Wear 434–435 (2019) 202972

Contents lists available at ScienceDirect

Wear
journal homepage: www.elsevier.com/locate/wear

Experimental investigation of centrifugal slurry pump casing wear handling T

solid-liquid mixtures
Rahul Tarodiya, Bhupendra K. Gandhi*
Department of Mechanical and Industrial Engineering, Indian Institute of Technology, Roorkee, 247667, India

A R T I C LE I N FO A B S T R A C T

Keywords: The rotating and stationary components of a centrifugal slurry pump develop complex flow field due to sus-
Centrifugal slurry pump pended solids in the liquid. An uneven wear of the wetted parts of the pump components was reported due to
Casing wear different particle impact conditions like the direct, random and directional. The present work is focused to
Flow field investigate the erosive wear profile of the pump casing experimentally and its relationship with numerically
Erosion mechanisms
simulated flow field to establish the effect of dominating factors affecting the erosion of the pump casing at
Operating condition
different operating conditions. Experiments were conducted in a pilot plant test rig with two equi-sized sand
particulate slurries at two pump speeds and two flow rates. Wear specimens (1.5 x 1.5 x 0.01 cm3) of AISI 304L
steel were affixed at different locations along the centerline of the casing wall. The erosion of the specimens was
determined based on the measurement of weight loss to study the wear pattern along the casing. Scanning
electron microscope (SEM) images of worn out samples were also examined in each case to investigate the
material failure at different locations. Further, the flow field inside the pump has been numerically simulated to
correlate the particle impact condition with the measured wear. Efforts have been made to identify the dominant
mechanism of erosion at different locations of the casing. It was observed that the wear at the volute tongue is
contributed by both the cutting and deformation whereas, at all other locations, the material is removed due to
cutting and ploughing.

1. Introduction erosion behaviour of the components handling solid-liquid mixture.

Different techniques were used to study the erosion along the compo-
Centrifugal slurry pumps are being widely used for the transporta- nents namely paint wear method [3,5], measurement of reduction in
tion of solids in the mining, cement, ceramics, power generation, che- the wall thickness [6–8], the weight loss measurement of individual
mical processing and dredging industries [1]. These pumps have been components [9–11] or wear pieces fixed inside the geometry through
developed with few modifications in the existing design of the con- suitable arrangements [2,12], or pasting thin wear samples inside the
ventional centrifugal pumps to ensure smooth passage of the solid-li- flow passages [13,14]. Want [15] studied the wear pattern with dif-
quid mixture [2]. Erosive wear is an important consideration in the ferent particle sized slurries and reported the region of failures like
design of these pumps for the longevity of the pump components. A puncturing of the volute adjacent to the cutwater, worn surface con-
better understanding of the erosion helps to determine the operational sisting of shallow scallops near the cut water etc. Roco et al. [3] used
life, hydraulic indices and the reliability of the pump [3]. The major the paint wear method and tested three different pump geometries for
components of the pump which fails due to erosion are the impeller and dilute and dense slurries of sand-water. They observed that the erosion
the casing [1]. The wear of the moving parts like the pump impellers is of the pump components is contributed by directional impact, random
nearly uniform [4] but this is not generally true for the high-efficiency impact, and sliding of the solid particles. The average wear rate of the
pump casing which is the most expensive and difficult to replace [3]. casing was found higher than that of the other components which re-
The casing wear showed generally non-uniform erosion pattern along duced with an increase in the flow rate. They reported that the max-
the flow path [4]. It may be due to the large variation in flow and imum average erosion rate of the casing along width at the rated flow
impact conditions. The measurement of material loss due to erosion rate occurred in the region between 285° to 315° from the volute tongue
during field test requires a long period of operation. Generally, ex- in the direction of impeller rotation. Gandhi et al. [2] measured the
periments in the pilot plant test rig were performed to investigate the average mass loss due to the erosion of the wear specimens to study the

*
Corresponding author.
E-mail address: [email protected] (B.K. Gandhi).

https://doi.org/10.1016/j.wear.2019.202972
Received 23 May 2019; Received in revised form 20 June 2019; Accepted 16 July 2019
Available online 17 July 2019
0043-1648/ © 2019 Elsevier B.V. All rights reserved.
R. Tarodiya and B.K. Gandhi Wear 434–435 (2019) 202972

Fig. 1. Slurry pilot plant test setup.

casing wear profile. They welded a mild steel strip inside the flow erosion profile may show different erosion behaviour with operating
passage of the casing and fixed the straight wear samples to cover the conditions as compared to the localized erosion profile. Also, the un-
complete casing width at different locations inside the casing. They derstanding of the mechanism of failure of the casing surface at dif-
observed that the location of the maximum wear was around 180° from ferent locations is developed only through the numerically predicted
the volute tongue, and it did not change with the change in flow rate flow field or by comparing the erosion profile variation with the erosive
and solid concentration. wear data generated through a laboratory test rig. In the present study,
Many investigators [16–19] performed numerical simulations to experiments are performed to investigate the casing erosion behaviour
identify the zones of high erosive wear in the pump and the factors under different operating conditions of the pump. Thin (100 μm) wear
affecting it. Minemura and Zhong [16] employed a two-dimensional specimens are used to paste on the casing wall to evaluate the erosion
Lagrangian finite volume approach to numerically predict the erosion with the minimum flow disturbance. The wear samples are pasted along
of a pump casing. They reported that the dominant mechanisms of the centerline of the casing in order to develop an understanding of the
erosion are deformation and cutting, and the location of maximum dominant parameters affecting the casing erosion with the change in
wear at BEP flow rate was at 300° from the volute tongue. Pagalthi- pump operating conditions. The centerline location is selected based on
varthi et al. [17] performed a two-dimensional numerical simulation of the higher erosive wear at this location as reported in literature [3]. To
the casing of a centrifugal slurry pump using finite element modeling. correlate the erosion rate of pump casing with the particulate flow field,
They observed that the distribution of wear is non-uniform along the numerical simulations of the pump has also been performed for similar
casing with two local peaks. The first peak was located at the tongue operating conditions. The effect of flow rate, particle size, and speed on
and the second peak was observed in the region of 285° to 315° from the the casing erosion are studied. Further to investigate the mechanism of
volute tongue. They reported that the flow conditions significantly af- casing surface failure, the surface of the worn out wear specimens at
fect the erosion profile of the casing. Roudnev et al. [18] predicted the different angular locations are analyzed using scanning electron mi-
wear rate along the casing using three-dimensional steady-state and croscope.
transient simulations using commercial CFD code Ansys CFX. They re-
ported the need of transient simulation of the complete pump model for
the better prediction of the casing wear. They observed that the location 2. Experimental setup and procedure
of maximum wear at the centerline of the casing was around 50o-140°
from the tongue. Recently, Noon and Kim [19] performed a steady 3D The experiments are performed in a slurry pilot plant test rig
frozen rotor simulation of a centrifugal slurry pump for handling lime available in the slurry research laboratory of Mechanical and Industrial
slurry using the Euler-Lagrangian model of Ansys CFX. They estimated Engineering Department, IIT Roorkee as shown in Fig. 1 (a,b). A short
the average erosion rate over the width of the volute casing at different loop of 50 mm pipe diameter is used in the present work for measure-
angular locations to show the effect of particle size and concentration. ment of “50 M WILFLEY” pump performance and wear characteristics
They observed two local peaks in the erosion profile under different of its casing. The details of important pump geometry are shown in
operating conditions. First was at the tongue of the casing and second Table 1. The pump is driven by 7.5 kW, 440 V and 14.5 A variable speed
was at around 265° from the tongue. induction motor. For the variation of motor speed, variable frequency
Literature shows that the efforts were made to study the erosion of drive (ABB make) is used. The hopper shaped mixing tank is used for
the pump casing under different operating conditions. It has been found the preparation of slurry which is equipped with an arrangement to
that the knowledge of the dominating factors affecting the erosion of keep the slurry in a well-mixed state during experimentation. The flow
the pump casing at different operating conditions is still not conclusive. rate is measured by an electromagnetic flowmeter (Magmeter, Iotaflow
Most of the studies [2,3,19] discussed the effect of operating conditions make) with an accuracy of ± 0.5%. An arrangement is made to deflect
on the average erosion rate along the complete width of the casing at the flow into either mixing tank or the measuring tank. The measuring
different angular locations. However, due to the complex nature of flow tank is used to calibrate the flowmeter during experimentation. The rise
inside the centrifugal slurry pump, the particle flow field may not be in the level of slurry in measuring tank for a known time interval is
uniform along the width of the casing and therefore, the average measured for calibration of an electromagnetic flowmeter. Plug valves
are provided to vary the flow rate. An efflux sampling tube fitted with a

2
R. Tarodiya and B.K. Gandhi Wear 434–435 (2019) 202972

Table 1 location 2 is close to the casing tongue (5°) which was reported as one
Geometry details of the centrifugal slurry pump. of the critical location of erosion in casing [17,19], location 3 is 15°
Specification downstream to location 2 to further understand the erosion close to the
tongue and other locations are equally spaced at 30° to each other to
Impeller Details determine the variation in erosive wear of the casing along its flow
(i) Type Closed
path. The erosive wear of the specimens is determined by measuring the
(ii) Material Ni-hard
(iii) Number of blades 5
weight loss of the specimens before and after running the pump for a
(iv) Impeller eye diameter (mm) 110 known period of time.
(v) Impeller outlet diameter (mm) 265 The performance of the pump is initially measured with water after
(vi) Impeller width at eye (mm) 44.2 affixing the specimens along the wetted wall of the casing. Thereafter,
(vii) Impeller width at the outlet (mm) 68.6
solid particles are added into the mixing tank and after proper mixing,
(viii) Impeller inlet vane angle (degree) 23
(ix) Impeller outlet vane angle (degree) 25 the test is started at the desired operating conditions. During the test,
(x) Blade thickness (mm) 22.4 slurry samples are collected at every half an hour duration to monitor
Casing Details the solid concentration. These samples are further analyzed to check the
(i) Type Semi-volute
particle attrition, if any. No significant change in the particle size is
(ii) Material Ni-hard
(iii) Base volute diameter (mm) 275
observed over the total duration of any tests. To evaluate the erosive
(iv) Volute width (mm) 85 wear at different locations of the casing centerline, weight loss of each
(V) Impeller-tongue radial gap (mm) 31 wear specimens is measured using an electronic balance (least count of
Inlet passage diameter (mm) 100 0.1 mg) by running the pump for a period of 1.5 hours.
Outlet passage diameter (mm) 50

3. Material and range of parameters

plug valve is used for the collection of the slurry sample to monitor the
solid concentration. Static pressure head developed by the pump is The solid-liquid mixture is prepared by mixing Indian standard sand
measured by installing two pressure transducers (LD301, Make: Smar) (IS 650:1991) with water. The physical and chemical properties of
with separation chambers at two diameters upstream and downstream Indian standard sand are given in Table 3, which shows that its main
of the suction and delivery flanges respectively. The absolute pressure constituent is silica. For the purpose of conducting the tests, the solid
transducer and gauge pressure transducer are used at suction and de- particles are sieved using available sieve sizes and the particles retained
livery sides, respectively. The accuracy of both the transducer between two successive sieves are used to prepare solid-liquid mixture.
is ± 0.75% of full scale. The pump speed is measured by a non-contact The mean of the sieve sizes of 400 μm (+300–500), and 605 μm
type tachometer having an accuracy of ± 1 RPM. For measurement of (+500–710) are used to designate the particle size for the experiments.
shaft power, the torque input is measured through a torque transducer The numbers in the parenthesis represent the successive sieve sizes. The
(T22-M, Make: M/s HBM LTD, Germany) installed between the pump scanning electron microscope (SEM) photographs of both the particles
and the motor shafts. The range of the torque transducer is 0.1 Nm to are shown in Fig. 3. The particles appear to be blocky in shape. The
200 Nm with an accuracy of ± 0.2%. cases of different operating conditions at which the experiments are
Erosive wear profile of the pump casing is determined by affixing performed to study the casing wear are listed in Table 4. The best ef-
100 μm thick wear specimens (1.5 x 1.5 cm2) of AISI 304L steel along ficiency point (BEP) of the pump at 1200 rpm and 1050 rpm is at 12 L/s
the centerline of the casing wall. The chemical composition and hard- and 10 L/s flow rate, respectively. The solid concentration at which the
ness of the specimens are listed in Table 2. The casing material of the experiments are performed is kept nearly 10% by weight. To determine
pump used is Ni-hard but the choice of material is not limited for its the effect of flow rate the experiments are conducted with 605 μm size
manufacturing and it may be of steel, cast iron, high chromium cast particles at 12 L/s and 9 L/s which is the BEP and 0.75 BEP flow rate
iron, ceramic, elastomer etc. [20]. The Ni-hard material has higher condition at 1200 rpm respectively. Experiments are also performed
erosion resistance and it is difficult to get measurable erosion loss in with 400 μm size particles at 1200 rpm and BEP flow rate condition to
short period of time in a laboratory under controlled conditions using investigate the effect of particle size. Further to determine the effect of
close circuit test loop. Therefore, experiments are performed to un- speed, experiments are performed at 1050 rpm with 400 μm size par-
derstand the casing wear by inserting specimens of low erosion re- ticles at the BEP flow rate condition.
sistance material. To minimise the flow disturbances, thin (100 μm)
wear specimens of 304L SS are pasted at selected locations inside the 4. CFD simulation
casing and the mass loss of the specimens are measured in a single test
run of 90 minutes to minimise the effect of particle attrition and 4.1. Modeling technique
rounding-off. Fresh wear specimens are used for each experiment. The
surface of the specimens is as received from the manufacturer (Bhandari The advances in the numerical techniques have facilitated to ana-
Foils and Tubes Ltd, Dewas, India) and no additional surface treatment lyze the flow field and particle impact conditions along the flow pas-
is performed. The average arithmetic mean roughness (Ra) of each sages. Roundev et al. [18] observed that three dimensional unsteady
specimen surface before the test is observed within 0.1 μm. The test modeling of the complete pump can reasonably predict the pump casing
specimens are cleaned with tap water, rinsed in acetone and dried with erosion which matches well with the experimental measurements. In
hot air blower before and after the test. The specimens are then glued at the present work, three dimensional unsteady modeling of the complete
14 different angular locations along the casing centerline as shown in pump is performed using commercial CFD code Ansys Fluent 17.2 to
Fig. 2. The locations 1 and 14 are at the discharge end of the casing, analyze the flow field in the casing. Ansys ICEM is used to develop the
mesh model of the complete pump. Due to rotating and stationary
components, separate mesh of tetrahedral elements is generated for
Table 2
different components, namely suction, impeller, and the casing which
Properties of wear specimens.
are connected through interfaces as shown in Fig. 4 (a). The complete
Material Chemical composition (wt. %) Hardness (Hv) mesh model of the pump is shown in Fig. 4b. Mesh independency test is
also performed using three different meshes, namely m1, m2 and m3
AISI SS304L C-0.021, Mn-1.76, P-0.030, S-0.003, Si-0.22, Cr- 160
18.08, Ni-8.06, Fe-bal. with numbers of elements as 2.02 million, 2.72 million and 4.93 mil-
lion, respectively. The simulation of different meshes are performed for

3
R. Tarodiya and B.K. Gandhi Wear 434–435 (2019) 202972

Fig. 2. Slurry pump casing showing the locations for wear specimen.

Table 3 Table 4
Physical and chemical properties of Indian standard Operating conditions of the experiments.
sand.
Case Flow Particle size Pump Solid Concentration Duration
Color Grayish White rate (L/ (μm) speed (by weight) (min)
s) (rpm)
Specific gravity 2.652
Shape of grains Sub angular I 12 605 1200 10.1% 90
SiO2 99.30% II 9 605 1200 9.4% 90
Fe2O3 0.10% III 12 400 1200 9.8% 90
IV 10 400 1050 9.5% 90

case III (see Table 4). The predicted particle flow field along the casing
centerline, particle velocity and volume fraction, with all the three 0.8 is set for pressure, momentum, volume fraction, turbulent kinetic
meshes are shown in Fig. 5. It is seen that the casing centerline profile energy and turbulent dissipation energy, respectively. The convergence
of particle velocity does not vary significantly with the increase in mesh criterion is set as 10−4 for each variable in every time step which was
density. However volume fraction profile showed variation for increase also taken by Pie et al [23] for unsteady simulation using discrete phase
in mesh elements from m1 to m2 which does not change further for m3. model (DPM). In this study, DPM is used for the multiphase simulations
Therefore, m2 mesh of 2.72 million elements with the minimum or- to investigate the particle flow field inside the pump casing. The si-
thogonal quality of 0.32 is selected for unsteady simulation in the mulated particle impact conditions on the casing surface are used to
present study. The time step of 4.76 x 10−4s and 4.16 x 10−4s which is correlate with the experimental measurements of erosive wear. Sand
equal to 3° impeller rotation is set to perform the simulation at particles (specific gravity = 2.65) are selected as the solid phase in the
1050 rpm and 1200 rpm, respectively. To get the stable periodicity in fluid domain. Surface type of injection is used to release the particles at
the results, the simulations are initially performed for first ten impeller the impeller inlet at the same velocity as water. The mass flow rate of
revolutions. The average data of the one complete impeller revolution the particle is calculated from the solid concentration. The number of
(i.e. eleventh rotation) is used for the analysis. The velocity at the inlet continuous phase iterations for each DPM iteration is set as 10. A total
and pressure as the outlet are set as boundary conditions. Standard k- of 15,000 particles are simulated during DPM iteration. Increase in the
epsilon model with standard wall function is used to model the turbu- number of particles for DPM simulation increases the computational
lent flow [21,22]. SIMPLE algorithm is used for pressure-velocity cou- time. Vieira et al. [24] reported that the simulated results from DPM
pling. Standard discretization schemes are used for the pressure terms modeling of the solid-liquid mixture are unaffected after 10,000 num-
and first-order upwind discretization scheme is used for the convection bers of particles. With the present number of particles, DPM setup,
and diffusion terms. The under-relaxation factor of 0.3, 0.7, 0.5, 0.8 and rotor-stator interaction, and mesh size, the time required to get one

Fig. 3. SEM micrographs of sand particles.

4
R. Tarodiya and B.K. Gandhi Wear 434–435 (2019) 202972

Fig. 4. (a). Computational domain of the centrifugal slurry pump with two interfaces namely suction to the impeller (Interface I) and impeller to the casing (Interface
II).

Fig. 6. Comparison of particle tangential velocity at the casing location 13 of

Furlan et al. [27] with the present CFD results.
Fig. 5. Mesh independency test of particle flow field.

erosion have been discussed in the following sections.

result with eleven impeller revolution is nearly 9 days with a 28 core
and 256 GB RAM workstation. Forder et al. [25] model is used to model
the particle-wall rebound characteristic which was also used by Messa 5.1. Effect of flow rate
and Malavasi [26]. For the particles, the wall boundaries are set as
‘reflect’ and the outlet boundary is set as ‘escape’. In order to get the The pump casing wear is experimentally evaluated at two flow
reasonable accuracy of the predictions, the following assumptions are rates, BEP, and 0.75 BEP. The mass loss of the specimens due to erosion
made: (a) Particles are spherical in shape with no attrition effect during at different locations of the casing is presented in Fig. 7 for the particle
simulation, and (2) there is no modification in the flow geometry size of 605 μm and pump speed of 1200 rpm at the two flow rates, 12 L/
during the simulation. s (BEP) and 9 L/s (0.75BEP). It is observed that the erosive wear along
the casing centerline is non-uniform in both the cases which may be

4.2. Model validation

The numerically estimated particle tangential velocity at the loca-

tion 13 (see Fig. 2a) is used to compare with experimental results of
Furlan et al. [27] as shown in Fig. 6. The simulation is performed with
similar input parameters at which the experiments were conducted.
Due to the difference in pump geometry, the results are compared by
plotting the variables in the non-dimensional form. The non-dimen-
sional velocity along the width is plotted by dividing the particle tan-
gential velocity with the particle tangential velocity at the middle
plane. A good agreement between the predicted non-dimensional par-
ticle velocity profile along the width of the experimental data is ob-
served.

5. Results and discussion

The average mass loss of the wear specimens affixed at different

locations at the centerline of the casing is evaluated and presented for Fig. 7. Experimentally measured mass loss of the specimens at different casing
different operating conditions. The effect of flow rate, particle size and locations for 12 L/s and 9 L/s operating flow rate with 605 μm particle size at
pump speed on the erosion, numerical flow field and mechanism of 1200 rpm.

5
R. Tarodiya and B.K. Gandhi Wear 434–435 (2019) 202972

Fig. 9. Experimentally measured mass loss of the specimens at different casing

Fig. 8. Numerically predicted particle impact velocity and volume fraction at locations for 12 L/s operating flow rate with 605 μm and 400 μm particle size at
the centerline of the casing for 12 L/s and 9 L/s operating flow rate with 605 μm 1200 rpm.
particle size at 1200 rpm.

attributed to the variation of particle flow field and the impact condi-
tions. The secondary flows developed in the space between the impeller
outlet and the casing also affects the flow field [2]. Further, the max-
imum material loss for both the flow conditions is observed at the lo-
cation 5 (θ = 80°) of the casing (see Fig. 7). It shows that the location of
the maximum erosion along the centerline of the pump casing remains
unaffected by the operating flow rate. Roudnev et al. [18] also reported
the zone of maximum erosion in the region 50o-140° from the casing
tongue in the direction of impeller rotation. To further understand the
casing erosion, the numerically predicted particle flow field along the
casing centerline, particle velocity and volume fraction, for the two
flow rates are presented in Fig. 8. It is observed that the particle velo-
city along the length of the casing does not vary much but a significant
variation in the volume fraction of the particles along the centerline of
the casing wall is observed. At 0.75 BEP flow rate, the mass loss of
specimens along the casing wall is lower than that of the BEP flow rate
at all the locations which may be attributed to the lower particle vo-
lume fraction. This observation is in line with Roco et al. [3] and
Walker et al. [7], however, Gandhi et al. [2] observed nearly same Fig. 10. Numerically predicted particle impact velocity and volume fraction at
weight loss for the flow rates of 9 L/s and 18 L/s. The variation of the centerline of the casing for 12 L/s operating flow rate with 605 μm and
particle flow field over the width of the casing is analyzed numerically 400 μm particle size at 1200 rpm.
and discussed in section 5.4.
change in the particle size at constant pump speed and flow rate do not
5.2. Effect of particle size significantly change the magnitude of particle impact velocity, how-
ever, the volumetric concentration of the particles increases with the
Particle size is one of the critical parameters during slurry handling reduction in particle size. This could be due to the increase in the
that affects the extent of erosion. The kinetic energy of a spherical uniformity of the particles along the casing width with the reduction in
particle increases as the cube of the particle size, however, the erosion particle size. Thus the increase in kinetic energy of the impacting par-
is generally reported to increase with an increase in particle size using ticles with an increase in particle size is partially balanced by the re-
power law relationship (E α dn) with a power-law coefficient (n) duction in particle volume fraction. It may be one of the reasons for the
varying between 0.3 and 2.0 [24]. The effect of particle size on casing lower increase in erosive wear as compared to the increase in kinetic
centerline erosion profile at nearly similar solid concentration (by energy due to the increase in particle size. However, the amount of
weight) is presented in Fig. 9 for a pump speed of 1200 rpm and flow mass loss of the specimens is dominated by the kinetic energy of the
rate of 12 L/s (BEP) for two particle sizes of 605 μm and 400 μm. It is particles impacting the wear specimens. Thus the bigger particle causes
observed that the erosion pattern with these two particle sizes is almost more erosion compared to the smaller size particles.
similar, however, the magnitude of mass loss with 400 μm size particles
is around 20–60% less as compared to that of the 605 μm size particles.
The number of particles impacting at the surface is lower for 605 μm 5.3. Effect of pump speed
particle size as compared to 400 μm particle size for the same solid
concentration [28], however, the kinetic energy of the former is higher The effect of pump operating speed on the casing wear is also in-
than the later due to 51.3% higher particle size. To further understand vestigated. Fig. 11 shows the effect of pump speed on the centerline
the effect of particle size on the casing erosion, the numerically simu- erosion profile of the casing operating at BEP flow condition. It is ob-
lated particle flow field data is presented in Fig. 10. It is observed that served that the mass loss of the specimens is normally decreased with

6
R. Tarodiya and B.K. Gandhi Wear 434–435 (2019) 202972

wear over the width at any cross-section [2,3,19]. To investigate the

particulate flow field inside the casing, the numerical simulation results
are plotted along the casing length and width as shown in Fig. 13(a–h)
for all the four cases studied in the present work (see Table 4). The
casing length is measured from the volute tongue and the data at each
angle along the flow direction are used to plot the velocity and con-
centration contours till the throat (see Fig. 2). To analyze the casing
wall flow, non-dimensionalised width is presented with ‘0’ for the front
side and ‘1’ for the back end.
Fig. 13a shows that the velocity is non-uniform across the width at
all the casing angles. Higher velocity in the center region is observed as
compared to both sides of the casing. The centerline velocity is max-
imum near 20° and 300° angle. The velocity between the center and
front side of the casing is observed higher than the other regions along
the whole length. The flow velocity is significantly low just after the
tongue due to the separation of flow which increases with increase in
the tongue angle. The maximum velocity is observed near 80° casing
Fig. 11. Experimentally measured mass loss of the specimens at different casing angle in the region between the centerline and front side. However, the
locations for BEP flow rate with 400 μm particle size at 1050 rpm and distribution of particle volume fraction is highly nonuniform along the
1200 rpm.
width of the casing at any volute angle as seen in Fig. 13b. The particle
concentration is significantly low near the front side as compared to the
back end.
To examine the change in the particle flow behavior at flow rate of
9 L/s (0.75 BEP) along the complete casing width, the contours of the
particle flow field on the casing wall have been plotted as shown in
Fig. 13(c and d). The flow velocity at this flow rate is also higher near
the center region similar to BEP flow rate (see Fig. 13a) however it
reduces near the front side and back end as compared to later. The
region of low concentration of particles from front side is increased and
reached to near centerline. The backside still has higher particle con-
centrations similar to the BEP flow rate (see Fig. 13b). Further to ob-
serve the particle flow behaviour along the casing wall with the re-
duction in the particle size, the contours of the particle flow field on the
casing wall have been plotted for the case III of 400 μm size particles,
BEP flow rate at 1200 rpm pump speed as shown in Fig. 13(e and f). It is
observed that the particle velocity is more near the casing front side
with 400 μm size particle compared to 605 μm size particles at the BEP
flow rate (see Fig. 13a). However, the particle velocity at the centerline
is not much changed as compared to the later. The reduction in particle
size increases the uniformity of the particle concentration along the
Fig. 12. Numerically predicted particle impact velocity and volume fraction at casing width as compared to higher size particles (see Fig. 13b). Further
the centerline of the casing for BEP flow rate with 400 μm particle size at to observe the particle flow field variation with the change in pump
1050 rpm and 1200 rpm.
speed, the contours of the particle flow field have been plotted for the
case IV of 400 μm size particles and BEP flow rate at 1050 rpm pump
the reduction in the pump speed. The magnitude of mass loss at speed as shown in Fig. 13(g and h). Comparing Fig. 13 (e) with Fig. 13
1050 rpm is around 25–65% less as compared to that of 1200 rpm pump (g) for the same particle size shows that the magnitude of particle ve-
speed. The dominant factor that may influence the mass loss of the locity impacting the casing wall is reduced with a reduction in the
specimens with the change in pump speed is analysed using the pre- pump speed. The particle volume fraction distribution over the width
dicted particle flow field. The particle velocity and volume fraction at appears nearly same for both the cases (see Fig. 13 (f) & (h)).
different locations of the casing is shown in Fig. 12. It shows that the
volume fraction at different locations is nearly same at both the speeds, 5.5. Mechanisms of material removal
however, the particle impact velocity is less at lower speed except the
small region just after the volute tongue. The decrease in particle im- To observe the mechanism of material removal of casing wall, the
pact velocity from 1200 rpm to 1050 rpm pump speed reduces the ki- scratches produced on the wear specimens due to particles impacts are
netic energy of the particles by around 10–35%. Thus the reduction in examined through scanning electron microscope (SEM) for all the cases
particle impact velocity with the reduction in pump speed dominantly under the present study. It is observed that the mechanism of material
affects the erosion of the target surface. The rate of increase in mass loss removal is not much changed at different locations with a change in
of the target surface with the increase in particle velocity is dependent operating conditions. For the sake of brevity the photographs of the
on the properties of the target surface and the impact conditions worn out specimens at six different casing locations for the cases I are
[29–31]. Truscott [29] reported that the functional dependence of shown in Fig. 14(a–f). It is observed that the worn out surfaces of the
erosive wear of materials with velocity can be represented by a power- specimens at different casing locations are not similar. At the location 2
law relationship with power index value of 3. which is near the casing tongue, the scratches and indentation marks
are seen on the wear specimens. The mechanism of material removal
5.4. Numerical flow field due to indentation is generally associated with the condition of high
impact angle of the particles. Minemura and Zhong [16] also reported
The casing erosive wear profile is generally reported as average the impact angle of the particles near the casing tongue close to 90°

7
R. Tarodiya and B.K. Gandhi Wear 434–435 (2019) 202972

Fig. 13. Contours of predicted particle impact velocity and volume fraction at various measuring locations of the pump casing.

8
R. Tarodiya and B.K. Gandhi Wear 434–435 (2019) 202972

Fig. 13. (continued)

9
R. Tarodiya and B.K. Gandhi Wear 434–435 (2019) 202972

Fig. 14. SEM micrographs of the eroded surface of the wear samples affixed at different angular locations along the casing centerline for the pump operating at 12 L/s
with 605 μm particle size, 10.1% weighted solid concentration at1200 rpm.

through numerical simulation. At other locations, the removal of ma- 6. Conclusions

terial is observed due to cutting and ploughing. The mechanism of
cutting and ploughing are generally associated with the low impact Based on the present experimental and numerical study, the fol-
angle and low particle velocity [32]. The intensity of surface failure due lowing conclusions can be drawn:
to cutting and ploughing is observed to vary with the location. It may be
due to the variation in the kinetic energy transferred by the particles on (1) The erosion of the pump casing is observed nonuniform along the
the specimen surface with the change in the flow field. At the location length and width of the casing flow path. The numerical results
3, the material loss is less (see Fig. 7) which is also depicted by a few show variation in velocity and particle volume fraction along the
ploughing marks and scratches on the surface in the SEM image of the width and length of the casing whereas the SEM analysis shows
specimen. This is attributed to the low velocity and less number of variation in the impact angle of particles along the flow path of the
particles hitting on the specimen surface at location 3 (see Fig. 13a). At casing.
location 5 the material loss is observed as the maximum which is also (2) It is observed that the flow velocity in the centerline of the casing
visible from the SEM image of the specimen surface showing large cut does not change much with the change in flow rate or particle size,
and deep ploughing marks (Fig. 14c). This is attributed to the increase however, the wear is affected by both parameters. The lower flow
in the particle velocity and a number of particles hitting at the target rate reduces the particle volume fraction which reduces the wear
surface (see Fig. 13a). Further on observing the SEM images of the whereas the bigger particles increase the wear due to higher kinetic
specimens at locations 7, 10, and 13, the mechanism of the mass loss energy, however, the smaller particles have a larger fraction of
appears to be cutting and ploughing of the surface. The observation of particles at the centerline.
the surface failure of the casing at different locations is in line with the (3) The reduction in pump speed reduces the flow velocity and thus the
observations of Gandhi et al. [2]. They also reported that the impact casing erosion is reduced significantly even if the particle volume
angle of the particle at the volute tongue is nearly 90° and the impact fraction does not change.
angle of particles decreases with increase in the volute angle. (4) The mechanism of erosion of the casing is different at different

10
R. Tarodiya and B.K. Gandhi Wear 434–435 (2019) 202972

locations. Around the casing tongue, cutting and deformation wear the basis of measurements in a pot tester, Wear 184 (1995) 169–178.
is dominant, whereas at other locations the cutting and ploughing [12] R. Mishra, S.N. Singh, V. Seshadri, Study of wear characteristics and solid dis-
tribution in constant area and erosion-resistant long-radius pipe bends for the flow
are the dominant mechanisms of casing surface failure. The location of multisized particulate slurries, Wear 217 (1998) 297–306.
of the maximum erosion at the casing centerline is around 80° from [13] M.K. Padhy, R.P. Saini, Study of silt erosion mechanism in Pelton turbine buckets,
the casing tongue in the direction of impeller rotation. Energy 39 (2012) 286–293.
[14] L. Zeng, G.A. Zhang, X.P. Guo, Erosion–corrosion at different locations of X65
carbon steel elbow, Corros. Sci. 85 (2014) 318–330.
Acknowledgment [15] F.M. Want, Centrifugal Slurry Pump Wear–Plant Experience, Hydrotransport 7,
BHRA Fluid Engineering, Sendai, Japan, 1980 Paper H1.
[16] K. Minemura, Y. Zhong, Numerical prediction of erosion wear on pump casing
The authors are thankful to the Ministry of Human Resource and under solid-water two phase flow, Proceedings of 2nd International Conference on
Development, Government of India and the Council of Scientific and Multiphase Flow, 95 1995 1A3-7.
Industrial Research, India for financial support (Grant no.: CSR-1098- [17] K.V. Pagalthivarthi, J.M. Furlan, R.J. Visintainer, Effect of particle size distribution
on erosion wear in centrifugal pump casings, ASME Fluids Engineering Division
MID; Funder ID: 10.13039/501100001412).
Summer Meeting, 2013V01CT20A005.
[18] A.S. Roudnev, R.J. Bourgeois, R.J. Kosmicki, Slurry pump casing wear prediction
Appendix A. Supplementary data using numerical multi-phase flow simulation, ASME Fluids Engineering Division
Summer Meeting, 2009, pp. 515–523.
[19] A.A. Noon, M.H. Kim, Erosion wear on centrifugal pump casing due to slurry flow,
Supplementary data to this article can be found online at https:// Wear (2016) 364–365 103–111.
doi.org/10.1016/j.wear.2019.202972. [20] Y. Xie, J. Jiang, K.Y. Tufa, S. Yick, Wear resistance of materials used for slurry
transport, Wear (2015) 332–333 1104–1110.
[21] R.J.K. Wood, T.F. Jones, J. Ganeshalingam, N.J. Miles, Comparison of predicted and
References experimental erosion estimates in slurry ducts, Wear 256 (2004) 937–947.
[22] C.Y. Wong, C. Solnordal, A. Swallow, J. Wu, Experimental and computational
[1] G. Wilson, Design aspects of centrifugal pumps for abrasive slurries, Cim. Bull. 66 modelling of solid particle erosion in a pipe annular cavity, Wear 303 (2013)
(733) (1973) 91–102. 109–129.
[2] B.K. Gandhi, S.N. Singh, V. Seshadri, Variation of wear along the volute casing of a [23] J. Pei, A. Lui, Q. Zhang, T. Xiong, P. Jianga, W. Wei, Numerical investigation of the
centrifugal slurry pump, JSME Int. J. Ser. B Fluids Therm. Eng. 44 (2) (2001) maximum erosion zone in elbows for liquid-particle flow, Powder Technol. 333
231–237. (2018) 47–59.
[3] M.C. Roco, G.R. Addie, J. Dennis, P. Nair, Modeling Erosion Wear in Centrifugal [24] R.E. Vieira, A. Mansouri, B.S. McLaury, S.A. Shirazi, Experimental and computa-
Slurry Pumps, Hydrotransport 9, BHRA Fluid Engineering, Rome, Italy, 1984 tional study of erosion in elbows due to sand particles in air flow, Powder Technol.
Paper G1. 288 (2016) 339–353.
[4] G.R. Addie, K.V. Pagalthivarthi, Prediction of dredge pump shell wear, Proceedings [25] A. Forder, M. Thew, D. Harrison, A numerical investigation of solid particle erosion
of 12th World Dredging Conference, 1989, pp. 481–504. experienced within oilfield control valves, Wear 216 (1998) 184–193.
[5] W. Widenroth, The Influence of Sand and Gravel on the Characteristics of [26] G.V. Messa, S. Malavasi, The effect of submodels and parameterizations in the si-
Centrifugal Pumps, Some Aspects of Wear in Hydraulic Transportation Installations, mulation of abrasive jet impingement tests, Wear 370–371 (2017) 59–72.
Hydrotransport 1, BHRA Fluid Engineering, Warwick, UK, 1970 Paper E1. [27] J.M. Furlan, M. Garman, J. Kadambi, R.J. Visintainer, K.V. Pagalthivarthi,
[6] W. Wiedenroth, The Evaluation of the Wear Distribution of a Dismountable Impeller Ultrasonic measurements of local particle velocity and concentration within the
in a Model-Dredge Pump, Coal & Slurry Technology Association, Washington DC, casing of a centrifugal pump, Proceedings of the Joint Fluids Engineering
United States, 1993930413. Conference AJKFluids 2015, Seoul, Korea, 2015, pp. 1–14. AJKFluids 2015-31217.
[7] C.I. Walker, P.J. Wells, G.C. Bodkin, The Effect of Flow Rate and Solid Particle Size [28] G.R. Desale, B.K. Gandhi, S.C. Jain, Particle size effects on the slurry erosion of
on the Wear of Centrifugal Slurry Pumps, American Society of Mechanical aluminium alloy (AA 6063), Wear 266 (2009) 1066–1071.
Engineers, New York, United States, 1994940659. [29] G.F. Truscott, A literature survey on abrasive wear in hydraulic machinery, Wear 20
[8] R.J.K. Wood, T.F. Jones, J. Ganeshalingam, N.J. Miles, Comparison of predicted and (1972) 29–50.
experimental erosion estimates in slurry ducts, Wear 256 (2004) 937–947. [30] I. Finnie, D.H. Mcfadden, On the velocity dependence of the erosion of ductile
[9] M.A. Rayan, M. Shawky, Evaluation of wear in a centrifugal slurry pump, Proc Inst metals by solid particles at low angles of incidence, Wear 48 (1978) 181–190.
Mech Eng A J Power Energy 203 (1) (1989) 19–23. [31] F.Y. Lin, H.S. Shao, The effect of impingement angle on slurry erosion, Wear 141
[10] M.K. Padhy, R.P. Saini, Study of silt erosion on performance of a Pelton turbine, (1991) 279–289.
Energy 36 (2011) 141–147. [32] V. Javaheria, D. Portera, V.T. Kuokkalab, Slurry erosion of steel – review of tests,
[11] R. Gupta, S.N. Singh, V. Seshadri, Prediction of uneven wear in a slurry pipeline on mechanisms and materials, Wear (2018) 408–409 248–273.

11