Iranian Journal of Chemical Engineering(IJChE)

Draft Manuscript for Review

"Improving the performance of a two-phase ejector using genetic

algorithm based on secondary fluid entrainment rate"

Journal: Iranian Journal of Chemical Engineering(IJChE)

Fo

Manuscript ID: IJCHE-2303-1488 (R1)

Manuscript Title: Improving the performance of a two-phase ejector using
genetic algorithm based on secondary fluid entrainment
rP

rate
ee
rR
ev
ie
w

Page 1 of 22
 Iranian Journal of Chemical Engineering(IJChE)
1 Responses to Reviewers
2
3 Manuscript Title
4
5 “Improving the performance of a two-phase ejector using genetic algorithm based on secondary fluid
6 entrainment rate”
7
8
9 The authors would like to thank the reviewers for their careful reading of the manuscript. The comments of the
10 reviewers were very fruitful and in essence all their recommendations have been implemented in the revised
11 version. Herein, we describe in details how we have considered the comments of the referees.
12
13 Responses to Reviewer 1
14
15 Comments to the Author
16 The paper titled "Improving the performance of a two-phase ejector using genetic algorithm based on secondary
fluid entrainment rate" is good one and the manuscript can be considered for publication “Iranian Journal of
Fo

17
18 Chemical Engineering (IJChE)” with minor revisions. The following suggestions are provided .
19
20 Comment:
rP

21 Symbols are not listed in the manuscript.

22 Response:
23 The Symbols are listed to the manuscript.
24
ee

25 Comment:
26 The novelty and originality of the study should be better highlighted in the academic perspective.
27 Response:
rR

28 As is mentioned in the other comment of the reviewer, the novelty of the present work is highlighted at the
29 last paragraph of the introduction section:
The current research investigates the use of gas-liquid ejectors in flare gas recovery through computational
30
fluid dynamics simulations. We optimized the throat diameter and length, as well as the nozzle diameter
31
and angles (convergent and divergent) using a multi-objective genetic algorithm to achieve maximum
ev

32
entrainment rate of the secondary fluid.
33
34 Comment:
ie

35 There are several typos and grammatical mistakes. It is anticipated to improve whole the text with a better
36 English writing style.
37 Response:
w

38 The manuscript was re-checked and its writing errors were corrected.
39
40 Comment:
41 Some sentences are not fluent and clear.
42 Response:
43 Upon closer inspection, these sentences were rewritten as much as possible.
44
45 Comment:
46 It is pretty found that the introduction of the present article ends with a precise evaluation of the novelty of
47 the proposed research article.
48 Response:
49 We would like to thank for your comment.
50
51 Comment:
52 Outlook and future perspectives can be included.
Response:
53
There are areas for future research including 3D modeling, some of which are of interest to the authors and
54
will be presented in the future.
55
56 Comment:
57 Why “Realizable-ε" method was used to model the turbulence? Why k-e standard did not use?
58 Comment:
59 Why is the “Realizable-ε" equation used for modeling? Have you fitted K and Epsilon at the entrance?
60 What equation is used to calculate their value?

P age |1
Page 2 of 22
 Iranian Journal of Chemical Engineering(IJChE)
1
Response:
2
Conforming to the high Reynolds number the flow inside the ejectors is turbulent. Many studies have
3 proposed different k-ε approaches for modelling turbulence flow in the two-phase ejectors [1 – 8]. Based
4 on the results of these references, we use the "Realizable -ε" method to model the turbulent flow in the
5 ejector.
6
7 Comment:
8 What is the means of the sentence in page 4 line 4: One-dimensional computational fluid dynamics
9 simulations ….?
10 Response:
11 One-dimensional mathematical model was employed in the past research to model ejectors.
12
13
14 Comment:
15 The quality of some figures is not acceptable (Figs. 4, 5, 6, 7 and etc.)
16 Response:
Fo

17 The previous figures were replaced with higher quality ones.

18
19 Comment:
20 It is recommended that throat length, nozzle radius, and etc. in pages of 8 and 9 list in a Table.
rP

21 Response:
The table 3 presents these quantities for three candidate points.
22
23
Comment:
24
ee

Highlights are too long. Please re-written them.

25
Response:
26 With great thanks, they have been shortened and stated as briefly as possible.
27
rR

28 Comment:
29 Is independence from the mesh done for CFD modeling?
30 Response:
31 Due to the disparity in CFD results, different grid sizes were generated and implemented in the domain.
ev

32 From the grid with a size of 0.25 mm, the results experienced better convergence and less fluctuations.
33 Therefore, this network is used in the following sections of this study.
34
ie

35 Comment:
36 The type of applied mesh must be mentioned.
37 Response:
w

38 The computational domain is meshed wielding the structured quadrilateral mesh.

39
40
41 Responses to Reviewer 2
42
43 Comments to the Author
44 The article is an original research work on optimizing a two-phase ejector using Computational Fluid Dynamics
45 and Genetic Algorithms. The results are precise, and the data is well understood in detail. In my opinion, the
46 manuscript is suitable for publication in "IJCHE", after addressing the following minor revisions:
47
48 Comment:
49 Abstract, Line 32: “In contrast, the throat and angle of the diverging section …” is that “throat diameter”
50 or “throat length”?
51 Response:
52 It is “throat diameter”. This was corrected in the manuscript.
53
Comment:
54
Page 2, line 28: “Using a multi-objective genetic algorithm, the optimal values for each parameter are
55
obtained”. This means each parameter has been optimized solely, while, all parameters have been optimized
56
simultaneously. It is better to be corrected.
57 Response:
58 This sentence was modified as follows. “Using a multi-objective genetic algorithm, the optimal values for
59 all parameters are obtained”.
60

P age |2
Page 3 of 22
 Comment:
Iranian Journal of Chemical Engineering(IJChE)
1
Page 2, line 58: “Some authors have dedicated to their focus to modifying”: “to” should be deleted.
2
Response:
3 It was corrected in the manuscript.
4
5 Comment:
6 In the results and discussion section, the changes in “turbulence intensity” of the mixture were monitored
7 for different cases. However, its effect on the target variable (absorption rate) was not clearly discussed.
8 Do higher mixture turbulences mean higher performance? How?
9 Response:
10 Increasing the turbulence intensity increases the amount of suction, but in terms of energy performance, it
11 has not necessarily led to an increase in efficiency.
12
13 Comment:
14 Page 9, line 43: “The results indicate that increasing the radius of the throat and the angle of the divergent
15 part, as well as decreasing the angle of the converging part, lead to an increase in the absorption rate of gas”
16 It is suggested to add a figure to show the mentioned changes graphically.
Fo

17 Response:
18 Refer to figure 5.
19
20 Comment:
rP

21 Table 4: the titles of the rows needed to be revised.

Response:
22
It was corrected in the manuscript.
23
24
ee

Comment:
25
Page 12, line 50 needed to be revised: “In contrast, the third design point not only increased the energy
26 efficiency but also decreased it.
27 Response:
rR

28 “In contrast, the third design point decreased the energy efficiency”.
29
30 Comment:
31 How the CFD simulation results have been validated? It has not been clearly illustrated in the “Verification
ev

32 of computational fluid dynamics results” section.

33 Response:
34 To validate the CFD simulation the geometry of the experimental model of Bhutada and Pangarkar (1987)
ie

35 is selected. In their model, water was the primary, and the air was taken as the secondary fluid. We validated
36 the CFD modelling (present work) for the case where the length to the diameter of the throat ratio and the
37 throat to the nozzle area ratio is 4. The diameter of the throat is 16 mm. The angle of the converging and
w

38 diverging area are 12 and 5 degrees, respectively. Bhutada and Pangarkar (1987) reported that using these
39 geometric conditions makes the air entrainment around 0.00294 kg/s as the axial symmetry in the ejector
40 geometry we used is based on 2D-axisymmetric model. Water and the air (with a diameter of 1 mm) are
41 considered as primary and secondary phases of the ejector, respectively. We use the following boundary
42 conditions: (i) Water inlet velocity is 21.2 m/s; (ii) The gauge pressure of the air inlet is zero; (iii) The
43 gauge pressure of the ejector outlet is zero; And (iv) the ejector and nozzle walls have non-slip condition.
44 The "Coupled" and "PRESTO" models are employed to connect the velocity-pressure and discrete the
pressure correspondingly. The second-order upwind model is used for turbulent kinetic energy, discrete
45
momentum, and turbulence dissipation rate. The under-relaxation coefficients are 0.1 for pressure, density,
46
momentum, slip velocity, and volume fraction.
47
The computational domain is meshed wielding the structured quadrilateral mesh, as shown in Fig. 1.
48
Various mesh sizes were generated and implemented in the domain due to the fluctuations in the CFD
49 results having better convergence and fewer alterations with a grid size of 0.25 mm. For that reason, such
50 a mesh is employed in the following sections of this study.
51
52
53
54 References
55
56 [1] Bhutada, S. and V. Pangarkar (1987). "Gas induction and hold-up characteristics of liquid jet loop
57 reactors." Chemical Engineering Communications 61(1-6): 239-258.
58
59
60

P age |3
Page 4 of 22
 Iranian Journal of Chemical Engineering(IJChE)
[2] Sriveerakul, T., S. Aphornratana and K. Chunnanond (2007). "Performance prediction of steam ejector
1
using computational fluid dynamics: Part 1. Validation of the CFD results." International Journal of
2
Thermal Sciences 46(8): 812-822.
3
4 [3] Yadav, R. L. and A. W. Patwardhan (2008). "Design aspects of ejectors: Effects of suction chamber
5 geometry." Chemical Engineering Science 63(15): 3886-3897.
6
7 [4] Ariyaratne, W. H., E. Manjula, C. Ratnayake and M. C. Melaaen (2018). CFD approaches for modeling
8 gas-solids multiphase flows–A review. Proceedings of The 9th EUROSIM Congress on Modelling and
9 Simulation, EUROSIM 2016, The 57th SIMS Conference on Simulation and Modelling SIMS 2016,
10 Linköping University Electronic Press.
11
12 [5] Parekh, J. and R. Rzehak (2018). "Euler–Euler multiphase CFD-simulation with full Reynolds stress
13 model and anisotropic bubble-induced turbulence." International Journal of Multiphase Flow 99: 231-
14 245.
15
16 [6] Ringstad, K. E., Y. Allouche, P. Gullo, Å. Ervik, K. Banasiak and A. Hafner (2020). "A detailed review
Fo

17 on CO2 two-phase ejector flow modeling." Thermal Science and Engineering Progress 20: 100647.
18
19 [7] de Oliveira Marum, V. J., L. B. Reis, F. S. Maffei, S. Ranjbarzadeh, I. Korkischko, R. dos Santos Gioria
20 and J. R. Meneghini (2021). "Performance analysis of a water ejector using Computational Fluid
rP

21 Dynamics (CFD) simulations and mathematical modeling." Energy 220: 119779.

22
[8] Suvarnakuta, N., K. Pianthong, T. Sriveerakul and W. Seehanam (2020). "Performance analysis of a
23
two-stage ejector in an ejector refrigeration system using computational fluid dynamics." Engineering
24
ee

Applications of Computational Fluid Mechanics 14(1): 669-682.

25
26
27
rR

28
29
30
31
ev

32
33
34
ie

35
36
37
w

38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

P age |4
Page 5 of 22
 Improving the performance
Iranian ofChemical
Journal of a two-phase ejector using genetic algorithm
Engineering(IJChE)
1
2 based on secondary fluid entrainment rate
3
4
5 Mehdi Moghaddasiaa, Mostafa Keshavarz Moravejibb* and Omid Alizadehcc
6
7
8 a
Department of Chemical Engineering, Borujerd Branch, Islamic Azad University, Borujerd, Iran;
9
b
10 Department of Chemical Engineering, Amirkabir University of Technology (Tehran Polytechnic), Tehran, Iran;
11 c
Department of Chemistry and Chemical Engineering, Rasht Branch, Islamic Azad University, Rasht, Iran
12
13 *Corresponding author: moraveji@aut.ac.ir
14
15
16 Abstract
Fo

17
18 Ejectors offer a cost-effective and practical solution for recovering flare gases, thereby reducing
19
20 greenhouse gases. Improving the entrainment rate of the secondary fluid can enhance ejector
rP

21 performance. The objective of this research is to identify the optimal ejector geometry to maximize
22
23 the absorption rate of the secondary fluid. Computational fluid dynamics is used to evaluate a two-
24
ee

25 phase ejector. Geometric parameters such as throat diameter and length, nozzle diameter, and
26
27
converging and diverging angles impact the absorption rate of the secondary fluid. Using a multi-
rR

28 objective genetic algorithm, the optimal values for all parameters are obtained. The results show
29
30 that reducing the throat length and angle of the converging section, as well as nozzle diameter,
31
leads to increased absorption. In contrast, the throat diameters and angle of the divergent section
ev

32
33 increase absorption. Additionally, energy efficiency is investigated under basic and optimized
34
ie

35 geometries. The findings reveal that increasing the soak range does not necessarily enhance energy
36
37 efficiency.
w

38
39
40 Keywords: Entrainment rate of secondary fluid; gas-liquid ejector; recovery of flare gas; geometry
41
42 design; multi-objective genetic algorithm optimization; computational fluid dynamics.
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Page 6 of 22
 Nomenclatures
Iranian Journal of Chemical Engineering(IJChE)
1 𝑎⃗ acceleration vector
2
3 𝐹𝑑 the drag force
4
5 𝑔 gravitational acceleration
6
𝐿𝑇 throat length
7
8 𝑚̇𝑔 gas flow rate
9
10 𝑚̇𝑙 fluid flow rate
11
12 forces measurements containing lift, drag, turbulent dispersion, wall lubrication, and
13 𝑀𝑘
14 virtual mass
15 𝑝𝑘 pressure of the phase 𝑘
16
Δ𝑃𝑔
Fo

17 the gas recovery pressure over suction to the diffuser outlet

18
19 Δ𝑃𝑙 the liquid pressure drop from the nozzle outlet to the ejector
20
𝑅𝑇
rP

21 throat radius
22
23
𝑅𝑁 nozzle radius
24 𝑅𝑊 liquid inlet radius
ee

25
26 Re the Reynolds number
27
𝑆𝑘
rR

28 the rate phase production between different phases

29
30
𝑡𝑠 relaxation time of secondary phase
31 𝑢𝑘 velocity of the phase 𝑘
ev

32
33 𝛼𝑘 volume fraction of the phase 𝑘
34
𝜂 the ejector energy efficiency
ie

35
36
37
𝜌𝑘 density of the phase 𝑘
w

38 𝜇𝑘 viscosity of the phase 𝑘

39
40 𝜑 diverging angle
41
42 𝜃 converging angle
43
44
45 Introduction
46
47 The term "flaring" encompasses two concepts: (a) the combustion of flammable gases from
48
49 equipment and parts removed from various oil and gas refineries and petrochemical complexes,
50
and (b) the continued burning of gas instead of its recovery. The flaring system is responsible for
51
52 emitting the largest quantity of environmental pollutants [1, 2].
53
54 Researchers have proposed different methods and techniques to reduce and recover the
55
56 gases sent to the flare. Some authors have dedicated to their focus modifying production in general,
57 i.e., gas production units in particular, to reduce the amount of gases transferred to the flare [3, 4].
58
59 Recovering flare gases is another approach to the reduction of greenhouse gas emission [5-13].
60

Page 7 of 22
 The use of certain systems for gas compression can reduce the flaring rate. Therefore, gases can
Iranian Journal of Chemical Engineering(IJChE)
1 be collected and recovered.
2
3 Of note, the application of an ejector is one way among many to recover flare gas. Leagas
4
5 et al. [14] focused on the cost-effectiveness and efficiency of the ejector technology for recovering
6 flare gas. Ainge [15] demonstrated that a compressor could best be used as an efficient technique
7
8 to recover flare gas. The author discussed the use of the ejector technology as an alternative to
9
10 rotary-type compression equipment. Mazumder et al. [16] presented a new approach that involved
11
12
combining the ejector-based flare gas recovery method with the developed thermal vapor
13 compression method. Eshaghi and Hamrang [17] utilized the gas-gas ejector as an alternative to
14
15 the compressor. Bashiri et al. [18] designed a syngas purification system derived from integrated
16
biomass with certain modifications. Ainge and Al-Khateeb [19] reviewed the latest advances in
Fo

17
18
ejector technology and studied its use as a compressor for flare gas recovery. Researchers have
19
20 used the ejector in these networks instead of other compressors and examined its cost-effectiveness
rP

21
22 from a technical point of view. The results suggest that utilizing ejectors for flare gas recycling
23
24 can provide economic benefits and energy savings. Nevertheless, researchers have not extensively
ee

25 investigated the influence of the ejector's geometrical parameters on its performance.

26
27 Ejectors are widely used in the chemical industry for suction, compression, transfer, or
rR

28
29 mixing of both fluids and solid particles [20-25]. Depending on requirements and demands, there
30
31
are different ways to design and optimize ejectors. One way is to increase the entrainment rate of
ev

32 secondary fluid, referred to as the intensity of the mixture of two substances, or the transfer of
33
34 fluid from a low-pressure area to a high-pressure area. The most important objective of this
ie

35
36 research is to increase the entrainment rate of recovery gas.
37
Initially, the performance of ejectors is analyzed based on one-dimensional classical gas
w

38
39 dynamics. Subsequently, this theory was further modified to incorporate loss coefficients in the
40
41 mixing chamber, nozzle, and diffuser. However, the geometry of the ejector remained unchanged
42
43 in this analysis [26]. The development of numerical methods has encouraged researchers to utilize
44 methods that involve computational fluid dynamics. Researchers utilized the computational fluid
45
46 dynamics (CFD) method to study the flow in ejectors and optimize their performance. Sriveerakul
47
48 et al. [26] studied the effects of functional conditions and geometrical properties of ejectors on
49
50
their performance. Yadav and Patwardhan [27] optimized the geometry of ejector suction chamber
51 using computational fluid dynamics method. Galanis and Sorin [20] presented a one-dimensional
52
53 thermodynamic model to investigate the behavior of ejectors and determine the highest possible
54
55 compression ratio for given inlet conditions, mass flow rate, and suction fluids. Wang et al. [21]
56 employed CFD to optimize the primary nozzle geometry and showed that the performance of the
57
58 ejector was highly dependent on the length and angle of divergence. However, their findings only
59
60 focused on optimizing the geometry of the nozzle section. Suvarnakuta et al. [23] employed

Page 8 of 22
 computational fluid dynamics to investigate the performance of a steam ejector in refrigeration
Iranian Journal of Chemical Engineering(IJChE)
1 systems and to enhance the operational flexibility and coefficient of performance.
2
3 After reviewing past research, the following points caught our interest:
4
5 ✓ One-dimensional mathematics was conducted in past research to model ejectors.
6 ✓ The efficiency of steam ejectors in refrigeration systems was optimized in past studies that
7
8 were limited to specific operating conditions.
9
10 ✓ Past research merely investigated specific parts of the ejector geometry, such as the suction
11
12
part.
13 ✓ The technical and economic effects of using ejectors instead of other compressors in flare
14
15 gas recovery were investigated.
16
The current research investigates the use of gas-liquid ejectors in flare gas recovery through
Fo

17
18
computational fluid dynamics simulations. We optimized the throat diameter and length, as well
19
20 as the nozzle diameter and angles (convergent and divergent) using a multi-objective genetic
rP

21
22 algorithm to achieve maximum entrainment rate of the secondary fluid.
23
24
ee

25 Simulation using computational fluid dynamics

26
27 In a two-phase (liquid-gas) ejector, a high-velocity liquid enters the ejector through the primary
rR

28
29 inlet and generates a low-pressure region in the throat. The pressure gradient between the nozzle
30
31
tip and the inlet section of the secondary fluid causes it to generate the force necessary to absorb
ev

32 gas (secondary fluid) [27]. Researchers frequently use Euler-Lagrange and Euler-Euler approaches
33
34 to model two-phase flows. In the second method, all phases are considered continuous [28, 29].
ie

35
36 Yadav and Patwardhan [27] proposed an Euler-Eulerian (mixed model) approach to simulate the
37
liquid-gas ejector. The governing equations for transient multiphase flow are as follows [30-33].
w

38
39 The mass conservation equation (continuity) for each phase is given below:
40
41 𝜕
(𝛼 𝜌 ) + ∇. (𝛼𝑘 𝜌𝑘 𝑢𝑘 ) = 𝑆𝑘 (1)
42 𝜕𝑡 𝑘 𝑘
43
44
45
46 Conservation of momentum for each phase:
47
48 𝜕
(𝛼 𝜌 𝑢 ) + ∇. (𝛼𝑘 𝜌𝑘 𝑢𝑘 𝑢𝑘 ) (2)
49 𝜕𝑡 𝑘 𝑘 𝑘
50 = −𝛼𝑘 ∇𝑝𝑘 + ∇. (𝛼𝑘 𝜇𝑘 (∇𝑢𝑘 + (∇𝑢𝑘 )𝑇 )) + 𝛼𝑘 𝜌𝑘 𝑔 + 𝑀𝑘
51
52
53 where 𝜌𝑘 , 𝛼𝑘 , and 𝑢𝑘 represent density, volume fraction, and phase velocity, respectively.
54
55 𝑆𝑘 indicates the rate phase production between different phases. The parameters, 𝑝𝑘 , 𝜇𝑘 , and 𝑔
56
57 express the pressure, viscosity, and the gravitational acceleration of the phase 𝑘, respectively. The
58 𝑀𝑘 is the sum of surface forces including lifting, drag, turbulence dispersion, wall lubrication, and
59
60

Page 9 of 22
 virtual mass. Virtual mass is negligible compared to other forces for multiphase flows in stable
Iranian Journal of Chemical Engineering(IJChE)
1 conditions [34]. Therefore, in the present work, the effects of this force are neglected.
2
3 By considering the continuity and momentum equations for all phases, the mixture
4
5 momentum conservation is obtained below:
6 𝜕 (3)
7 (𝜌 𝑢 ) + ∇. (𝜌𝑚 𝑢𝑚 𝑢𝑚 ) = −∇𝑝𝑚 + ∇. (𝜇𝑚 (∇𝑢𝑚 + (∇𝑢𝑚 )𝑇 )) + 𝜌𝑚 𝑔 + 𝑀𝑚
8 𝜕𝑡 𝑚 𝑚
9
10 where the subscript 𝑚 indicates the quantity related to the mixture. The secondary velocity (s)
11
12 corresponding to the primary phase (p), also known as the relative velocity or slip to face, is
13
14 defined [27, 35].
15
16 𝑢𝑠𝑝 = 𝑢𝑠 − 𝑢𝑝 (4)
Fo

17
18
19 In this research, Fluent software is used to simulate the two-phase flow in the ejector. In
20
rP

21 this software, the sliding speed provided by Manninen et al. [36] is employed here.
22
𝑡𝑠 𝜌𝑠 − 𝜌𝑚 (5)
23 𝑢𝑠𝑝 = ( ) 𝑎⃗
24 𝐹𝑑 𝜌𝑠
ee

25
26
27 The drag force (𝐹𝑑 ) can be calculated through the following equation [21, 29].
rR

28
29 1 + 0.15 (Re)0.687 , Re ≤ 1000 (6)
𝐹𝑑 = {
30 0.0183 , Re > 1000
31
ev

32
33 where Re represents the Reynolds number. This equation, which utilizes user-defined functions
34
and is introduced to Fluent software, takes the form of DEFINE_VECTOR_EXCHANGE_
ie

35
36
PROPERTY.
37
w

38
39
40 Verification of computational fluid dynamics results
41
42 To validate the calculation results of the fluid dynamics simulation, the geometry of the
43 experimental model proposed by Bhutada and Pangarkar [37] was selected. In their model, water
44
45 is the primary phase while air is treated as the secondary fluid. The validation results are for a case
46
47 where the diameter of the throat and the area of the throat to the nozzle are equal to 16 and 4 mm,
48
49
respectively. The angles of the converging and diverging areas are 12 and 5 degrees, respectively.
50 Bhutada and Pangarkar [37] reported that the entrainment rate of secondary fluid (air) was equal
51
52 to 0.00294 kg/s in these geometrical conditions.
53
54 Due to the axial symmetry of the ejector's geometry, a two-dimensional axial symmetry
55
model was used. Air (with a diameter of one millimeter) was assumed to be the primary and water
56
57 was the secondary phase of the ejector, respectively. The following boundary conditions were
58
59 used: (1) The water inlet velocity was set to 21.2 m/s; (2) The relative pressure at the air inlet was
60
set to zero using a pressure gauge; (3) The gauge pressure at the ejector outlet was set to zero; and

Page 10 of 22
 (4) Non-slip condition was maintained on the walls of both the ejector and the nozzle. "Coupled"
Iranian Journal of Chemical Engineering(IJChE)
1 models and "PRESTO" were used for pressure-velocity and discrete pressure connections,
2
3 respectively. The second-order upwind model was applied to determine the turbulence kinetic energy,
4
5 discrete momentum, and turbulence dissipation rate. The under-relaxation coefficient of 0.1 was
6 considered for pressure, density, momentum, sliding speed, and volume fraction.
7
8 Due to the disparity in CFD results, different grid sizes were generated and implemented
9
10 in the domain. From the grid with a size of 0.25 mm, the results experienced better convergence
11
12
and less fluctuations. Therefore, this network is used in the following sections of this study. The
13 flow inside the ejectors due to the high Reynolds Number is turbulent. Many studies have adopted
14
15 different approaches to flow modeling turbulence in two-phase ejectors [23, 26-28, 37-39]. In the
16
present research, the "Realizable -ε" method was employed to model the turbulent flow.
Fo

17
18
After performing calculations, it was found that the absorption rate of the secondary fluid
19
20 (air) was 0.002863 kg/s, which differs from the experimental data of Ref. [37] by 2.6%. However,
rP

21
22 this level of error is considered acceptable for numerical results.
23
24
ee

25 Multi-objective genetic algorithm

26
27 The multi-objective genetic algorithm is a search algorithm that is used to optimize multiple
rR

28
29 objectives simultaneously. It is a powerful optimization tool that utilizes genetic operators to
30
31
accurately explore the entire solution space. The Genetic Algorithm identifies the parameters
ev

32 needed to achieve the best results, or optimal mode. The remarkable feature of the Genetic
33
34 Algorithm is its ability to avoid converging to local optima, thereby finding the overall optimal
ie

35
36 point in the entire domain. Furthermore, the Genetic Algorithm can optimize more than one
37
parameter in a single problem, as reported in [40].
w

38
39 In the classical genetic algorithm, if a problem has multiple objective functions, the user
40
41 must evaluate their importance by assigning weights to each of them. However, with the multi-
42
43 objective genetic algorithm, a new approach is introduced to evaluate candidate parameters
44 without the need to define relative importance coefficients. This provides a more efficient and
45
46 accurate optimization method for multi-objective problems [40-42].
47
48 The candidates in a multi-objective problem are Pareto optimal, meaning that there is a
49
50
chance of having both local and global optimal scores, just as in a single-objective problem. After
51 the first evaluation of a population, a set of solutions, or candidate points, is classified into different
52
53 non-dominant levels, creating the initial and best non-dominant set population. This set is
54
55 evaluated once again, and the non-dominated solutions form the second level, which is the next
56 best. The same method is used again to identify non-dominant solutions of the third level, and this
57
58 process continues until all members are at a non-dominant level. The minimum state of a non-
59
60 dominated surface occurs when no solution dominates any other solution, and all candidates of the

Page 11 of 22
 initial population fall into one category. At most, N non-dominant levels arise when a hierarchy of
Iranian Journal of Chemical Engineering(IJChE)
1 mastery exists, where any solution is exactly dominated by another level in the set [42].
2
3
4
5 Results and discussion around them
6 The primary fluid is the liquid phase, consisting of amine, while the secondary fluid is the gas
7
8 phase, consisting of flare gas. Table 1 provides their respective physical properties. As depicted in
9
10 Figure 1, a specific geometry is considered as the base state.
11
12
13 Table 1. Properties of amine and flare gas
14
15 density heat capacity thermal conductivity Viscosity
16
Fo

17
18 𝜌 [kg/m3 ] 𝑐𝑃 [J/kg ∙ K] 𝑘 [W/m ∙ K] 𝜇 [Pa ∙ s]
19
20 amine 1064 3193 0.2557 0.0103
rP

21
22 flare gas ideal gas 1978 0.0509 1.228e-5
23
24
ee

25
26 The problem was modeled using the computational fluid dynamics method in Fluent
27
rR

28 software. The absorption rate of flare gas was calculated to be 0.0024846 kg/s. Figure 2 shows the
29
30 axial velocity for flare gas and amine, as well as the turbulence intensity of the mixture in the
31
ejector, as depicted by the lines.
ev

32
33
34
ie

35
36
37
w

38
39
40
41
42
43
44
45 Figure 1. Geometrical parameters of the base case
46
47
48
49 The objective of this study is to achieve an optimal design of the ejector geometry and
50 improve the absorption rate of the secondary fluid (flare gas). The design parameters to be
51
52 optimized include the length and diameter of the throat, the diameter of the nozzle, the diameter
53
54 of the liquid inlet, and the converging and diverging section angles, as shown in Figure 3. Target
55
56
variables (absorption rate) and design are parametrically defined in Fluent software. Design
57 variables cannot take any arbitrary value. For instance, the radius of the nozzle should not exceed
58
59 the radius of the inlet. This leads to the creation of a constrained optimization problem, which can
60
be tackled using the investigated multi-objective genetics algorithm.

Page 12 of 22
 Iranian Journal of Chemical Engineering(IJChE)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Fo

17
18
19
20
rP

21
22
23
24
ee

25
26
27
rR

28
29
30
31
ev

32
33 Figure 2. a) Flare gas; b) amine axial velocity contours; and c) the turbulence intensity of the mixture
34
ie

35
36
37
w

38
39
40
41
42
43
44
45
46 Figure 3. Design parameters of the two-phase ejector
47
48
49
50 In meta-heuristic optimization algorithms, an acceptable range of variation should be
51
52 considered for each design parameter. The researchers have reported values for the nozzle and
53 throat geometry, as well as the converging and diverging angles, in which the absorption rate is
54
55 maximum. Based on these values, the range of changes for each design parameter was determined
56
57 as follows:
58
59
57.6 mm < throat length < 70.4 mm
60

Page 13 of 22
 3.6 mm < throat radius < 20.0 mm
Iranian Journal of Chemical Engineering(IJChE)
1
2 1.8 mm < nozzle radius < 8.8 mm
3
4 2.7 mm < liquid inlet radius < 14.4 mm
5
6 152 < converging angle < 170
7
8
9 170 < diverging angle < 178
10
11
12
13 The constants of the genetic algorithm are set, as shown in Table 2. After 111 optimization
14
points, the design has converged, and the design parameters have been optimized, resulting in the
15
16 determination of three candidate solutions. The results are shown in Table 3 .
Fo

17
18
19
Table 2. The constants of the multi-objective genetic algorithm
20
rP

21
22
The constant parameter value
23
24 Estimated number of design points 1050
ee

25
26 Population size 100
27
rR

28
29
Number of samples per iteration 50
30
31 Minimum allowed percentage of Pareto 70
ev

32
33 Convergence stability percentage 2
34
ie

35 Maximum of iteration 20
36
37
w

38 Number of selected points after optimization 3

39
40
41
42 Figure 4 illustrates the schematic of the candidate design point compared to the base case.
43
The results indicate that increasing the radius of the throat and the angle of the divergent part, as
44
45 well as decreasing the angle of the converging part, lead to an increase in the absorption rate of
46
47 gas. Figures 5 and 6 show the axial velocity of the flare gas and liquid, respectively. As depicted
48
49 in Figure 5, the velocity of the secondary fluid is significantly higher in all candidate points
50 compared to the base case. The first design point exhibits a higher velocity than the other candidate
51
52 points. Moreover, the results presented in Figure 6 confirm that all candidate points enhance the
53
54 axial velocity of the amine compared to the base state.
55 Table 3. The geometry of optimal points and the entrainment of the flare gas
56
57
58
The geometry parameter base case 1st point 2nd point 3th point
59
60 throat length, 𝐿𝑇 [mm] 64.0 62.99 58.56 63.48

Page 14 of 22
 radius, 𝑅Journal
throatIranian 𝑇 [mm] of Chemical
8.0 15.29 16.44
Engineering(IJChE) 18.88
1
2 nozzle radius, 𝑅𝑁 [mm] 4.0 4.26 4.66 6.85
3
4 diverging angle (𝜑) 168 176.99 177.38 177.63
5
6
7
converging angle (𝜃) 175 162.79 165.63 153.24
8
9 liquid inlet radius, 𝑅𝑊 [mm] 5.5 14.23 11.33 13.26
10
11 entrainment rate [kg/s] 0.002485 0.06340 0.03991 0.03172
12
13
14
15
16
Fo

17
18
19
20
rP

21
22
23
24
ee

25
26
27
rR

28
29
30
31
ev

32
33
34
ie

35
36
37
w

38
39
40
41
42
43
44 Figure 4. The geometry of three optimal candidates: (a) base model, (b) 1st candidate, (c) 2nd candidate, and
45
46 (d) 3th optimal geometry
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Page 15 of 22
 Iranian Journal of Chemical Engineering(IJChE)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Fo

17
18
19 Figure 5. The axial velocity of the flare gas: (a) base model, (b) the 1st, (c) the 2nd, and (d) the 3th optimal
20
rP

21 points
22
23
24
ee

25
26
27
rR

28
29
30
31
ev

32
33
34
ie

35
36
37
w

38
39
40
41
42 Figure 6. The axial velocity of amine: (a) base model, (b) the 1st, (c) the 2nd, and (d) the 3th optimal points
43
44
45
46 Among the candidate design points, the first point exhibits a significantly higher axial
47
48 velocity for the liquid phase compared to the other modes. On the other hand, the second and third
49 design points have the same performance in terms of the axial velocity of the liquid phase. The
50
51 turbulence intensity of the mixture inside the ejector is depicted in Figure 7, which shows the base
52
53 mode and the three design points. The first design point shows a higher turbulence intensity
54
compared to the other two design points. However, the turbulence intensity in the second and third
55
56 design points is comparable.
57
58
59
60

Page 16 of 22
 Iranian Journal of Chemical Engineering(IJChE)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Fo

17
18
19 Figure 7. The turbulence intensity of the mixture: (a) base model, (b) the 1st, (c) the 2nd, and (d) the 3th
20
rP

optimal points
21
22
23
24 Ejector energy efficiency
ee

25
26 Based on the obtained results, the first candidate point appears to be the most suitable choice for
27
rR

28 ejector design due to its higher axial velocities of both phases and turbulence intensity. However,
29
30
it is essential to calculate and evaluate the energy efficiency of the ejector as these increases may
31 come at the cost of higher energy consumption, reducing productivity. The energy efficiency of
ev

32
33 the ejector can be calculated using the following equation [37].
34
𝑚̇𝑔 Δ𝑃𝑔
ie

35 𝜂=( )( ) (13)
36 𝑚̇𝑙 Δ𝑃𝑙
37
w

38
39
where Δ𝑃𝑔 represents the gas recovery pressure from the suction to the diffuser outlet, and Δ𝑃𝑙 is
40
41
the liquid pressure drop from the nozzle outlet to the output of the ejector, measured in pascals.
42
The parameters 𝑚̇𝑔 and 𝑚̇𝑙 represent the flow rate of gas and liquid, respectively, and they are
43
44
45
measured in cubic meters per second.
46 The efficiency of the ejector was calculated using equation (13) for the base model and the
47
48 three design points, and the results are presented in Table 4. The second design point exhibited the
49
50 highest energy efficiency, with an improvement of approximately 3.5 times compared to the base
51 model. The first design point can increase the ejector efficiency up to 78 percent. In contrast, the
52
53 third design point decreased the energy efficiency by 47 percent.
54
55 Table 4. The ejector efficiencies of the three optimal points compared to the base case
56
57 Optimal Point
58 The base case 1st
2nd 3th
59
60

Page 17 of 22
 The energy efficiency
Iranian Journal of 17.79 31.73 60.21
Chemical Engineering(IJChE) 9.40
1
2 percentage of change +78.36% +238.42% -47.16%
3
4
5
6 Conclusion
7
8 This study investigated a two-phase ejector (liquid-gas) and employed Euler-Euler approach and
9 mixed method to simulate it. Then, the flow of two-phase fluids was modeled using Fluent
10
11 software. The "Coupled" and "PRESTO" models were used for pressure-velocity connection and
12
13 discrete pressures, while a second-order upwind model was used for turbulent kinetic energy,
14
15
discrete momentum, and rate perturbation loss. The domain was gridded using a structured
16 quadrilateral grid. Different grid sizes were achieved due to disparity in the results of
Fo

17
18 computational fluid dynamics simulations. In the current study, a numerical solution was obtained
19
20 using a 0.25 mm grid, resulting in better convergence and fewer fluctuations in the results. The
rP

21 "Realizable" method was used to model turbulent flow in this research. The primary and secondary
22
23 fluids considered were amine and flare gas, respectively. The primary aim of this study is to design
24
ee

25 ejector geometry that maximizes flare gas absorption. To achieve this goal, several design
26
27
parameters must be optimized, including throat length and diameter, nozzle diameter, liquid inlet
rR

28 diameter, and converging and diverging section angles. In order to define the target variables (i.e.,
29
30 addition rate) and design, the researchers employed Fluent software and parametric analysis. To
31
optimize the ejector geometry, a multi-objective genetic algorithm was utilized. After reaching
ev

32
33
111 points in the convergent design stage, three candidate points were identified where the flare
34
ie

35 gas absorption rate exceeded the base case. The optimization process revealed that decreasing the
36
37 throat length and angle of the converging section resulted in an increased absorption rate of the
w

38
39 secondary fluid. Additionally, increasing the nozzle diameter led to an increase in the throat
40 diameter and angle of the divergent section, resulting in a higher gas absorption rate. The design
41
42 outcomes yielded absorption rates for the secondary fluid of 25.5, 16.06, and 12.76 times,
43
44 respectively, compared to the base case. The energy efficiency of the ejector was calculated for
45
46
both the base model and all design points. The second design point exhibited the best performance,
47 improving energy efficiency by approximately 3.5 times when compared to the base case. The first
48
49 design point demonstrated potential for improving energy efficiency by up to 78%, whereas the
50
51 third candidate resulted in a 47% decrease in energy efficiency.
52
53
54 Acknowledgment
55
56 This manuscript is prepared based on PhD thesis of the first author at Borujerd Branch, Islamic
57
58 Azad University, Borujerd, Iran.
59
60

Page 18 of 22
 Iranian Journal of Chemical Engineering(IJChE)
1
2
3
4
5 References
6
7
8 1. Emam, E.A., GAS FLARING IN INDUSTRY: AN OVERVIEW. Petroleum & coal, 2015.
9 57(5).
10 2. Soltanieh, M., et al., A review of global gas flaring and venting and impact on the
11
12
environment: Case study of Iran. International Journal of Greenhouse Gas Control, 2016.
13 49: p. 488-509.
14 3. Davoudi, M., et al., The major sources of gas flaring and air contamination in the natural
15 gas processing plants: A case study. Journal of Natural Gas Science and Engineering, 2013.
16 13: p. 7-19.
Fo

17 4. Elhagar, M., et al., Increase flared gas recovery and emission reduction by separator
18
optimization. International Journal of Energy and Environmental Engineering, 2021. 12: p.
19
20 115-130.
rP

21 5. Rahimpour, M., et al., A comparative study of three different methods for flare gas recovery
22 of Asalooye Gas Refinery. Journal of Natural Gas Science and Engineering, 2012. 4: p. 17-
23 28.
24 6. Sangsaraki, M.E. and E. Anajafi. Design criteria and simulation of flare gas recovery
ee

25 system. in Proceedings of the International Conference on Chemical, Food and

26
27
Environment Engineering (ICCFEE’15), Dubai, UAE. 2015.
7. Zolfaghari, M., V. Pirouzfar, and H. Sakhaeinia, Technical characterization and economic
rR

28
29 evaluation of recovery of flare gas in various gas-processing plants. Energy, 2017. 124: p.
30 481-491.
31 8. Hajizadeh, A., et al., Technical and economic evaluation of flare gas recovery in a giant
ev

32 gas refinery. Chemical Engineering Research and Design, 2018. 131: p. 506-519.
33 9. Hamidzadeh, Z., et al., Development of a multi-objective decision-making model to recover
34
flare gases in a multi flare gases zone. Energy, 2020. 203: p. 117815.
ie

35
36 10. Mousavi, S.M., et al., Technical, economic, and environmental assessment of flare gas
37 recovery system: a case study. Energy Sources, Part A: Recovery, Utilization, and
w

38 Environmental Effects, 2020: p. 1-13.

39 11. Xue, M., et al. A Pilot Demonstration of Flaring Gas Recovery During Shale Gas Well
40 Completion in China. in International Petroleum Technology Conference. 2022. OnePetro.
41
12. Khalili-Garakani, A., M. Nezhadfard, and M. Iravaninia, Enviro-economic investigation of
42
43 various flare gas recovery and utilization technologies in upstream and downstream of oil
44 and gas industries. Journal of Cleaner Production, 2022. 346: p. 131218.
45 13. Zaresharif, M., A. Vatani, and M. Ghasemian, Evaluation of different flare gas recovery
46 alternatives with exergy and exergoeconomic analyses. Arabian Journal for Science and
47 Engineering, 2021: p. 1-20.
48 14. Leagas, T., G. Seefeldt, and D. Hoon. Ejector Technology for efficient and cost effective
49
50
flare gas recovery. in Proc. GPA-GCC 24 th Annual Technical Conference, Kuwait City
51 [online], May. 2016.
52 15. Ainge, P. Ejector Technology for Flare Gas Recovery as an Alternative to Rotating
53 Equipment. in Abu Dhabi International Petroleum Exhibition & Conference. 2019.
54 OnePetro.
55 16. Mazumder, M., L. Chen, and Q. Xu, Integrated Ejector‐Based Flare Gas Recovery and
56
On‐Site Desalination of Produced Water in Shale Gas Production. Chemical Engineering
57
58 & Technology, 2020. 43(2): p. 200-210.
59 17. Eshaghi, S. and F. Hamrang, An innovative techno-economic analysis for the selection of
60 an integrated ejector system in the flare gas recovery of a refinery plant. Energy, 2021.
228: p. 120594.

Page 19 of 22
 18. Bashiri, H., O. Ashrafi, and P. Navarri, Energy-efficient process designs of integrated
Iranian Journal of Chemical Engineering(IJChE)
1 biomass-derived syngas purification processes via waste heat recovery and integration of
2 ejector technology. International Journal of Greenhouse Gas Control, 2021. 106: p.
3 103259.
4 19. Ainge, P. and E. Al-Khateeb. Selection and Evaluation of Ejectors for Flare Gas Recovery
5 and Upstream/Midstream Gas Compression Applications. in International Petroleum
6 Technology Conference. 2022. OnePetro.
7
8
20. Galanis, N. and M. Sorin, Ejector design and performance prediction. International Journal
9 of Thermal Sciences, 2016. 104: p. 315-329.
10 21. Wang, L., et al., Numerical study on optimization of ejector primary nozzle geometries.
11 International Journal of Refrigeration, 2017. 76: p. 219-229.
12 22. Śmierciew, K., J. Gagan, and D. Butrymowicz, Application of numerical modelling for
13 design and improvement of performance of gas ejector. Applied Thermal Engineering,
14
2019. 149: p. 85-93.
15
16
23. Suvarnakuta, N., et al., Performance analysis of a two-stage ejector in an ejector
refrigeration system using computational fluid dynamics. Engineering Applications of
Fo

17
18 Computational Fluid Mechanics, 2020. 14(1): p. 669-682.
19 24. Li, H., et al., Numerical investigation of the nozzle expansion state and its effect on the
20 performance of the steam ejector based on ideal gas model. Applied Thermal Engineering,
rP

21 2021. 199: p. 117509.

22
25. Xu, J., X. Liu, and M. Pang, Numerical and experimental studies on transport properties
23
24 of powder ejector based on double venturi effect. Vacuum, 2016. 134: p. 92-98.
ee

25 26. Sriveerakul, T., S. Aphornratana, and K. Chunnanond, Performance prediction of steam

26 ejector using computational fluid dynamics: Part 1. Validation of the CFD results.
27 International Journal of Thermal Sciences, 2007. 46(8): p. 812-822.
rR

28 27. Yadav, R.L. and A.W. Patwardhan, Design aspects of ejectors: Effects of suction chamber
29 geometry. Chemical Engineering Science, 2008. 63(15): p. 3886-3897.
30
31
28. Ariyaratne, W.H., et al. CFD approaches for modeling gas-solids multiphase flows–A
review. in Proceedings of The 9th EUROSIM Congress on Modelling and Simulation,
ev

32
33 EUROSIM 2016, The 57th SIMS Conference on Simulation and Modelling SIMS 2016.
34 2018. Linköping University Electronic Press.
ie

35 29. Suh, J.-W., et al., Development of numerical Eulerian-Eulerian models for simulating
36 multiphase pumps. Journal of Petroleum Science and Engineering, 2018. 162: p. 588-601.
37
30. Song, X., et al., Numerical investigation of a liquid-gas ejector used for shipping ballast
w

38
39
water treatment. Mathematical Problems in Engineering, 2014. 2014.
40 31. Alshebani, M.E., et al., Numerical Investigation of a Liquid-Gas Ejector in Marine Ships.
41 International Journal of Applied Engineering Research, 2017. 12(14): p. 4663-4674.
42 32. Parekh, J. and R. Rzehak, Euler–Euler multiphase CFD-simulation with full Reynolds
43 stress model and anisotropic bubble-induced turbulence. International Journal of
44 Multiphase Flow, 2018. 99: p. 231-245.
45
33. Kwiatoń, P., et al., Multiphase Modeling of Droplet Fall using Euler-Euler Approach. Acta
46
47 Physica Polonica A, 2020. 138(2): p. 276-279.
48 34. Frank, T., et al., Validation of CFD models for mono-and polydisperse air–water two-
49 phase flows in pipes. Nuclear Engineering and Design, 2008. 238(3): p. 647-659.
50 35. Selvakumar, R.D. and S. Dhinakaran, Forced convective heat transfer of nanofluids
51 around a circular bluff body with the effects of slip velocity using a multi-phase mixture
52 model. International Journal of Heat and Mass Transfer, 2017. 106: p. 816-828.
53
54
36. Manninen, M., V. Taivassalo, and S. Kallio, On the mixture model for multiphase flow.
55 1996.
56 37. Bhutada, S. and V. Pangarkar, Gas induction and hold-up characteristics of liquid jet loop
57 reactors. Chemical Engineering Communications, 1987. 61(1-6): p. 239-258.
58 38. Ringstad, K.E., et al., A detailed review on CO2 two-phase ejector flow modeling. Thermal
59 Science and Engineering Progress, 2020. 20: p. 100647.
60

Page 20 of 22
 39. de Oliveira Marum, V.J., et al., Performance analysis of a water ejector using
Iranian Journal of Chemical Engineering(IJChE)
1 Computational Fluid Dynamics (CFD) simulations and mathematical modeling. Energy,
2 2021. 220: p. 119779.
3 40. Zolpakar, N.A., et al., Application of multi-objective genetic algorithm (MOGA)
4 optimization in machining processes. Optimization of manufacturing processes, 2020: p.
5 185-199.
6 41. Murata, T. and H. Ishibuchi. MOGA: multi-objective genetic algorithms. in IEEE
7
8
international conference on evolutionary computation. 1995. IEEE Piscataway, NJ, USA.
9 42. Deb, K., Multi-objective genetic algorithms: Problem difficulties and construction of test
10 problems. Evolutionary computation, 1999. 7(3): p. 205-230.
11
12
13
14
15
16
Fo

17
18
19
20
rP

21
22
23
24
ee

25
26
27
rR

28
29
30
31
ev

32
33
34
ie

35
36
37
w

38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Page 21 of 22
 Iranian Journal of Chemical Engineering(IJChE)
1
2
3
4
5
6 Research Highlights
7
8 1- Flare Gas Recovery: The significance of flare gas recovery is remarkably high, considering
9 the harmful environmental consequences associated with burning gases dispatched to flare and
10
the gradual depletion of valuable fossil fuel reserves.
11
12
13
14 2- Using ejectors for flare gas recovery: The significant importance of ejectors in the process of
15
16
flare gas recovery stems from their ease of use and lower costs of operation and maintenance,
particularly when compared to other compressor types
Fo

17
18
19
20
rP

21 3- Using liquid amine as a motive fluid with an ejector: The importance of using liquid amine as
22 a motive fluid to eliminate sour gas components (like H2S) in operational-refining units lies in
23 the necessity to remove toxic and non-reusable materials. However, using water vapor has
24 drawbacks, such as high wear and tear costs and the failure to effectively separate harmful
ee

25
26
substances at the end of the recovery process.
27
rR

28
29 4- Optimization of important geometric parameter sizes of the ejector: Optimizing parameters
30
31 such as the nozzle radius, throat radius, as well as convergent and divergent angles of the nozzle
ev

32 are crucial as they play a significant role in maximizing the secondary gas intake.
33
34
ie

35
36 5- Utilizing Computational Fluid Dynamics: Given the high costs and frequency of testing
37 involved in optimizing the geometric parameters of the ejector, the use of CFD holds substantial
w

38 significance.
39
40
41
42 6- Utilizing multi-objective genetic algorithms: To achieve maximum suction flow, it is crucial
43 to employ multi-objective genetic algorithms to simultaneously optimize the important geometric
44
parameters of the ejector.
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Page 22 of 22