Research Article

Advances in Mechanical Engineering

2018, Vol. 10(1) 1–10
Ó The Author(s) 2018
The mechanism of emitter clogging DOI: 10.1177/1687814017743025
journals.sagepub.com/home/ade
analyzed by CFD–DEM simulation and
PTV experiment

Liming Yu1, Na Li1, Jun Long2, Xiaogang Liu1 and Qiliang Yang1

Abstract
Small but complicated labyrinth channel emitters are easily clogged. In this study, computational fluid dynamics–discrete
element method coupling approach was employed to investigate the mechanism of emitter clogging caused by particles
in size of 65, 100, and 150 mm. Computational fluid dynamics used Navier–Stokes equation to analyze flow characteris-
tics of continuous phase. Discrete element method used Newton’s laws of motion to measure single particle motion and
group distribution of disperse phase. Particle tracking velocimetry was also utilized to follow the trajectories and velocity
of single particle. Our results indicated that the smaller the particle size, the less the total force. Tiny sands were mainly
influenced by drag forces. The amplitude between tooth tips was small. Particles moved basically in the main stream with
fast velocity and short travel distance, thereby having good following performance. It took shorter time to reach micro-
dynamic balance. Meanwhile, the amount of sediments in the labyrinth channel was less. Particles in size of 150 mm were
mainly affected by inertial forces. They can easily enter vortex areas. Sands staying longer than 0.1 s in the labyrinth chan-
nel accounted for 37.9% of total number. Sand groups were mainly distributed at the inlet of labyrinth channel. The more
sands trapped in vortex areas, the easier it was to precipitate and cause emitter clogging.

Keywords
Clogging, emitter, labyrinth channel, particle size, two-phase flow

Date received: 22 February 2017; accepted: 11 October 2017

Handling Editor: Francisco Denia

Introduction the channels. Nowadays, more and more researches

have been focused on whether those sands will precipi-
So far, as the most water-conserving precision irriga- tate and cause emitter clogging or not.
tion technique, drip irrigation can effectively alleviate Nakayama and Bucks4 considered suspended particles
the shortage of water resources. Emitter is the most as a sensitive factor that had brought about the clogging
important part of the irrigating system. Besides, labyr- of emitters. Camp5 thought total suspended solid (TSS)
inth channel emitter is presently one of the most exten-
sively applied drippers.1 However, the flow passage of
the labyrinth channel emitter is narrow, resulting in 1
Faculty of Modern Agricultural Engineering, Kunming University of
easily clogging by solid particles, organics, and Science and Technology, Kunming, China
microbes.2 For example, the average sediment concen- 2
School of Hydraulic Engineering, Changsha University of Science &
tration of the main stream of the Yellow River reaches Technology, Changsha, China
up to 35 kg/m3 over the years, which brings great risks
Corresponding author:
to the currently running irrigating systems in that area.3 Liming Yu, Faculty of Modern Agricultural Engineering, Kunming
Although sediment content is considerably reduced University of Science and Technology, Kunming 650500, Yunnan, China.
using filtration apparatus, sand granules can still enter Email: [email protected]

Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License
(http://www.creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without
further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/
open-access-at-sage).
2 Advances in Mechanical Engineering

Figure 1. CAD model of the emitter with geometrical parameters: (a) structure of tooth shape labyrinth channel, (b) dimensions
of tooth shape labyrinth channel, and (c) hexahedron mesh.

was closely related to emitter blocking. Many studies have Materials and methods
been done on the movement of solid particles in the flow
channels including computational fluid dynamics (CFD)
Physical model and boundary conditions
numerical simulation of the two-phase flow and particle The structure of tooth shape labyrinth channel was
tracking velocimetry (PTV) particle movement measure- indicated in Figure 1(a). The flow path was closely
ment. CFD modeling has been a major interest of against the inner wall of drip irrigation pipe. Water in
researchers in engineering field.6,7 Li et al.8 investigated the drip line entered channel through filter grid, then
water and sediment two-phase flow in the channel of converged at the exit after energy dissipation and finally
emitter with CFD. Our research optimized the structure effused from the drip line after punching in the corre-
of labyrinth channel and significantly reduced emitter sponding positions of pipe. There are a variety of chan-
clogging using external factors.9 Wei et al.10 measured nel dimensions. The dimensions adopted in this study
solid particle movement in the flow channel using PTV are illustrated in Figure 1(b). The flow path had a
and confirmed the results of CFD simulation. But CFD length (S) of 20.62 mm, and depth (D) of 1.41 mm. The
simulation is based on discrete phase mode (DPM) in sectional area (A) was 1.33 mm2. The angle of the tooth
Lagrangian coordinate system. Particles were added after prong was 65°. The tooth space width (L) was 2.2 mm,
flow field calculation without considering the influence of and the tooth height (H) was 1.2 mm. In line with the
particles on the flow field or the collisions among particles actual operational condition of the drip irrigation sys-
or between the particles and the wall. Although grain flow tem, pressure boundary conditions for the inlet and out-
in the emitter is dilute-phase flow, the velocity and direc- let were defined. If the inlet pressure is 50,000 Pa and
tion of water changes frequently. Therefore, either the col- 0 Pa is the outlet pressure, others are wall boundary
lision among particles or that between the particles and conditions. By adopting the finite volume method, the
the wall is not to be neglected. Classical discrete element better simulated results can be got from the small size
method (DEM) model was proposed by Cundall and of flow path and the grid spacing set at 0.1 mm. As
Strack.11 In recent years, CFD-DEM coupling system shown in Figure 1(c), the mesh number of hexahedron
was developed to determine the collision among granules was 39,830 after gridding, little influence on the final
or between the granules and the wall. However, no results was found when using even finer grids.
numerical simulation of particle movement in the emitter
has been reported. In this study, CFD-DEM coupling
approach was adopted to investigate movement and dis- Mathematical model and simulation method
tribution of sediment in the flow and the possible block- The flow in channel can be considered as a viscous
ing positions caused by granules. PTV was employed to incompressible fluid, running under normal temperature.
determine the direction and velocity of movement. Our The effect of gravity and channel wall roughness was
study revealed the mechanism of emitter clogging and regarded. The influence of surface tension was ignored.
provided theoretical foundation for flow channel structure At present, Reynold’s number of emitters (Re) is roughly
optimization. The CFD and DEM software used in this among 78–1284. Nishimura found that laminar turbulent
research was Fluent12.0 and EDEM2.2, respectively. transition occurred when Re was between 200 and 350,
Yu et al. 3

Table 1. Parameters used in the model.17

Phase Parameter Symbol Units Value

Solid Density distribution r kg/m3 2500

Particle diameter di mm 65 mm, 100 mm, 150 mm
Rolling friction coefficient mr 0.005
Sliding friction coefficient ms – 0.3
Poisson’s ratio V – 0.4
Young’s modulus E N/m2 2 3 107
Coefficient of restitution C – 0.545
Particle velocity at inlet – m/s 1.02
Water Density r kg/m3 998.2
Viscosity m kg/m/s 0.001
Velocity at inlet – m/s 1.02
Pressure at inlet – Pa 50,000
Pressure at outlet – Pa 0.000

which was far less than the critical Reynold’s number of time increment size is 0.001 s. The total record is 1000
the conventional straight path.12,13 Other similar experi- times. The discrete approach was employed to simulate
mental results were also reported at a number of times.14 sand movement, collisions among granules and between
There are two reasons for this: first, because of the small granules and the wall, the effect of sand movement to
cross-sectional area of the flow passage which can be eas- the surrounding continuous phase, and energy and
ily affected by the roughness of the wall and so on; sec- momentum exchange. Collisions among sands and
ond, because of the labyrinth’s complicated flow channel between sands and the wall did not lead to significant
boundaries, the structures are bending and changeable plastic deformation, thereby attributing to hard particle
which is more likely to turn laminar flow to turbulent contact, which is a wet grain contact model. ‘‘Hertz–
flow as shown in equation (1)15 Mindlin (no slip) built-in’’ was utilized in this study.
The overriding consideration in the process of calcu-
rVR lation was viscous drag force and gravity. Other addi-
Re = ð1Þ
m tional forces such as pressure gradient force, virtual
mass force, and Saffman force had smaller order of
where Re is Reynold’s number, r (kg/m3) is the density magnitude compared with the former; thus, they were
of fluids, V (m/s) is the average velocity, R (m) is the not considered.18 The referring equations19 can be seen
hydraulic radius of flow, and m (kg/(m s)) is the viscos- in Table 2.
ity coefficient of fluids. The two-way coupling process of CFD and DEM is
The working pressure of emitter was 50,000 Pa, the as follows: continuous phase was solved by CFD to
rated flow was 4.85 L/h, and Reynold’s number Re was acquire fluid drag force in sediments, which was trans-
286. The unsteady standard k 2 e turbulence model was formed from flow field information through drag force
adopted for the simulation of continuous phase. model. The stress state of sediments was measured by
The continuity equation is16 DEM to obtain new information such as position and
∂ velocity of sands and counter reflect to the flow field.
(ruj ) = 0 ð2Þ CFD was used to renew the flow and produce the latest
∂xj
stress condition of sediments. The two approaches were
The momentum equation is coupled by a certain model that carried out transmis-
sion of mass, momentum, and energy.
  
∂ ∂p ∂ ∂ui ∂uj
r (ui uj ) =  + m + ð3Þ
∂xi ∂xj ∂xi ∂xj ∂xi PTV flow test platform and experimental sediments
r is the density of fluid; ui and uj are both fluid velocity, PTV experimental platform consisted of continuous
which is the sum of average fluid velocity and turbulent light source, high-speed camera, and VS-M0910 magni-
velocity; and m is dynamic viscosity. fying lens (Figure 2). Movias Pro Viewer 1.63 analysis
Sediments used belonged to dilute phase flow software can calculate the velocity of particles and dis-
with the maximum particle volume fraction of 1%. play direction of movement according to the distance
Lagrangian coupling system was adopted in this study. they move per unit time. The test platform was engraved
Table 1 indicates the specific parameter settings. The in plexiglass using numerical control machines, covered
total time of numerical simulation is 1.0 s, and every by Lucite plate with good light transmission to seal the
4 Advances in Mechanical Engineering

Table 2. The referring equations and specifications.

S/N Name Equations Specifications

1 The normal force (Fn) 4   1=2 3=2 R is the equivalent radius

Fn = E (R ) a a is the normal overlap
3
2 The equivalent elastic modulus E 1 1  v12 1  v22 E1 , v1 and E2 , v2 are elastic modulus
= + and Poisson’s ratio of sand 1 and sand 2
E Er
1 ffiffiffi E2
3 The damping force Fnd 5 pffiffiffiffiffiffiffiffiffiffi vnrel is the normal relative velocity
Fnd =  2 b Sn m vnrel Sn is the normal stiffness
6
b is coefficient
4 The equivalent mass m m1 m2 m1 and m2 are the mass of sand 1 and sand 2
m =
m1 + m2
5 The tangential force among the sands Ft Ft =  Stp
d ffiffiffiffiffiffiffiffi d is the tangential overlap
6 The tangential stiffness St St = 8G R a
7 The equivalent shear modulus G 2  v12 2  v22 G1 and G2 are shear modulus of sand 1 and sand 2
G = +
G1 G2
rffiffiffi v1 and v2 are velocity of sand 1 and sand 2
8 The tangential damping force 5 pffiffiffiffiffiffiffiffiffi vtrel is the tangential relative velocity
among sand particles Ft Ft =  2 b St m vtrel
6
9 The rolling friction (Ti) Ti =  mr Fn Ri vi mr is the coefficient of rolling friction
Ri is the distance between the center of
mass to the point of contact
vi is the unit angular velocity vector of
object at the contact point

maximum of particle size was adopted such as 150, 100,

and 65 mm.

Results and analyses

Characteristics of flow field of continuous phase
Figure 3(a) shows the pressure distribution chart at half
of the channel depth under 50,000 Pa working pressure.
There was no energy exchange with the outside.
Friction loss and minor loss were provided by pressure
drop. A linear reduction of pressure in the direction of
flow was found. Besides, the pressure drop in each unit
was approximately the same. Figure 3(b) indicates the
velocity distribution chart at half of the channel depth
Figure 2. PTV experimental platform. under 50,000 Pa working pressure. Except the first unit
1: Continuous light; 2: labyrinth channel; 3: amplifying lens; 4: high-speed at the inlet, velocity distribution at other inlets was
camera. basically the same with the speed range of 0.18–
2.91 m/s. The flow velocity reached maximum near the
tooth tip, and the minimum appeared approaching the
flow channel. In this experiment, 100 sieve and 120 sieve center of wall and vortex. Figure (4) shows the velocity
screen meshes were utilized to sift out grains at a diameter streamlines of the fourth unit under 50,000 Pa working
of 150–125 mm. In all, 150 sieve and 180 sieve screen pressure. A represented the main flow. B represented
meshes were employed to sift out sands at a diameter of the vortex area. In Figure 3(b), each vortex area was
100–90 mm. In all, 270 sieve and 220 sieve screen meshes labeled with a number with a total of 17 vortex areas.
were used to sift out grains at a diameter of 65–58 mm. In The flow channel consisted of an inlet, a middle part
total, 10 continuous grains of sand were collected to ana- and an outlet. (1)–(6) indicated inlet. (7)–(11) repre-
lyze their movement. Particles were delivered at a concen- sented the middle part, and (12)–(17) indicated outlet.
tration of 12 g/L and stirred constantly in the water tank It suggested that there was an obvious boundary
to ensure a moderate sediment content and even distribu- between the main stream and vortex area. The main
tion in flow. For CFD numerical simulation, the stream was close to the tooth tip, which changed with
Yu et al. 5

Analysis of representative sediment movement

Sediments entered labyrinth channel after releasing at
random location of inlet. The initial velocity was the
average flow rate which was 1.02 m/s. Figure 5 shows
the direction of movement of representative sediments
in a diameter of 65, 100, and 150 mm, respectively.
Particles in size of 65 mm mainly moved in the main
stream without entering vortex area, thereby indicating
very good following performance. When the diameter
increased to 100 mm, particles moved primarily in the
main stream. However, the motion amplitude in the
Y-axis increased with four times entering the 16th vor-
tex area. Compared with 65 mm sands, their paths are
Figure 3. The distribution chart of pressure and velocity at half
of the channel depth under 50,000 Pa working pressure: in disorder. For particles in size of 150 mm, the moving
(a) pressure distribution chart and (b) velocity distribution chart. trajectory was more disturbed with larger amplitude in
the Y-axis and multiple times of departure from the
main stream entering several vortex areas. The reasons
are that the drag force on sediments is proportional to
the square of the particle size. But the inertial force is
proportionate to the cube of the diameter of sands. The
force condition of different sediments varies in different
sizes. For particles in the size of 65 mm, the drag force
was far beyond the inertial force. Hence, sands moved
only in the main stream. With the increase in diameters,
the drag force and inertial force tended to equilibrium.
For sediments in the size of 100 mm, the drag force was
close to the inertial force, which enhanced the probabil-
ity of sands into the vortex. When the diameter went
up to 150 mm, the inertial force became the major force.
Particles tended to leave the main stream and enter vor-
Figure 4. The velocity streamlines of the fourth unit under tex areas under the effect of inertial force. Thus, the
50,000 Pa working pressure. trajectories of sediments were disarranged. All sand
movement displayed various degrees of disturbance
due to rapid change of flow directions in the channel,
the position shift of the tooth tip. The instantaneous especially in the vortex area with flow velocity at
velocity was approximately 1.1–2.91 m/s. The vortex approximately 0.18–0.53 m/s. Both velocity and sand
areas were away from the tooth tip and the main movement were slow. Sediments underwent endless cir-
stream, which formed some independent closed areas cular low-speed motion if they could not leave the vor-
with instantaneous velocity around 0.18–0.53 m/s. tex area. With the increase in time and numbers, they

Figure 5. Numerically obtained particle trajectories in the channel: (a) particles with a diameter of 65 mm, (b) 100 mm, and
(c) 150 mm.
6 Advances in Mechanical Engineering

Figure 6. Velocity and force condition of representative sand particles: (a) particles with a diameter of 65 mm, (b) 100 mm, and
(c) 150 mm.

could easily form adhesive aggregates and cause emitter PTV tracing of sediments
clogging. Figures 7(a)–(c) and 8(a)–(c) indicate the trajectories
Figure 6 shows the velocity and total force of the rep-
and velocity of representative sand movement in dia-
resentative sediments in Figure 5. For particles in the
meters of 65, 100, and 150 mm measured by PTV. In all,
size of 65 mm, the force was even with small variations.
10 grains of sand were traced, respectively. Due to lim-
The force was small with the magnitude at 1027–1026.
itations of viewing region and clarity, only the fourth
The corresponding velocity change was small. The range
and fifth channel units, including the main stream of
of velocity was 0.76–2.37 m/s. The average velocity was
the fourth and fifth channel units and the 7th, 8th, 9th,
1.48 m/s. For particles in the size of 100 mm, the varia-
and 10th vortex areas, were measure in the PTV experi-
tion of force was complex with magnitude at 1026. The
ment in this study.
corresponding velocity changed considerably. The range
Figures 7(d)–(f) and 8(d)–(f) indicate the trajectories
of velocity was 0.08–2.05 m/s. The average velocity was
and velocity of representative sand movement in diameters
0.77 m/s. For particles in the size of 150 mm, the varia-
of 65, 100, and 150 mm measured by numerical simulation,
tion of force was complicated and huge. The absolute
they are the same fourth and fifth channel units measured
value of force was also great with 1026–1025 in magni-
by PTV. The movement trajectories of sands, namely the
tude. The frequency of velocity changed drastically. The
representative sands’ trajectories in the fourth and fifth
range of velocity was 0.07–1.80 m/s. The average velo-
channel units, are shown in Figure 5(a)–(c). The velocity
city was the lowest, which was 0.55 m/s.
Yu et al. 7

Figure 7. PTV traced and numerical obtained of particle trajectories: (a) and (d) particles with a diameter of 65 mm,
(b) and (e) 100 mm, and (c) and (f) 150 mm.

Figure 8. PTV traced and numerical obtained of particle velocity: (a) and (d) particles with a diameter of 65 mm,
(b) and (e) 100 mm, and (c) and (f) 150 mm.

of the representative sands is what presented in the detected. The amplitude in the Y-axis was small. The
fourth and fifth channel units of Figure 6(a)–(c). The following performance of sand was great. A representa-
motion laws of sands measured by PTV and numerical tive sand trajectories 7(a), exactly similar 7(d). The
simulation are almost the same. With the increase in average velocity of sands from the PTV was 1.32 m/s,
sand sizes, sediments turned to be more and more dis- the average velocity of sands from the numerical simu-
turbed in the channel. lation was 1.82 m/s, and almost all velocities were larger
For particles of 65 mm in size, 10 sand trajectories than 1.0 m/s. Therefore, the conclusion is that the sand
were approximately the same. No circular motion was movement mostly stays in the main flow area and
8 Advances in Mechanical Engineering

Table 3. Characteristics of sand groups in flow channel at t = 1 s.

Characteristics of sand groups Diameter (mm)

150 mm 100 mm 65 mm

Input velocity (per 0.01 s) 76 257 362

Time to reach input and output micro-balance (s) 0.50 0.05 0.03
No. of sands in channel after micro-balance 1155 671 632
Variation of sand No. in channel after micro-balance (%) 0.6 1.1 2.1
No. of sands staying . 0.3 s in channel (%) 0.6 0 0
No. of sands staying . 0.2 s in channel (%) 5.1 0 0
No. of sands staying . 0.1 s in channel (%) 32.2 0 0
No. of sands staying in channel longer than those not involved in circulation (%) 33.2 4.9 0.2

passes through the fourth and the fifth units at higher respectively, as shown in Table 3). However, the velo-
velocity (see Figure 8(a) and (d)). city of particles in the size of 150 mm was low. Hence,
For particles of 100 mm in size, 1 of 10 grains under- the input speed was slow. It took a long time to attain
went one time of circular movement. The amplitude in micro-dynamic balance (0.50 s as shown in Table 3).
the Y-axis increased. A representative sand trajectories Therefore, the time points in Figure 9(a)–(c) were dif-
is shown in Figure 7(b). The same situation also ferent. For particles in size of 65 mm, all sand clusters
occurred in numerical simulation and it is shown in moved in the main stream without entering any vortex
Figure 7(e). The average velocity measured by PTV area. The trajectories were almost the same. In the
was 0.99 m/s, whereas the numerical simulation mea- effect of the drag force, particles got over the inertial
sured number was 1.14 m/s. More than half of the time, force and exhibited very good following performance.
the movement velocity was between 0.5 and 1.0 m/s For particles in size of 100 mm, most of sand clusters
which was apparently lower than the velocity of the moved in the main stream with lesser grains getting
65 mm sand, which explained that although the sand into the vortex area. The trajectories were similar. The
movement was in the main flow area, the larger the drag force and inertial force were close, but the drag
amplitude of vibration, the more declining of the velo- force was dominant. Sediments had good follow-up
city (see Figure 8(b) and (e)). ability. For particles in size of 150 mm, the vast major-
For particles of 150 mm in size, 9 of 10 grains under- ity of sands got into vortex areas, all grains had distinct
went circular movement, among which one grain had trajectories, and the inertial force was dominant, parti-
endless circular motion within limited tracing time. A cles showed bad following performance.
representative sand trajectories 7(c). In numerical simu- After sand clusters reached dynamic equilibrium, the
lation, the results showed that sands entered into vortex number of sand particles in the size of 65, 100, and
area to have a circulation movement which can be seen 150 mm in the channel was 632, 671, and 1155, respec-
in Figure 7(f). The average velocity of sands measured tively. A certain fluctuation may happen. As indicated
by PTV was 0.68 m/s, while the average velocity from in Table 2, more particles occurred in the channel with
the numerical simulation was 0.57 m/s. For most of the bigger diameter and longer time. Part of sediments had
time, the velocity of the sands was lower than 0.5 m/s comparatively longer run time, for example, for parti-
and that is apparently lower than the velocity of 65 mm cles in size of 150 mm, at t = 1.0 s, 0.6% of sediments
and 100 mm. Sands moved in vortex area for most of moved longer than 0.3 s in the channel. 5.1% of sedi-
the time with low velocity and they even could not ments moved longer than 0.2 s. 32.2% of sediments
escape from vortex area (see in Figure 8(c) and (f)). moved longer than 0.1 s. 33.2% of sediments moved
longer than 0.044 s. The reason of this phenomenon is
due to participation of particles in the circulation. On
Analysis of movement of sand clusters the contrary, it took only 0.044 s for particles not
Figure 9 shows the distribution of sand clusters in the involved in the circulation to pass through the same
size of 65, 100, and 150 mm at different time. The sedi- channel, which accounted for merely 0.84% of total
ment volume was calculated at 1%. The time was 1 s. sand clusters (Table 4). Sediments were difficult to
The total number of grains detected in 1 s was 36,215, escape from vortex areas in the effect of inertial forces.
25,658, and 7592, respectively. In combination with This is the major cause of emitter clogging.
Figures 6–8, they suggested that particles in the size of The characteristics of sand clusters running through
65 mm and 100 mm moved fast. The input speed was the whole channel in 1 s are indicated in Table 4. The
also fast. It reached micro-dynamic balance inside the circulating times increased with enlarging particle sizes.
flow channel in a short time (0.03 s and 0.05 s, The corresponding circulating times for particles in size
Yu et al. 9

Table 4. Characteristics of sand groups through flow channel

in 1 s.

Characteristics of sand Diameter (mm)

groups
150 mm 100 mm 65 mm

No. of sands through 6437 24,987 36,215

channel
No. of sands through 54 24,059 35,581
channel (not entering
circulation)
Average travel distance 28.67 25.93 23.23
(mm) of sands (not
entering circulation)
Average time (s) of sands 0.044 0.024 0.019
(not entering circulation)
Average velocity (m/s) of 0.65 1.08 1.36
sands (not entering
circulation)
Total times entering 71,047 1124 76
vortex areas:
(1) 0 0 0
(2) 1788 4 0
(3) 2723 62 6
(4) 5123 169 12
(5) 5295 38 3
(6) 4757 86 7
(7) 5105 40 4
(8) 4644 91 8
(9) 4898 44 0
(10) 4714 90 5
(11) 4970 41 3
(12) 5009 128 10
(13) 4720 37 2
(14) 4769 111 8
(15) 4923 33 1
Figure 9. Distribution of sand groups at different moments: (a)
(16) 4660 105 2
particles with a diameter of 65 mm, (b) particles with a diameter (17) 4949 45 0
of 100 mm, and (c) particles with a diameter of 150 mm.

of 65, 100, and 150 mm were 76, 1124, and 71,047,

Conclusion
respectively. The circulating times in different vortex
areas were arranged in descending order as outlet, mid- In this study, the forces, trajectories, velocity, and dis-
dle, and outlet. The velocity reduced with rising num- tribution of sand clusters of particles in size of 65, 100,
bers of circulation. Under the influence of surface and 150 mm in the labyrinth channel were measured by
charges, particles tended to accumulate and precipitate CFD-DEM coupling approach. The results were con-
to cause the clogging of emitters. Our results were simi- firmed by the trajectories and velocity of single grain of
lar to those in other reference.20 The distribution of cir- sand through PTV. The conclusions are as follows:
culating times not only had a certain regularity, which
occurred many times at the inlet, but it also showed 1. Tiny sands had small amplitude between tooth
some randomness. Particles in size of 150 mm per- tips. They moved basically in the main stream.
formed 5295 times of circular motion in vortex area 5 Their velocity was fast. The travel distance and
in 1.0 s. Grains of sands collided many times. The colli- time were short. Particles were mainly influ-
sion among particles consumed immense kinetic energy, enced by drag forces. The total force was small,
making them unable to escape from the inertial force in thereby having better following performance.
the vortex areas. The probability of precipitation enor- With the increase in particle sizes, the amplitude
mously increased with prolonged running time and between tooth tips enlarged. The velocity
occurrence of flocculent materials. It consequently led decreased but the frequency of change was
to emitter clogging. raised. The travel distance and time rose.
10 Advances in Mechanical Engineering

Sediments easily entered vortex areas from the 5. Camp CR. Subsurface drip irrigation: a review. T ASAE
main stream. The inertial force gradually built 1998; 41: 1353–1367.
up. So the total force increased. As a result, the 6. Gücüyen E, Erdem RT and Gökkusx Ü. FSI analysis of
following performance turned to be bad. submarine outfall. Brodogradnja 2016; 67: 67–80.
2. The smaller the particle sizes, the shorter time it 7. Gücüyen E and Erdem RT. Corrosion effects on struc-
tural behaviour of jacket type offshore structures. Grad--ev
took to reach micro-dynamic balance. Thus, the
2014; 66: 981–986.
amount of sediments in the labyrinth channel was
8. Li YK, Yang PL, Xu T, et al. CFD and digital particle
less. For particles in size of 150 mm, sands staying tracking to assess flow characteristics in the labyrinth
longer than 0.1 s in the channel accounted for flow path of a drip irrigation emitter. Irrigation Sci 2008;
37.9% of total sediments. Particles not involved 26: 427–438.
in circulation merely accounted for 0.84% of total 9. Zhang J, Zhao WH, Wei ZY, et al. Numerical and
sands. Therefore, it is necessary to have filters or experimental study on hydraulic performance of emitters
even to have multi-stage filtration system to with arc labyrinth channels. Comput Electron Agr 2007;
reduce the entering of larger particles which are 56: 120–129.
more than 100 mm into the labyrinth channel. 10. Wei ZY, Cao M, Xia L, et al. Flow behaviour analysis
3. Sand groups were mainly located at the inlet of and experimental investigation for emitter micro-chan-
the labyrinth channel. Meanwhile, they showed nels. Chin J Mech Eng 2012; 25: 729–773.
certain randomness. A large number of sands 11. Cundall PA and Strack ODL. A discrete numerical model
for granular assemblies. Geotechnique 1979; 29: 147–165.
were trapped in vortex areas. Collisions among
12. Nishimura T, Bain YN and Matsumoto Y. Fluid flow
particles used kinetic energy, making sands
and mass transfer characteristics in a sinusoidal wavy-
unable to escape from vortex areas, which even- walled tube at moderate Reynolds numbers for steady
tually led to precipitation and emitter clogging. flow. Heat Mass Transfer 2003; 39: 239–248.
13. Tatsuo N, Shinichiro M, Shingho A, et al. Flow obser-
Declaration of conflicting interests vations and mass transfer characteristics in symmetrical
wavy-walled channels at moderate Reynolds numbers
The author(s) declared no potential conflicts of interest with for steady flow. Int J Heat Mass Tran 1990; 33:
respect to the research, authorship, and/or publication of this 843–845.
article. 14. Li YK, Yang PL and Ren SM. Experimental study on
flow characteristics in labyrinth path emitters. J Hydraul
Funding Eng 2005; 36: 886–890.
15. Wei QS, Shi YS, Dong WC, et al. Study on hydraulic
The author(s) disclosed receipt of the following financial sup-
performance of drip emitters by computational fluid
port for the research, authorship, and/or publication of this
dynamics. Agr Water Manage 2006; 84: 130–136.
article: This work was supported by the National Natural
16. Zhang J, Zhao WH and Tang YP. Numerical investi-
Science Foundation of China (NSFC) (Nos 51379024 and
gation of the clogging mechanism in labyrinth channel
51769009).
of the emitter. Int J Numer Meth Eng 2007; 70:
1598–1612.
References 17. Chu KW, Wang B, Yu AB, et al. CFD-DEM modelling
of multiphase flow in dense medium cyclones. Powder
1. Wei QS, Shi YS, Dong WC, et al. Advanced methods to
develop drip emitters with new channel types. Am Soc Technol 2009; 193: 235–247.
Agric Biol Eng 2006; 22: 243–249. 18. Tang XL, Yu X and Ren SC. Solid-liquid two phase liquid
2. Chandra A and Madramootoo JM. Advances and chal- hydrodynamics and its application in hydraulic machinery.
lenges with micro-irrigation. Irrig Drain 2013; 62: 255–261. Zhenghou: The Yellow River Water Power Press, 2006.
3. Jiang RQ and Cao WH. Research for sediment taken 19. Asakura K, Asari T and Nakajima I. Simulation of solid-
from the Yellow River. Beijing: China Water & Power liquid flows in a vertical pipe by a collision model. Powder
Press, 2012. Technol 1997; 94: 201–206.
4. Nakayama FS and Bucks DA. Water quality in drip/ 20. Liu L and Niu W. The impact of sediment particle size to
trickle irrigation: a review. Irrigation Sci 1991; 12: the labyrinth emitter clogging. Trans Chin Soc Agric Eng
187–192. 2012; 28: 87–93 (in Chinese with English abstract).