You are on page 1of 19

263

Wear Prediction in Fully Developed


Multi-Size Particulate Flow in Horizontal
Pipelines
K. V. Pagalthivarthi1, J. S. Ravichandra2, S. Sanghi3 and P. K. Gupta4#
1CFDResearch Leader, GIW Industries Inc., Grovetown, GA — 30813, USA
2Research Engineer, Applied Combustion Lab, GE, Bangalore — 560066, India
3Professor, Dept. of Applied Mechanics, IIT Delhi, New Delhi - 110016, India
4#Assoc. Prof., Dept. of Mech. Engg., Raghu Engineering College, Visakhapatnam,

A.P. - 531162, India

Received: 14th April 2009, Accepted: 12th August 2009

Abstract
Erosion wear prediction in fully developed multi-size particulate flow through hori-
zontal pipes is carried out numerically using Galerkin finite element methodology. The
predicted multi-size particulate flow field near the pipe wall is correlated with wear
rates via empirically determined wear coefficients. The wear models for particle
impact and sliding take into account the broad particle size distribution. The effects of
various operating parameters such as flow rate, solids concentration, solids density,
particle size, particle size distribution and so forth has been studied. Normalized pre-
dicted wear rates are compared with published experimental results and are found to be
within an average error of 16.5% with respect to experimental observations. Different
wear trends with different flow parameters indicate the existence of a very complex
relationship between wear rate and the flow governing parameters. A major conclusion
is that accurate wear prediction could be obtained by characterizing the particle size
distribution into suitable number of size classes, and accounting for the particle-size
dependence in the wear model(s).

Keywords: Multi-size Particulate Flow, Horizontal pipes, Impact wear rate, Sliding
wear rate, Particle size distribution, Mixture velocity

1. INTRODUCTION
Hydrotransport of solids in the form of slurries has many advantages over conventional modes of
transport, especially in reliability and economy. Thus it finds numerous applications in commercial
industries. In long pipelines used in this kind of transportation, the flow attains fully developed
state after a relatively short distance. In slurry pipelines, the superficial mixture velocity often
exceeds 10 m/s. Furthermore, the slurries have a wide distribution of particle sizes, with fair
percentage of coarse particles. Together these effects can lead to severe erosion of pipe material.
Erosion wear is a complex phenomenon and a large number of independent parameters govern
the complex process of erosion wear. Wood et. al. (2004), Meng and Ludema (1995) quote 33
independent parameters used in the currently available 22 erosion models and predictive equations.
Some of the parameters are the local solids concentration, mixture velocity, solid velocity, as well
as material properties of the carrier, particulate material and the wearing surface [Wood et al.
(2004)].
Furthermore, several [Elkholy (1983); Clark (1991); Lin and Shao (1991); Lynn et al. (1991);
Gandhi et al. (1999); Clark and Hartwich (2001); Gandhi and Borse (2004)] studies have
corroborated that wear increases with increase in the particle size. Erosion wear prediction

#Corresponding author: pankajkgupta@gmail.com


264 Wear Prediction in Fully Developed Multi-Size Particulate Flow
in Horizontal Pipelines
becomes particularly difficult for multi-size particulate slurries. Commercial slurries usually
consists of particle sizes varying over 2-3 orders of magnitude [Gandhi and Borse (2004); Kaushal
et al. (2002)]. Representing such slurries using a single size such as nominal particle size, weighted
mean diameter or D50 particle size might lead to serious errors in erosion wear prediction
[Pagalthivarthi and Gupta (2009); Gandhi and Borse (2004); Ravichandra and Pagalthivarthi
(2003)].
The near-wall phenomenon of particulate erosion wear takes place by the interaction of solid
particles with the pipe wall. These interactions are understood to be due to the combined effect of
directional impact of solid particles, random impact of solid particles due to turbulent fluctuations,
and sliding bed of solids [Roco et al. (1984); Roco and Cader (1988)]. In pipe flows [Roco and
Cader (1988)], solid particles are either in turbulent suspension or in Coulombic contact with the
pipe wall. Interaction of the particles with the pipe wall occurs when their convective velocity is
directed towards the wall (direct impingement) or when the particles in the contact-load slide along
the pipe wall. Interaction is also caused by the random impingement of particles onto the pipe wall.
Wear due to impingement mechanism (either directed or random), is generally negligible in straight
pipelines [Roco and Cader (1988)]. The interaction between particles and the wall leads to erosion
wear, when the local stresses exceed a critical value, leading to ductile or brittle failure or their
combination. Roco et. al. (1984), Roco and Cader (1988) relate the two types of interactions
(directional impingement and sliding bed) to specific stresses acting in the dense slurry flows.
Erosion wear due to particle-sliding is related to supported load stress and wear due to particle
impact is related to dispersive stress. Alternatively, it is possible to quantify the dominant sliding
wear in particulate flows by considering the local particulate shear stress and velocity at the wall.
The particle impact wear is correlated with the local mass flux and kinetic energy of the solid
particles.
A numerically predictive tool to obtain the two-phase flow field enables the determination of
particle concentration and velocity near the wear surface. As there is no universal theory that
relates the particulate flow conditions to the wear rate, a certain degree of empiricism is
unavoidable in the computation of wear rate. Empirically determined wear coefficients are used to
relate the particulate flow conditions to the wear rate. For a specific wear mechanism, the wear
coefficients (with units of J/m3) represent the amount of energy lost by the particles while
removing unit volume of material [Pagalthivarthi and Helmly (1992)]. Various studies [Roco and
Addie (1983); Roco, Addie, Visintainer, Ray (1986); Pagalthivarthi and Veeraraghavan(1998);
Tuzson and Clark (1998)] have corroborated that the specific energy for wear (or wear coefficients)
due to impact or sliding depends on particle size [Pagalthivarthi and Gupta (2009); Pagalthivarthi
and Veeraraghavan (1998); Pagalthivarthi and Addie (2001)], particle shape, particle size
distribution, material properties, hardness and microstructure of eroding material [Tian, Addie,
Pagalthivarthi (2005)] and so forth.
The experimentally obtained material wear coefficients takes into account the effect of the
properties of the wearing surface material (viz., hardness, ductility, toughness, microstructure etc.)
and those of the particulate material. Existing models for erosion wear are based on these
experimentally determined coefficients (which vary depending on the type of wear mechanism) and
flow pattern. Generally, the empirical coefficients are determined by simple laboratory tests and
the flow details are obtained by a numerical simulation of the flow in the application in context.
Wear test cases in pilot plant test loops are very expensive and time consuming. Therefore,
different bench scale tests that give measurable wear in short time are used to study the wear
characteristics. A detailed review of experiments conducted by various researchers for wear rate
measurements in pilot plant test loops is given by Gupta (1994). The development of erosion
models and similarities between them are reviewed by Finnie (1995), Meng and Ludema (1995),
Wood et. al. (2001).
The Coriolis wear tester [Tuzson and Scheibe-Powell (1984), Pagalthivarthi and Helmly (1992)]
simulates sliding wear under the action of a normal force caused by Coriolis acceleration. This
device produces measurable wear within a short duration of device operation. The sliding wear
coefficients can be determined (for a specified wear material and slurry combination) using this
device. Sliding wear coefficients obtained using Coriolis wear tester are very relevant to fully
developed multi-size particulate flow, as wear due to particle impact is comparatively small
[Pagalthivarthi and Gupta (2009)]. Experiments [Roco and Cader (1988)] suggest that for small

Journal of Computational Multiphase Flows


K. V. Pagalthivarthi, J. S. Ravichandra, S. Sanghi and P. K. Gupta 265

impact angles, the impact wear coefficient (in J/m3) is usually at least an order of magnitude larger
than sliding wear coefficient (in J/m3) for the same slurry/wear material combination. Slurry pot
testers [Gupta et al. (1995); Gandhi and Borse (2004)] have also been employed to measure uneven
wear in slurry pipelines.
The erosion wear prediction methodology comprises essentially of three steps [Tian, Addie,
Pagalthivarthi (2005); Pagalthivarthi and Helmly (1992)]: (1) computation of the two-phase flow
field, (2) relating the local flow velocity and concentration near the wear surface to the wear rate
via a suitable wear model, and (3) empirical determination of wear coefficient relevant to the wear
model. Such methodologies have been successfully used in predicting erosion wear in centrifugal
pump impellers [Pagalthivarthi and Addie 2001], casings [Roco et a. (1986); Addie and
Pagalthivarthi (1989)] and straight rotating channels [Pagalthivarthi and Gupta (2009)]. The
present work extends the methodology used in [Pagalthivarthi and Gupta (2009); Ravichandra and
Pagalthivarthi (2003)] to erosion wear prediction due to multi-size particulate flows in pipelines.
Thus, the objective of the present work is to compute and study erosion wear in fully developed
multi-size particulate flows in horizontal pipes. Erosion wear depends on the material properties
of pipe, solids content, carrier fluid and the flow parameters. It varies along the pipe circumference
as a function of near wall concentration, solids velocity, wall shear stress and to a lesser degree on
the impact angle.
Fully developed pseudo-homogeneous horizontal pipe flow of slurry with multi-size particles is
modelled via Galerkin finite element method using nine-noded isoparametric elements. The
momentum equation for the slurry mixture is based on a pseudo-fluid model. For each particle size
(termed as species), a convection-diffusion equation is solved to obtain the volumetric
concentration distribution for that species. A mixing length model, with a concentration-dependent
eddy viscosity, is used to account for turbulence. Wall functions are used to prescribe mixture
velocity at the wall. For complete details of mathematical and numerical modeling of multi-size
particulate flow in pipes, please refer [Ravichandra et al. (2004)]. The effect of material properties
(of a given pipe and particulate pair) is introduced through material coefficients obtained from the
open literature. Erosion wear might be accompanied by corrosion, scaling, changes in surface layer
properties or cavitation. These effects are not considered in the present study.

2. EROSION WEAR RATE MODELS


Sliding abrasion and directional impingement are major erosive wear mechanisms in slurry flows.
These mechanisms are schematically shown in Figure 1. In addition to these two, random
impingement can also cause wear. However, in fully developed pipe flow, its effect is quite small.

Figure 1. Schematic drawing of sliding and impact wear mechanisms in pipes.

The wear model is based on correlating the local mass flux and kinetic energy of the solid
particles to sliding .and impact wear rates through empirically determined wear coefficients. The
sliding wear rate (WSL) is modeled as

τ eff wm
W SL = , (1)
Esp

Volume 1 · Number 3 · 2009


266 Wear Prediction in Fully Developed Multi-Size Particulate Flow
in Horizontal Pipelines
where Esp is the specific energy for a given material slurry combination, wm is the axial velocity
of the particle, and τeff is the effective shear stress of particles modeled as

(
τ eff = Cτ S = τ m − 1 − C τ L , ) (2)

where C is the overall solids concentration, τS, τL and τm are the shear stresses in the solid, liquid
and mixture, respectively. The mixture shear stress,τm, is given as

τ m = ρmum2τ , (3)

and the liquid shear stress is given as

τ L = ρ L uL2τ , (4)

Carrier phase friction velocity uLτ and mixture friction velocity umτ are computed using wall
functions [Ravichandra et al. (2004)]. .
For multi-size particulate flow, the overall impact wear rate (WI) is modeled as the summation
of the impact wear rates due to each of the species. It can be written as [Ravichandra and
Pagalthivarthi (2003)]

3
N ρSV pk Ck
W I = ∑ , (5)
k =1 ( )
EI α k

where,

2 2
V pk = wm + vk , (6)

is the particle impact velocity of species ‘k’ and E1(αk) is the specific energy of impact for a
specified angle of impact, αk = tan-1vk/wm . The specific energy of impact is highly dependent
on the angle of impact and is also a function of particle diameter. From the experimental results of
Roco et. al. (1984), the specific energy of impact may be written for ductile or brittle materials as
[Pagalthivarthi and Gupta (2009)],

Eo
( )
EI α =
4α  α 
, (7)
1−
π  π 

where E0 is coefficient of the pipe wall for normal impact (α = π/2). Wear coefficients (Esp and
E0) are taken from the open literature [Roco and Cader (1988), Pagalthivarthi and Ramanathan
(1998)]. They are usually functions of particle size, as noted by Pagalthivarthi and Ramanathan
(1998). Their relative magnitudes are in the sequel.
The Coriolis wear tester [Pagalthivarthi and Helmly (1992); Tuzson and Scheibe-Powell (1984)]
has been widely used to determine the specific energy of sliding wear in dense slurry flows. The
impact wear coefficient is usually measured using wedges of specified angles placed head-on in a
stream of slurry flow [Pagalthivarthi and Helmly (1992)] of known velocity and concentration. The
amount of wear (and hence the wear rate) may be computed from measured weight loss.
Alternatively, profilometer measurements may be made to directly obtain the wear depth.

Journal of Computational Multiphase Flows


K. V. Pagalthivarthi, J. S. Ravichandra, S. Sanghi and P. K. Gupta 267

3. RELATIVE MAGNITUDES OF SLIDING AND IMPACT WEAR RATES


Impact wear depends on material coefficients, local concentration, solid velocity and impact angle
[Pagalthivarthi and Addie (2001)]. In Figure 2(a), variation of impact angle along the pipe wall is
shown for particles of various sizes. Particles studied are in the range of 50-650 µm. For all the
particle sizes, the impact angle is less than one degree. Therefore, the contribution of impact wear
rate to the overall wear rate could be expected to be significantly smaller than due to sliding wear
component. Figure 2(a) shows that the impact angle increases with θ from the pipe bottom. As
shown in Figure 2(b), this is due to increase in the y-component of the solid velocity along the pipe
wall (with increasing o). Solid velocity (or mixture velocity) increases in magnitude along the pipe
wall due to decrease in concentration [Ravichandra et al. (2004)], as shown in Figure 2(c). It can
be noted that due to decrease in concentration, the mixture velocity increases along the pipe wall.
Due to increase in impact angle and decrease in concentration, exposure of pipe wall to the
impacting particles increases along the pipe wall. Due to this, impact wear rate could be expected
to increase along the pipe wall (i.e., along the circumference).

Figure 2. Variation of (a) impact angle, (b) solid velocity (vk), (c) concentration and (d)
mixture velocity (wm) along the pipe wall in flows with various uniform size particles.

Roco and Cader (1988) and Roco and Addie (1987) give experimentally calculated values for
Esp and E0 for several pipe-particle pairs. These values are summarized in Table 1. Table 1
indicates that there is more than an order of magnitude difference in the specific energy values of
sliding and impact wear mechanisms in solid-liquid horizontal pipe flows. Furthermore, the angle
of impact is quite small (typically, ~1o). The difference between Esp and E0 along with the
observed small impact angles indicates that the contribution of impact wear would be much smaller
than the contribution of sliding wear rate to the total wear rate. An example of the relative
magnitudes of sliding and impact wear rates is shown in Figure 3.

Volume 1 · Number 3 · 2009


268 Wear Prediction in Fully Developed Multi-Size Particulate Flow
in Horizontal Pipelines
Table 1. Values of Esp and EI for various pairs of pipe and particulate material
[Roco and Cader (1988)]

Particulate
Pipe material D50 (mm)
material (J/m3) (J/m3)

Carbon steel
Sand 0.67 7.20E11 9.47E16 1.32E05
(JISG 3454)

Epoxy resin Sand 0.25/0.15 8.78E07 1.38E09 1.58E01


Carbon steel Sand 0.17 3.75E09 1.16E11 3.10E01

Carbon steel
Phosphate 0.24 7.50E10 2.00E12 2.67E01
(STPG-38)

Figure 3. Comparison of sliding and impact wear rates for several mono-size particles.

4. COMPARISON OF COMPUTED AND EXPERIMENTALLY DETERMINED WEAR


RATES
Test cases of Roco and Cader (1988) and Gupta et. al. (1995) are simulated for comparison of the
computed wear rate predictions with corresponding experimental observations. The operating
conditions for these test cases are presented in Table 2. Case W1 consists of mono-size particles,
while Case W2 consists of five size classes with equal contributions to the overall concentration.
Test cases W3-W8 correspond to multi-size particulate slurries with six size classes; the individual
concentrations of the six size classes are shown in Table 2.
For test cases W1 and W2 [Roco and Cader (1988)], the values of Esp and E0 are provided, and
these are shown in Table 2. However, the experimental data for cases W3-W8 do not include Esp
and E0 values. Thus Esp for these cases are taken from the open literature [Pagalthivarthi and
Ramanathan (1998)], E0 is assumed to be two orders of magnitude greater than Esp. The assumed
values of Esp and E0 are included in Table 2 for cases W3-W8.
In the literature [e.g., Roco
. . and Addie (1987) and Roco and Cader (1988)], it is common to plot
normalized wear rates (W / Wmax) rather than the actual wear rates. This enables the comparison
of wear trends between computation and experiment more easily. As seen in the foregoing, the
actual wear rate depends on the wear coefficients (Esp and E0) for a specified slurry/wear material
combination. In many experimental studies [e.g., Gupta et. al. (1995)], the actual values of Esp and
E0 are not included in the data. Even when included, some uncertainties due to particle break-up
in closed loop systems can significantly contribute to wear rate differences. Hence, in general, the
trends of wear rate are compared using plots of normalized wear.

Journal of Computational Multiphase Flows


K. V. Pagalthivarthi, J. S. Ravichandra, S. Sanghi and P. K. Gupta 269

Table 2. Details of test cases for comparing present computed wear results
with experimental observations.

D (µm) %C of each
Case Authors
(mm) Species (m/s) (J/m3) (J/m3)

Roco and Cader


W1 200 250 10.0 10.0 7.92 7.20E11 9.47E16
(1988)

Roco and Cader 300, 200,


W2 750 5.69 1.14 6.5 3.75E09 1.16E11
(1988) 150, 140, 40
0.51, 0.99,
Gupta et. al., 438, 235, 168,
W3 55 6.9 0.50, 1.69, 1.95 1.56E11 1.56E13
(1995) 127, 91, 38
1.07, 2.09
0.51, 0.99,
Gupta et. al., 438, 235, 168,
W4 55 6.9 0.50, 1.69, 2.75 1.56E11 1.56E13
(1995) 127, 91, 38
1.07, 2.09
438, 235, 0.82, 1.44,
Gupta et. al.,
W5 55 168, 127, 10.8 0.74, 2.50, 1.95 1.56E11 1.56E13
(1995)
91, 38 1.63, 3.67
438, 235, 0.82, 1.44,
Gupta et. al.,
W6 55 168, 127, 10.8 0.74, 2.50, 2.75 1.56E11 1.56E13
(1995)
91, 38 1.63, 3.67
438, 235, 1.09, 2.27,
Gupta et. al.,
W7 55 168, 127, 15.8 1.58, 3.63, 1.95 1.56E11 1.56E13
(1995)
91, 38 2.67, 4.54
438, 235, 1.09, 2.27,
Gupta et. al.,
W8 55 168, 127, 15.8 1.58, 3.63, 2.75 1.56E11 1.56E13
(1995)
91, 38 2.67, 4.54

The computed and experimentally determined normalized wear rates are compared for Cases
W1-W8 of Table 2 in Figures 4-7. Test Cases W1, W2 are dealt with in Figure 4. The maximum
and average difference between prediction and experiment are 18.3% and 10.5%, respectively. The
discrepancy may be due to the generation of fine particles due to attrition in closed loop systems in
the laboratory setup. Typically, erosion tests are performed over long time durations, and in a
closed loop system fine particles tend to accumulate while coarse ones break up. Thus, the actual
slurry flow experienced by the wear surface can be quite different from the nominal slurry data used
in computations. Experiments in field setups involving ‘once-through’ slurry flow would possibly
yield better match. According to Roco and Cader (1988), errors are also possible in particle
sampling and particle size measurement.

Figure 4. Comparison of normalized predicted wear rates with experiments of Roco and
Cader (1988): Cases W1 and W2.

Volume 1 · Number 3 · 2009


270 Wear Prediction in Fully Developed Multi-Size Particulate Flow
in Horizontal Pipelines

Figure 5. Comparison of normalized predicted wear rates with experiments


of Gupta et. al. (1995): Cases W3 and W4.

Figure 6. Comparison of normalized predicted wear rates with experiments


of Gupta et. al. (1995): Cases W5 and W6.

Figure 7. Comparison of normalized predicted wear rates with experiments


of Gupta et. al. (1995): Cases W7 and W8.

Journal of Computational Multiphase Flows


K. V. Pagalthivarthi, J. S. Ravichandra, S. Sanghi and P. K. Gupta 271

Another reason for the discrepancy could be due to the method of measuring actual wear rate.
In several studies, mass loss of embedded wear specimen is used to calculate the wear rate. This
method has been used, for example, by Gupta et. al. (1995). They used mild steel and brass pieces
inserted (embedded) in the pipe wall. Experiments were run typically. for 20 hours with slurry of
specified broad PSD. The mass loss (∆m) is related to the wear rate W as

∆m
W = , (8)
TP ρ wm Awp

where TP is the test duration, πwm is the density . of wear material and Ap is the area (exposed to
slurry) of the wear sample. For convenience, W is usually expressed in µm/hr.
The normalized wear rate distributions along the pipe wall are shown in Figures 5-7 for Cases
W3-W8. All six of these cases involve the multi-size particulate slurries of Gupta et. al. (1995).
The predicted wear rates are in reasonable agreement with experimental observations; the
maximum difference for all cases is 31%, while average difference is 16.5%. The discrepancy is
larger in the upper portion of the pipe. This may be due to the larger percentage of fines near the
top of the pipe due to particle attrition. Other sources of error are possibly due to particle sampling
and size measurement.
The computed wear rate trends indicate that the maximum wear does not necessarily occur at
the pipe bottom (θ = 0o), but could occur on the lateral side of the pipe wall. In some cases, the
maximum point is shifted to even 60o from the pipe bottom. In experiments involving mass loss
(for wear depth calculation), the exact location of the maximum is difficult to pin down. For
example, in the experiments of Gupta et. al. (1995), the sample width is 15 mm, which subtends an
angle of 15.62o at the centre of the pipe of diameter 55 mm (Figure 8). Since the calculated wear
depth is an average representation over the sample width, it is difficult to locate the exact position
of maximum wear.

Figure 8. Angle subtended by wear sample at pipe center.

Erosion literature shows many examples [Hisamitsu et. al. (1981), Roco and Cader (1988), Roco
and Addie (1987), Wood et al. (2004)] with maximum wear occurring at θ = θ∗ ≠ 0. Theoretically,
this conclusion is easy to arrive from the plots of concentration and solid velocity along the wall
[Ravichandra et al. (2004)]. One such example is shown in Figure 3. From Equations (1) and (2),
it is clear that the sliding wear rate (the dominant wear component) depends both on the local
concentration and local velocity. While the concentration decreases along the pipe wall, the
mixture velocity (and also the shear stress) increases with θ. Thus, it is expected that the maximum
wear should occur away from the pipe bottom (θ = 0o).

Volume 1 · Number 3 · 2009


272 Wear Prediction in Fully Developed Multi-Size Particulate Flow
in Horizontal Pipelines
5. PARAMETRIC STUDY OF WEAR TRENDS IN MULTI-SIZE PARTICULATE FLOWS
Various flow parameters governing the wear rate in solid-liquid flows are pipe diameter, solids
concentration, average flow rate, particle density and particle size distribution. In this section, the
effect of these parameters on wear in uniform and multi-size particulate flows is presented.

5.1 Wear Rate in Uniform Size Particulate Flows


5.1.1 Effect of Particle Diameter
Parametric study on particle size is performed with particle of size 50 µm, 200 µm, 350 µm, 500 µm
and 650 µm in a pipe of 300 mm diameter at W0 = 5.5 m/s and Ca = 15%. The computed sliding
wear rate along the pipe wall is shown in Figure 9.

Figure 9. Variation of sliding wear rate along the pipe wall for several mono-size particles.

Except for the flow with 50 µm particles, maximum sliding wear rate is located away from the pipe
bottom. The maximum value and the non-uniformity in sliding wear rate increased with particle
size. Sliding wear rate at the pipe bottom increased with particle size up to 350 µm and then
decreased with further increase in particle size. This anomalous behavior is explained as follows.
As particle size increases, the concentration at the pipe bottom increases due to increase in settling
tendency. This reduces the mixture velocity at the pipe bottom. The decrease in sliding wear rate
at pipe bottom could thus be attributed to the combined effect of increase in concentration and
decrease in mixture velocity with increase in particle size.
Returning for a moment to Figure 3, impact wear rate is seen to increase with increase in particle
size. As vk increases with increase in dp, impact angle increases with dp. According to Equation
(7), E1(α) decreases as α increases. Hence impact wear rate increases with increase in particle
diameter. Impact wear rate is practically constant along the pipe wall for 50 µm and 200 µm
particles. For particles larger than 200 µm, impact wear rate increases along the pipe wall. Non-
uniformity in impact wear rate increases with increase in particle size. These observations are in
accordance with the variations in concentration distributions with increase in particle size (see
Figure 2(c) for example).

5.1.2 Effect of Average Solids Concentration


Figure 10 shows the results of parametric study involving the effect of average concentration on
sliding wear rate along the pipe wall. Results are presented for four particle sizes (50 µm, 200 µm,
500 µm and 650 µm) at five average concentrations (5% to 25% in steps of 5%) in a pipe of 300
mm diameter at a velocity of 6.5 m/s.

Journal of Computational Multiphase Flows


K. V. Pagalthivarthi, J. S. Ravichandra, S. Sanghi and P. K. Gupta 273

Figure 10(a), shows sliding wear rate by 50 µm particles along the pipe wall. Sliding wear rate
increases with Ca and is practically constant all along the pipe wall. The uniformity increases with
increase in Ca. This is due to the practically uniform suspension of 50 µm particles in the pipe
cross-section. Thus, both the velocity and concentration near the pipe wall are nearly constant, and
so is the wear rate.

Figure 10. Effect of average concentration on sliding wear rates with particles
of various sizes.

Figure 10(b) shows sliding wear rate of 200 µm particles. Similar to 50 µm particles, sliding wear
rate of 200 µm increases with Ca, but its variation along the pipe wall is not uniform. Non-
uniformity is more at lower concentrations and it decreases with increase in concentration. Further,
with increase in Ca, the location of maximum sliding wear rate shifts to the right (away from pipe
bottom).
Figure 10(c) shows sliding wear rate of 500 µm particles. Unlike 50 µm and 200 µm particles,
sliding wear rate is highly non-uniform in the case of 500 µm. The location of maximum wear rate
is found to be shifted much more laterally than in the case of 200 µm particles. Another typical
observation in the case of 500µm particles is that, the sliding wear rate increases at a lesser rate
with increase in concentration, particularly near the pipe bottom. For example, sliding wear rate at
the pipe bottom is almost the same at 20% and 25% concentrations. As dp increases, the
concentration at pipe bottom tends to saturate, regardless of the average concentration.
Figure 10(d) shows sliding wear rate trends of 650 µm particles. Sliding wear trends are much
more non-uniform than with 500 µm particles. The location of maximum sliding wear rate on the
pipe wall shifts further to the right. The location of maximum wear rate is almost at θ = 90˚ in the
case of flow at 25% concentration. Furthermore wear rate at 10% and 15% concentrations and
wear rate at 20% and 25% concentrations are practically equal.
The plots of Figure 10 show that wear rate trends have a complex relationship with flow
parameters. The complexity increases enormously for multi-size particulate flows.
Results for impact wear were also similarly generated. However, impact wear rate is quite small
in comparison with sliding wear rate, and hence results of impact wear rate are not shown here.

Volume 1 · Number 3 · 2009


274 Wear Prediction in Fully Developed Multi-Size Particulate Flow
in Horizontal Pipelines
5.1.3 Effect of Average Mixture Velocity
Sliding wear rate are computed with 200 µm and 650 µm particles at 15% concentration in a pipe
of 300 mm diameter at average flow rates of 5.5 m/s and 6.5 m/s. They are presented in Figure 11.
The wear rate increases with increase in average flow rate. Both wear rates (sliding and impact)
appear to become more uniform with increase in average flow rate. This is expected theoretically
as an increase in flow rate tends to more uniformly distribute the particles. Thus, there are more
particles near the pipe top as flow rate increases.

Figure 11. Effect of average mixture velocity on sliding wear rate


with two mono-size slurries.

5.2 Wear Rate in Multi-Size Particulate Flows


Parametric study of multi-size particulate flows is performed on three slurries. The compositions
of the slurries and the range of various governing variables are given in Table 3. Slurry A is small
particle dominated, Slurry B has particles of various sizes with equal contribution to the overall
concentration and Slurry C is large particle dominated. Effect of various parameters on the wear
rates of these slurries are discussed in this section.

Table 3. Operating parameters used in finite element solutions.

Density of water = 1000 kg/m3


Viscosity of water = 0.001 N-s/m2
Overall concentration = 5% to 25% (in steps of 5%)
Pipe diameters = 100, 200, 300, 400 and 500 mm
Particulate densities = 1520 kg/m3, 2650 kg/m3, 3600 kg/m3 and
4200 kg/m3
738 235 180 128 91 38 %
Slurry/d
(µm) (µm) (µm) (µm) (µm) (µm) (µm) (µm)

A 3.5 10.0 5.7 19.3 13.9 47.6 116.9 6.58

B 16.67 16.67 16.67 16.67 16.67 16.67 238.4 42.63

C 47.6 13.9 19.3 5.7 10.0 3.5 439.1 137.16


= weighted mean diameter
= standard deviation

Journal of Computational Multiphase Flows


K. V. Pagalthivarthi, J. S. Ravichandra, S. Sanghi and P. K. Gupta 275

5.2.1 Effect of Average Mixture Velocity and Average Concentration


The effect of average mixture velocity on sliding wear rate for multi-size Slurry A is shown in
Figure 12. A pipe diameter of 0.3 m and average flow velocities of 3.5, 4.5 and 5.5 m/s are
considered. As in the case of uniform size slurries, sliding wear rate increases with increasing W0.
The effect of average concentration is also shown in the same figure. The following generalities
are observed in the studied range:
1. Wear rate at any given θ increases with Ca and with W0.
2. The non-uniformity of wear rate persists with increases in W0; this trend is similar to that
observed for mono-size slurries. For instance, in Figure 12, W0 = 3.5 m/s, Ca = 25%, the
minimum wear rate occurs at θ = 0˚ and it. is ~0.25 µm/hr, and . the maximum wear rate is
~0.5 µm/hr. However, at W0 = 5.5 m/s, Wmin ≅ 0.9 µm/hr, Wmax ≅ 1.4 µm/hr.
3. At the smaller W0 (= 3.5 m/s), the wear rate decreases along θ for Ca = 5%; at W0 = 5.5
m/s, the wear rate is nearly. constant.
4. The θ -location at which Wmax occurs tends to increase with Ca for a given W0.

Figure 12. Effect of average mixture velocity on sliding wear rates of Slurry A at several
average concentrations.

Although impact wear rate is expected to be much smaller than sliding wear rate (see Figure 3), for
the sake of completion, the variation of impact wear rate vs θ is shown in Figure 13 for multi-size
Slurry A of Table 3. Since particles have a downward settling motion, impact wear is significant
only for 0 ≤ θ ≤ 90˚, and hence impact wear rate is plotted for this range. The following
observations emerge from Figure 13:
1. At W0 = 3.5 m/s and θ = 0˚, maximum impact wear rate occurs with Ca = 5%, and
minimum impact wear rate occurs with Ca = 25%.
2. At W0 = 5.5 m/s and θ = 00, maximum impact wear rate occurs with Ca = 10%, and
minimum impact wear rate occurs with Ca = 5%.
3. The maximum wear rate is 4-6 times the minimum wear rate for a given W0 at Ca = 5%.
As Ca increases the wear rate distribution tends to become more uniform.

Volume 1 · Number 3 · 2009


276 Wear Prediction in Fully Developed Multi-Size Particulate Flow
in Horizontal Pipelines
4. The impact wear rates are much smaller than sliding wear rates for the same operating
conditions.

Figure 13. Effect of average mixture velocity on impact wear rate of Slurry A at several
average concentrations.

5.2.2 Effect of Solids Density


Sliding and impact wear rates of Slurry A are computed with three particulate densities (Ss =
1.52,2.65 and 3.6). Sliding wear rates are shown in Figure 14. Sliding wear rates at all
concentrations are nearly uniform along the pipe wall for Ss = 1.52. However, sliding wear
distribution becomes non-uniform as Ss is increased to 2.65 and 3.6. The non-uniformity due to
increase in Ss increases with increase in Ca.
Impact wear rates in flows with the three specific gravities are shown in Figure 15. Along with
non-uniformity in impact wear distribution, the magnitude of impact wear rate also increases
significantly with increase in Ss. Impact wear rates with Ss = 3.6 are almost 10 times those with
Ss = 1.52.
Again, sliding wear rates are much greater than impact wear rates for the same operating
conditions.

5.2.3 Effect of Particle Size Distribution


Sliding and impact wear rates for Slurries A, B and C are compared in Figure 16 and 17,
respectively. The results are with D = 0.3 m, W0 = 5.5 m/s and Ss = 2.65.
Slurry A is small particle-dominated. Figure 16(a) shows variation of its sliding wear rate along
the pipe wall. Sliding wear rate variation could be considered quasi-uniform, except for Ca = 25%.
Slurry B has particles of various sizes with each size class having equal contribution to the overall
concentration. Unlike Slurry A, sliding wear rate variation is not uniform in the case of Slurry B.
Notably, at θ = 0˚ (pipe bottom), sliding wear rate of slurry B is almost the
. same for Ca = 5% to
Ca = 25 %. Unlike in slurry A, for slurry B, the wear rate distribution (W SL vs θ) is non-uniform
for all five values of Ca. The location of the maximum
. wear rate shifts along the pipe wall (away
from θ = 00) as Ca increases. Furthermore, WSL tends to increase with Ca at any specified θ.

Journal of Computational Multiphase Flows


K. V. Pagalthivarthi, J. S. Ravichandra, S. Sanghi and P. K. Gupta 277

Figure 14. Effect of solids density on sliding wear rates of Slurry A at several average
concentrations.

Figure 15. Effect of solids density on impact wear rates of Slurry A at several average
concentrations.

Volume 1 · Number 3 · 2009


278 Wear Prediction in Fully Developed Multi-Size Particulate Flow
in Horizontal Pipelines

Figure 16. Sliding wear rates distributions of Slurries A, B and C at several concentrations.

Figure 17. Impact wear rate distributions of Slurries A, B and C at several concentrations.

.
For slurry C, which is dominated by large particles, the trends of WSL are similar to those for slurry
B. However, the wear rate magnitudes are higher for slurry C than for slurry B.

Journal of Computational Multiphase Flows


K. V. Pagalthivarthi, J. S. Ravichandra, S. Sanghi and P. K. Gupta 279

In Figure 17, the distribution of impact wear rate is shown for Slurries A, B and C at average
concentrations in the range 5% to 25%. The observations from this figure are
1. For the fine particle dominated slurry (Slurry A), the maximum impact wear rate occurs
at the pipe bottom; for Slurries B and C, the maximum impact wear rate occurs at θ values
increasing with Ca.
2. For Slurry A, no definitive trend in wear rate can be seen as a function of Ca; for Slurries
B and C, impact wear rate decreases with increasing Ca at the pipe bottom.
3. Again, impact wear rates are much less than sliding wear rates.
Impact wear rate distribution has a complex relationship with flow parameters as in previous
cases.

6. CONCLUSIONS
The code for predicting fully developed multi-size particulate flows in horizontal pipes
[Ravichandra et al. (2004)] accurately predicts the values of local concentration, solids velocity,
wall shear stress and local impact angle. These local variables are translated into wear rate along
the pipe wall using relevant specific energy data collected from the open literature. The following
conclusions are drawn from the present study:
1. Wear is assumed as the sum of sliding and impact wear rates. However, impact wear rate
is found to be at least two orders of magnitude less than sliding wear rate due to
• negligible angle of particle impact on the pipe wall (<1˚), and
• specific energy coefficient (in J/m3) of impact wear is 1-3 orders of magnitude
greater than the corresponding coefficient of sliding wear.
2. Normalized wear rates predicted by the code are compared with published experimental
results.
3. Wear predictions are found to be within an average error of 16.5% with respect to
experimental observations.
4. In uniform size particulate flows,
• non-uniformity of wear rate increases with increase in dp,
• depending on the operating conditions, maximum wear rate is located either at the
pipe bottom or laterally away from pipe bottom,
• the lateral shift of the location of maximum erosion wear increases with dp and Ca,
and
• wear rate increases and tends to become uniform with increase in W0.
5. In multi-size particulate flows,
• wear rate is a strong function of PSD,
• in small particle dominated slurries, wear is quasi-uniform,
• non-uniformity increases with the proportion of larger particles,
• effect of increase in Ca on wear rate is proportional to the corresponding increase in
the larger particles, and
• increase in solids density increases the wear rate and laterally shifts the location of
maximum wear rate.
6. Different wear trends with different flow parameters indicate the existence of a very
complex relationship between wear rate and the flow governing parameters.

A major conclusion relevant to the present work is that wear rate prediction using a single
representative diameter may be seriously in error, particularly if the particle size distribution is
broad. So for slurries with broad PSDs, accurate wear prediction could be obtained by
characterizing the particle size distribution into suitable number of size classes, and accounting for
the particle-size dependence in the wear model(s). Suitable correlations can be developed for an
“equivalent particle size” based on the predictions for several slurries with various PSDs. This
would yield the same wear rate as that obtained for the actual particle size distribution. Such an
“equivalent particle size” could be useful in treating the slurry as consisting of mono-size particles
in designing for erosion wear.

Volume 1 · Number 3 · 2009


280 Wear Prediction in Fully Developed Multi-Size Particulate Flow
in Horizontal Pipelines
REFERENCES
Elkholy, A., Prediction of Abrasion Wear for Slurry Pump Materials, Wear, 1983, 84, 39–49.
Addie, G.R. and Pagalthivarthi, K.V., Prediction of Dredge Pump Shell Wear, Proc. WODCON XII, 12th World Dredging
Conference, World Organization of Dredging Associations, Arlington, VA, 1989, 481-504.
Gandhi, B.K., Singh, S.N., Seshadri, V., Study of the Parametric Dependence of Erosion Wear for the Parallel Flow of
Solid–Liquid Mixtures, Tribol. Int., 1999, 275–282.
Lin, F.Y., Shao, H.S., Effect of Impact Velocity on Slurry Erosion and a New Design of a Slurry Erosion Tester, Wear, 1991,
147, 231– 240.
Finnie, I., Some Reflections on the Past and Future of Erosion, Wear, 1995, 186-187, 1-10.
Gandhi, B.K. and Borse, S.V., Nominal Particle Size of Multi-Sized Slurries for Evaluation of Erosion Wear and Effect of
Fine Particles, Wear, 2004, 257 (1-2), 73-79.
Gupta, R., Singh, S.N. and Seshadri, V., A Study on the Uneven Wear Rate in a Slurry Pipeline, Bulk Solids Handling, 1995,
15(4), 603-607.
Gupta, R., Studies on Erosion Wear in Slurry Pipelines, Ph.D. Thesis, 1994, Indian Institute of Technology, New Delhi,
India.
Clark, H.M., On the Impact Rate and Impact Energy of Particles in a Slurry Pot Erosion Tester, Wear, 1991, 147, 165–193.
Clark, H.M. and Hartwich, R.B., A re-examination of the particle size effect in slurry erosion, Wear 248 (2001) 147–161.
Hisamitsu, N., Iseh, T., Honda, Y. and Takeiski, Y., An Experimental Study on Pipe Erosion by Sand Slurry,” Proceedings
of 6th Conference on Slurry Transportation, STA, 1981, 319-332.
Kaushal, D.R., Seshadri, V. and Singh, S.N., Prediction of Concentration and Particle Size Distribution in the Flow of Multi-
Sized Particulate Slurry through Rectangular Duct, Applied Mathematical Modeling, 2002, 26(10), 941-952.
Meng, H.C. and Ludema, K.C., Wear Models and Predictive Equations – Their Form and Content, Wear, 1995, 181-183,
443-457.
Pagalthivarthi, K.V. and Addie, G.R., Prediction Methodology for Two-Phase Flow and Erosion Wear in Slurry Impellers,
Proceedings of the International Conference on Multiphase Flow, New Orleans, May 27-June 22, 2001, Ref. No.
905, Paper No. EC2.
Pagalthivarthi, K.V. and Gupta, P.K., Comparison of Three Turbulence Models in Erosion Wear Prediction of Multi-Size
Particulate Flow through Rotating Channel, Fluid Dynamics and Materials Processing, 2009, 5(1), 93-122.
Pagalthivarthi, K.V. and Helmly, F.W., Applications of Materials Wear Testing to Solids Transport via Centrifugal Slurry
Pumps, Wear Testing of Advanced Materials, 1992, ASTM STP 1167, eds. Divakar, R. and Blau, P.J., 114-126.
Pagalthivarthi, K.V. and Ramanathan, R., Numerical Insight into Experimental Results of Particle Size Effect in Coriolis
Wear Tester, Proc. of the First International FMFP Conference, 15-17 Dec., 1998, IIT Delhi, New Delhi.
Lynn, R.S., Wong, K.K., Clark, H.H., On the Particle Size Effect in Slurry Erosion, Wear, 1991, 149, 55–71.
Ravichandra, J.S., Pagalthivarthi, K.V. and Sanghi, S., Finite Element Study of Multi-Size Particulate Flow in Horizontal
Pipes, Progress in Computational Fluid Dynamics, 2004, 6(4), 299-308.
Ravichandra, J.S., Pagalthivarthi, K.V. ,Sanghi, S., Numerical Study of Particulate Impact Wear in Horizontal Pipe Flow,
IMPLAST ’03, 2003, New Delhi, India, 972-978.
Roco, M.C. and Addie, G.R., Analytical Model and Experimental Study on Slurry Flow and Erosion in Pump Casings,” STA,
1983, 8, 263-271.
Roco, M.C. and Addie, G.R., Erosion Wear in Slurry Pumps and Pipes, Powder Technology, 1987, 50, 35-46.
Roco, M.C. and Cader, T., Numerical Method to Predict Wear Distribution in Slurry Pipelines, Advances in Pipeline
Protection, 1988, eds. Jones, G. and Thorn, J., BHRA, Cranfield, UK.
Roco, M.C., Addie, G.R., Visintainer, R. and Ray, E.L., Optimum Wearing High Efficiency Design of Phosphate Slurry
Pumps, Proceedings of the 11th International Conference on Slurry Technology, Slurry Transport Association, 1986,
Washington, DC, 277-285.
Roco, M.C., Nair, P., Addie, G.R. and Dennis, J., Erosion of Concentrated Slurry in Turbulent Flow, ASME FED, 1984, ed.
Roco, M.C., 13, 69-77,
Tian, H.H., Addie, G.R., Pagalthivarthi, K.V., Determination of Wear Coefficients for Prediction through Coriolis Wear
Testing, Wear, 2005, 259, 160-170.
Tuzson, J.J. and Clark, H.Mcl., The Slurry Erosion Process in the Coriolis Wear Tester, Proc. FEDSM ’98, ASME Fluids
Engg. Division Summer Meeting, June 21-25, 1998, Washington, D.C., Paper no. FEDSM98-5144.

Journal of Computational Multiphase Flows


K. V. Pagalthivarthi, J. S. Ravichandra, S. Sanghi and P. K. Gupta 281

Tuzson, J.J. and Scheibe-Powell, K.A., Slurry Erosion Tests with Centrifugal Erosion Tester, Liquid-Solid Flows and
Erosion Wear in Industrial Equipment, ASME-FED, 1984, ed. Roco, M.C., 13, 84-87.
Wood, R.J.K., Jones, T.F., Ganeshalingam, J. and Miles, N.J., “Comparison of Predicted and Experimental Erosion
Estimates in Slurry Ducts,” Wear, 2004, 256 (9-10), 937-947.
Wood, R.J.K., Jones, T.F., Ganeshalingam, J. and Miles, N.J., Upstream Swirl Induction for Reduction of Erosion Damage
from Slurries in Pipeline Bends, Wear, 2001, 250(1-10), 771-779.

Volume 1 · Number 3 · 2009

You might also like