Zhu 2018

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Original Article

Proc IMechE Part C:


J Mechanical Engineering Science
The effects of the lower outlet on the flow 0(0) 1–9
! IMechE 2018

field of small gas–liquid cylindrical cyclone Reprints and permissions:


sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0954406218768838
journals.sagepub.com/home/pic

Weibing Zhu, Liang Hu and Xiaobin Zhang

Abstract
In this study, the effects of the lower outlet on the flow field of small gas–liquid cylindrical cyclones are investigated using
Reynold stress turbulence model. Under the same operating conditions, four configurations with different outlet styles
and angles are established. The time-averaged tangential velocity, axial velocity, and root mean square velocity are
compared, respectively. It is shown that many local secondary flow patterns are present in small gas–liquid cylindrical
cyclones, and those flow patterns may cause serious energy losses. The lower outlet mainly influences the gas–liquid
cylindrical cyclones flow field in the central region. The small gas–liquid cylindrical cyclones with single rectangular outlet
provides a steady flow field and a large backflow zone, which are helpful in improving the separation efficiency. According
to the simulations, a single rectangular lower outlet is the optimal one for small gas–liquid cylindrical cyclones.

Keywords
Numerical simulations, small gas–liquid cylindrical cyclone, lower outlet, flow field, aero engines

Date received: 17 September 2017; accepted: 2 March 2018

Introduction
They found that a 27 inlet angle is an optimal angle
Due to the advantages of its simple geometry, for the separator, which is helpful in decreasing the
small space requirement, and low maintenance costs, phenomenon of liquid carry-over (LCO). Therefore,
gas–liquid cylindrical cyclones (GLCC) separators are this optimal angle is widely used for GLCC separators
widely applied in industry.1 For instance, GLCC with different diameters.13,14 Erdal15 studied a GLCC
separators are used as alternative devices to replace separator by LDV experimentally. A strongly
gravity separators in petroleum industry,2 and they nonaxisymmetric flow was obtained, and local LDV
are also used in lubrication systems of aero engines measurements were used to improve the existing
to separate the air–oil mixture collected from the correlations for swirling, axial, and tangential veloci-
lubrication sumps.3 The complex flow patterns ties. Molina et al.16 studied the effects of an annular
taking place in GLCC body significantly influence film extractor on a GLCC. They concluded that the
the separation efficiency, so many scholars investi- performance of the modified GLCC is improved in
gated the GLCC flow field numerically and the new structure. Hreiz et al.14,17,18 performed
experimentally.4–9 numerical and experimental analyses for the swirling
With no moving parts, the geometry of GLCC is hydrodynamics of GLCC separators, and studied the
simple. It is composed of a separator body, a tangen- effects of inlet geometry on flow pattern and perform-
tial inlet, and two outlets located at the upper and ance. The results showed that inlet with a suitable
lower positions.10 Figure 1 shows the structure of a convergence can improve the separation performance.
GLCC separator. The gas–liquid mixture enters tan- Recently, Van Sy19 simulated the GLCC flow field of
gentially through the inlet pipe and then rotates twelve GLCC separators with different inlet angles.
downward in the GLCC body. Because of the differ-
ence of density between the gas and the liquid, gas
then leaves the wall of separator, and is collected by
the reversed flow. The liquid is pulled to the wall, and College of Aerospace and Civil Engineering, Harbin Engineering
leaves through the lower outlet. Arpandi et al.11 tested University, Harbin, China
the GLCC with a diameter of 76.2 mm, and developed
Corresponding author:
a mechanistic model for predicting the distribution of Xiaobin Zhang, College of Aerospace and Civil Engineering, Harbin
the GLCC flow field.11 Kouba et al.12 studied the Engineering University, Harbin 150001, China.
effects of the inlet angle on a GLCC experimentally. Email: zhangxiaobin@hrbeu.edu.cn
2 Proc IMechE Part C: J Mechanical Engineering Science 0(0)

The results showed that the inlet angle significantly Numerical model and simulation method
influences the flow field.
For the GLCC separator used in petroleum industry,
Lower outlet configuration of GLCC separators
most researchers mainly focus on the influence of geom- Figure 1(a) shows a schematic plot of a small GLCC
etry, angle, position and number of inlets, because the separator. The positions of the analyzed plane are
inlet is one of the most relevant parameters influencing performed in section Y1–Y4, as shown in
the GLCC flow pattern and performance, as reported Figure 1(b), and the center of the GLCC bottom is
in many articles.18–22 The height of the vertical pipe is the origin of coordinates. Those sections are located
so big that the lower outlet has only small influences on in the zones of the upper outlet, the inlet, the middle
the main flow field. However, the GLCC separator, and the lower outlet in the GLCC body. According to
used in lubrication systems of aero engines to separate our previous study,26 Table 1 gives the detail of the
air bubbles from oil,3,23 has small geometrical dimen- geometrical dimensions, where a and b are the height
sions due to the inlet flow rate and the restriction of and width of the rectangular inlet, respectively.
available installation space. The lower outlet is very To evaluate the effects of the lower outlets on the
close to the main flow field, and it will influence the flow field in small GLCC separators, the numerical
flow field in GLCC separator directly. Although some investigation was performed on four separators with
studies have investigated the flow field in GLCC separ- different lower outlets, as shown in Figure 2.
ator with different lower outlets,13,14,24,25 all of them The lower outlets of A1 and A2 are rectangular, and
have not considered the effects of the lower outlet on the lower outlets of A3 and A4 are circular. To avoid
the GLCC flow field systematically. other random factors, A1, A2, and A3 have a lower
The present paper aims at numerically studying the outlet with the same area and axial height.
effects of the lower outlet on the single-phase flow field The detailed dimensions of the lower outlets are
of a small GLCC separator. Firstly, four GLCC sep- shown in Table 2, where c and d are the height and
arators with different lower outlets will be constructed. width of the rectangular outlet respectively and D3 is
Secondly, Reynolds stress model (RSM) will be chosen the diameter of the circular outlet.
for all numerical simulations, and the features of the
flow field in small GLCC will be carried out. Finally,
the effects of the lower outlet on the flow field of small
Turbulence model
GLCC separators will be studied in detail. Some rec- A suitable turbulence model is vital for the swirling
ommendations are provided for a structure of optimal flow in GLCC separators, since the complex flow
design of small GLCC in industrial applications. pattern taking place in such a separator.27 Previous

Figure 1. Schematic plot of a small GLCC separator: (a) geo-


metrical dimensions, (b) positions of the analyzed plane (mm). Figure 2. Separators with different lower outlets.

Table 1. Geometrical dimensions of the small GLCC separator.

D1 (mm) L1/D1 L2/D1 L3/D1 L4/D1 L5/D1 D2/D1  ( ) a/D1 b/D1

16 1.56 3.25 0.81 0.63 0.94 0.19 15 0.75 0.19


Zhu et al. 3

studies have used various turbulence models for the Results and discussion
GLCC flow field.15,17,19,24 As having the greatest
potential for predicting the behavior of strongly swir-
Validation of the numerical results
ling flows,19,28,29 RSM model is chosen to investigate The following method is used to simulate a GLCC
the effects of the lower outlet on the small GLCC separator, as the velocity profiles measured by
flow field.30 Erdal.15 Figure 4 compares the simulated results
with the experimental data of Erdal.15 Because the
flow patterns are extremely complex in the GLCC
Boundary conditions separator, we consider the simulated results as
The working fluid is oil in this study with the liquid acceptable.
density r ¼ 963.9 kg/m3 and the dynamic viscosity
 ¼ 0:001467 Pa  s. A constant velocity of 2.778 m/s
The features of the flow field
at the inlet to keep a constant flow rate, and an
absolute pressure (101,325 Pa) at the outlet boundary To determine the features of the flow field in small
surface is used. In addition, a no-slip boundary GLCC separators, the contour plots and the time-
condition is used on the walls. The time step of averaged velocity profiles for A1 are analyzed.
5e–5s is used for the unsteady simulations. The positions of the four sections are shown in
Figure 1(b).
The tangential velocity significantly influences the
Grid performance of GLCC under the operating conditions
All numerical simulations are generated into struc- where gas and liquid exist simultaneously, because the
tured grids with the value of yþ 5 30, as shown in centrifugal force is the most important factor in
Figure 2. We study the grid independence for the GLCC, which is helpful in increasing the separation
simulated separators with different grid numbers. performance. The contour plots of the time-averaged
Three levels of grids, using respectively 120,222, tangential velocity for model A1 are shown in
248,472, 369,698 cells, are used for A1, as shown in Figure 5. It can be seen that two flow patterns with
Figure 3. It can be seen that the separator with opposed directions are present in the measured plane.
120,222 cells is acceptable, as the difference between This is caused by the swirling flow in the GLCC flow
the maximum and minimum grids is less than 5%. field. The tangential velocity decreases towards the
To eliminate the random factors, simulations are car- central region, and the minimum value appears
ried out by meshing the separator with 248,472 cells. around the horizontal axis. Due to the energy loss
in GLCC, the maximum value of tangential velocities
is decayed in the direction of the negative axis.
As shown in Figure 1, the position of Y2 is the nearest
Table 2. Geometrical dimensions of different lower outlets. to the inlet. The reason why there is an obvious
difference of the time-averaged tangential velocity
A1 A2 A3 A4 between Y2 and other positons is less energy losses
c/D1 0.75 0.38 in Y2.
d/D1 0.19 0.19 The axial velocity reflects the flow direction in
D3/D1 0.43 0.43
GLCC flow field, which determines the ratio of the
volume flow from the upper outlet to that of the inlet.

Figure 3. Computational velocities of the three grid levels: (a) Tangential velocity, (b) Axial velocity.
4 Proc IMechE Part C: J Mechanical Engineering Science 0(0)

Figure 4. Comparisons of the simulated results with the experimental results: (a) Tangential velocity, (b) Axial velocity (Re ¼ 9285,
X ¼ 46.7 cm).

significantly influences the performance of GLCC.


Figure 6(c) shows the circulatory flow at the bottom
of GLCC, which affects the flow field near the lower
outlet, and causes vortex breakdown.
Figure 7 presents the velocity distributions of A1
from section Y1 to Y4. Due to the interference of the
upper outlet pipe, the distributions of tangential
and axial velocity are discontinuous in section Y1.
The trend of the axial velocity profiles in section Y1
is different from others, which is related to the effects
of circulatory flow at the top of GLCC, as shown in
Figure 6(a). The maximum tangential velocity in sec-
tion Y2 is larger than the values in other sections.
The phenomenon is related to the effect of the inlet.
The magnitudes of tangential and axial velocities in
different sections are similar except section Y1, which
means the small GLCC separator has a low energy
loss in the flow field. As shown in Figure 7, the
upward and downward flow streams present in
GLCC flow field. In the outer region, the axial
velocity increases with increasing diameter.
However, in the inner region, the axial velocity
decreases with increasing diameter. It is an inverted
V-shaped axial velocity profile for A1 in sections Y2,
Y3, and Y4. The distributions of the axial velocity
Figure 5. Time-averaged tangential velocity contours for profiles at different sections are similar, but the
model A1 (m/s). radial positions of the highest axial velocity in the
inner region are not fixed, which range from 0.2R
The contour plots of the time-averaged axial velocity to 0.2R.
for A1 are shown in Figure 6. The axial velocity indi-
cates that two flow streams exist in the GLCC body: a
downward flow directs to the lower outlet and an
Effects of the lower outlet on the flow field
upward flow directs to the upper outlet. It can be To evaluate the effects of the lower outlet on the
seen that the vortex helical pitch is clearly predicted separator flow field, four models with different lower
by the numerical results. In addition to the flow fea- outlets are carried out under the same operating
tures described above, some other flow patterns are conditions.
present in the GLCC flow field. Figure 6(a) gives the As shown in Figure 8, the magnitudes of tangential
circulatory flow at the top of GLCC. This flow pat- velocity near the wall are almost equal, whereas the
tern will cause serious energy losses in the flow field. tangential velocities in the inner region show obvious
Figure 6(b) shows the short-circuit flow near the differences. It is observed that the difference increases
upper outlet. It indicates that parts of fluid from the with decreasing distance to the lower outlet. This phe-
inlet are collected by the upper outlet directly, which nomenon means that the lower outlet of small GLCC
Zhu et al. 5

Figure 6. Time-averaged axial velocity and local flows for model A1: (a) The circulatory flow at the top, (b) The short-circuit flow,
(c) The circulatory flow at the bottom (m/s).

Figure 7. Radial profiles of the time-averaged tangential and axial velocity for model A1 at different sections: (a) Tangential velocity,
(b) Axial velocity (Z ¼ 0).

mainly influences the central region of the flow field, of the profiles is obviously changed, in which the
and has no significant effect on the tangential veloci- forced vortex region is narrowed down and the free
ties near the wall. The tangential velocity distributions vortex region is expanded. The maximum tangential
of A2 are different from that of A1, in which the dis- velocity in A4 is larger than that of other separators.
tribution symmetry of A2 is better than A1. The dis- Similar to the tangential velocities, the magnitudes
tributions of A1 and A3 are similar at each section. of axial velocity near the wall are almost equal, but
The tangential velocity profile of A4 has significant they have obvious differences in the inner region. The
difference from that of other separators. The shape differences increase along the direction of the negative
6 Proc IMechE Part C: J Mechanical Engineering Science 0(0)

Figure 8. Comparison of the radial profiles of the time-averaged tangential velocity at different sections: (a) Y1, (b) Y2, (c) Y3, (d) Y4.

Figure 9. Comparison of the radial profiles of the time-averaged axial velocity at different sections: (a) Y1, (b) Y2, (c) Y3, (d) Y4.
Zhu et al. 7

axis of GLCC, as shown in Figure 9. In comparison to operating conditions, the horizontal circular outlet
A1, the maximum axial velocities of central reversed may increase the probability of bubble breakage,
flow for A2 and A3 are smaller, and the ranges of and negatively affect the performance of the separ-
reversed region (capture diameter) are narrowed ator. Compared with the RMS velocity values of A1
down. However, the distribution symmetry of A2 is and A4, the values of A2 are larger. The RMS velocity
better than other separators. Different from other values of A2 in section Y4 are larger than the values in
models, the axial velocity distribution of A4 is a section Y3. It indicates that double rectangular outlets
double backflow reversal at section Y3, with down- led to a disturbance in the fluid field, and the outlets
ward flow near the wall and double backflow appears increase the turbulence intensity. The trends of the
in the central region. RMS velocity profiles for A1 and A4 are similar,
The study above analyzes the effects of the lower and the values are smaller than with other separators.
outlet on the tangential and axial velocity profiles. The It shows that both single rectangular outlet and ver-
turbulence intensity will change as the velocity changes tical circular outlet can decrease the turbulence inten-
in the flow field. At any moment, the root mean square sity, and increase the stability of the flow fields in
(RMS) velocity value at the point reflects the turbulence small GLCC than the other two outlet types.
intensity in the flow field. To evaluate the effects of the It can be seen that the effects of the lower outlet on
lower outlet on the turbulence intensity, the RMS vel- the flow field of small GLCC separator are significant.
ocity profiles are plotted in Figure 10. As shown in Figures 8 and 9, the lower outlet has no
As shown in Figure 10, the RMS velocity value obvious effects on the tangential and axial velocity near
near the wall is small, and increases towards the the wall, but it has significant influence in the central
center. The phenomenon is related to the reversed region. The distribution symmetry of A2 is better than
flow and the swirling flow, those complex fluid pat- A1, which means that double rectangular outlets can
terns taking place in the separator increases the tur- eliminate the asymmetry in small GLCC. However, the
bulence intensity in its central region. In addition to capture diameter of the central reversed flow is smaller,
A1, the RMS velocity values of A3 are much larger and the turbulence intensity of A2 in the flow field is
than the values of other separators, which means that stronger than that of A1. The flow pattern will reduce
a horizontal circular outlet leads to high turbulence the probability of bubble capture under the operating
intensity in the separator flow field. Under the conditions. The capture diameter of A3 is smaller than

Figure 10. Comparison of the radial profiles of the RMS velocity at different sections: (a) Y1, (b) Y2, (c) Y3, (d) Y4.
8 Proc IMechE Part C: J Mechanical Engineering Science 0(0)

A1, and the RMS velocity values of A3 are much larger Funding
than with other models. It indicates that horizontal The author(s) received no financial support for the research,
circular outlet leads to a strong turbulence intensity, authorship, and/or publication of this article.
which negatively affects the separation performance
of small GLCC separators. The velocity distribution References
of A4 has obvious difference with other models, in 1. Kouba G and Shoham O. A review of gas-liquid cylin-
which the free vortex region expanded and the force drical cyclone (GLCC) technology. In: International
vortex region narrowed down. Meanwhile, double conference on production separation systems, Aberdeen,
reversed flow pattern appears in the central region. UK, April 1996, pp.23–24.
Although the maximum tangential velocity of A4 is 2. Kouba GE, Wang S, Gomez LE, et al. Review of the
larger than others and the turbulence intensity is state-of-the-art gas-liquid cylindrical cyclone (GLCC)
weaker, the backflow zone is too small to capture the technology-field applications. In: International oil &
bubbles under the operating conditions. A small back- gas conference and exhibition in China. Beijing, China,
December 2006.
flow zone will lead to a higher gas carry-under (GCU)
3. Gruselle F, Steimes J and Hendrick P. Study of a two-
phenomenon. Different from other separators, the phase flow pump and separator system. J Eng Gas
backflow zone of A1 is wide, and the turbulence inten- Turbines Power 2011; 133: 062401.
sity is weak. As mentioned in Hreiz et al.,17 the turbu- 4. Chang F and Dhir VK. Turbulent flow field in tangen-
lence intensity of the flow field significantly influences tially injected swirl flows in tubes. Int J Heat Fluid Flow
bubble turbulent dispersion, coalescence and breakup. 1994; 15: 346–356.
Thus, a weaker turbulence intensity leads to an easier 5. Ghasemi A, Shams M and Heyhat MM. A numerical
bubble coalescence process. Meanwhile, a larger cap- scheme for optimizing gas liquid cylindrical cyclone sep-
ture diameter means that the bubbles could be collected arator. Proc IMechE, Part E: J Process Mechanical
more easily. Thus, it should be helpful in increasing the Engineering 2017; 231: 836–848.
probability of bubble capture. In consequence, the sep- 6. Gomez L, Mohan R and Shoham O. Swirling gas-liquid
two-phase flow-experiment and modeling, Part I:
arator with the single rectangular outlet is the optimal
Swirling flow field. J Fluids Eng 2004; 126: 935–942.
one for small GLCC separators, among A1–A4.
7. Gomez L, Mohan R and Shoham O. Swirling gas-liquid
two-phase flow-experiment and modeling, Part II:
Conclusions Turbulent quantities and core stability. J Fluids Eng
2004; 126: 943–959.
In this study, we numerically investigate the effects of 8. Pereyra E, Gómez L, Mohan R, et al. Design and per-
the lower outlet on the flow field of small GLCC sep- formance of gas–liquid cylindrical cyclone/slug damper
arators, in which the RSM model is used to manage system. In: Constanda C and Harris PJ (eds) Integral
turbulence. The following conclusions are obtained. methods in science and engineering. Basel, Switzerland:
Birkhaüser, 2011, pp.299–310.
1. A large number of local secondary flow patterns 9. Yang J, Zhang X, Shen G, et al. Modeling the mean
are present in the flow field of small GLCC separ- residence time of liquid phase in the gas–liquid cyclone.
ator, and those flow patterns may cause serious Ind Eng Chem Res 2015; 54: 10885–10892.
energy losses. 10. Erdal FM, Shirazi SA, Shoham O, et al. CFD simula-
2. For small GLCC separators, although the lower tion of single-phase and two-phase flow in gas-liquid
cylindrical cyclone separators. SPE J 1997; 2: 436–446.
outlet has no significant effects on the flow field
11. Arpandi IA, Joshi AR, Shoham O, et al.
near the wall, the central region is clearly influenced
Hydrodynamics of two-phase flow in gas-liquid cylin-
by the lower outlet. By comparison with other drical cyclone separators. SPE J 1995; 1: 427–436.
lower outlets, the single rectangular outlet is the 12. Kouba GE, Shoham O and Shirazi S. Design and per-
most suitable one due to the steady flow field and formance of gas-liquid cylindrical cyclone separators. In:
the large backflow zone, which is helpful in increas- Proceedings of the BHR Group 7th international meeting
ing the possibility of bubble capture. on multiphase flow, Cannes, France, 1995, pp.307–327.
13. Erdal FM and Shirazi SA. Local velocity measurements
In summary, our studies show that lower outlet has and computational fluid dynamics (CFD) simulations
a direct influence on the single phase fluid field, and a of swirling flow in a cylindrical cyclone separator.
single rectangular lower outlet is the optimal one for J Energy Resour Technol 2004; 126: 326–333.
small GLCC separators. 14. Hreiz R, Gentric C, Midoux N, et al. Hydrodynamics
and velocity measurements in gas–liquid swirling flows
in cylindrical cyclones. Chem Eng Res Des 2014; 92:
Declaration of Conflicting Interests 2231–2246.
The author(s) declared no potential conflicts of interest with 15. Erdal FM. Local measurements and computational fluid
respect to the research, authorship, and/or publication of dynamics simulations in a gas-liquid cylindrical cyclone
this article. separator. PhD Thesis, The University of Tulsa, USA,
2001.
Zhu et al. 9

16. Molina R, Wang S, Gomez LE, et al. Wet gas separ- lubricant-air separator. J Propul Technol 2016; 37:
ation in gas-liquid cylindrical cyclone separator. 852–857 (in Chinese).
J Energy Resour Technol 2008; 130: 130–134. 27. Erdal FM, Shirazi SA, Mantilla I, et al. CFD study of
17. Hreiz R, Gentric C and Midoux N. Numerical investi- bubble carry-under in gas-liquid cylindrical cyclone sep-
gation of swirling flow in cylindrical cyclones. Chem arators. SPE Prod Facil 1998; 15: 217–222.
Eng Res Des 2011; 89: 2521–2539. 28. Gao X, Chen J, Feng J, et al. Numerical investigation of
18. Hreiz R, Lainé R, Wu J, et al. On the effect of the the effects of the central channel on the flow field in an
nozzle design on the performances of gas–liquid cylin- oil–gas cyclone separator. Comput Fluids 2014; 92:
drical cyclone separators. Int J Multiphase Flow 2014; 45–55.
58: 15–26. 29. Xu YX, Liu Y, Zhang YH, et al. Effect of shear stress
19. Van Sy L. Influence of inlet angle on flow pattern and on deoiling of oil-contaminated catalysts in a hydrocy-
performance of gas-liquid cylindrical cyclone separator. clone. Chem Eng Technol 2015; 39: 567–575.
Particul Sci Technol 2017; 35: 555–564. 30. Elsayed K and Lacor C. The effect of cyclone inlet
20. Erdal FM and Shirazi SA. Effect of inlet configuration dimensions on the flow pattern and performance.
on flow behavior in a cylindrical cyclone separator. In: Appl Math Model 2011; 35: 1952–1968.
ASME 2002 engineering technology conference on
energy, Houston, TX, US, February 2002, p. 521-9.
21. Erdal FM and Shirazi SA. Effect of the inlet geometry
on the flow in a cylindrical cyclone separator. J Energy
Appendix
Resour Technol 2006; 128: 62–69. Notation
22. Uvwo I. Expanding the operational envelope of compact
cylindrical cyclone gas/liquid separators using a variable a height of the rectangular inlet
inlet-slot configuration. Thesis, Texas A&M University, b width of the rectangular inlet
USA, 2006. c height of the rectangular outlet
23. Steimes J and Hendrick P. Dimensional analysis of an d width of the rectangular outlet
integrated pump and de-aerator solution in more elec- D1 separator diameter
tric aero engine oil systems. Aeronaut J 2017; 121: D2 upper outlet diameter
803–820. D3 diameter of the circular outlet
24. Gupta A and Kumar R. Three-dimensional turbulent
L1 length from top wall to inlet
swirling flow in a cylinder: Experiments and computa-
L2 length from inlet to lower outlet
tions. Int J Heat Fluid Flow 2007; 28: 249–261.
25. Petty CA and Parks SM. Flow structures within mini- L3 length from lower outlet to bottom wall
ature hydrocyclones. Miner Eng 2004; 17: 615–624. L4 length from upper outlet to top wall
26. Zhu WB, Wang DL, Zhang XB, et al. Experimental L5 length of central channel
study on performance of dynamic pressure type
 the angle of inclination of the inlet

You might also like