International Journal of ISSI, Vol. 21(2024), No. 1, pp.

75-84

Journal Homepage: https://journal.issiran.com/

Research Article

Enhancing the Production Rate of Mobarakeh MIDREX Direct Reduction Mega

Module Unit Using Numerical Simulation and Geometric Parameter Study of its
Ejector Stack
M. Darbandi *1, M. Fazeli 2, D. Rashidi 3, B. Basiriani 4, S. M. Hosseinian 5, M. Allah Dadi 6

1
Faculty of Aerospace EngineeringSharif, Sharif University of Technology, Tehran, Iran
Esfahan’s Mobarakeh Steel Company, Esfahan, Iran
2, 3, 4, 5, 6

ARTICLE INFO ABSTRACT

Keywords: This study uses the numerical simulation to successfully improve the performance
Ejector, Stack, Nozzle, of a subsonic ejector stack, which is used for the expulsion of flue gas in a Direct
Computational Fluid Dy- Reduction Iron unit located in Esfahan’s Mobarakeh Steel Company. The current
namics, Sensitivity Analysis, extended computational fluid dynamics (CFD) software uses the finite-volume
Geometrical Parameters. method to solve the governing equations representing the incompressible turbulent
flow through the ejector stack. The study specifically focuses on evaluating the
influence of key geometric parameters, such as the mixing chamber length, diffuser
length, and the nozzle ejector and diffuser angles, on enhancing the deteriorated
Article history: performance of the ejector. The performed simulations show that the mixing cham-
Received 20 May 2024 ber length cannot be treated as a good geometry parameter to enhance the present
Received in revised form 06 performance. As another important parameter, it is shown that the diffuser length
September 2024 can reliably enhance the ejector performance. It needs about 38% increases in this
Accepted 26 November 2024 length to achieve an optimum performance value. Examining the convergence and
divergence angles of the nozzle and diffuser of the ejector, respectively, it is shown
that the optimal angles are 4.9 degrees for the divergence part and 12.5 degrees
for the convergence part. Indeed, this study provides a novel and fundamental
approach, which helps the industry people to improve the efficiency of their
defective ejector stack systems.

1. Introduction electrical, and chemical sectors. Notably, the ejector sys-

tem is free from any moving part, which consequently
Indeed, the ejector is a device with so many appli- leads to a lower maintenance cost [1]. There are many
cations across various industries including the steel, different parameters, which can affect the performance
of an ejector stack, of which the geometric parameters
emerge as the most impactful ones. Contrary to the nu-
* Corresponding Author merous studies that have been performed on the super-
Email: [email protected] sonic ejectors [2-7], there has been less research to pay
Address: Faculty of Aerospace EngineeringSharif, Sharif attention to the effect of stack geometry parameters on
University of Technology, Tehran, Iran the subsonic ejector performances. Among the leading
1. Professor, 2. M.S., 3. M.S., 4. M.S., 5. M.S., 6. M.S.
works, Watanabe [8] experimentally concluded that the
DOI: http://10.22034/IJISSI.2024.2029716.1293 maximum ejector performance would occur when the
Published by ISSI (Iron & Steel Society of Iran) distance from the nozzle to the mixing chamber inlet was

75
 M. Darbandi et al. / International Journal of ISSI, Vol. 21(2024), No. 1, 75-84

2.56 times the diameter. Additionally, he found that the numerical modeling of air ejectors covering supersonic,
optimal ratio of mixing chamber length to the diameter subsonic, and closed-port operations. Kumar, et al. [24]
would be between 5.8 to 6.0. Vias and Kar [9] experi- have provided a detailed review of the research made in
mentally investigated the impact of various geometrical the field of ejector systems including the design meth-
parameters on the ejector’s performance. They showed odology, geometrical parameters, operating parameters
that the length of the mixing chamber should exceed effect, CFD studies, turbulence model selection, working
44 times the ejector nozzle diameter for optimal perfor- fluid, and irreversibility of the ejector system.
mance. However, this has been corrected by the next re- The primary objective of this study is to investigate
searchers including the present study. an ejector stack for expelling the hot flue gas generated
The researchers in the past twenty years have used by a Direct Reduced Iron (DRI) unit into the atmosphere.
both experimental approaches and numerical tools to The DRI is located in Esfahan’s Mobarakeh Steel Com-
evaluate the impacts of geometric and applied parameters pany. Therefore, it is very crucial to explore how the geo-
on the resulting ejectors’ performances. Riffat and Omer metrical parameters of the ejector stack would influence
[10] explored the nozzle's position and its consequential its overall performance and efficiency. The DRI Unit
effect on the ejector system's performance. They report- indicates that the ejector stack does not perform normal-
ed that the substantial impact of the nozzle's location on ly and that it performs much less than its in-design per-
the ejector system's performance would be attributed to formance. Therefore, the current research explores the
its influence on the length of the primary and secondary impact of geometric parameters, namely mixing chamber
fluid mixing zones. Riffat, et al. [11] have presented a length, diffuser length, diffuser divergence angle, ejector
complete review of the ejector technology development. nozzle angle, and mixing entry nozzle angle, on enhanc-
Yadav and Patwardhan [12] investigated the influence of ing the performance of the available subsonic ejector
parameters such as the suction chamber diameter, diver- stack. This is the simplest way to improve the perfor-
gence angle, and nozzle position on the ejector system's mance of an already defective ejector stack.
performance. Meakhail, et al. [1] assumed a constant ra-
tio of diffuser exit diameter to mixing chamber diameter 2. Introducing the Present Ejector Stack
and investigated the effects of variations in geometrical As known, the ejector stack can play a crucial role
parameters such as the number of inlets, nozzle location, in enhancing the efficiency and performance of indus-
length of the mixing chamber, and the divergence angle trial exhaust systems. Fig. 1. presents a general over-
of the diffuser section on the achieved performances. view of an ejector stack's configuration. As illustrated in
Zhang, et al. [13] conducted a careful research to explore this figure, the motive air enters the nozzle to expel the
the nozzle location's effect on the ejector system's per- secondary flow, which is the flue gas in the present
formance, utilizing both numerical simulation results and work. Additionally, the stack comprises four other es-
laboratory data. Their findings indicated that the optimal sential components including a suction chamber, a throat
geometric composition of the ejector would vary under or mixing entry, a mixing chamber, and a diffuser. The
different operating conditions. Li, et al. [14] investigat- main objective of this ejector is to generate a suction
ed the configuration dependence and the optimization of force within the primary process and facilitate the ex-
the entrainment performance for the gas–gas and gas– traction of flue gases from the original gas storage such
liquid ejectors. Yang, et al. [15] performed a numerical as a combustion chamber. The ejector stack system con-
investigation to study the mixing process in a steam ejec- sists of a fan that propels air into the stack and an ejector
tor with different nozzle structures. Kong and Kim [16] nozzle, which converts the air into a high-speed jet. The
studied the performance of a two-stage ejector–diffuser high-speed jet generated by the ejector nozzle induces a
system both analytically and computationally. Wang, et pressure reduction precisely at the nozzle outlet, resulting
al. [17] used numerical tools to optimize the ejector pri- in a substantial pressure gradient between the inlet of the
mary nozzle geometries. Chen, et al. [18] used numerical flue gas and the interior of the stack. This pressure gradi-
methods and enhanced the ejector performance via using ent sucks the flue gas and enhances the mixing of gases
the combined adjustable geometry and bypass methods. within the stack. Subsequently, the mixed gas traverses a
Tashtoush, et al. [19] have presented a comprehensive diffuser, where its kinetic energy undergoes conversion
review of the ejector design, performance, and applica- into pressure energy and causes pressure recovery before
tions. Yan, et al. [20] performed numerical investigations being discharged into the atmosphere.
to optimize the ejector primary nozzle geometry with As said previously, the primary objective of the
fixed/varied nozzle exit position. Hadi, et al. [21] used present work is to explore the impact of geometric pa-
the CFD method to optimize the hydrocarbon ejector. rameters, namely mixing chamber length (LMC), diffuser
Tavakoli, et al. [22] used numerical approaches and in- length (R), diffuser divergence angle (θ), ejector nozzle
corporated a fluidic oscillator as the primary nozzle and angle (β), and mixing entry nozzle angle (α), on enhanc-
enhanced the performance of a subsonic ejector. Schil- ing the performance of present DRI ejector stack using
laci, et al. [23] presented a detailed investigation of the the CFD. It should be noted that the internal diameter of

76
 M. Darbandi et al. / International Journal of ISSI, Vol. 21(2024), No. 1, 75-84

the refractory has been carefully taken into account in all the Navier-Stokes equations, can be expressed as
the present calculations and simulations. As was said pre-
viously, the current ejector stack is to expel the hot flue 𝛻𝛻. 𝑉𝑉 = 0 Eq.(1)
gas generated by a DRI Unit into the atmosphere. The
DRI Unit is located in Esfahan’s Mobarakeh Steel Com- (𝑉𝑉. 𝛻𝛻)𝑉𝑉 = −𝛻𝛻𝛻𝛻 + 𝜇𝜇𝛻𝛻 2 𝑉𝑉 + 𝛽𝛽 Eq.(2)
pany. This ejector is designed for a discharge of 327,190
(Nm3/h) of flue gas, which is fed from two separate
inlets, into the atmosphere. Eq.(3)
|𝑉𝑉 |2 𝑉𝑉
2.1. The Governing Equations and the CFD 𝛻𝛻 ⋅ ( + e𝑉𝑉 + 𝑝𝑝𝑝𝑝 − 𝜏𝜏 ⋅ 𝑉𝑉 − 𝑞𝑞) = 0
2
Methodology
where V is the velocity vector, p the pressure, μ the
The CFD software is used to simulate the physics of dynamic viscosity, β the thermal body forces, e the total
flow through the ejector stack in expelling the flue gas, energy, τ the stress tensor, and q is the conduction heat
obtain its performance at design conditions, and evaluate transfer. Back to our past experiences, see Refs. [4, 6,
the variation of its performance with the variation of some 25-27], this research also employs the widely accepted
geometric parameters. The fundamental assumptions are standard k-ε turbulence model to reproduce the true tur-
that the turbulent flow through the ejector stack is steady, bulent airflow dynamics within the ejector stack. Darban-
incompressible, and viscous. The continuity, momentum, di et al. [25-27] have conducted an in-depth exploration
and energy conservation equations, commonly known as in this context. The standard k-epsilon model involves

Fig. 1. A typical configuration close to the current ejector stack.

77
 M. Darbandi et al. / International Journal of ISSI, Vol. 21(2024), No. 1, 75-84

two transport equations, of which one governs the tur- governing equations. References [4, 6, and 25-27] pro-
bulent kinetic energy (k) and the other one the turbulent vide details of computational modeling and the CFD
kinetic energy dissipation rate (ε). The conservation laws method. As is raised in these references, a pressure-based
for the turbulent kinetic energy and its dissipation rate solver is employed to solve the aforementioned steady-
are given by state governing equations. This solver integrates the
Eq.(4) energy equation and the species transport equation
to simulate the mixture of H2O, O2, N2, and CO2 with
𝜕𝜕(𝑘𝑘𝑢𝑢𝑖𝑖 ) 𝜕𝜕 𝜇𝜇𝑡𝑡 𝜕𝜕𝜕𝜕 varying volume fractions in distinct sections of the ejec-
= [ ] + 2𝜇𝜇𝑡𝑡 𝐸𝐸𝑖𝑖𝑖𝑖 𝐸𝐸𝑖𝑖𝑖𝑖 − 𝜀𝜀
𝜕𝜕𝑥𝑥𝑖𝑖 𝜕𝜕𝑥𝑥𝑖𝑖 𝜎𝜎𝑘𝑘 𝜕𝜕𝑥𝑥𝑖𝑖 tor stack. Moreover, the semi-implicit method for the
pressure-linked equations (SIMPLE) scheme is utilized
to couple the pressure and velocity fields suitably. As
𝜕𝜕(𝜀𝜀𝑢𝑢𝑖𝑖 ) 𝜕𝜕 𝜇𝜇𝑡𝑡 𝜕𝜕𝜕𝜕 𝜀𝜀 known, this scheme is a widely accepted approach for
= [ ] + 𝐶𝐶1𝜀𝜀 Eq.(5) the steady-state solution algorithms.
𝜕𝜕𝑥𝑥𝑖𝑖 𝜕𝜕𝑥𝑥𝑗𝑗 𝜎𝜎𝜀𝜀 𝜕𝜕𝑥𝑥𝑗𝑗 𝑘𝑘
2.2. Geometry, Meshing, and the Applied
𝜀𝜀 2 Boundary Conditions
2𝜇𝜇𝑡𝑡 𝐸𝐸𝑖𝑖𝑖𝑖 𝐸𝐸𝑖𝑖𝑖𝑖 − 𝐶𝐶2𝜀𝜀
𝑘𝑘
Fig. 2. illustrates the actual geometry of the ejector
The above equations are simply called in the k-ε tur- stack, which is under investigation in the current study.
bulence model. In these equations, ui is the mean velocity The present ejector stack has six separate parts, namely
component, μt the turbulent or eddy viscosity. The other the ejector nozzle serving as the inlet for the motive air
variables are treated as the k-ε turbulent model constants. propelled by a fan, a suction chamber, two inlets for the
These constants are given by Cμ= 0.09, σk=1.00, σε= secondary flow (or the flue gas), a mixing entry nozzle, a
1.30, C1ε= 1.44, and C2ε= 1.92, which have been derived mixing chamber, and a diffuser. In this figure, the ejector
through extensive data fitting for diverse turbulent flows. nozzle is represented by the red color, the flue gas in-
The present developed numerical tool uses the fi- lets are colored in green, and the remaining components
nite-volume method to treat the aforementioned forming the stack walls are shown in blue.

Fig. 2. The geometric model of the current ejector stack consisted of six separate parts.

78
 M. Darbandi et al. / International Journal of ISSI, Vol. 21(2024), No. 1, 75-84

As was mentioned in earlier sections, the objective better-optimized geometry parameters. This means that the
of this research is to assess the impact of variations in optimum geometry choices are very close to the original
several geometric parameters on the performance of the design cases. So, wider ranges for the selected parameters
present ejector stack. We have chosen the flue gas flow do not necessarily lead to more optimum values for them.
rate as a key parameter, which can directly quantify the Fig. 3. illustrates the entire computational domain,
ejector stack's performance. So, the other interesting the mesh distributed over the computational domain,
parameters such as the ejector entrainment ratio will not and the major associated boundary conditions. As a
be addressed in this work. part of the computational domain, a large red cylinder
The geometric parameters investigated in this is positioned at the top of the stack outlet. It is because
research include the lengths of the mixing chamber and we intend to accurately implement the effect of the sur-
diffuser and the angles of the nozzle ejector, the noz- rounding ambient in the simulations. One simple idea is
zle diffuser, and the mixing entry, see Fig. 1. Table 1. to implement the outlet boundary conditions right at the
describes their symbols and units. The table also exit plane of the ejector diffuser. It will certainly reduce
provides the current sizes. They are referred to as the design the number of grid points. However, one should note
values. that the simulation will miss the correct buoyancy effect,
Table 2. describes the names of the chosen param- enforced by the warm or hot air at the top of the stack
eters and their ranges of variations. The selected rang- and its surroundings. On the other hand, one will certain-
es have been determined after a long investigation on ly miss implementing the influences of environmental
choosing different values for them and a long discussion winds on the flow field inside the ejector stack. So, this
with the ejector stack operator. In other words, we have study considers a large volume of space around the stack
considered all the limits determined by the industry spe- as the computational domain. Generally, four boundary
cialists, specifically, those who work in the DRI Unit. On conditions are implemented around the computational
the other hand, the original ranges have been much wid- domain. 1- The air passing through the ejector stack is
er than those, which are reported in Table 2. However, subject to an intake fan boundary condition. 2- A pressure
the wide ranges do not necessarily help to achieve boundary condition is applied at the flue gas flow intake.

Table 1. The selected geometrical parameters.

item symbol the selected parameter design value unit
1 𝐿𝐿𝑀𝑀𝑀𝑀 the length of the mixing chamber 7700 mm
2 R the length of the diffuser 18915 mm
3 α the angle of the nozzle ejector 19.8 deg
4 β the angle of the diffuser 15 deg
5 θ the angle of the mixing entry 5.39 deg

Table 2. The ranges of variations for the selected geometrical parameters.

α
Simulatio θ β LMC R
degree
n Number degrees degrees mm mm
s
1 4.5 0 14.85 5000 2125.28
2 4.9 10 16.15 5555.56 5077.06
3 5.3 12.5 18.15 6111.11 8028.84
4 5.39 15 19.8 6666.67 10980.6
5 5.7 17.5 23.1 7222.22 13932.4
6 6.1 20 26.4 7700 16884.18
7 6.9 22.5 29.7 7777.78 18915
8 - - - 8333.33 19836
9 - - - 8888.89 22787.7
10 - - - 9444.44 25739.5
11 - - - 10000 28691.3

79
 M. Darbandi et al. / International Journal of ISSI, Vol. 21(2024), No. 1, 75-84

3- An ambient pressure boundary condition is applied that this number of grid points is sufficiently low to avoid
around the red color surface, which lets us observe the considering y+>1, e.g. y+=30, in our calculations.
exit jet from the stack and phenomena beyond the design
point at the stack exit. 4- The wall boundary condition
is applied to the ejector stack and ejector nozzle walls.

Fig. 4. Grid-independent study via monitoring the

volumetric flow rate value of the flue gas versus the grid
resolutions.

3. Results and Discussions

This section presents the results of present simulations
and discusses the resulting achievements. The first step is
to validate the CFD software code. The best validation is
to model the ejector stack in the real working condition,
i.e., off-design conditions, and compare the achieved
numerical solutions with the online collected data, i.e., the
Fig. 3. The computational domain, the outer boundary grid
measured data. However, it should be noted that there is
distribution, and the implemented boundary conditions.
no reliable measurement tool and no available measuring
probe installed in the DRI unit of Esfahan’s Mobarakeh
Furthermore, a high-quality mesh is generated to Steel Company to measure the off-design conditions. So,
effectively capture all the physics occurring within the it is not possible at all to validate the results of present
domain. To achieve this, the grid undergoes compression numerical simulations experimentally. Fortunately and
in regions characterized by narrow geometry, high curva- as was discussed previously, the DRI Unit has provided
ture, proximity to walls, and a substantial gradient of flow the ejector stack design working conditions, of which the
parameters. It is noteworthy that more pronounced mesh volumetric flow rate of flue gas shows the most important
compression occurs at locations where the cross-section one. The given design data indicates that the volumet-
of the ejector stack and the ejector nozzle change. ric flow rate of the flue gas is about 327,000 (Nm3/h) at
To assess grid independence, the volumetric flow rate the design working conditions. As was discussed previ-
of the flue gas is monitored under design conditions for ously in the mesh-independent study section, the present
the ejector stack. Back to our past experiences, see Refs. grid-independent study demonstrated that the volumetric
[4, 6, 25-27], this research also employs y+<1 to ensure flow rate of the flue gas would be around 327,000 (Nm3/h)
the accuracy of the achieved numerical solutions. This at design working conditions. So, the predicted numeri-
should be fair if and only if this consideration does not cal solution agrees well with the design value provided
lead to a grid with a tremendously large number of grid by the designer of the DRI Unit. So, the validation is car-
cells. So, it is necessary to carry on an accurate grid-inde- ried out successfully. Despite the achieved validation, it
pendent study at this stage. In this regard, seven different is further possible to validate the current numerical solu-
meshes with varying sizes are generated to perform the tions against the analytical solutions provided by Ander-
present grid-independent study. Fig. 4. presents the vol- son [27] and White [28]. This has been carried on in this
umetric flow rate of the flue gas considering a different research. Fig. 5. illustrates the temperature distributions
number of cells. As is depicted in the figure, the solution and Fig. 6. demonstrates the pressure magnitudes along
is nearly independent of the mesh size when the num- the vertical direction, i.e., along the centerline of the
ber of cells is over 4.8 million. The concluding remark is stack. The presented plots in these two figures are broken

80
1
 M. Darbandi et al. / International Journal of ISSI, Vol. 21(2024), No. 1, 75-84

into two parts by the three drawn vertical lines, which maximum velocity magnitude occurs right at the outlet of
respectively indicate the beginning of the mixing cham- the nozzle ejector. So, it produces the necessary motive
ber, the end of the mixing chamber, and the stack exit. As pressure gradient to effectively suck the flue gas into the
is seen in these two figures, the differences between the stack. The current calculations indicate that the flue gas
CFD solutions and the analytical solutions are minimal. flow rate is approximately 144 kg/s in design conditions.
The accuracy is more pronounced in the temperature dis- We choose this flow rate as the reference flow rate to
tribution plot. The slight differences can be attributed to evaluate the positive/negative impact of the resized ejec-
the many simplifying assumptions, which are necessary tor stack on its resulting performance, see Tables 1 and 2.
to apply in deriving the analytical relations. The numer- In other words, this reference flow rate establishes a suit-
ical solution would be more accurate than the analytical able frame to determine the positive or negative impacts
relations due to its much less simplifying assumptions. of the geometry parameter variations. It is said there will
Since the analytical relations are long, they are not pre- be an improvement in the ejector stack performance if
sented here. It helps to shorten the length of this paper. the geometric parameter variation results in a stronger
flue gas rate through the stack. These two plots indicate
that the fluid flow pattern is not axisymmetric inside the
ejector stack. It is quite evident because the ejector stack
geometry and the implemented boundary conditions are
not also axisymmetric.

Fig. 5. Presenting the present numerical temperature

distribution along the centerline of the stack and
comparison with the analytical solution.

Fig. 7. Contours of the velocity fields in two longitudinal

cross-sections of the ejector stack.

The next step is to examine the geometric parame-

Fig. 6. Demonstrating the present numerical pressure ter variation impact on the ejector stack performances,
distribution along the centerline of the stack and which is determined by calculating the flue gas flow
comparison with the analytical solution. rate. All simulations are carried on at in-design working
conditions. As said before, these geometric parameters
Fig. 7. displays the present numerical velocity con- include the lengths of the mixing chamber and the diffus-
tours at two longitudinal cross-sections of the ejector er, the angles of the nozzle ejector, the diffuser, and the
stack. The results are derived for the ejector stack in mixing entry. These parameters are suitably varied with-
design conditions. As illustrated in these plots, the in their specified ranges of variation, see Table 2. The

81
 M. Darbandi et al. / International Journal of ISSI, Vol. 21(2024), No. 1, 75-84

results are illustrated in Figs 8 to 11. To have a better

frame of comparison, the effects of different aforemen-
tioned geometrical parameters are evaluated separately.
Fig. 8. illustrates the impact of the mixing chamber
length on the flue gas flow rate magnitude. The figure
shows the design value as well, see Table 1. As is seen,
increasing or decreasing this length does not lead to a

design value
contiguous trend of increasing or decreasing. In other
words, there are several different maximum and mini-
mum flue gas rates as the chamber length either increases
or decreases. So, it is not generally advised to focus on
this parameter as a serious remedy to improve the ejector
stack performance. Hence, it is better to pay attention to
the other geometric parameters. Fig. 9. Variation of the flue gas flow rate with the diffuser
length changes.

The next step is to investigate the diffuser diver-

gence angle impact on the achieved performance. Fig.
10. demonstrates the impact of the diffuser divergence
angle changes on the flue gas flow rate value. The figure
also shows the design value for this parameter. As before,
the diffuser divergence angle has been either decreased
or increased to see its consequences on the achieved per-
design value

formances. The figure shows that increasing the diffuser

angle does not sound good because the achieved perfor-
mance decreases continuously. This can be attributed
to the flow separation occurrences in the ejector stack
system. In contrast, a smaller diffuser divergence angle
value can lead to a better ejector performance. However,
Fig. 8. Variation of the flue gas flow rate with the mixing there is a lower limit for the diffuser angle because the
chamber length changes. ejector stack performances show an abrupt decrease with
little decrease in the diffuser length. This point should be
very risky and should be avoided very cautiously. The
The next step is to study the impact of the diffuser
figure indicates that the optimum diffuser divergence
length variation on the achieved ejector stack perfor-
angle occurred at about 4.9 degrees. This angle enhances
mance. Contrary to Fig. 8. Fig. 9. shows that there is al-
the ejector stack performance by about 4%. So this little
most a smooth variation in the flue gas rate as the diffus-
increase in the performance suggests that this strategy
er length increases. Considering the design point value,
does not sound great, specifically considering the insta-
the figure shows that a decrease in the diffuser length
bility of the ejector stack flow around this risky point.
will cause a continuous decrease in the performance,
which is not in favor of this study. However, increas-
ing the diffuser length, first shows a little decrease in
the performance, which sounds inappropriate. Howev-
er, the performance starts increasing at higher diffuser
lengths. This can be attributed to the improvement in
the pressure recovery magnitude. Inspecting the figure,
indicates that a diffuser length of about 25 m can suit-
ably enhance the achieved performance. This diffuser
design value

length will lead to a 10% increase in the flue gas rate,

which is in favor of the present study. Unfortunately, the
story will not be so good if one wishes to increase the
diffuser length more. Due to the formation of flow vor- 1
ticities and flow separation in the ejector stack, the rate
of increase of the ejector stack performance becomes
less and less. A diffuser length of 15 m will be very
compromising if one does not wish to improve the per- Fig. 10. Variation of the flue gas flow rate with the angles
formance of the present ejector stack. of diffuser changes.

82
 M. Darbandi et al. / International Journal of ISSI, Vol. 21(2024), No. 1, 75-84

The last step is to perform the sensitivity analysis on the DRI Unit. After a careful study, five important geom-
the impact of nozzle divergence angle variation on the etry parameters were selected to perform the present sen-
achieved ejector stack performance. Fig. 11. illustrates sitivity analyses, which were carried out using the CFD
the ejector stack performance in terms of the nozzle software and simulating the fluid flows passing through
divergence angle. Again, the design value is demonstrated the ejector stack. The geometric parameters included the
in this figure. The trends of changes are very similar to that length of the mixing chamber, the length of the diffuser,
in Fig. 10. The plot shows that the ejector stack performance the angle of the nozzle ejector, the angle of the diffus-
gradually reduces as the nozzle divergence angle gradually er, and the angle of the mixing entry. These parameters
increases, which sounds bad. However, a reduction in the were suitably changed within their selected ranges and
nozzle angle improves the ejector stack efficiency. Unfor- their impacts on the performance of the re-sized ejector
tunately, there is a limit in decreeing this angle because stack were carefully monitored by calculating the volume
there will be an abrupt decrease in the performance as the rate of the flue gas, which was sucked by the re-sized
nozzle angle is further decreased, which is unsatisfacto- ejector stack system. To validate the CFD software, the
ry. Generally speaking, the optimum angle for the diver- present numerical results were compared with the analyt-
gence of the nozzle ejector is about 12.5 degrees. It leads ical solutions and the data provided by the manufacturer
to the highest ejector performance achievement, which of the ejector stack. The agreements were excellent. The
is about 6.25%. It can be counted on this improvement. present simulations revealed the effects of varying the
selected geometric parameters on the flue gas flow rate.
It was shown that the increase or decrease in the mix-
ing chamber length would not lead to a continuous trend
of increase or decrease in the performance. So, it is not
generally advised to focus on this parameter as a serious
remedy to improve the ejector stack performance. How-
ever, the increase in the diffuser length would reliably
enhance the performance. It was shown that if the diffus-
design value

er length is increased by about 38%, the ejector perfor-

mance will increase by about 10%. Further study showed
that the optimal divergence angles would be about
4.9 degrees for the diffuser and 12.5 degrees for the noz-
zle ejector. It was shown that if the diffuser angle be-
comes about 4.5 degrees, the flue gas rate will increase by
Fig. 11. Variation of the flue gas flow rate with the angles about 4%. On the other hand, if the nozzle ejector angle
of the nozzle ejector changes. is adjusted at 12.5 degrees, the ejector performance will
increase by about 6.25%. The study concluded that the
The sensitivity analysis performed in Figs. 8-11. best suggestion would be to increase the diffuser length;
results in several conclusions. Fig. 9. indicates that if the however, this must be consulted with industry specialists
diffuser length is increased by about 38%, the ejector very carefully. The present literature review showed that
performance will increase by about 10%. Fig. 10. indi- the present study is novel and that it had not been fulfilled
cates that if the diffuser angle becomes about 5 degrees, similarly in the other industries. So, the present study can
the flue gas rate will increase by about 4%. Fig. 11. also be considered as the novelty of this work. Considering
shows that if the nozzle ejector angle is adjusted at 12.5 this novelty, the presented procedure can be equally
degrees, the ejector performance will increase by about applied in other industries, which suffer from their
6.25%. So, the best suggestion is to increase the diffuser defective or inefficient ejector stacks.
length; however, this must be consulted with the industry
specialists very carefully. However, it should be noted Acknowledgment
that this may require changing the fan because of a higher
pressure loss along the stack. So, this should be checked The authors would like to acknowledge the financial sup-
very carefully before changing the diffuser heights. port received from Esfahan’s Mobarakeh Steel Company.

4. Conclusions References

Back to the inefficiency of an ejector stack system [1] Meakhail T, Zien Y, Elsallak M, AbdelHady S,
located in a Direct Reduction Iron (DRI) Unit, the main Experimental study of the effect of some geometric
objective of this work was to possibly improve the effi- variables and number of nozzles on the performance of
ciency of this system by improving the attributed geo- a subsonic air–air ejector, Proceedings of the Institution
metric parameters of the ejector stack. The main purpose of Mechanical Engineers, Part A: Journal of Power and
of this subsonic ejector stack is to expel the flue gas from Energy. 2008; 222: 809-818.

83
 M. Darbandi et al. / International Journal of ISSI, Vol. 21(2024), No. 1, 75-84

[2] Ariafar K, Buttsworth D, Sharifi N, Malpress R, nozzle structures, International Journal of Thermal
Ejector primary nozzle steam condensation: Area ratio Sciences. 2012; 56: 95-106.
effects and mixing layer development, Applied thermal [16] Kong F, Kim H.D, Analytical and computation-
engineering. 2014; 71(1): 519-527. al studies on the performance of a two-stage ejector–
[3] Ariafar K, Buttsworth D, Al-Doori G, Sharifi N, diffuser system, International Journal of Heat and Mass
Mixing layer effects on the entrainment ratio in steam Transfer. 2015; 85: 71-87.
ejectors through ideal gas computational simulations, [17] Wang L, Yan J, Wang C, Li X, Numerical study
Energy. 2016; 95: 380-392. on optimization of ejector primary nozzle geometries,
[4] Darbandi M, Sabzpoushan S.A, Schneider G.E, International Journal of Refrigeration. 2017; 76: 219-229.
Numerical study to evaluate the effect of phase change [18] Chen W, Huang C, Chong D, Yan J, Numerical
on performance calculation of a steam ejector, CFDSC assessment of ejector performance enhancement by means
Paper 2018-0539. The proceedings of the 26th Annu- of combined adjustable-geometry and bypass methods,
al Conference of the CFD Society of Canada, (Ottawa, Applied Thermal Engineering. 2019; 149: 950-959.
ON, Canada), CFDSC 2018, University of Manitoba, [19] Tashtoush B.M, Al-Nimr M.A, Khasawneh M.A, A
Winnipeg, Manitoba, Canada. 10-12, 2018. comprehensive review of ejector design, performance,
[5] Ariafar K, Cochrane T, Malpress R, Buttsworth and applications, Applied Energy. 2019; 240: 138-172.
D, Pitot and static pressure measurement and CFD [20] Yan J, Li S, Liu Z, Numerical investigation on
simulation of a co-flowing steam jet, Experimental optimization of ejector primary nozzle geometries with
Thermal and Fluid Science. 2018; 97: 36-47. fixed/varied nozzle exit position, Applied Thermal
[6] Darbandi M, Sabzpoushan S, Schneider G.E, Engineering. 2020; 175.
Geometrical optimization of a steam jet-ejector using [21] Hadi M, Arshad A, Shaik N.B, Benjapolakul W,
the computational fluid dynamics, ASME-FEDSM Paper Gillani Q.F, Optimization of hydrocarbon ejector using
2018-83203, Proceedings of the ASME 2018 5th Joint computational fluid dynamics, Engineering Journal.
US-European Fluids Engineering Summer Conference 2022; 6: 5.
(FEDSM2018), Montreal, Quebec, Canada. 2018; 2. [22] Tavakoli M, Nili-Ahmadabadi M, Joulaei A, Ha
[7] Xiao J, Wu Q, Chen L, Ke W, Wu C, Yang X, Yu M.Y, Enhancing subsonic ejector performance by
L, Jiang H, Assessment of different CFD modeling incorporating a fluidic oscillator as the primary
and solving approaches for a supersonic steam ejector nozzle: a numerical investigation, International Journal of
simulation, Atmosphere. 2022; 13: 144. Thermofluids. 2023; 20: 100429,
[8] Watanabe I, Experimental investigations concerning [23] Schillaci E, Vera J, Oliet C, Vemula J.B,
pneumatic ejectors, with special reference to the effect of Duponcheel M, Bartosiewicz Y, Numerical modeling of
dimensional parameters on performance characteristics, air ejectors covering supersonic, subsonic and closed-port
Symposium on Jet Pumps and Ejectors. 1972: 97-120. operations, Energy. 2024; 302: 131627.
[9] Vyas B.D, Kar S, Study of entrainment and mixing [24] Kumar A, Yadav S, Kumar V, Kulkarni A, A
process for an air to air jet ejector, The 2nd Symposium comprehensive exploration of ejector design, operational
on Jet Pumps and Ejectors and Gas Lift Techniques, factors, performance metrics, and practical applications,
BHRA Fluid Engineering, Cambridge, England. 1975. Journal of the Brazilian Society of Mechanical Sciences
[10] Riffat S.B, Omer S, CFD modelling and and Engineering. 2024; 46(39).
experimental investigation of an ejector refrigeration [25] Darbandi M, Jalali R, Schneider G.E, Robust
system using methanol as the working fluid, International 1-D fluid flow and heat transfer predictions in gas tur-
Journal of Energy Research. 2011; 25: 115-128. bine cooling passages,” AIAA Paper 2019-4057, AIAA
[11] Riffat S.B, Jiang L, Gan G, Recent development in 2019 Propulsion and Energy Forum and Exposition,
ejector technology- A review, International Journal of Indianapolis, Indiana, USA, August. 2019: 19-22.
Ambient Energy. 2005; 26(1): 13-26. [26] Darbandi M, Jalali R, A cost-effective approach to
[12] Yadav R.L, Patwardhan A.W, Design aspects of generate accurate correlations via analyzing a minimum
ejectors: Effects of suction chamber geometry, Chemical number of data points; Case study on the convection heat
Engineering Science. 2008; 63: 3886-3897. transfer problem, The Canadian Journal of Chemical
[13] Zhang X, Jin S, Huang S, Tian G, Experimental Engineering. 2022; 101: 3634-3647.
and CFD analysis of nozzle position of subsonic ejec- [27] Darbandi M, Jalali R, Internal cooling
tor, Frontiers of Energy and Power Engineering in China. sensitivity analysis to improve the thermal performance
2009; 3: 167-174. of gas turbine blade using a developed robust conjugate
[14] Li C, Li Y, Wang L, Configuration dependence and heat transfer method, International Journal of Engine
optimization of the entrainment performance for gas–gas Research. 2022; 24: 949-964.
and gas–liquid ejectors, Applied Thermal Engineering. [28] Anderson J, Fundamentals of Aerodynamics, 6th
2012; 48: 237-248. ed., McGraw-Hill Education, 2016.
[15] Yang X, Long X, Yao X, Numerical investigation [29] White F.M, Fluid Mechanics, 7th ed., McGraw-Hill
on the mixing process in a steam ejector with different Education, 2011.

84