Anand, Hydraulic transportation of mineral ores, a state of
the art report, Internal Report, Fluid Mechanics Laboratory, I.I.T. Delhi, 1975; V. Seshadri, R.C. Malhotra, Rheological properties of iron ore
slimes slurries at high concentrations, Internal Report, Fluid Mechanics Laboratory, I.I.T. Delhi, 1980] data with multisized particulate iron
ore slimes, copper tailings and zinc tailings slurries flowing through 105 and 55 mm diameter pipelines have been considered. Based on
extensive analysis of experimental data, the Karabelas [AIChE J. 23 (1977) 426] model for prediction of concentration profile has been
modified. On the basis of predicted concentration profiles, it is concluded that solid concentration at the pipe bottom reaches approximately
3.0 times the product of efflux concentration and static settled concentration by volume at deposition velocity. D 2002 Elsevier Science B.V.
All rights reserved.
Keywords: Deposition velocity; Slurry transportation pipeline; Concentration profile; Particle size distribution; Static settled concentration; Particle diffusion
coefficient
1. Introduction
The deposition velocity is one of the important parameter
that must be accurately known for the optimized design of a
slurry transportation pipeline. The significance of this
velocity is that it represents the lowest speed at which slurry
pipelines can operate. The magnitude of this velocity is
strongly dependent on particle size, particle density, solid
concentration and pipe diameter.
Many attempts have been made to propose a generalized
correlation for the prediction of deposition velocity. At the
last count, there are close to 60 correlations for the pre-
diction of the deposition velocity [6]. Most of them are
based on the Froude criterion F
L
and are limited in range of
applicability. Most of them are based on considering uni-
form size particles. One of the earliest correlation is the one
proposed by Durand [7] is as follows:
V
D
F
L
2gD
q
s
q
l
q
l
_ _ _ _
1=2
1
where D is the pipe diameter, q
s
the particle density, q
l
the
liquid density, g the acceleration due to gravity and F
L
is the
constant for a given system but varies from system to
system as a function of particle size d and solid concen-
tration C
v
. Durand [7] established the manner in which F
L
varied with concentration and particle size by plotting
particle size vs. F
L
curves for different efflux concentra-
tions. He concluded that for particle sizes up to 1 mm, both
concentration and particle size have an effect on the value of
F
L
. For particles greater than 1 mm, the dependence
becomes weaker. The sediment size and concentration have
no effect on the deposition velocity in case of uniform
sediments provided the material is coarser than 2.0 mm. For
0032-5910/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved.
PII: S0032- 5910( 02) 00031- 1
*
Corresponding author. Tel.: +81-93-884-3160; fax: +81-93-871-8591.
E-mail address: kaushal@mech.kyutech.ac.jp (D.R. Kaushal).
www.elsevier.com/locate/powtec
Powder Technology 125 (2002) 89101
a given system of uniform particles larger than 2 mm, F
L
becomes 1.34.
Durand [7] has also concluded that F
L
is smaller for
nonuniform sediments than for uniform sediments and there
is a significant effect of concentration on F
L
(even for coarse
material) in case of nonuniform sediments.
Wasp et al. [8,9] compared the Durand correlation given
by Eq. (1) for the prediction of deposition velocity with sand
water slurries data of Durand [7], Sinclair [10] and Wicks
[11]. Wasp et al. concluded that Durand correlation gives
excellent prediction with the exception of those Wicks,
which were for low solid concentration (1% by volume)
and a varying fluid density and viscosity. Wasp et al. derived
the following equation for the deposition velocity of dilute
suspensions in fully turbulent flow on the basis of Wicks
[11] data.
V
D
1:87
d
D
_ _
1=6

2gD
q
s
q
l
1
_ _

2
The above equation describes the behaviour of 1% dilute
suspension. For higher concentration, Wasp et al. [9] modi-
fied the Durand relation as follows
V
D
4:0
d
D
_ _
1=6
C
v

1=5

2gD
q
s
q
l
1
_ _

3
where C
v
is the volumetric concentration of the suspension.
Sinclair [10] is one of the few investigators who have
systematically varied the density of the suspended particles
over a wide range. He presented a plot of F
L
versus
concentration C
v
for suspensions of coal and iron particles
and compared it with the Durand correlation. On the basis of
his plot, Sinclair concluded that the Durand correlation is
inadequate.
Hence, the present work is envisaged on the deposition
velocity considering the experimental data of deposition
velocity for the flow of multisized particulate slurries with
wide range of particle diameters and solid density.
It is well known that at very high transportation veloc-
ities, the solid particles are more or less uniformly dis-
tributed across the cross section of the pipe. As the
transportation velocity is decreased, the solid distribution
becomes more and more nonuniform. The solid concen-
tration at the bottom increases monotonically as the trans-
portation velocity reduces. Simultaneously, the solid
particles at the bottom start slowing down due to strong
interaction effects present at high concentration levels.
Thus, it is logical to expect that when the solid concen-
tration of the bottom reaches a limiting value, the solid
particles come to rest and deposition start occurring. This is
the mechanism behind the methodology adopted in the
present study.
For the computation of deposition velocity, it is essential
to establish the criteria of relating the concentration at the
bottom of the pipe with the static settled concentration. In
the absence of any reliable correlation for the prediction of
deposition velocity for multisized particulate slurries, an
attempt has been made to develop a methodology for
estimation of deposition velocity based on the concentration
limit of the slurry in the bed. To prove this hypothesis,
concentration profiles have been predicted based on modi-
fied Karabelas [5] model at the measured deposition veloc-
ity. For this purpose, Kaushal [1], Mukhtar [2] and Seshadri
[3,4] data with multisized particulate iron ore slimes, copper
tailings and zinc tailings slurries flowing through 105 and
55 mm diameter pipelines have been considered.
2. Brief description of Karabelas model
OBrien [12] and Rouse [13] presented following diffu-
sion equation for the prediction of concentration distribution
e
s
dC
dy
wC 0 4
where w is the settling velocity, C the concentration, e
s
the
particle diffusivity and (dC/dy) is the concentration gradient.
Eq. (4) simply states that the flux of solid particles due to
gravity wC is balanced by the upward flux due to turbulent
diffusion e
s
(dC/dy).
The completeness of the model represented by Eq. (4)
has been questioned by Hunt [14]. He has shown that the
main deficiency of this model lies in the fact that the law of
mass conservation has been essentially disregarded by not
taking into account the volume of particles in suspension.
Hunt has developed a more complete set of diffusion
equations, in which he has taken into account the liquid
volume displaced upwards by an equal volume of particles
moving in the lower strata of a horizontal flow field as given
below:
e
s
@C
j
@y
C
j
w
j
v
y
0, j 1,2, . . . ,n 5
where v
y


n
i1
w
i
c
i
is the liquid velocity in vertical
direction, C
j
the concentration of the jth particle size and
w
j
is the settling velocity of the jth particle size.
Obviously, Eq. (5) proposed by Hunt [14] are more
suitable than Eq. (4) for studies of relatively concentrated
suspensions. Another major advantage of Hunts equations
is that they can be directly applied to suspensions of wide
particle size distribution.
Karabelas [5] obtained the general solution of system of
Eq. (5) proposed by Hunt [14] for the flow of multisized
particulate slurries through horizontal pipe as given below:
C
j
y
G
j
expw
j
f y
1

n
i1
G
i
expw
i
f y
; j 1,2, . . . ,n 6
D.R. Kaushal et al. / Powder Technology 125 (2002) 89101 90
where f( y) = mdy/e
s
( y), G
j
is a set of coefficients characteristic
of each size fraction but independent of space coordinates and
w
j
is the settling velocity of mean diameter particle of the jth
size fraction. In order to proceed in the development of the
distribution function C
j
( y), the following two assumptions
are made:
(i) The dimensionless eddy diffusivity n is a constant
independent of solid concentration and space coordinates,
that is
e
s
nRu* 7
where R is the radius of pipe, u* the shear velocity and e
s
the
particle diffusivity.
(ii) The solids concentration C
j
( y) is a function of
vertical coordinate y only.
Substitution of value of e
s
given by Eq. (7) into Eq. (6)
leads to
C
j
y*
G
j
expk
j
y*
1

i
G
i
expk
i
y*
; j 1,2, . . . ,n 8
where k
j
=(w
jo
)/(nu*), w
jo
the unhindered settling velocity of
mean diameter particle of the jth size fraction and y*=( y/R)
varies from 1 to + 1.
A transformation is required in order to determine the
parameter G
j
. For that, first, the following local and mean
relative concentrations are defined as:
v
j
y
C
j
y
1

i
C
i
y
; v
j

C
vjf
1

i
C
vif

C
vjf
1 C
vf
;
j 1,2, . . . ,n 9
where C
vf
is the efflux or overall average concentration by
volume, C
vjf
and C
j
( y) are the average and local concen-
tration of the jth size fraction, respectively. In terms of the
local relative concentration v
j
( y), solution (8) can be
expressed as follows:
v
j
y* G
j
expk
j
y*; j 1,2, . . . ,n: 10
Now it is assumed that the flow is steady, the mean
concentration of each particle size C
vjf
in the pipe cross
section is constant and already known. Thus, the mean
relative concentration v
j
is constant too. Therefore, the
integration of Eq. (10) over the pipe cross-sectional area A
leads to:
v
j

1
A
_
A
v
j
y*dA G
j
1
A
_
A
expk
j
y*d A
G
j

v
j
Ek
j

; j 1,2, . . . ,n 11
where Ek
j

1
A
_
A
expk
j
y*d A.
The final solution for the concentration distribution for
each size fraction as a function of y* is:
C
j
y*
v
j
Ek
j

expk
j
y*
_ _
1

n
i1
v
j
Ek
j

expk
i
y*
_ _
1
;
j 1,2, . . . ,n: 12
A computer program for the Karabelas final solution
given by Eq. (12) was developed. The solid particles were
divided into six size fractions and the unhindered settling
velocity w
jo
for the mean diameter of each size fraction was
calculated using the drag relationships given in Table 1,
where R
ed
is the particle Reynolds number. The value of
shear velocity u
*
required in the Karabelas model is
calculated using measured pressure drop.
3. Description of experimental facilities used in the
collection of experimental data for the present study
The schematic layout of the pilot plant test loop used in
the present study is shown in Fig. 1. The major components
are mixing tank, slurry pumps, two pipe loops of 60 m
length, bypass line and measuring tank. The test loop is laid
horizontally in the Fluid Mechanics Laboratory of I.I.T.
Delhi.
The mixing tank, a hopper shaped container, was used
for preparing the slurry. The tank is fabricated from 4-mm-
thick mild steel sheets and has an overall height of 2 m. It
has a square shape at the top (1.25 1.25 m) and has an
overall capacity of 1.65 m
3
. This tank is also provided with
a stirrer arrangement, which kept the slurry well mixed
during experimentation. The slurry from this tank was
pumped in either of the two pipe loops, namely the 105
mm loop and the 55 mm loop by separate slurry pumps. The
slurry in the 105 mm loop was pumped by a slurry pump
with Ni hard casing (WILFLEY, Model 100 K, Manufac-
tured by M/s Dorr Oliver) driven by a 50 hp motor through a
V belt drive system. The flow in the 55 mm loop was
achieved by a similar type of pump having a smaller
Table 1
Drag relationships
Fall regime and range of particle
Reynolds number (R
ed
)
Relation for drag
coefficient (C
D
)
Stokes law
R
ed
V1.0 C
D
= 24R
ed
1
Intermediate
1 <R
ed
V1000 C
D
= 24R
ed
1
(1 +0.15R
ed
0.687
)
Newtons law
1000 < R
ed
V2 10
5
C
D
= 0.44
D.R. Kaushal et al. / Powder Technology 125 (2002) 89101 91
capacity. The pumps had the capacity to achieve the desired
head and discharge needed for simulating the conditions in
the prototype pipeline. A bypass line with a valve is
provided in each loop, which allowed better control over
the flow rates in the main pipe loops while at the same time
ensuring that the deposition did not occur in the suction
pipes during experiments at low velocities in the pipe loop.
The flow through bypass line also ensured thorough mixing
of the slurry in the mixing tank especially when the flow
rate through the pipe loop was small. The pipe loops started
just downstream of the pumps and were provided with an
AUDCO valve (plug type) for controlling the flow rate.
These pipe loops contained bends and observation chamber.
At the exit of the pipe loop, a diverter has been provided
which facilitates the diversion of the flow to the measuring
tank. The measuring tank is again in the shape of a hopper.
The height of the tank is 1.5 m with the top being a square
of 1 1 m with total volume capacity of 1.25 m
3
. This tank
is also connected to the suction of the pump through a 50
mm plug valve to facilitate the transfer of slurry from the
measuring tank to the mixing tank. The measuring tank is
also provided with a stirrer arrangement. The diverter
system at the end of the pipe loop facilitated the diversion
of slurry flowing out of the pipeline into the measuring tank
for any given length of time. The duration for which the
flow is diverted is measured by an electronic stop watch
having a resolution of 0.01 s. The volume of slurry collected
during this interval is measured by noting the rise in the
level of the slurry accurately after allowing sufficient time
for the level to stabilize. The area of the cross section of the
measuring tank was accurately determined and from this
data, the flow rate of slurry could be determined to an
accuracy better than F1.0%. This method was used to
calibrate the magnetic flow meter installed in the pipeline
for regular monitoring of the flow rate.
For measuring pressure drop, pressure taps along with
separation chambers are provided at distance of 19.4 m in
the pipeline. Separation chambers are provided at each
pressure tap for interface separation of slurry and mano-
metric fluid, water being the intermediate fluid. For better
accuracy, the pressure drop along the pipeline is measured
by an inclined mercury manometer.
The concentration distribution has been measured using a
sampling tube having a 4 6 mm rectangular slot, 3 mm
above the end to collect representative samples in the pipe
line. Samples are collected from different heights from
bottom of the pipe in the vertical plane of the cross section
to determine the concentration profile under near isokinetic
conditions. During the collection of concentration samples
at various locations, it was ensured that the flow of the
slurry through the sampling tube outlet is continuous and
uniform. If the tube got choked, high-pressure water was
used to open it. Further sufficient time was allowed before
sample collection in order to ensure steady state conditions.
The sampling tubes are mounted on vernier type of travers-
ing mechanism to enable traversing of the tubes from the top
to the bottom in vertical plane and its location can be
accurately measured.
At the end of the pipe loop, a sampling point is provided in
the vertical portions for collecting the efflux sample. The
efflux concentration of the slurry flowing through the pipe-
line was monitored by measuring the density of efflux
sample. The efflux sample was stored in a bottle for deter-
mination of particle size distribution.
In the straight pipeline, a small length of perspex pipe
was provided (designated as observation chamber) to estab-
Fig. 1. Schematic diagram of pilot plant test loop.
D.R. Kaushal et al. / Powder Technology 125 (2002) 89101 92
lish the deposition velocity of the slurry in the pipeline by
observing the motion of the particles at the bottom of the
pipeline.
The static settled concentration is measured by allowing
the slurry of intermediate concentration to settle in a meas-
uring jar till the level of the settled solids becomes static.
Table 2
Experimental data used in the present study
Author Material Specific
gravity, G
Diameter
of pipe
(mm), D
Static settled
concentration
by volume
(%), C
vss
Mean particle
diameter of
efflux sample
(Am), d
50
Efflux
concentration
range (%), C
vf
Flow velocity
range
(m/s), V
m
Pressure drop
range
(m/mwc),
H
f
/L
Kaushal [1] Zinc tailings 2.82 105 33.9 34.95 3.826 23.5 0.0620.214
Mukhtar [2] Zinc tailings 2.597 105 28.6 69.24 4.0925.8 1.483.08 0.0360.147
Seshadri [3,4] Iron ore slimes 4.2 105 and 55 48.58 13.5432.13 518 1.72.36 0.07140.1032
Copper tailings 2.7 45.6 23.0342.66 527 1.221.71 0.0340.065
Zinc tailings 2.6 46.5 36.9363.54 1013 1.31.38 0.05320.0636
Fig. 2. Measured (by Kaushal [1]) and predicted (by modified and Karabelas [5] model) concentration profiles at different efflux concentrations for the flow of
zinc tailings slurry through a 105-mm-diameter pipe with a flow velocity of 2 m/s.
D.R. Kaushal et al. / Powder Technology 125 (2002) 89101 93
4. Experimental data used in the present study
In the present study, 15 sets of Kaushal [1] data, 15 sets
of Mukhtar [2] data and 20 sets of Seshadri [3,4] data with
multisized particulate iron ore slimes, copper tailings and
zinc tailings slurries flowing through 105 and 55 mm
diameter pipelines have been considered. The particle size
range used in all these data was wide enough to cover the
range expected in commercial slurries. All these data are
tabulated in Table 2. In Kaushal [1] and Mukhtar [2] data,
concentration profiles have been reported for different efflux
concentrations at different flow velocities, always higher
than the deposition velocity. These data are used for
proposing the modified Karabelas [5] model for the pre-
diction of concentration profile and are shown graphically in
Figs. 29. However, Seshadri [3,4] measured only the
deposition velocity for slurries with different particle size
distribution and efflux concentrations. In the present study,
Seshadri [3,4] data along with modified Karabelas model
have been used for correlating bottom concentration with
efflux concentration and static settled concentration at the
deposition velocity.
5. Comparison between measured and predicted con-
centration profiles based on Karabelas [5] model
Figs. 29 present the comparison between measured and
predicted concentration profiles for Kaushal [1] and Mukh-
tar [2] data, where y V= y/D is the reduced vertical coor-
dinate from pipe bottom, C
v
( y V) the concentration by
volume at y Vand V
m
the flow velocity of slurry. It is seen
that for almost all the data except for a few data with lower
efflux concentrations, the Karabelas model fails to predict
Fig. 3. Measured (by Kaushal [1]) and predicted (by modified and Karabelas [5] model) concentration profiles at different efflux concentrations for the flow of
zinc tailings slurry through a 105-mm-diameter pipe with a flow velocity of 2.75 m/s.
D.R. Kaushal et al. / Powder Technology 125 (2002) 89101 94
accurately. The agreement between measured and predicted
concentration profiles at lower efflux concentration was
good for all the velocities. It is also found that the errors in
Karabelas model prediction are systematic, overestimating
the concentration at the bottom and underestimating the
concentration at the top.
From the quantitative comparison, it is clear that the
Karabelas model is not taking into consideration the
changes in fluid and flow properties that occur with increase
in efflux concentration. The causes for failure of Karabelas
model at higher efflux concentrations are identified as
(i) In his final solution given by Eq. (12), he used
unhindered settling velocity in the calculations not account-
ing for the effect of concentration, particle size distribution
and pipe walls. Richardson and Zaki [15] have already
given a correlation for hindered settling velocity by taking
into consideration the above factors.
(ii) Particle diffusivity, e
s
, is assumed as constant and
equal to liquid diffusivity, e
l
. Further, e
s
is assumed to be
independent of solid concentration, particle size and shape.
Hence, an attempt has been made in the present study to
modify the model to take into account some of these
effects.
6. Description of modified Karabelas model
The general solution given by Eq. (6) is used instead of
approximately closed form solution given by Eq. (12). A
computer program for Karabelas general solution with
Fig. 4. Measured (by Kaushal [1]) and predicted (by modified and Karabelas [5] model) concentration profiles at different efflux concentrations for the flow of
zinc tailings slurry through a 105-mm-diameter pipe with a flow velocity of 3.5 m/s.
D.R. Kaushal et al. / Powder Technology 125 (2002) 89101 95
modifications has been developed. The modifications incor-
porated are given below.
(1) The hindered settling velocity w
j
for the mean
diameter of each size fraction is calculated using the formula
given by Richardson and Zaki [15] as follows:
w
j
w
jo
1 C
vf

Z
13
where,
for 0:002 < R
ed
V0:2; Z 4:65 19:5
d
j
D
_ _
0:2 < R
ed
V1:0; Z 4:35 17:5
d
j
D
_ _ _ _
R
0:03
ed
1:0 < R
ed
; Z 4:45 18:0
d
j
D
_ _ _ _
R
0:1
ed
where d
j
is the mean diameter of the jth size fraction.
In the current modifications, the provision for allowing
any variation of w
j
across the pipe cross section has also
been incorporated. It is to be noted that w
j
is a function of
local solid concentration that is not known a priori. Hence,
in the first iteration, w
j
corresponding to efflux concentra-
tion is used to compute the overall concentration profile. In
the subsequent iterations, the computed values of local
concentrations are used to calculate w
j
( y) at each point.
This procedure is repeated until convergence is obtained.
(2) Karabelas has assumed the variation of particle
diffusivity (or sediment diffusion coefficient) as
e
s
be
l
where b is the dimensionless particle diffusivity (or dimen-
sionless particle diffusion coefficient) and e
l
is the liquid
diffusivity (or momentum transfer coefficient).
Further, he also assumed that e
l
and b are constant across
the cross section of the pipe. The assumption of liquid
diffusivity and b being constant is not valid. The liquid
diffusivity is not constant across the pipe cross section due
to turbulent motion and its variation depends on several
parameters like pipe diameter, flow conditions, etc. In order
to improve the accuracy of prediction, realistic variation of
e
l
across the cross section of the pipe has been incorporated.
As suggested by Longwell [16], the liquid diffusivity in
turbulent pipe flow is given by:
e
l
0:369u*
y
R
1
y
R
_ _
for 0Vy=DV
e
l
0:0775Ru* for 0:337Vy=DV0:663
e
l
0:369u*
y
R
1
_ _
2
y
R
_ _
for 0:663Vy=DV1:0:
14
Fig. 5. Measured (by Mukhtar [2]) and predicted (by modified and Karabelas [5] model) concentration profiles at different flow velocities for the flow of
zinc tailings slurry through a 105-mm-diameter pipe with an efflux concentration of 4.09% by volume.
D.R. Kaushal et al. / Powder Technology 125 (2002) 89101 96
Fig. 7. Measured (by Mukhtar [2]) and predicted (by modified and Karabelas [5] model) concentration profiles at different flow velocities for the flow of
zinc tailings slurry through a 105-mm-diameter pipe with an efflux concentration of 10.52% by volume.
Fig. 6. Measured (by Mukhtar [2]) and predicted (by modified and Karabelas [5] model) concentration profiles at different flow velocities for the flow of
zinc tailings slurry through a 105-mm-diameter pipe with an efflux concentration of 8.83% by volume.
D.R. Kaushal et al. / Powder Technology 125 (2002) 89101 97
Fig. 9. Measured (by Mukhtar [2]) and predicted (by modified and Karabelas [5] model) concentration profiles at different flow velocities for the flow of
zinc tailings slurry through a 105-mm-diameter pipe with an efflux concentration of 25.8% by volume.
Fig. 8. Measured (by Mukhtar [2]) and predicted (by modified and Karabelas [5] model) concentration profiles at different flow velocities for the flow of
zinc tailings slurry through a 105-mm-diameter pipe with an efflux concentration of 16.84% by volume.
D.R. Kaushal et al. / Powder Technology 125 (2002) 89101 98
To determine the variation of particle diffusivity b with
concentration across the pipe cross section, the following
method has been followed.
For a given efflux concentration and flow velocity, the
value of b
opt
is obtained by iteration method so that the
deviation between measured concentration and predicted
concentration profiles, i.e.,

9
i1
C
imeasured
C
ipredicted

2
is
minimum. The value of b
opt
for the different sets of
experimental data reported by Kaushal [1] and Mukhtar
[2] have been evaluated and its variation with C
vf
/C
vss
is
presented in Fig. 10, where C
vss
is the static settled con-
centration.
It shows that b
opt
increases monotonically with C
vf
/C
vss
and has a minimum value equal to 1.0. The trend in the
variation of b is clear. At low values of C
vf
/C
vss
(less than
0.1), b is approximately equal to unity. The values of b
increase as C
vf
/C
vss
increases, and at higher values of C
vf
/
C
vss
( > 0.4), the rise is almost exponential. Further, b
opt
was
found to be insensitive to flow velocity at any given efflux
concentration. Therefore, the expression for b in terms of
C
vf
/C
vss
was obtained by least square curve fitting method
and is given below:
b 1:0 0:12504 e
4:22054C
vf
=C
vss
: 15
C
vf
/C
vss
is used as the correlating parameter since C
vss
represents the highest achievable concentration by gravity
settling. Physical meaning and the reason for such a func-
tional relationship for b as shown in Fig. 10 and given by
Eq. (15) is discussed below.
The value of b approaches unity as C
vf
approaches zero
and hence the assumption of Karabelas that e
s
is equal to
e
l
is valid. Now as the value of C
vf
increases, the value of
the ratio C
vf
/C
vss
approaches unity, which implies that e
s
increases as concentration increases. An increase in the
value of b would result in the increase of e
s
and hence a
reduction in the concentration gradient in Eq. (5). Thus, at
higher concentrations, solid distribution across the pipe
cross section would tend to get uniform, which is borne
out from experimental observations also. Hence, the
empirical correlation for b takes into account the inhib-
ition of settling of solids due to interference effect at
higher concentrations.
Eq. (15) has been used to calculate the local values of
b in the modified model considering C
vf
as the local
concentration. It is to be noted that in Eq. (6), both w
j
and e
s
are complicated function of local solid concen-
tration, and hence the calculation procedure has to be
iterative.
7. Comparison between measured and predicted
concentration profiles based on modified
Karabelas model
The overall concentration profiles are predicted by
modified Karabelas model for 30 sets of experimental
data reported by Kaushal [1] and Mukhtar [2] and are
shown in Figs. 29. It is observed that for almost all the
data, the modified Karabelas model gives an almost exact
fit between the measured and predicted overall concen-
tration profiles. It may be recalled that in the earlier
predictions by the Karabelas [5] model (Figs. 29), the
deviations were systematic and were very large particu-
larly at higher concentrations. It shows that the predictions
by modified Karabelas model for overall concentration
profile are better than the predictions by Karabelas [5]
model.
8. Development of an equation for bottom
concentration of pipe at deposition velocity
The steps involved in development of an equation for
concentration at pipe bottom at deposition velocity are as
follows:
8.1. Step 1
Concentration by volume at y V= 0.1, C
v(0.1)
is calcu-
lated at measured deposition velocity by using modified
Karabelas model proposed in the present study for all 20
sets of Seshadri [3,4] data for deposition velocity. These Fig. 10. C
vf
/C
vss
vs. b
opt
curve for Kaushal [1] and Mukhtar [2] data.
D.R. Kaushal et al. / Powder Technology 125 (2002) 89101 99
data are given in Table 2 and have already been dis-
cussed.
8.2. Step 2
The ratio of volumetric concentration at the bottom of
pipe and volumetric static settled concentration of the solid
particles C
v(0.1)
/C
vss
is calculated and are plotted against
C
vf
in Fig. 11.
8.3. Step 3
From Fig. 11, it is observed that C
v(0.1)
/C
vss
varies
linearly with C
vf
at deposition velocity. Hence, a linear
equation on the basis of least square curve fitting method
was determined for this variation as follows
C
v0:1
=C
vss
3C
vf
16
or at deposition velocity, C
v(0.1)
= 3(C
vf

vss
).
Hence, the deposition velocity for transportation of
commercial slurries through pipeline is the flow velocity
corresponding to the bottom concentration equal to three
times the product of efflux concentration and static settled
concentration by volume.
9. Conclusions
As a result of this study, the following conclusions have
been drawn:
(1) The modifications incorporated in the Karabelas
model result in a significant improvement in the accuracy
of prediction of overall concentration profile in pipe. At
high solid concentrations, it is essential to incorporate the
effect of solid concentration on particle settling and turbu-
lent diffusivity.
(2) From the modified Karabelas model, an overall
concentration profile C
v
( y V) across the pipe cross section
at the measured deposition velocity is obtained. From the
analysis of predicted concentration profiles, it is concluded
that the ratio of concentration at the bottom to the static
settled concentration can be used effectively as a tool to
evaluate the deposition velocity.
(3) Deposition occurs when the solid concentration at the
bottom (C
v(0.1)
) reaches approximately three times the
product of efflux concentration and static settled concen-
tration by volume.
(4) It is recommended for further study to find out the
dependence of particle diffusion coefficient (b) on particle
size, particle shape and other fluidparticle relationships
besides efflux concentration of slurry.
List of symbols
C
D
Drag coefficient
C
j
Concentration by volume of the jth size fraction
C
v
( y V) Concentration by volume at y V
C
v
Concentration by volume
C
vf
Efflux or overall average concentration by volume
C
vjf
Average concentration by volume of the jth size
fraction
C
vss
Static settled concentration by volume
d Diameter of particle
D Diameter of pipe
g Acceleration due to gravity
R Pipe inside radius
R
ed
Particle Reynolds number
u* Shear velocity
v
y
Liquid velocity in vertical direction
V
D
Deposition velocity
V
m
Flow velocity of slurry
w
j
Hindered settling velocity of the jth size particle
w
jo
Unhindered settling velocity of the jth size
particle
y Vertical coordinate
y V y/D, reduced vertical coordinate from pipe bot-
tom
y* y/R, reduced vertical coordinate from pipe centre
Greek letters
b Ratio of particle to liquid diffusivity (particle
diffusion coefficient)
e
l
Liquid diffusivity
e
s
Particle diffusivity
q
l
Density of liquid
q
s
Density of solid
n Dimensionless eddy diffusivity
Fig. 11. Variation of C
v(0.1)
/C
vss
with C
vf
for the Seshadri [3,4] data.
D.R. Kaushal et al. / Powder Technology 125 (2002) 89101 100
References
[1] D.R. Kaushal, Prediction of particle distribution in the flow of multi-
sized particulate slurries through closed ducts and open channels, PhD
Thesis, Department of Applied Mechanics, I.I.T. Delhi, 1995.
[2] A. Mukhtar, Investigations of the flow of multi-sized heterogeneous
slurries in straight pipe and pipe bends, PhD Thesis, Department of
Applied Mechanics, I.I.T. Delhi, 1991.
[3] V. Seshadri, R.C. Malhotra, S. Anand, Hydraulic transportation of
mineral ores, a state of the art report, Internal Report, Fluid Mechanics
Laboratory, I.I.T. Delhi, 1975.
[4] V. Seshadri, R.C. Malhotra, Rheological properties of iron ore slimes
slurries at high concentrations, Internal Report, Fluid Mechanics Lab-
oratory, I.I.T. Delhi, 1980.
[5] A.J. Karabelas, Vertical distribution of dilute suspensions in turbulent
pipe flow, AIChE J. 23 (4) (1977) 426434.
[6] R.R. Dighade, Estimation of deposition velocity in the flow of multi-
sized particulate slurries through pipes, M.Tech. (Hydraulic Engineer-
ing) Project Thesis, VRCE Nagpur, 1999.
[7] R. Durand, The hydraulic transportation of coal and other materials in
pipe, Paper IV, Proc. Colloquium on the Hydraulic Transport of Coal,
Held by National Coal Board, London, Nov., 1952, pp. 3952.
[8] E.J. Wasp, T.C. Aude, R.H. Seiter, J.P. Kenny, R.B. Jacques, Depo-
sition velocity, transition velocities and spatial distribution of solids
in slurry pipelines, Paper H4, Proc. Hydrotransport 1, BHRA Fluid
Engineering, Coventry, England, 1970, pp. 5376.
[9] E.J. Wasp, J.P. Kenny, R.L. Gandhi, Solid Liquid Flow Slurry Pipeline
Transportation, first ed., Trans. Tech. Publications, Clausthal, Ger-
many, 1977.
[10] C.G. Sinclair, The limit deposit velocity of heterogeneous suspensions
interactions between fluid particles, International Conference on
Chem. Eng., London, 1962.
[11] M. Wicks, Transportation of solids at low concentration in horizontal
pipes, ASCE Intl. Symp. on SolidLiquid Flow in Pipe, University
of Pennsylvania, 1968.
[12] M.P. OBrien, Review of the theory of turbulent flow and its relations
to sediment transportation, Trans., Am. Geophys. Union 14 (1933)
487491.
[13] H. Rouse, Modern conceptions of the mechanics of fluid turbulence,
Trans. ASCE 102 (1937) 463505.
[14] J.N. Hunt, The turbulent transport of suspended sediment in open
channels, R. Soc. London, Proc., Ser. A 224 (1158) (1954), pp.
322335.
[15] J.F. Richardson, W.M. Zaki, Sedimentation and fluidization, Part I,
Trans. Inst. Chem. Eng. 32 (1954) 3553.
[16] P.A. Longwell, Mechanics of Fluid Flow, McGraw Hill, New York,
1977.
D.R. Kaushal et al. / Powder Technology 125 (2002) 89101 101

Вам также может понравиться