-
SLURRY-FLOW PRESSURE DROP IN PIPES WITH MODIFIED WASP METHOD
Tanaji Mali, Andritz Technologies Pvt. Ltd, Bangalore, India
Vijay Khudabadi, Andritz Technologies Pvt. Ltd, Bangalore,
India
Rana A.S, Indian School of Mines, Dhanbad, India Arihant Vijay,
Indian School of Mines, Dhanbad, India
Adarsh M.R, University of Petroleum and Energy Studies,
Dehradun, India
Presented at SME Annual Meeting/Exhibit, February 24-26, 2014,
Salt Lake City, UT, USA
Abstract
Over the last many decades, a significant amount
of research has gone into the domain of slurry transport.
However, design engineers still face many challenges with respect
to prediction of pressure drop, critical velocity and other design
parameters as a function of Solids % and Particle Size Distribution
(PSD). The industry requirement is to transfer the slurry at the
maximum concentration as possible (above 30% (volume %)) to make
slurry transport more economically viable and to reduce water
con-sumption. To facilitate the design and scale-up of slurry
transport in pipelines and in process plants, there is a need for a
correlation that can predict slur-ry pressure drops over a wide
range of operating conditions and physical properties of different
slur-ries. The objective of this study is to overcome the limited
range of applicability and validity of existing correlations and to
develop a generalized but more rigorous correlation applicable to a
wider range of slurry systems. The existing Wasp et al. (1977)
meth-od is based on multi-phase flow modeling approach. This study
attempts to modify this approach by con-sidering material-specific
values of Durands equa-tion co-efficient and by defining flow
regimes based on particle Reynolds number. When compared with
experimental data, the modified Wasp method pro-posed in this study
predicts the pressure drop for slurry flows more accurately than
other available correlations. Also, the proposed method requires
minimal test/experimental data for a particular slurry system and
can be extended over different input con-ditions.
An iterative computer algorithm is developed to calculate the
critical settling velocity and pressure drop in a pipe as a
function of Solids % and PSD. The solution method can easily be
implemented in design-ing slurry pipes, design validation, and
studying the different slurry transport scenarios. The modified
method can also be extended to accurately predict pressure drops in
dynamic pressure flow networks used in commercial process
simulators.
Keywords: Particle size distribution, Slurry flow, volumetric
concentration, Pressure drop, Wasp Method, Drag Coefficient,
friction factor, critical velocity, Durands equation
Introduction
Slurry transport involves huge capital investment. Therefore, at
present many organizations throughout the world are carrying out
research and development to abate these costs. Literature survey
reveals that studies on slurry transport have followed one of these
three major approaches:
(a) The empirical approach (b) The rheological based continuum
approach (c) The multiphase flow modeling approach. Amongst the
above mentioned approaches, the
empirical approach is the simplest, and hence has been widely
used and applied. This has led to formu-lation of the correlations
for prediction of pressure drop and for delineation of flow
regimes. The rheo-logical approach is best applicable to slurries
of ultra-fine non-colloidal particles. The multiphase flow modeling
approach, which considers liquid, particle and boundary interaction
effects, requires significant computational effort and is best
suitable for describ-ing heterogeneous solid-liquid mixture
flows.
In this study, multiphase flow modeling approach has been
followed. It considers various important design parameters such as
particle size distribution, volumetric concentration and pipe
roughness in pre-dicting the pressure drop. Slurry Flow in Pipe
For a pure liquid, the pressure drop in a pipe de-pends on the
flow velocity. The change of pressure drop with respect to flow
velocity is monotonic in nature. However, in case of slurries, it
is not mono-tonic (Vanoni 1975; Govier and Aziz 1977), as shown in
Figure 1. When the flow velocity is sufficiently high, all solid
particles are suspended with the parti-
www.andritz.com 1 of 13
-
2
cle distribution being homogeneous. As the velocity decreases
below V4 (see Figure 1) (Govier and Aziz 1977), all of the solids
are still suspended, but their distribution becomes heterogeneous.
When the ve-locity further decreases to the critical velocity V3,
some solids start to move along the pipe bottom as a bed load. At
this point, the pressure drop is usually minimum. When the velocity
decreases further very few solids are transported as the suspended
load, and more amount of solid is transported as the bed load. At
further reduced velocity, V2, the bed load starts to generate a
stationary bed. The stationary bed further increases the apparent
pipe fraction factor, resulting in increased pressure drop.
Finally, at further reduced velocity, V1, all solids stop
moving.
Figure 1. Plot of transitional mixture velocity with pressure
drop
To achieve the optimum performance i.e. mini-mum pump pressure
requirement, the slurry should be transported at critical velocity
(V3). When a higher proportion of particles start to move as bed
load, a higher pump pressure is required to move them. If the pump
pressure is not high enough, danger of plugging the pipeline
arises. Thus, to overcome the risk of pipeline plugging, slurry
transfer through the pipeline must be operated above the critical
velocity, V3. Flow Regimes
In slurry transport, different patterns of solid movement are
observed depending upon the nature of the slurry and the prevailing
flow condition. As shown in Figure 2, the patterns are dependent on
the particle size, volumetric concentration of the solids and the
flow velocity. In horizontal pipes, the patterns can be
conveniently be classified into the following four regimes:
Homogeneous flow
This regime is also called symmetric flow. In this flow regime
there is a uniform distribution of solids about the horizontal axis
of the pipe, although it may not be exactly uniform. In this
regime, turbulent and other lifting forces are capable of
overcoming the net
body forces as well as the viscous resistance of the particles.
Heterogeneous flow
When the slurry velocity decreases, intensity of turbulence and
lift forces also decreases due to which, there is distortion of the
concentration profile of the particles. In this flow regime more of
the solids, particularly the larger particles are contained in the
lower part of the pipe. Thus, there is a concentration gradient
across the pipe cross section with a larger concentration of solids
at the bottom. This flow is also called asymmetric flow Saltation
flow
In this regime the slurry velocity is low and the solid
particles tend to accumulate on the bottom of the pipe, first in
the form of separated dunes and then as a continuous moving bed.
Stationary bed flow
In this regime the slurry velocity is further re-duced which
leads to the lowermost particles of the bed being nearly
stationary. Thus, the bed thickens and the bed motion is due to the
movement of the uppermost particles tumbling over one another
(salta-tion).
Figure 2. Flow regimes of heterogeneous flows in terms of
particle size vs. velocity (after Shen,1970) and pictorial
representation of flows
www.andritz.com 2 of 13
-
3
Where
A represents saltation flow regime B represents heterogeneous
flow regime C represents homogeneous flow regime
Research Work & Different Methods
Over the years, two principal research works have
been developedone around the DurandCondolios approach and the
other around the Newitt approach. The former evolved gradually and
Wasp modified it for multilayer compound systems (Abulnaga, BE
2002). The latter gradually evolved to yield the two-layer model
(Abulnaga, BE 2002).
Wasp and Durand methods are useful tools for concentrations of
coarse particles up to 20% (volume fraction) (Abulnaga, BE 2002).
This covers, in fact, most dredged gravels and sands, coal in a
certain range of sizes, as well as crushed rocks (Abulnaga, BE
2002). It is also worth noting that Zandi and Govatos (1967) worked
on sand samples up to 22% (volume concentration).
The two-layer models have made it possible to work with
volumetric concentrations of 30% (Abulnaga, BE 2002). But these
models have many limitations and still considerable amount of work
has to be done to overcome these limitations.
Wasp Method
Wasp et al. (1977) method is the most widely used method for
slurry transport applications around the world because it is
applicable for all kinds of flow regime and accurately predicts the
pressure drop, considering the PSD of the slurry which is an
im-portant parameter accounting for pressure drop calcu-lation.
Wasp method is an improvement of Durand-Condolios approach which
predicts pressure drop accurately for both slurry systems in which
particles have narrow as well as wide size range.
Wasp method accounts for large particle size dis-tributions and
pressure drop by dividing the slurry into homogenous (due to
vehicle) and heterogeneous (due to bed formation) fractions. The
solids in the homogenous fraction increase the density and
viscosi-ty of the equivalent liquid vehicle. Wasp method also
considers the effect of pipe diameter and pipe rough-ness on
pressure drop in slurry pipes.
The iterative method proposed by Wasp et al. (1977) is
summarized as follows:
1. By using Durands equation, the total size fraction is divided
into a homogeneous and heterogeneous fraction.
2. The friction losses of the homogeneous frac-tion are
calculated based on the rheology of the slurry, assuming Newtonian
flow.
3. The friction losses of the heterogeneous frac-tion are
calculated using Durands equation.
4. A ratio, C/CA, is defined for the size fraction of solids
based on friction losses estimated in steps 2 and 3 (where C/CA is
the ratio of volumetric concentration of solids at 0.08D from top
to that at pipe axis).
5. Based on the value of C/CA, the fraction size of solids in
homogeneous and heterogeneous flows is determined.
Steps 2 to 5 are re-iterated until convergence of
the friction loss.
Although this method works well for water-coal mixture, it over
predicts pressure drop for mineral and rock slurry systems. To
overcome this limitation, the following two modifications have been
proposed in the paper:
1. The value of k, i.e., Durands equation coef-ficient, is
material-specific. For different ma-terials, different values of k
should be used and the value is calculated as described in
algorithm given in next section (step 9). This modification is
needed because the pressure drop predicted by the heterogeneous
part does not match with the experimental re-sults.
2. Different flow regimes have been defined based on the
assumption that particle size having Reynolds number less than 2
will al-ways contribute toward homogeneous losses and particle size
having Reynolds number greater than 525 will always contribute
to-ward heterogeneous losses in horizontal slur-ry flow pipe for
all flow velocities as pro-posed by Duckworth (Jacobs 2005).
These modifications have been described in the following
section. We refer to these modifications as the Modified Wasp
Model.
Algorithm for Modified Wasp et al. (1977) Method
The example below illustrates the modifications
proposed in this paper to the Wasp et al (1977) meth-od. This
algorithm has been developed in MS Excel and all the cases
described later in this work have been validated using this newly
developed algorithm. Example 1
Nickel ore slurry was tested in a 159-mm pipe-line with a
roughness coefficient of 0.045 at a weighted concentration of
26.3%. The results of pres-sure drop versus velocity are presented
in Table 1.
Table 1. Pressure drop versus Speed in a 159-mm ID Steel Pipe at
a Weight Concentration of
26.3% Velocity
(m/s) Pressure drop
(Pa/m) 1.5 175
www.andritz.com 3 of 13
-
4
1.9 270 2.3 360 2.7 525 3.1 688 3.5 847 4.0 1046
The particle size distribution of the ore is presented in Table
2.
Table 2. Particle Size versus Wt. % solids in the slurry
Particle Size (m)
Wt. %
-450 1.88 -200 2.2 -95 1.65 -61 1.17 -44 93.1
Step 1: Plot the PSD curve.
Based on the Particle size distribution (PSD) of solids present
in the slurry, plot the PSD curve. Parti-cle size vs. cum. wt. %
passing is presented in Table 3 and Figure 3 shows the PSD
curve.
Table 3. PSD Data of the solids presents in the slurry
Particle Size (mm)
Weight %
Cumulative Weight % Passing
-0.85 + 0.40 1.88 100 -0.40 + 0.20 2.20 98.12
-0.20 + 0.105 1.65 95.92 -0.105 + .044 1.17 94.27
-0.044 93.1 93.1 Total 100.00
Figure 3. Particle size distribution curve
Step 2: Determine cut size. dcut using Wasps modified Durand
equation.
This is the most important step as it determines the flow regime
of the slurry. The following is the correlation given by Wasp et
al. (1977) to calculate the cut size, dcut, i.e., the maximum size
of the particle that will remain suspended in the slurry
(pseudo-
homogeneous) at a given operating flow velocity of the slurry.
Any particle with size greater than dcut will settle down
(heterogeneous):
[ ]
= 340416
6
25253 LLS.
vcut
/)(gD*C*).(VDd
(1)
Where
v = Operating velocity of slurry (m/s) D = Pipe inside diameter
(m) Cv = Volumetric concentration of solid dcut = Cut size
s = Specific gravity of solid
L = Specific gravity of the liquid
Based on the cut size, i.e., dcut, four cases have been
considered. They are as following: Case 1: Cut size - dcut > dRe
max and dRe max < d100
Figure 4. Cut size for flow regime Case 1
Where, dRe max is the particle size having Reynolds number 525
and d100 is the maximum particle size present in the slurry.
In this case, particles having size greater than dRe max will
remain in the heterogeneous part and contrib-ute moving bed losses,
whereas particles having size less than dRe max will remain
suspended and contribute toward homogeneous losses.
This is based on the assumption proposed by Duckworth that the
minimum Reynolds number at which particles will settle by saltation
without contin-uous suspension is approximately 525 (Jacobs 2005).
Hence particles having Reynolds number greater than or equal to 525
will always be in the heterogeneous part of the mixture. This is
the proposed modification in the existing wasp method. As a result
of this modi-fication, the new cut size dcut will now be dRe max.
Case 2: Cut size - dcut < dRe min
Figure 5. Cut size for flow regime Case 2
Where, dRe min is the particle size having Reynolds
number 2.
www.andritz.com 4 of 13
-
5
In this case, particles having size less than dRe min will
remain suspended in the slurry and contribute toward the losses,
whereas particles having size greater than dRe min will settle down
and contribute toward the heterogeneous losses.
This is another proposed modification in the exist-ing wasp
method. As a result of this modification, the new cut size dcut
will now be dRe min. Case 3: Cut size - dcut lies between dRe max
and dRe min.
Figure 6. Cut size for flow regime Case - 3
In this case, particles having size less than dcut
will remain suspended and contribute toward homo-geneous losses,
whereas particles with size greater than dcut will settle down and
contribute toward heter-ogeneous losses. Case 4: Cut size - dcut is
> d100 and dRe max > d100
Figure 7. Cut size for flow regime Case 4
In this case, no particles will settle, and the total
loss will be due to the homogeneous part only and losses due to
the heterogeneous part will be zero. No iteration is needed in this
case; therefore, loss is cal-culated directly considering
pseudo-homogeneous flow. This usually happens when velocities are
very high and particle size present in the slurry is very
small.
For this particular example, the mean slurry ve-locity under
consideration is 1.9 m/s, solid specific gravity = 4.074, liquid
specific gravity = 1, and pipe diameter is 0.159 m.
Thus, calculated cut size, dcut = 152 micron. Step 3: Calculate
dRe max and dRe min to determine the flow regime of the slurry.
In order to calculate the particle size having Reynolds number 2
and 525, the following equation are used:
The drag co-efficient cD is calculated using the Reynolds number
by the following equation (Manfred Weber):
40424 .)Re
()Re
(cD ++= (2)
The single particle settling velocity vt is calculat-ed using
the Stokes equation:
)/(*)c
gd(*v LLS
D
Ret = 3
4 (3)
The Particle Reynolds number Re is then calcu-
lated :
L
Ret
v
dvRe = (4)
Where
dRe =Particle size having Reynolds number Re (Re min = 2 and Re
max = 525) L = kinematic liquid viscosity
To calculate the value of particle size having
Reynolds number 2 and 525, substitute the values in equations
(4), (3) and (2) and calculate dRe.
For this particular example, dRe max = 2.2 mm and dRe min =
0.150 mm; therefore, Case 3 of the flow re-gime will be considered.
Step 4: Calculate the friction losses of the homogeneous fraction
based on the rheology of the slurry, assuming Newtonian flow.
Friction losses of the homogeneous fraction are calculated using
Darcys formula:
Loss (Pa/m) = D 2
Vf m2
D (5)
Where
fD = Darcy friction factor. v = Mean velocity of slurry
(m/s).
m = Density of carrier fluid (kg/m3) (including
the particles less than dcut). D = Inside diameter of pipe
(m).
The Swain-Jain equation may be used in the range
of 5000 < Re < 107 to determine the friction coeffi-cient
of the homogeneous part of the mixture:
{ }29074573250
].mRe/../)i/Dlog[(
.Df
+= (6)
Where = Roughness coefficient (m). Di = Inside diameter of the
pipe (m).
Rem is the Reynolds number for the slurry which is
calculated using Thomas (1965) correlation for slurry viscosity
m as given below:
www.andritz.com 5 of 13
-
6
m
mmm
)DV(Re
= (7)
)6.16exp(00273.0
+05.10+5.2+1= 2l
m
vf
vfvf
C
CC
(8)
Where
Cvf = volumetric concentration of solids in ho-mogeneous part of
mixture
l = liquid viscosity m = slurry viscosity
For this particular example, dcut, as determined in step 2 is
152 micron. Therefore, from the PSD curve, 95% by weight solids are
less than 152 micron; hence, they contribute to the homogeneous
losses. Next, calculate the volumetric concentration Cvf of solids
in the homogeneous part of the mixture as (Abulnaga, BE 2002):
Cv total = Cw*( m / s ) (9)
Cv total = 26.3*(1244/ 4072) = 8.035%
Cvf = 0.95*Cv total = 7.633% Where
m = slurry density
s = solid density Cv total = total volumetric concentration of
sol-ids Cw = total solids concentration by weight
Therefore, out of total 8.035% solids by volume in
slurry, 7.633% solids by volume are in suspension and contribute
toward the homogeneous losses.
Next, m can be calculated using the Cvf and it is calculated as
1.2598 and corresponding Rem is 2956192. Substituting the value of
Rem in Swain-Jain equation to determine the fD, which is equal to
0.017.
Therefore, loss (Pa/m) due to homogeneous part of the mixture
from Darcys formula is calculated as 238.46 Pa/m. The lab test
measured 270 Pa/m; the losses due to the moving bed are therefore
31.54 Pa/m. Step 5: Calculate the redistributed particle size
considering the cut size dcut.
To calculate the losses due to the moving bed or heterogeneous
part of the mixture, the size distribu-tion of the heterogeneous
part (size greater than dcut) is divided into various size fraction
and then loss due to each size fraction is calculated and sum of
all the
size fraction loss gives the combined loss due to the moving bed
or heterogeneous part of the mixture.
In this particular example, dcut is 152 micron, therefore,
particle size between 850 micron and 152 micron contributed toward
heterogeneous losses. Redistributed particle size is presented in
Table 4.
Table 4. Redistributed particle size for heteroge-neous part of
the mixture
Size ( mm)
Average Particle
size (mm)
Cv in Sol-
ids (%)
Cv bed in the slurry
(at overall
solids Cv of mixture at 8.035%)
(%) -0.85
+ 0.40 0.63 1.88 0.151
-0.40 + 0.20
0.30 2.20 0.177
-0.20 +0.152
0.18 0.92 0.074
Total 5 0.402
Step 6: Determine the particle Reynolds number and drag
coefficient for each size range.
It is essential first to determine the drag coeffi-cient and the
particle Reynolds number for each size fraction to calculate the
loss due to each size fraction in the heterogeneous part of the
mixture.
To calculate the particle Reynolds number, the density of the
slurry m, viscosity of the slurry m and the speed of the carrier
fluid V are used. The equation to calculate the particle Reynolds
number is as fol-lowing:
m
mPP
VdRe
= (10)
Where
dp = the average particle size of each size frac-tion.
To calculate the drag coefficient, CD, of a sphere,
the Turton equation is used:
CD = ((PRe
24 )*(1+0.173*Rep0.657)) +
)(Re*11630+1413.0
09.1P
(11) Results are summarized in Table 5.
www.andritz.com 6 of 13
-
7
Table 5. Drag Coefficient for single particle size
in Example 1 Particle size (mm)
Av-erage
Particle Size (mm)
Particle Reynolds number
(Rep)
Drag coefficient
(Cd)
-0.85 + 0.40
0.63 1160 0.455
-0.40 + 0.20
0.30 557 0.550
-0.20 + 0.152
0.18 327 0.662
Step 7: Calculate the friction losses of the heterogeneous
fraction using Durands equation.
Wasp et al. (1977) recommend using Durands equation for each
size fraction of solids to determine the increase in pressure
losses due to moving bed:
51
2
.
D
LLsivbedLbed
CV/)(gD
CP82P
=
(12) Where
PL = Pressure drop due to flow of carrier liquid. Cv bed =
Volumetric concentration of bed por-tion of a particular size
fraction. Di = Inside diameter of the pipe v = Mean slurry
velocity. CD = Drag coefficient of a particular size frac-tion.
Ellis and Round (1963) indicated that Durands
equation coefficient of 82 is too high for nickel sus-pensions.
Therefore, for this particular example, a value of 23 is used as
the modified Durands equation coefficient. The value of the
constant is determined by iteration based on the test results. Step
9 explains how to determine the value of this constant.
Results of the calculation for this particular case are
presented in Table 6:
The total friction loss is therefore 239.46 Pa/m + 34.61 Pa/m =
274.07 Pa/m. This value is compared with the experimental pressure
drop, i.e., 270 Pa/m, it is observed that these values do not
match. Therefore, an iterative technique is used to refine the
value of pressure drop. This is explained in the following
steps.
Table 6. Calculated Losses for each fraction of
solids in moving bed Av-
erage Particle
Size (mm)
Parti-cle
Reyn-olds
number (Rep)
Drag coeffi-cient
(Cd)
Cv bed
(%)
Pres-sure Loss
(Pa/m)
0.63 1160 0.455 0.151
14.51
0.30 557 0.550 0.177
14.74
0.18 327 0.662 0.074
5.36
Total 34.61 Step 8: Based on the value of C/CA, determine the
fraction size of solids in homogeneous and heterogeneous flows.
By comparing with the measured 270 Pa/m, the calculations for
the bed are higher and can be refined by the method of
concentration1 using equation (Abulnaga, BE 2002):
fx
t
A UKV.
]CC[log
81
10 = (13)
Where
C/CA = the ratio of volumetric concentration of solids at 0.08D
from top to that at pipe axis,
= the dimensionless particle diffusivity and is taken as 1.0
Kx = 0.4 and is defined as von Karman coeffi-cient
Uf = the friction velocity calculated from the pressure drop in
first iteration Vt = settling velocity of a particular size
parti-cle calculated using standard drag relationships.
Where Vt is defined as the single particle size terminal
velocity using the Stokes equation:
DL
gLSt C
gd)(V
34
= (14)
The equivalent fanning factor (Abulnaga, BE 2002) is calculated
as:
Total loss = if /DVf 22 (15)
2itotal
N 2VDP
f
= (16)
To calculate Uf, the following Equation (Abulnaga, BE 2002) is
used:
Uf = V (fN /2) (17)
www.andritz.com 7 of 13
-
8
By determining the value of C/CA ,new Cv bed for
each size fraction is calculated directly by multiplying
original Cv bed by C/CA. Iterated pressure drop for each size
fraction is calculated using this new Cv bed. This is repeated till
the value converges.
For this particular example, the refined pressure loss after
1st, 2nd and 3rd iteration is represented in table 7, 8 and 9
respectively.
After iteration, pressure loss due to heterogene-ous part
converges to 29.93 Pa/m, and hence total loss due to the slurry
flow is 238.46 + 29.93 = 268.4 Pa/m which is very close to
experimental value, i.e., 270 Pa/m. Step 9: Determine the value of
Durands equation coefficient, k.
The value of the Durands equation coefficient is
material-specific. To determine its value, just one test result
data is required and the modified value of k is calculated using
iteration. For the 1st iteration, pres-sure drop using k=82 is
calculated. If the total calcu-lated pressure drop does not match
with the experi-mental pressure drop then a modification in the
value of k is required. To determine the modified value of k, a
ratio R is defined.
R = total experimental loss / total calculated loss (for
k=82).
An Iterated value of k is determined by multiply-ing k by ratio
R, which is re-iterated until the value of R converges to 1. After
iteration, a value of k is ob-tained for which R = 1, which is then
defined as the modified Durands equation coefficient.
Validation of Modified Wasp et al (1977) Method
Case study 1: Verification of Modified Wasps method for
different flow regime.
The modified Wasp method was verified for all the cases of flow
regime, i.e., case 1, case 2, case 3, and case 4 and results are
presented in Table 10. The calculated pressure drop was very
accurate and within 10% error in prediction.
Case study 3: Verification of data sets available from open
literature.
In this case study, various experimental points have been
collected from several sources spanning the years 1942-2002 [refer
to Table 12]. This wide range of databases includes experimental
information from different physical systems. Table 13 suggests the
wide range of the collected databank for pressure drop. Results of
the case study are presented in Table 14. Case study 4: Kaushal and
Tomita {2003} data for different volumetric concentration of glass
beads
in slurry flow was collected and Modified Wasp pressure drop was
calculated and compared with the experimental results.
To validate this method for different volumetric concentration
of solids in slurry, data points from the open literature was
collected and pressure drop using the Modified Wasps method was
calculated. The Cv value ranged from 9.4% to 51.7% and the speed
ranged from 1 m/s to 5 m/s. The Pipe inside diameter for this case
was 54.9 mm and the solid specific grav-ity was 2.47. Results of
this case study are presented in Table 15
Vertical Slurry Flow in Pipes
Solid particles can be moved upward when the flu-id velocity (V)
exceeds the hindered settling velocity of the solids (wS).
SwV >> Where
)c(vw VtS = 1 (18)
Sw = Hindered settling velocity of particle = constant. vt =
Single particle terminal velocity (from
equation 14)
Hindering effect of solids concentrations must be taken into
account while calculating the pressure drop. Fig.4 gives the
influence of the concentration according to Maude and Whitmore
(1958).
Figure 8. Influence of the concentration on settling velocity
according to Maude and Whitmore (1958) replotted from Weber
(1974).
Reynolds number is calculated using equations (2),(3) and (4)
corresponding to particle diameter d100 (=dRe). Using this Reynolds
number, the value of is calculated from the above curve. Hindered
settling velocity of the particle is calculated corresponding to
the value of obtained, using equation (18). If the fluid velocity
is lesser than hindered settling velocity of particle (ws),
chocking condition occurs and there is no movement of solids along
the vertical section. If
www.andritz.com 8 of 13
-
9
the fluid velocity is greater than the hindered settling
velocity of particle (ws), then the pressure drop is calculated as
described in the next section.
Pressure Drop Calculation
The flow regime of the vertical section depends on critical
deposition velocity (ucri), which is calculat-ed by Wasp Modified
Durands equation (19).
L
Ls/
.vcri DgD
dc.u
= 25253
611002340
(19)
Homogeneous or pseudo-homogenous flow re-
gime occurs when mean slurry velocity (V) is greater than the
critical deposition velocity (ucri). At this con-dition there is
almost no slip between particles and fluid, therefore particle
velocity is approximately equal to mean slurry velocity. Pressure
drop for the vertical homogenous regime is calculated using the
following equation (20).
vertmvertm
D LgDL
Vfp +
= 22 (20)
Where
p =total pressure drop for vertical slurry flow us = velocity of
solids
vertL = length of vertical section under con-sideration
www.andritz.com 9 of 13
-
Table 7. 1st iteration results Average Particle Size
(mm)
Drag Coeffi-cient
(Cd)
Terminal Velocity
(m/s)
-1.8 vt/KxUf
C/CA Iterated Cv bed (%)
Iterated Pressure loss (Pa/m)
0.63 0.455 0.201 -0.0844 0.823 0.125 11.94 0.30 0.550 0.127
-0.0532 0.884 0.157 13.03 0.18 0.662 0.089 -0.0371 0.918 0.068 4.92
Total 29.89
Table 8. 2nd iteration results Average Particle Size (mm)
Drag Coef-ficient
(Cd)
Terminal Velocity
(m/s)
-1.8 vt/KxUf
C/CA Iterated Cv bed (%)
Iterated Pressure loss (Pa/m)
0.63 0.455 0.201 -0.0837 0.824 0.1248 11.96 0.30 0.550 0.127
-0.0528 0.886 0.1569 13.05 0.18 0.662 0.089 -0.0368 0.919 0.068
4.92 Total 29.89
Table 9. 3rd iteration results
Average Particle Size (mm)
Drag Coef-ficient
(Cd)
Terminal Velocity
(m/s)
-1.8 vt/KxUf
C/CA Iterated Cv bed (%)
Iterated Pressure loss (Pa/m)
0.63 0.455 0.201 -0.0837 0.824 0.1248 11.96 0.30 0.550 0.127
-0.0528 0.886 0.1569 13.05 0.18 0.662 0.089 -0.0368 0.919 0.068
4.92 Total 29.89
Table 10. Pressure drop vs. speed in a 159-mm ID Steel pipe at a
weight concentration of 26.3% at 20o C
Cv (%)
Veloci-ty
(m/s)
Di (m) S
(kg/m3)
L (kg/m3)
L (mPa.s)
Experi-mental Loss
(Pa/m)
Calculat-ed
Loss(Pa/m)
% Error Flow Regime
8 1.5 .159 4074 1000 1 175 195 11 Case 2 8 1.9 .159 4074 1000 1
270 268 -0.7 Case 3 8 2.3 .159 4074 1000 1 360 354 -1.6 Case 3 8
2.7 .159 4074 1000 1 525 472 -10 Case 4 8 3.1 .159 4074 1000 1 688
616 -10 Case 4 8 3.5 .159 4074 1000 1 847 779 -8 Case 4 8 4.0 .159
4074 1000 1 1046 1008 -3.6 Case 4
Table 11. Calculated pressure drop for case study 2
Cv (%)
Velocity (m/s)
Di (m) S
(kg/m3)
L (kg/m3)
L (mPa.s)
Experi-mental Loss
(Pa/m)
Calculated Loss
(Pa/m)
% Error Flow Regime
4 1.2 .105 2820 1000 1 232 227 -1.81 Case 3 4 2.8 .105 2820 1000
1 1006 897 -10.9 Case 4 4 3.2 .105 2820 1000 1 1316 1165 -11.4 Case
4 4 3.6 .105 2820 1000 1 1652 1469 -11.0 Case 4 4 4 .105 2820 1000
1 2013 1808 -10.1 Case 4
8.1 1.6 .105 2820 1000 1 438 366 -16.5 Case 3 8.1 2 .105 2820
1000 1 619 500 -19.5 Case 4 8.1 3.6 .105 2820 1000 1 1781 1579
-11.3 Case 4 8.1 4 .105 2820 1000 1 2168 1942 -10.4 Case 4
12.8 1.6 .105 2820 1000 1 468 448 -4.3 Case 3
www.andritz.com 10 of 13
-
11
12.8 3.6 .105 2820 1000 1 1953 1700 -12.9 Case 4 12.8 4 .105
2820 1000 1 2317 2092 -9.7 Case 4 19.1 2 .105 2820 1000 1 722 638
-11.6 Case 3 19.1 3.6 .105 2820 1000 1 1962 1868 -4.74 Case 4 19.1
4 .105 2820 1000 1 2375 2298 -3.22 Case 4 26 2 .105 2820 1000 1 774
741 -4.23 Case 3 26 3.2 .105 2820 1000 1 1703 1636 -3.98 Case 4 26
3.6 .105 2820 1000 1 2104 2059 -2.14 Case 4 26 4 .105 2820 1000 1
2529 2531 0.04 Case 4
Table 12. Literature sources for pressure drop
No Author No Author 1 Wilson (1942) 11 Gillies et al. (1983) 2
Durand & Condolios(1952) 12 Roco & Shook(1984) 3 Newitt et
al. (1955) 13 Roco & Shook(1985) 4 Zandi & Govatos (1967)
14 Ma (1987) 5 Shook et al.(1968) 15 Hsu (1987) 6 Schriek et al.
(1973) 16 Doron et al. (1987) 7 Scarlett & Grimley (1974) 17
Ghanta (1996) 8 Turian & Yuan (1977) 18 Gillies et al. (1999) 9
Wasp et al. (1977) 19 Schaan et al.(2000) 10 Govier & Aziz
(1982) 20 Kaushal and Tomita(2002)
Table 13. Slurry system[1] and parameter range from the
literature data
Pipe Dia (m)
Particle Dia (micron)
Liquid den-sity
(kg/m3)
Solids den-sity
(kg/m3)
Liquid viscosity (mPa.s)
Velocity U (m/s)
Solids Conc. (fraction) f
0.019-0.495 38.3-13000 1000-1250 1370-2844 0.12-4 0.86-4.81
0.014-0.333 [1] Slurry system: coal/water, copper ore/water,
sand/water, gypsum/water, glass/water, gravel/water.
Table 14. Results of data points collected from open literature
having flow regime of case 4 Cv Veloci-
ty (m/s)
Di (m)
S (kg/m3)
L (kg/m3)
L (mPa.s)
Experi-mental
Loss (Pa/m)
Calcu-lated Loss
(Pa/m)
% Error
Flow Regime
0.3 3 0.0549 2470 1000 0.85 1990 2210 -11.1 Case 4 0.3 4 0.0549
2470 1000 0.85 3430 3726 -8.6 Case 4 0.3 5 0.0549 2470 1000 0.85
5350 5599 -4.7 Case 4 0.4 3 0.0549 2470 1000 0.85 2230 2706 -21.3
Case 4 0.4 4 0.0549 2470 1000 0.85 3790 4536 -19.7 Case 4 0.4 5
0.0549 2470 1000 0.85 6390 6786 -6.2 Case 4 0.5 3 0.0549 2470 1000
0.85 3410 3645 -6.9 Case 4 0.1 1.1 0.019 2840 1000 0.85 1250 1114
10.9 Case 4 0.1 1.11 0.0526 2330 1000 1 294 321 -9.2 Case 4 0.3 1.3
0.0526 2330 1000 1 543 609 -12.2 Case 4 0.3 2.59 0.2085 1370 1000 1
267 318 -19.1 Case 4 0.3 2.34 0.2085 1370 1000 1 226 263 -16.4 Case
4 0.3 2.01 0.2085 1370 1000 1 177 196 -10.7 Case 4 0.3 1.78 0.2085
1370 1000 1 147 158 -7.5 Case 4 0.3 1.59 0.2085 1370 1000 1 123 129
-4.9 Case 4
www.andritz.com 11 of 13
-
12
0.3 1.37 0.2085 1370 1000 1 99 98 1.8 Case 4 0.1 1.66 0.0515
2650 1000 1 666 671 -0.8 Case 4 0.2 1.66 0.0515 2650 1000 1 900 791
12.1 Case 4 0.3 1.66 0.0515 2650 1000 1 1136 930 18.1 Case 4 0.1
2.9 0.263 2650 1000 1 261 272 -4.0 Case 4 0.1 3.5 0.263 2650 1000 1
334 392 -17.4 Case 4 0.2 2.9 0.263 2650 1000 1 305 313 -2.4 Case 4
0.2 3.5 0.263 2650 1000 1 382 443 -15.9 Case 4 0.3 2.9 0.263 2650
1000 1 355 353 0.5 Case 4 0.3 3.5 0.263 2650 1000 1 453 505 -11.4
Case 4 0.3 2.9 0.263 2650 1000 1 414 392 5.2 Case 4 0.3 3.5 0.263
2650 1000 1 526 560 -6.4 Case 4 0.1 3.16 0.495 2650 1000 1 143 151
-5.7 Case 4 0.1 3.76 0.495 2650 1000 1 186 211 -13.6 Case 4 0.2
3.07 0.495 2650 1000 1 157 161 -2.5 Case 4 0.3 3.76 0.495 2650 1000
1 254 271 -6.4 Case 4 0.1 2.5 0.1585 2650 1000 1.3 475 421 11.4
Case 4 0.3 2.5 0.1585 2650 1000 1.3 630 532 15.6 Case 4 0.1 3
0.1585 2650 1000 0.12 648 526 18.9 Case 4 0.1 1.9 0.0507 2650 1000
1 1175 1140 3.0 Case 4 0.2 2.8 0.04 2270 1250 4 3926 4022 -2.4 Case
4 0.1 2.7 0.04 2270 1250 4 3580 3487 2.6 Case 4 0.1 2.01 0.04 2270
1250 4 2217 1939 12.5 Case 4
Table 14. Results of the case study
Cv Velocity (m/s)
Di(m) S (kg/m3)
L (kg/m3)
L (mPa.s)
Experimental Loss (Pa/m)
WASP modified
loss (Pa/m)
% Error
Flow Regime
9.4 1 0.0549 2470 1000 1 261 222 14.9 Case 4 10.06 2 0.0549 2470
1000 1 847 781 7.7 Case 4 10.41 3 0.0549 2470 1000 1 1754 1642 6.3
Case 4 10.44 4 0.0549 2470 1000 1 2868 2783 3.0 Case 4 10.93 5
0.0549 2470 1000 1 4153 4230 -1.9 Case 4 19.22 1 0.0549 2470 1000 1
341 263 22.6 Case 4 20.48 2 0.0549 2470 1000 1 1051 932 11.3 Case 4
20.4 3 0.0549 2470 1000 1 1981 1937 2.2 Case 4
19.52 4 0.0549 2470 1000 1 3263 3225 1.1 Case 4 20.45 5 0.0549
2470 1000 1 4666 4927 -5.6 Case 4 30.3 1 0.0549 2470 1000 1 373 323
13.4 Case 4
30.02 2 0.0549 2470 1000 1 1037 1101 -6.2 Case 4 31.19 3 0.0549
2470 1000 1 2037 2330 -14.4 Case 4 30.75 4 0.0549 2470 1000 1 3291
3889 -18.2 Case 4 30.24 5 0.0549 2470 1000 1 4851 5783 -19.2 Case 4
38.95 3 0.0549 2470 1000 1 2420 2731 -12.9 Case 4 40.64 4 0.0549
2470 1000 1 3865 4760 -23.2 Case 4 39.56 5 0.0549 2470 1000 1 5761
6933 -20.0 Case 4 51.7 2 0.0549 2470 1000 1 2099 2002 4.6 Case
4
www.andritz.com 12 of 13
-
13
49.24 3 0.0549 2470 1000 1 3082 3689 -19.7 Case 4 Heterogeneous
vertical flow regime occurs when mean slurry velocity (V) is less
than critical deposi-tion velocity (ucri). Although the
distribution of solids is homogeneous, considerable slip exists.
Therefore, the local concentration cV has to be calculated for each
velocity with respect to the delivered concentra-tion and given
solid mass flow rate, before calculating the pressure drop.
+
+=
Vv
c4Vv
11Vv
V2Vc tvtotal
2tt
tV
(21) The new value of volumetric concentration is used to
calculate the slurry viscosity by using Thomas corre-lation (eq.
8). This results in a modified value of slur-ry Reynolds number
(eq. 7) and Darcy friction factor (eq. 6). This value of Darcy
friction factor (fD) is used to calculate pressure drop using
equation (20).
Results and Discussion
The parity plot for experimental and predicted pressure drop for
all the data points considered in this study is shown in Figure 9.
It is observed that pres-sure drop calculation by the Modified Wasp
method gives better prediction and is in good agreement with
experimental data. The best fit is the straight line having slope =
1, which means the predicted values are equal to the experimental
values. It is observed that most of the predicted values lie very
close to straight line and maximum error observed is less than 15%,
which confirms that prediction is better using this method.
Figure 9. Experimental vs. Predicted pressure drop
k is corrected (Durands equation coefficient) for
calculating pressure drop using this method. The k value for any
specific slurry system can be deter-mined with method described by
Wasp.
This method requires one experimental value of pressure drop for
a unit length at a given solid % and PSD. Once this slurry specific
k is available the same value can be used for predicting the
pressure
drop for different pipe sizes, solid % and PSD. The
applicability of this method is thus not constrained due to
unavailability of extensive experimental data. It can be easily
applied to new slurry systems with minimal laboratory effort. The
Modified Wasp Model:
i. Is easy to implement, has direct calculations involving
iterations.
ii. Accurately predicts the pressure drop over a wide range of
input parameters for all types of slurry system like coal/water,
nickel ore/water, sand/water, copper ore/water, etc.
iii. Effectively predicts pressure drop as a func-tion of
Particle Size Distribution (PSD) and solid percentage.
Thus, with the modifications proposed in this
study, this method delivers a comparatively smaller error
percentage as compared to other existing models for prediction of
pressure drop in slurry flows.
References
1. Abulnaga, Baha E (2002), Slurry Systems Handbook,
McGraw-Hill.
2. Shou George, Solid-liquid flow system simulation and
validation, Pipeline Systems Incorporated, USA
3. Manfred Weber, Liquid-Solid Flow 4. Lahiri SK and Ghanta
KC,Prediction of
Pressure drop of slurry flow in Pipeline by Hybrid Support
vector Regression and Ge-netic Algorithm Model, Department of
Chemical Engineering, NIT, Durgapur, In-dia.
5. Kaushal and Tomita, Solids concentration profiles and
pressure drop in pipeline flow of multisized particles slurries,
Department of Mechanical Engineering, Kyushu Institute of
Technology, Japan.
6. Kaushal and Tomita, Effect of particle size distribution on
pressure drop and concentra-tion profile in pipeline flow of highly
con-centrated slurry, Department of Mechanical Engineering, Kyushu
Institute of Technolo-gy, Japan.
7. Jacobs BEA (2005), Design of Transport Systems, Elsevier
Applied Science, England.
www.andritz.com 13 of 13