Documente Academic
Documente Profesional
Documente Cultură
PII: S1383-5866(17)31221-2
DOI: https://doi.org/10.1016/j.seppur.2017.09.040
Reference: SEPPUR 14049
Please cite this article as: W.I. Mazyan, A. Ahmadi, J. Brinkerhoff, H. Ahmed, M. Hoorfar, Enhancement of cyclone
solid particle separation performance based on geometrical modification: numerical analysis, Separation and
Purification Technology (2017), doi: https://doi.org/10.1016/j.seppur.2017.09.040
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Enhancement of cyclone solid particle separation performance
based on geometrical modification: numerical analysis
W. I. Mazyan1 , A. Ahmadi2 , J. Brinkerhoff1 , H. Ahmed3 , M. Hoorfar1 *
1
University of British Columbia, School of Engineering, Kelowna, Canada
2
University of Prince Edward Island, School of Sustainable Design Engineering, Canada
3
American University of Sharjah, School of Engineering, UAE
*Corresponding Author:
Mina Hoorfar
School of Engineering
University of British Columbia
Email: mina.hoorfar@ubc.ca
Tel: (250) 807–8804
Fax: (250) 807-9850
1
Abstract
Computational fluid dynamic (CFD) methods are used to investigate the enhancement of the
conventional cyclone separator. It is shown that the addition of the tangential chamber enhances
the separation of the particles near the conical section wall, and hence the overall separation
efficiency, particularly for small particles (1-3 m). This enhancement occurs with only an 8%
increase in the pressure drop between the inlet and outlet of the cyclone. The effects of the inlet
velocity and the number of the tangential chambers on the separation efficiency are studied to
find the optimum conditions. It is shown that the increase in the velocity enhances the efficiency
at the expense of an increase in the pressure drop (the increase in the inlet velocity from 14 to 20
m/s doubles the pressure drop). The model also shows that the addition of one tangential
chamber produces a lower dissipation rate of turbulence in the cyclone as compared to multiple
chambers, and hence a higher separation efficiency. The efficiency of the proposed geometrical
modification is also compared against the conventional cyclone design and that with another
geometrical modification reported in the literature (i.e., a cyclone with an elongated conical
length). The results reveal that the proposed modification in this paper enhances the separation
efficiency for small particles (less than 3 m) up to 50% compared to the conventional and the
elongated designs and 15% for large particles (larger than 6 m) compared to the conventional
design. The proposed modification was also compared to the conventional design in terms of the
Keywords: Solid-gas separation, tangential collecting chamber, lateral forces, cyclone separators,
2
1. Introduction
Natural gas is becoming one of the most desired fossil fuel energy sources. Two main reasons
have led to the rapid increase in the demand for natural gas (NG) [1]: (1) natural gas is a more
efficient energy source than oil (e.g. gas energy content is 51.6 kJ/gas compared to 43.6 kJ/g
contained in petroleum) [2]; and (2) the amount of CO2 emitted during NG combustion is lower
than that of oil or other common fossil fuels. The increasing demand for natural gas has led to
many technical innovations within the natural gas supply stream to reduce the overall energy
technological advances has been made in the treatment processes for the removal of solid
particles (such as fine sand particles and black powder [3]), a key process to reduce pitting on the
downstream equipment.
One of the most efficient methods for solid particle removal from the gas stream is the cyclone
separator (which is used extensively in the NG industry). A cyclone relies on the centrifugal
forces to guide the particles towards the walls, reducing the momentum of the particles and
hence causing their separation at the bottom, while the clean gas is removed from the top of the
cyclone. During this process, however, small particles rebounding from the walls reduce the
efficiency as they can be carried away by the gas to the clean side of the cyclone. The most
significant factor affecting the centrifugal force, and hence the particle separation/collection
efficiency, is the particle momentum. Of course, an increase in the inlet velocity enhances the
separation efficiency, but at the expense of an increased pressure drop in the cyclone and thus
requires higher energy for transportation in the downstream processes. Another way to increase
the velocity of the particles is to decrease the diameter of the cyclone at the bottom, referred to as
the conical section, which enhances the impact of the finer particles to the wall. At the end of the
3
process, however, there are always very small particles escaping with the clean gas. There have
been numerous studies based on either active methods, such as applying magnetic (e.g.,
attracting black powder mixed with the gas stream [4, 5]), mist injection [6] or electrical field
(e.g., attracting charged particles mixed with the gas stream [7, 8]), or passive methods such as
geometrical modifications (e.g., varying the conical section dimensions [9, 10] of the cyclone)
that have been proposed to reduce the number of these escaping particles. The latter is the focus
of this paper in which the proposed geometrical modifications reported in the literature are
discused in the following paragraphs. These studies include numerical modeling (and some
experimental studies) to show the enhancement in the particle separation efficiency as a result of
Park et al. [11] investigated numerically the separation efficiency using cyclones in series. In
their study, one cyclone separator is divided into three sections. Each section is composed of a
different diameter decreasing from the top cyclone to the bottom. This setup allows for larger
particles with higher inertial forces to be separated without interfering with the smaller ones
which are separated at a later stage. Their results showed that the first cyclone segregates
particles in the range of 4.5 – 11.0 m, whilst the second setup traps the particles in the size of
3.4 – 7.9 m, and the third cyclone collects the particles in the range of 1.8 – 4.3 m. This design
provides an overlap in terms of the particle size between the three stages, so if a certain particle
size is missed in the earlier stage it will be trapped in the following stages. However, the series
configuration of the cyclones increases the pressure drop across the setup as compared to one
cyclone. Kim et al. [12] implemented helical guiding vanes inside the cyclone to experimentally
study their effect on the separation efficiency. Their design with 6 revolutions of helical vanes at
the inlet flow rate of 15 l/min led to a 27% enhancement in the separation efficiency of 4 m
4
particles. Despite the general enhancement in the efficiency, the design provides a complexity in
the manufacturing process of the cyclone. In addition, only a certain size of the particles follows
the same helical path as provided by the fixed design proposed by Kim et al. [12]. For the sizes
outside this range, the particles impact the vanes leading to an increased turbulence inside the
cyclone. Brar et al. [13] numerically studied the effects of increasing the conical length of the
cyclone on the separation efficiency. This approach increased the separation efficiency by 9.5%
for 3 m particles by increasing the major cyclone cylinder length by 5.5 times of the cyclone
diameter. Moreover, the proposed increase in the conical length reduces the pressure drop by
34% as compared to the conventional design. They also showed that by further increase in the
conical length (6.5 times of the cyclone diameter) the separation efficiency increases by 11%
while the pressure drop reduces to 29% of the conventional design. This is due to the longer
conical section which means a larger angle between the main barrel and the conical section
walls, which makes the flow transition of the gas smoother before it redirects towards the exit.
Xiang et al. [14] also experimentally investigated the effect of the dimensions of the conical
section on the separation efficiency. At the inlet flow rate of 30 l/min, their experimental results
showed that reducing the opening at the bottom of the conical section from 19.4 mm to 11.6 mm
enhances the 4 m particle separation efficiency from 70% to 86%. By increasing the inlet flow
rate to 40 l/min, the efficiency for the same size of the particles was further increased to 92%.
Similar to previous geometrical modifications, this reduction in the cone diameter increases the
pressure drop through the cyclone separator by 15%. In another study, Chuah et al. [15] studied
numerically the effect of the conical dimensions on the cyclone performance. By reducing the
bottom conical diameter from 19.4 mm to 11.6 mm, they achieved a 40% increase in the
efficiency for 1.5 m particles. Despite this enhancement in the separation efficiency, the main
5
disadvantage of this method is an increase in the pressure drop by 42%. Similarly, Wasilewski
[16] investigated the effect of an additional counter cone at the bottom of the cyclone on the
configurations of the proposed counter cone. In each of the studied configurations, the maximum
separation efficiency was enhanced by 3%. However, the maximum pressure drop showed an
increase of 4.1%. In a similar study, Misiulia et al. [17] conducted a numerical study on the
effect of four deswirler on the separation efficiency. Their design reduced the pressure drop by
43% at the cost of increasing the separation cutsize from 1.5 m to 1.72 m. Furthermore,
Parvaz et al. [18] studied the use of vortex finders (in an eccentric way) and their impact on the
flow pattern inside the cyclone. Their numerical results indicate that increasing the eccentricity
up to 10% increases the pressure drop by 32%. Similarly, the separation efficiency was enhanced
analysis on the effect of the conical heights on the pressure and velocity fields inside the cyclone
separator. They conclude that the conical height should be limited to 1.5 times the barrel
Another important parameter affecting the separation efficiency in the cyclone separators is the
length of the down comer (which has also been referred to as the vertical tube in literature [20]).
Bryant et al. [21], Zhu and Lee [22] and Mothes [23] emphasized that the down comer controls
the natural vortex length of the flow and particle capture. Qian et al. [20] studied the
the prolonged vertical tube attached to the bottom of the conical section at the dust outlet. Their
study showed that an increase of 0.5 m in the length of the vertical tube increases the separation
efficiency by a maximum value of 15% for 3 m particles. Another example is the CFD analysis
6
conducted by Bogodage and Leung [24] on the effect of the down comer height on the
efficiency. Their results showed that increasing the down comer height by 381 mm increases the
separation efficiency by 20% for particles smaller than 3 m. A similar study conducted by Gil
et al. [25] showed the effect of the increase in the down comer height (without the use of the
hopper) enhancing the separation efficiency to 87% for particles smaller than 5 m.
solid-gas separation in cyclone scrubbers, alternative methods aiming to further increase the
efficiency in “existing” cyclone separation systems must be further explored. The aim of this
study is to increase the separation efficiency of fine particles by including tangential collection
chambers at the conical section in addition to the existing bottom collection chamber. Using a
CFD model, the optimum number of the tangential collection chambers that yield the maximum
collection efficiency for a range of particle sizes is identified, along with the corresponding
pressure losses. Using this model, we identify that the solid-gas separation efficiency in cyclones
2. Computational Setup
2.1 Computational domain
In this study, a cyclone with the dimensions shown in Figure 1 is analyzed numerically. The inlet
verify the results of the numerical model (explained in Section 3), the separation efficiency is
verified against the experimental values obtained by Ji et al. [26] for the same cyclone
dimensions. After verification, the model is used to study the effect of the addition of the
tangential chamber on the separation efficiency. The proposed additional tangential chamber is
introduced to create an additional collecting pot for all sized particles. In this method, solid
7
particles in the gas stream reach the conical section by which their rotational velocity increases.
Hence, the particles experience larger centrifugal forces pushing the particles much faster
Figure 1: The cyclone dimensions (side and front views) used for numerical modeling. The
parameters are: 240 mm, 115 mm, 740 mm, 190 mm, 65 mm, 70
mm, 90 mm (height of the inlet chamber), and 40 mm (width of the inlet chamber).
Figure 2 shows the 3-D model of the conventional cyclone under investigation with the
additional tangential chamber. The tangential chamber is placed in the middle of the conical
section. Since the centrifugal forces on the particles are inversely proportional to their rotational
radius, the centrifugal forces applied on the particles would increase due to the reduced rotational
radius of the conical section. Thus, there are higher chances of smaller particles to reach the
8
Figure 2: The 3D model of the proposed cyclone with the tangential chamber
During the CFD analysis, the solver approximates the medium results of a continuous solution
space using discrete elements [11-14]. A hybrid mesh type is used in the analysis that consists of
unstructured tetrahedral cells in the far- field and clustered triangular prisms near the walls
(inflation layers). Near the boundary layer of the cyclone walls, the inflation-layer growth rate is
set to 1.2 with 5 layers. The cell size has a minimum value of 0.09 mm and a maximum size of
5.7 mm. The size of each element determines the number of mesh cells which the solver will
calculate the pressure and velocity of the flow at. If the mesh cells are sized too large, the error in
pressure and velocity in that cell will be large. On the other hand, reducing the cell sizes to an
infinitesimal dimension will provide more accurate results but will increase the computational
time. Therefore, it is essential that the results obtained from the solver be consistent regardless of
the size and number of the meshed cells (mesh independent results). In this study, the meshing
9
elements number was increased until the separation efficiency curves have consistent results with
different mesh grids. In this paper, the mesh count was varied from 400,000 to 800,000 elements
3. Numerical Method
This study uses ANSYS Fluent with the particle injection scheme. For the cyclone investigated,
the Reynolds stress turbulence model (RSTM) (i.e., the most complicated of the Reynolds-
Averaged Navier Stokes (RANS) models [27]) is used. The advantage of using this model is that
it more accurately accounts for anisotropy of the turbulence, which is the case of the turbulent
flow inside the cyclone due to the influence of flow swirl [27]. The analysis was discretized
using finite volume method with transient simulation having a varying time step from 0.1 to 10 -4
sec. Convective terms are discretized using of the second-order QUICK type scheme, and the
PRESTO method [28] was used to interpolate the cell pressures in an incompressible flow
regime. This method evaluates the pressure of each meshing element at its center rather than
interpolating the center pressure from the corner values, resulting in smaller errors. The
boundary conditions (as shown in Figure 3) include the inlet velocity of 14 m/s as an initial
condition, an outlet pressure of 0 Pa, stationary walls, and a collecting chamber (similar to the
stationary wall but capturing the particles). The scheme used in this study consists of SIMPLEC
scheme [28], which is more accurate than SIMPLE scheme [28]. Solution of the algebraic system
of equations is performed using a W-type algebraic multigrid method in which the convergence
is achieved when the root- mean-square residual decreases below 10-4 to ensure a small error
between subsequent iterations. The total flow time of the simulated flow ranges between 8-12
seconds, which corresponds to approximately 11,000 time steps of the transient solver,
depending on the mesh element count and the shape o f the cyclone (e.g. conventional against the
10
chamber design). Particle injection uses the discrete phase model for particles sizes from 1-6 m
(with a particle density of 1500 kg/m3 ). The discrete phase model couples the particle motion to
the fluid motion by applying kinematic equations on the particle subjected to flow and pressure
forces from the surrounding fluid. Solution of the particle motion commences after the flow
solution has achieved a steady-state. The transient kinematic equations for the particle motion are
integrated for 107 steps, which is sufficient for the fate of all particles to reach a steady state.
Whereas, the flow time step varies from 0.1 to 10 -4 sec. The particles used in the analysis have a
uniform size in the simulations, and cases are simulated with sizes 1-6 m.
To study mesh independency, the separation efficiency was evaluated for mesh element counts
ranging from 400,000 to 800,000. The time required for the computations varied between 3
consecutive days with 400,000 mesh elements to 10 consecutive days with 800,000 mesh
11
Figure 3: The 3D model boundary conditions
12
4. Results and Discussions
The separation efficiency, sp , pressure drop through the conical separator (defined as the
difference in the average inlet and exit static pressures), and material erosion rate (defined based
on the mass of the material removed per unit area per second) at the cyclone separator walls
made from carbon steel are used as criteria to evaluate the performance of the proposed
geometrical modification. The larger the value of sp , the larger the enhancement in particulate
capturing by the cyclone. On the other hand, the smaller the pressure drop and the erosion rate,
the more the enhancement in the cyclone lifetime. The separation efficiency is calculated as
nOutlet
SP 1 (1)
n Inlet
the conventional (referred to as Stairmand cyclone) design, and ii) the cyclone with the proposed
geometrical modification. As indicated by Cortes and Gil [29], the conventional cyclones have
generally higher efficiency for larger particles. In essence, larger particles experience larger
centrifugal forces due to their larger inertia, and hence, are expected to reach the outer walls
faster. As a particle hits the outer walls of the cyclone, its momentum declines and it falls down
the collecting chamber. Fluent deals with the reflection of particles when impacting the walls by
specifying the impact coefficient value (e.g. an elastic impact coefficient is set to 1). Similarly,
smaller particles can also reach outer walls by increasing the angular velocity and hence
centrifugal forces through the conical section. However, very fine particles may escape with the
gas stream as they can be picked up by the gas easily. The goal of the proposed geometrical
modification is to capture these fine particles prior to their escape through the gas stream.
13
Model verification and mesh independency analysis – To verify the CFD modeling results, the
separation efficiency values obtained using the model are compared against the experimental
results reported by Ji et al. [26] (see Figure 4). The modeling results were determined for three
different mesh cell counts to determine mesh independency. The results show that the maximum
error between the experimental [26] and CFD results (regardless of the number of the meshing
elements) occurs for 3 m particles. The smaller particles (1-2 m) have high tendency to escape
as they reach the exit; while larger particles (4-6 m) have high tendency to settle at the bottom.
The 3 m particles, on the other hand, have high tendency to rotate inside the cyclone before
they either escape or are captured. This can result in the observed difference between the CFD
14
To conduct the mesh independency analysis, the difference between the CFD and experimental
results for each mesh resolution is quantified. The difference between the 400,000 and 600,000
1.0
0.9
0.8
Collection Efficiency
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 1 2 3 4 5 6 7
Particle Diameter ( m)
element cases are too large to consider the results mesh independent. For example, for 5 m
particles, the difference between the errors obtained for two meshing elements is 10% which is in
the same order of the magnitude of the separation efficiency enhancement being sought.
Increasing the mesh counts from 600,000 to 800,000 elements reduces the magnitude of the error
by less than 2.1% for 4-m particles but at the expense of increasing significantly the
computational time (by 150%). Table 1 presents the relative error between particle sizes against
the mesh size. Thus, 600,000 meshing elements is chosen for further analysis.
Figure 4: Comparison between the experimental separation efficiency (Ji et al. [26]), CFD
modeling conducted at 400,000 meshing elements, 600,000 meshing elements and 800,000
15
Table 1: Relative errors for particle size against mesh size
1 0 0 0 0
Number of the tangential chambers – Using 600,000 meshing elements, the separation efficiency
determined for the range of the particle sizes mentioned above. The addition of the tangential
chamber at the conical section of the cyclone is expected to enhance the separation efficiency of
smaller particles as they reach the walls. As particles enter the conical section, their radial
velocities increase due to the decrease of the rotational radius. Furthermore, the centrifugal
forces applied at the particles increase due to the increased radial velocities. Thus, particles from
all ranges have higher tendencies to reach the outer walls of the conical section. Moreover, the
location to place the tangential chamber is chosen to be just above the lowest part of the conical
section. Further optimization of the location of the tangential chamber is needed and will be
16
performed in the future studies. In this proposed modification, the small- sized particles are
trapped and collected inside the tangential chamber before they have a chance to gain their
momentum again. Figure 5 shows the separation efficiency of the proposed modification as
compared to those reported by Ji et al. [26]. Also, the effect of the number of the tangential
chambers number on the separation efficiency is shown in this figure. Figures 6 shows the
turbulence eddy dissipation rate inside the cyclone separator space. In general, the addition of the
tangential chambers increases the separation efficiency as compared to that reported by Ji et al.
[26]. However, the number of the chambers affects the separation efficiency; it is observed that
adding one tangential chamber to the conical section results in higher efficiency than the case of
two or three tangential chambers. This is attributed to the increased turbulence at the inlet of
each of the chambers. Figures 6 shows the turbulence eddy dissipation rate at the cross section of
the chambers. It is observed that for the case of one chamber, the maximum value of the
turbulence dissipation rate is smaller (12770 m2 /s3 ) than the two and three chamber cases
(15580 m2 /s3 and 14220 m2 /s3 , respectively). For the case of three chambers, the eddy value is
smaller, and hence the separation efficiency is higher, than those values obtained for the case of
adding two chambers. This is likely due to the reduced distance between the openings.
Figure 7 shows that at the inlet of the chambers the largest velocity corresponds to the case of
one chamber (7.686 m/s), followed by three and two chambers (6.52 m/s and 4.8 m/s,
respectively). It is obvious from the velocity contours that the fluid flow has the farthest
propagation inside the chamber for the case of one chamber. For the case of adding two
chambers, on the other hand, the fluid flow seems to have the least propagation inside the
chamber. Due to the increased turbulence dissipation rate at the inlets of the chambers in the
two- and three-chamber cases, it seems the velocity near the walls and the chambers decreases,
17
yielding the reduced flow penetration into the chamber in those chases. As noted before, the
decrease in the velocity directly reduces the centrifugal forces. The reduction in centrifugal
forces reduces the ability of solid particles (especially smaller ones) to reach the outer walls and
1.0
0.9
Collection Efficiency
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 1 2 3 4 5 6 7
Particle Diameter ( m)
Figure 5: The comparison between the experimental results reported by Ji et al. [26], CFD results
of a conventional design, CFD results for the proposed geometrical modification involving the
addition of one tangential chamber, CFD results for the case of two chambers, CFD results for
18
Figure 6: The turbulence eddy dissipation for the cases of adding one, two, or three tangential
chambers
19
Figure 7: The fluid velocity contours for the cases of adding one, two, or three tangential
chambers
20
Velocity at the inlet of the cyclone - Increasing the inlet velocity increases the separation
efficiency as mentioned in the introduction. Using the numerical model, this effect was studied
for two inlet velocity values of 14 and 20 m/s applied to the geometrical modification with one
tangential chamber. These results, shown in Figure 8, were compared against the experimental
results that Ji et al. [26] reported for these inlet velocities. The results show that increasing the
inlet velocity from 14 to 20 m/s increases the separation efficiency for the design with one
tangential chamber for all particle sizes. The largest increase of 25% is achieved for the 2 m
particles and the smallest increase of 1% is achieved for the 6 m particles. However, this
improvement incurs a significant rise in the pressure drop, with increases of up to 110% as
Length of the conical section – Among different geometrical modifications reported in literature
(see Introduction), the vertical length of the conical section seems to yield the highest separation
efficiency gains [13, 14]. The CFD model is used to compare the separation efficiency for the
range of the particles studied here of the cases of: i) the conventional cyclone design, ii) the
design with an elongated vertical cone, and iii) the design with one tangential chamber. Figure 9
demontrates that elongating the cone enhances the efficiency for particle sizes larger than 5 m
as compared to the conventional and elongated designs. The addition of one tangential chamber,
however, enhances the efficiency for all ranges of particles (especially the small particles)
compared to the conventional and elengated designs. Specifically, the addition of a tangential
chamber at the conical section of the cyclone increases the separation efficiency for particles less
than 3 m by 50% compared to the conventional and elongated designs and 15% for particles
21
1.0
0.9
0.8
Collection Efficiency
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 1 2 3 4 5 6 7
Particle Diameter ( m)
Figure 8: The effect of the velocity at the inlet of the cyclone on the separation efficiency for the
22
1.0
0.9
0.8
Collection Efficiency
0.7
0.6
0.4
Geometrical modification by
0.3
elongating the cone
0.2
Geometrical modification by
0.1 adding one chamber
0.0
0 1 2 3 4 5 6 7
Particle Diameter ( m)
Figure 9: The comparison between the separation efficiency obtained numerically for the
conventional design, conventional design with an elongated cone length, and the conventional
investigated here using the developed CFD model. Here, the pressure drop is defined as the
difference in the average static pressure between the inlet and exit boundaries of the cyclone
separator. Table 2 gives the pressure drop of i) the experimental results of Ji et al. [26] reported
for the conventional cyclone at the inlet velocity of 14 m/s, ii) the CFD modeling result for the
conventional design at 14 m/s, iii) the CFD result for the case of adding one tangential chamber
at 14 m/s, and iv) the CFD result obtained for the case of adding one tangential chamber at 20
23
m/s. The efficiency enhancement using one tangential chamber at 14 m/s inlet velocity is 50%
for 1 m particles (Figure 5) at the cost of only an 8% increase in the pressure drop. On the other
hand, the efficiency enhancement using one tangential chamber at 20 m/s inlet velocity is 74%
for 1 m particles size, but with a significant pressure drop increase of 132%.
separator walls, which eventually cause erosion of the wall material [29]. The effect of the
proposed model on the material erosion rate inside the cyclone separator is investigated
numerically to ensure that it will not negatively affect the erosion rate. Figures 10 and 11
represent the erosion rates of the walls due to the particles impact and the velocity profiles inside
the cyclone, respectively. The results are presented for the cases of the conventional and
modified designs. The addition of the tangential chamber reduces the maximum erosion rates
inside the cyclone from 5.410-8 kg/m2 s for the conventional design to 2.510-8 kg/m2 s for the
tangential modification. This decrease in the erosion rate can be explained due to the decrease in
the absolute velocity at the walls of the cyclone from 5.6 m/s to 5.2 m/s (see Figure 11). The
decrease of the absolute velocity at the cyclone walls is a result of the turbulence created near the
chambers. The solid impact erosion rate can be evaluated using Eq. (2) [30]
24
1 vp
2
1 vw
2
Er (2)
Ew Ep
where v w, Ew, v p and Ep are the cyclone wall and particle poison’s ratio and Young’s modulus,
respectively
Figure 12 represents the radial and axial velocity contours inside the cyclone separator. The
tangential velocity shows a decrease at the location of the chamber (shown in the figure with a
square box) due to the backflow caused by the tangential chamber. The axial velocity, in
addition, shows a decrease in its value near the chamber walls due to the same reason. Both of
these reductions explain the absolute velocity reduction near the walls in the overall velocity
25
Figure 10: Locations of erosion in (a) conventional , and (b) the proposed design including one
tangential chamber. The maximum erosion rates for (a) and (b) designs are 5.410-8 kg/m2 s and
26
(a) (b)
Figure 11: Velocity profile for (a) conventional, and (b) the design chamber added
Figure 12: Tangential and axial velocities contours for conventional and chambered cyclones
27
5. Cost Analysis
Although cyclone separators are less efficient than other types of filters such as bag filters, they
are commonly used to remove the bulk of the solid particles in the upstream of the NG plants due
to their low maintenance cost as compared to bag filters [3, 11, 12]. For example, cartridge filters
(bag filters) cause an increase in the pressure drop over time due to the blockage of the filter
element [3, 11, 12]. This blockage requires an additional annual maintenance cost as well as the
installation of a second cartridge in parallel to the main one to avoid shutdown of the pipeline
[3]. On the other hand, cyclone separators have a fixed pressure drop over time while providing a
reliable filtration mechanism [3, 11, 12]. One major change to the cost of the cyclone due to the
6. Conclusions
The feasibility of adding tangential chambers at the conical section of a Stairmand-type cyclone
was investigated in terms of its effect on the efficiency of the overall solid-gas separation
efficiency, pressure drop, and erosion rate. The CFD modeling results indicate that the addition
of one tangential chamber increases the particle separation efficiency by a maximum of 50%
over current cyclone designs, with the largest improvements obtained for particle sizes below 3
m. The proposed modification increases the pressure drop by 8% at the designed inlet velocity
while reducing the erosion rate by 54%. Thus, the proposed modification makes the cyclone a
powerful and at the same time low cost tool for particle separation that is critical for minimizing
damage to the downstream equipment in oil and gas, refinery, chemical and polymer industries.
Besides the addition of a tangential chamber, current industries can also benefit from an
increased separation efficiency by increasing the inlet velocity and/or having an elongated
28
conical section. Further optimization of the location of the tangential chamber is needed and will
29
References
1- Mei, M. (2011) North American Natural Gas Market Dynamics. Canadian Energy
2- Energy from Fossil Fuels. Online source. Retrieved on January 2017 from
https://www.wou.edu/las/physci/GS361/Energy_From_Fossil_Fuels.htm
4- Svoboda, J.; Coetzee, C.; Campbell, Q. (1998) Experimental Investigation into the
11: 501-509.
5- Mazyan, W.; Ahmadi, A.; Jesus, R. D.; Ahmed, H.; Hoorfar, M. (2016). Use of ferrous
powder for increasing the efficiency of solid particle filtration in natural gas cyclones.
6- Yang, K.; Yoshida, H. (2004) Effect of mist injection position on particle separation
8- Mazyan, W.; Ahmadi, A.; Ahmed, H.; Hoorfar, M. Enhancement of solid particle
30
9- Mazyan, W.; Ahmadi, A.; Ahmed, H.; Hoorfar, M. Increasing efficiency of natural gas
cyclones through addition of tangential chambers. Journal of Aerosol Science. 2017; 110:
36-42.
10- Xiong, Z.; Ji, Z.; Wu, X. (2014) Investigation on the Separation Performance of a
Multicyclone Separator for Natural Gas Purification. Aerosol and Air Quality Research,
14: 1055–1065.
11- Park, C. W.; Song, D. H.; Yook, S. J. (2015). Development of a single cyclone separator
with three stages for size-selective sampling of particles. Journal of Aerosol Science, 89:
18-25.
12- Kim, H. T.; Lee, K. W.; Kuhlman, M. R. (2001). Exploratory design modifications for
13- Brar, L. S.; Sharma, R. P.; Elsayed, K. (2015). The effect of the cyclone length on the
14- Xiang, R.; Park, S. H.; Lee, K. W. (2001). Effects of cone dimension on cyclone
15- Chuah, T. G.; Gimbun, J.; Choong, T. S. Y. (2006). A CFD study of the effect of cone
17- Misiulia, D.; Elsayed, K.; Andersson, A. Geometry optimization of a deswirler for
cyclone separator in terms of pressure drop using CFD and artificial neural network.
31
18- Parvaz, F.; Hosseini, S.; Ahmadi, G.; Elsayed, K. Impacts of the vortex finder
eccentricity on the flow pattern and performance of a gas cyclone. Separation and
19- Demir, S.; Karadeniz, A.; Aksel, M. Effects of cylindrical and conical heights on pressure
20- Qian, F.; Zhang, J.; Zhang. M. (2006). Effects of the prolonged vertical tube on the
21- Bryant, H.; (1983). How dust in gas affects cyclone pressure drop. Hydrocarbon Process.
62.
22- Zhu, Y.; Lee, K. (1999). Experimental study on small cyclones operating at high
23- Mothes, H. (1982). Bewegung und Abscheidung der Partikeln im Zyklon (PhD. Thesis).
24- Bogodage, S. G.; Leung, A. Y. T. (2015). CFD simulation of cyclone separators to reduce
25- Gil, A.; Cortes, C. (2002). Gas-particle flow inside cyclone diplegs with pneumatic
26- Ji, Z.; Xiong, Z.; Wu, X.; Chen, H.; Wu, H. (2009) Experimental investigations on a
27- CFD Online. Reynolds Stress Model. Online source. Retrieved on November, 2016 from
https://www.cfd-online.com/Wiki/Reynolds_stress_model_(RSM)
32
28- ANSYS. Introductory Fluent Training-Solver Settings. Fluent User Services Center.
www.fluentusers.com
29- Cortes, C., & Gil, A. (2007). Modeling the gas and particle flow inside cyclone
30- Aquaro, D. Erosion due to the impact of solid particles of materials resistant at high
33
Cyclone separation efficiency is enhanced using tangential chamber.
34