Professional Documents
Culture Documents
Erosion Failure of Horizontal Pipe Reducing Wall in Power-Law Fluid Containing Particles Via CFD-DEM Coupling Method
Erosion Failure of Horizontal Pipe Reducing Wall in Power-Law Fluid Containing Particles Via CFD-DEM Coupling Method
(2017) 17:1067–1080
DOI 10.1007/s11668-017-0340-1
TECHNICAL ARTICLE—PEER-REVIEWED
Abstract A CFD–DEM-based two-phase flow model and experiment, and it shows a good agreement. The calculated
a test-based erosion model are used to obtain the specific results show that the erosion rate of the reducing wall is
erosion on the reducing wall of sudden contraction sec- mainly determined by the flow velocity, and the erosion
tion. The dimensionless filtered governing equations are area is affected by liquid viscosity. The serious erosion
adopted for incompressible power-law fluid flow, and the region is located in the inner edge of the sample lower part,
Hertz–Mindlin (no-slip) model for particle–particle and and this region expends to the outer circumference with the
particle–wall contact. The annular reducing wall is divided increasing flow velocity and the reducing liquid viscosity.
into two erosion areas in radial direction based on erosion The increase in flow velocity expands the flow region
form and divided into four parts in the circumferential where the particle can impact the wall and thus increases
direction. The calculated result is verified with a full-scale the particle impact numbers.
N. Zhang
Department of Petroleum Engineering, Xi’an Shiyou University,
Xi’an 710065, People’s Republic of China
Z. Li Y. Dou Y. Cao
Department of Mechanical Engineering, Xi’an Shiyou
University, Xi’an 710065, People’s Republic of China
123
1068 J Fail. Anal. and Preven. (2017) 17:1067–1080
Graphical Abstract
123
J Fail. Anal. and Preven. (2017) 17:1067–1080 1069
123
1070 J Fail. Anal. and Preven. (2017) 17:1067–1080
Badr et al. [7] have proven inlet velocity and particle size
exerted significant effects on erosion rate of reducing wall.
However, as for power-law fluid containing particles,
the time-varying fluid flow inevitably causes various par-
ticle distributions in the pipe and erosion patterns on metal
surface, which requires measurements to be taken in a very
short time due to the time-varying liquid viscosity. But
such a short test time reduces the surface erosion depth,
thus increasing the measuring errors.
In the present work, the CFD–DEM (computational fluid
dynamics and discrete element method) coupling method,
which is based on the test coefficients, was used to predict
the erosion rate on the reducing wall at different flow
velocities and liquid viscosities, and the calculated data
were verified by comparing with the experimental result.
Finally, the erosion-affected flow region and the erosion
pattern of reducing wall have been discussed in detail.
Method
Governing Equations
8.94 9 108 0.35 1.57 5.9 9 105 7.2 9 105 0.75 0.21 0.83 1.2 70
123
J Fail. Anal. and Preven. (2017) 17:1067–1080 1071
pseudoplastic, Newtonian, and dilatant fluids. Shear stress for incompressible non-Newtonian flow. These equations
s is given by s ¼ Kðou=oyÞn . When the exponent n is less are as follows [9]:
than one, the power law predicts that effective viscosity o
ui
will decrease with indefinitely increasing shear rate [8]. ¼0
oxi
The fully developed pipe flow of power-law fluids in a
o
ui oui uj op 1 o n1 o ui o
uj o
sij
sudden contraction can be treated as governing equations þ þ ¼ c þ
ot oxj oxi Re oxj d oxj oxi oxj
ðEq 1Þ
Table 2 CFD–DEM simulation parameters and geometric
dimensions of domain The apparent viscosity g of fluid is given by g ¼ Kcn1 ,
DEM parameter Value and the Reynolds number is defined as Re ¼ ql u2n n
x D =K.
The dimensionless shear rate is defined as: cd ¼ cD=ux .
Liquid phase The fluid is assumed to enter the upstream region with fully
Liquid density (kg/m3) 1020 developed velocity.
Liquid viscosity (mPa s) 200–300
Solid phase Discrete Element Model
Diameter (mm) 0.65
Mass (mg) 0.26 The discrete element method (DEM) is based on the inte-
Sphericity 0.85 gration of Newton’s second law to obtain the evolution in
Particle density (kg/m3) 1850 time of the (translational and rotational) velocity and
Number of particles in each calculation 1000 position of the particles. The translation and rotational
Geometry motions of each and individual particle can be expressed by
Upstream pipe length in the axial direction (mm) 2000 [10]:
Downstream pipe length in the axial direction (mm) 1000
dvi X N
Circumferential phase angle for each erosion area (°) 60 Ft;i ¼ mi ¼ ðFt;ij þ Fn;ij Þ þ Fd;i þ mi g ðEq 2Þ
The critical erosion area width in radial direction (mm) 4
dt j¼1
Discretization length for critical erosion area Dr (mm) 0.02
and
123
1072 J Fail. Anal. and Preven. (2017) 17:1067–1080
Fig. 4 Schematic diagram and actual picture of experimental setup: 1 electrical control cabinet; 2 liquid storage tank containing electric heater
(2 m3); 3 screw pump; 4, 6 cutoff valve; 5 flow meter; 7 computer; 8 high-speed camera; 9 test section; 10 pressure transducer
123
J Fail. Anal. and Preven. (2017) 17:1067–1080 1073
The overlap Dx between two particles is given by: Fs = 1.0 for sharp, 0.53 for semi-rounded, or 0.2 for
fully rounded sand particles [17], and f(a) is the function
Dx ¼ ri þ rj xj xi ðEq 5Þ
of the impact angle that is given by the following
where xi and xj are the positions of particle i and j, equations:
respectively. The normal unit vector is defined as:
xj xi aa2 þ ba for a h
nij ¼ ðEq 6Þ f ðaÞ ¼
xj xi x cos2 a sinðwaÞ þ y sin2 a þ z for a [ h
ðEq 14Þ
The relative velocity of particle contract point is
expressed by: According to the previous studies using jet flow system
at the Xi’an Shiyou University [18], the erosion rate of
Dv ¼ vi vj þ Li xi þ Lj xj nij ðEq 7Þ super 13Cr stainless steel decreases with increasing impact
The normal component of the relative velocity is: angle when the angle is less than 70°, and the new erosion
sub-model parameters used in Eq 13 are listed in Table 1.
Dvn ¼ Dvnij nij ðEq 8Þ
~ 1
F d ¼ Cd ql Ap ju vjðu vÞ ðEq 10Þ
2
where the following equation for the drag coefficient is
obtained:
( .
24 1 þ 0:15Re0:687 Rep Rep \1000
Cd ¼ p
0:44 Rep [ 1000
ðEq 11Þ
where the particle Reynolds number Rep is defined as:
ql ju vjdp
Rep ¼ : ðEq 12Þ
Kcn1
Erosion Model
123
1074 J Fail. Anal. and Preven. (2017) 17:1067–1080
123
J Fail. Anal. and Preven. (2017) 17:1067–1080 1075
Experimental Description
123
1076 J Fail. Anal. and Preven. (2017) 17:1067–1080
Erosion Distribution
Fig. 9 Erosion rates and distributions of reducing wall obtained by numerical calculation at different flow velocities and liquid viscosities, each
column: the inlet flow velocities are 1.5, 2.5, and 3.5 m/s, respectively, each row: the liquid viscosities are 300, 250, 200 mPa s, respectively
123
J Fail. Anal. and Preven. (2017) 17:1067–1080 1077
impact the outer circumference, which causes some dis- accounts for 25 or 35% of the width of TEA, and all of
continuous erosion areas on the surface. their widths increase slowly with the increasing flow
When the inlet flow velocity is increased, the change in velocity (Fig. 11).
erosion rate is more significant than that in erosion area, as
shown in each column of Fig. 9. Most particle impact
positions focus on the inner edge of the reducing wall and Discussion
expend gradually to the outer circumference. Meanwhile,
the difference in erosion rate among different erosion areas Particle Dynamical States
is also reduced. Erosion area 4 has the maximum erosion
depth when the flow velocity is less than 2.5 m/s (Fig. 10); According to the particle impact properties, as shown in
however, most serious erosion area becomes erosion area 3 Fig. 12, the flow region in sudden contraction section can
when the flow velocity is greater than 2.5 m/s. be divided as follows:
When the liquid viscosity is decreased, as shown in each
Erosion-unaffected flow region In the flow region 1,
row of Fig. 9, the erosion area extends to the outermost
most of the particles flow into the downstream without
circumference, especially in erosion area 3, which is
impacting the sample surface. These particles will rarely
caused by the impingement between deposited particles the
erode the sample surface.
lower surface of the reducing wall. The width of FEA
Erosion-affected flow region This region includes flow
regions 2 and 3. In flow region 2, some particles directly
impact the inner edge of sample and are carried by the
incoming flow into the flow region 1. And other particles
will move into the viscous sub-layer region (flow region
3) due to particle–particle collisions and subsequently
impact the sample surface.
If a particle is located in flow region 2, as shown in
Fig. 13, the particle movement will exhibit three states
including horizontal, curvilinear, and radial movements.
The increase in inlet flow velocity directly causes the growth
of liquid radial velocity, thereby increasing the erosion-af-
fected flow region and causing more particles impact the
reducing wall. Figure 14 shows that the number of impact
particles at the flow velocity of 3.5 m/s is almost twice at the
flow velocity of 1.5 m/s (c = 300 mPa s), and this change is
Fig. 11 Erosion widths of different erosion areas vs. flow velocity more significant at the viscosity of 200 mPa s.
123
1078 J Fail. Anal. and Preven. (2017) 17:1067–1080
If a particle is located in flow region 3, the fluid will Additionally, Fig. 16 shows that the impact angles at
either calm the particle or drive it to new flow region. In outer circumference are several times larger than that the
this region, as shown in Fig. 15, there is a vortex that varies inner edge of sample, and the impact angles in erosion
with the upstream flow velocity. When the drag force area 3 are generally smaller than those in other areas. As
acting on a particle is far greater than its own gravity, the shown in our previous experimental results [18], the ero-
particle will be carried to the flow region 2 and impact the sion rate decreases with increasing impact angle when the
sample or move to the downstream pipe. The drag force is angle is less than 70° for super 13Cr stainless steel.
0.03, 0.06, and 0.07 N, respectively, at the inlet velocity of Therefore, the low angle impact is the main reason for
1.5, 2.5, and 3.5 m/s, which are far greater than the gravity causing a serious erosion at the inner edge of sample
(2.6 9 106 N). Therefore, the vortex in flow region 3 lower part.
carries the settling particles into the main flow region and
makes particle impact on the wall again.
Fig. 13 Influence of flow velocity on erosion-affected flow region in the sudden contraction section
123
J Fail. Anal. and Preven. (2017) 17:1067–1080 1079
Flow velocity
(m/s) u=1.5 m/s u=2.5m/s u=3.5m/s
Fig. 15 Flow streamlines in different inlet flow velocities in sudden contraction section
1. The simulation results show a good agreement with the 1. C. Rivard, D. Lavoie, R. Lefebvre et al., An overview of Cana-
experiment in the area (fatigue erosion area) where the dian shale gas production and environmental concerns. Int. J.
Coal Geol. 126(5), 64–76 (2013)
particle impact repeatedly, but the accuracy of the 2. A.K. Mahmood, A.A. Khadom, Erosion–corrosion of low-carbon
results in few and scattered erosion areas (transition steel in the absence and presence of slurry in saline water: kinetic
erosion area) is relatively poor. Although the width of and mathematical views. J. Fail. Anal. Prev. 16, 1071–1081
transition erosion area is almost three times the width (2016)
3. R. Malka, S. Nešić, D.A. Gulino, Erosion–corrosion and syner-
of fatigue erosion area, the depth of transition erosion gistic effects in disturbed liquid-particle flow. Wear 262(7–8),
area is less than one-tenth of maximum erosion depth. 791–799 (2007)
123
1080 J Fail. Anal. and Preven. (2017) 17:1067–1080
4. C.Y. Wong, C. Solnordal, A. Swallow et al., Experimental and 12. P.A. Cundall, O.D.L. Strack, A discrete numerical model for
computational modelling of solid particle erosion in a pipe granular assemblies. Géotechnique 29(18), 47–65 (1979)
annular cavity. Wear 303(1–2), 109–129 (2013) 13. X. Chen, B.S. McLaury, S.A. Shirazi, Application and experi-
5. Z. Lin, H. Xu, Y. Wang et al., Experimental study of particle mental validation of a computational fluid dynamics (CFD)-based
erosion in a cavity with a height difference between its walls. erosion prediction model in elbows and plugged tees. Comput.
Powder Technol. 286, 378–384 (2015) Fluids 33(10), 1251–1272 (2004)
6. M.A. Habib, H.M. Badr, R.B. Mansour et al., Erosion rate cor- 14. H.C. Meng, K.C. Ludema, Wear models and predictive equa-
relations of a pipe protruded in an abrupt pipe contraction. Int. J. tions: their form and content. Wear 181(95), 181–183 (1995)
Impact Eng. 34(8), 1350–1369 (2007) 15. B. McLaury, Predicting solid particle erosion resulting from
7. H.M. Badr, M.A. Habib, R.B. Mansour et al., Numerical inves- turbulent fluctuations in oilfield geometries, Master’s Thesis, The
tigation of erosion threshold velocity in a pipe with sudden University of Tulsa, 1996
contraction. Comput. Fluids 34(6), 721–742 (2005) 16. K. Ahlert, Effects of particle impingement angle and surface
8. S. Dhinakaran, M.S.N. Oliveira, F.T. Pinho et al., Steady flow of wetting on solid particle erosion of AISI 1018 steel, Master’s
power-law fluids in a 1:3 planar sudden expansion. J. Non Thesis, The University of Tulsa, 1994
Newton. Fluid 198(8), 48–58 (2013) 17. I. Finnie, An experimental study of erosion. Wear 3(2), 76 (1960)
9. P.S. Gnambode, P. Orlandi, M. Ould-Rouiss et al., Large-Eddy 18. Y.A. Zhao, W.B. Cai, L. Cui, et al, Erosion of premium con-
simulation of turbulent pipe flow of power-law fluids. Int. J. Heat nection cross-over joint in solid-liquid flow, in Proceedings of the
Fluid Flow 54, 196–210 (2015) International Conference on Engineering Technology and
10. N. Iqbal, C. Rauh, Coupling of discrete element model (DEM) Application, May 29–30, Xia Men, China, EDP Sciences, 2015
with computational fluid mechanics (CFD): a validation study. 19. C. Huang, S. Chiovelli, P. Minev et al., A comprehensive phe-
Appl. Math. Comput. 277, 154–163 (2016) nomenological model for erosion of materials in jet flow. Powder
11. P. Gupta, J. Sun, Y.J. Ooi, CFD-DEM simulation of a dense Technol. 187(3), 237–279 (2008)
fluidized bed: wall boundary and particle size effects. Powder
Technol. 293, 37–47 (2016)
123