Professional Documents
Culture Documents
2018 Book CFDTechniquesAndEnergyApplicat
2018 Book CFDTechniquesAndEnergyApplicat
Driss
Brahim Necib
Hao-Chun Zhang Editors
CFD
Techniques
and Energy
Applications
CFD Techniques and Energy Applications
Zied Driss Brahim Necib
•
Hao-Chun Zhang
Editors
123
Editors
Zied Driss Hao-Chun Zhang
Department of Mechanical Engineering School of Energy Science and Engineering
National School of Engineers of Sfax Harbin Institute of Technology
Sfax Harbin
Tunisia China
Brahim Necib
Faculty of Sciences and Technology
University of Constantine 1
Constantine
Algeria
This book focuses on CFD (Computational Fluid Dynamics) techniques and the
recent developments and research works in energy applications. It is also devoted to
the publication of basic and applied studies broadly related to this area. The
chapters present the development of numerical methods, computational techniques,
and case studies in the energy applications. Also, they offer the fundamental
knowledge for using CFD in energy applications through new technical approaches.
Besides it describes the CFD process steps and provides benefits and issues for
using CFD analysis in understanding the flow complicated phenomena and its use
in the design process. The best practices for reducing errors and uncertainties in the
CFD analysis are further described. The Book is expected not only to reveal the
recent advances and future research trends of CFD Techniques but also to provide
the reader with valuable information about energy applications.
The chapters present the development of numerical methods, computational
techniques, and energy application case studies. Such studies and development
approaches aim to provide the readers, engineers and Ph.D. students with the
fundamentals of CFD prior to embarking on any real simulation project.
Additionally, engineers supporting or being supported by CFD analysts can take
advantage from the information of the Book’s different chapters.
In the first chapter, a numerical analysis on the performance of a solar chimney
power plant using steady state Navier-Stokes and energy equations in cylindrical
coordinate system was presented. The fluid flow inside the chimney is assumed to
be turbulent and simulated with the k–e turbulent model, using FLUENT software
package. The computed results were in good agreement with the experimental
measurements of the Spanish Manzanares power plant. The numerical model was
then coupled with a mathematical one for a geothermal heat exchanger to inves-
tigate the option of coupling solar and geothermal sources for a continuous day and
night operation. The results show the benefits of the hybrid solar-geothermal plant
compared to the single solar chimney plant for day and night periods.
The second chapter presents the behavior of the air flow characteristics inside the
Solar Chimney Power Plant (SCPP). A two-dimensional (2D) steady model was
carried out using the commercial computational fluid dynamics (CFD) code
v
vi Preface
Ansys-Fluent 17.0. In this chapter, five turbulence models were tested to present the
air flow characteristics distribution such as magnitude velocity, temperature, pres-
sure and turbulence. The above work showed that the turbulence model types have
a direct effect on the numerical results. Its computational results were compared to
the experimental data found by Kasaeian et al. (2014) to choose the adequate
model.
In the third chapter, the behavior of a negatively buoyant jet in laminar condi-
tions that results from the injection of lighter fluids downwards into a large con-
tainer of homogeneous fluid of denser density was studied numerically using Open
Foam with the finite volume method. The fluid characteristics effect on the evo-
lution of the pure water injected in a tank full of salt water was investigated
particularly the molecular diffusivity that affects the mixing layer, the density rel-
ative difference between two liquids and the salt water kinematic viscosity that has
an important effect on the transient phase as well as the subsequent steady state in
terms of stationary penetration depth and jet profile.
The fourth chapter aimed to perform a numerical simulation of the liquid
sloshing using the “Fluent” software. The simulation of the two phase application
was achieved using the Volume Of Fluid (VOF). The container was subjected to a
sinusoidal excitation. To impose the external excitation, a user defined function was
developed and interpreted in “Fluent”. Four numerical simulations were developed
with different turbulence models; standard k–e, RNG k–e, Realizable k–e and
standard k–x. The fluid flow characteristics for the different simulation cases were
presented and discussed. The numerical results were compared with the experi-
mental data. The comparison results show a good agreement with the turbulence
model standard k–e.
This fifth chapter presents a numerical model in order to capture the flow fields
within a vanned volute under steady conditions. Numerical simulations were
conducted using the CFX 17.0 package to solve Navier–Stokes equations by means
of a finite volume discretization method. The good agreement between the exper-
imental and numerical results of the turbine performance confirms the validation
of the numerical model. Then, many computed flow discharge parameters, such as
the averaged volute exit flow angle, were plotted to understand the behavior of the
volute under different turbine expansion ratios. Furthermore, several loss coeffi-
cients distribution and entropy contours were plotted to characterize the occurring
losses. In addition, pressure distributions, velocity and turbulence parameters as
well as streamlines were numerically obtained to analyze the flow behavior within
the turbine volute.
The sixth chapter presents some CFD simulation results of the hydrodynamic
structure around a modified anchor system. Using the CFD code Ansys-FLUENT
17.0, the finite volume method was used to solve the Navier-Stokes equations. This
study was carried out using the standard k-e turbulence model. The comparison
between the numerical and the experimental results found in the literature shows a
good agreement.
The seventh chapter focuses on the numerical study of forced convection, for a
thermo dependent Newtonian fluid in an eccentric horizontal annular duct. The
Preface vii
inner and outer cylinders were heated with a constant heat flux. The governing
equations were solved numerically by a finite difference method with implicit
scheme. The dynamic profile was assumed to be fully developed while the tem-
perature profile was assumed uniform at the entrance. The aim of this work was to
present the eccentricity effects on the dynamic and thermal fields along the duct.
The thermo dependency effect of the fluid was also examined, and some interesting
results regarding the reduction of the dynamic blocking phenomenon of the flow in
the narrow part of the duct for large eccentricities were presented. These results
reduce the strictness of precautions for neglecting the axial diffusion when making
computations in such geometries.
The eighth chapter investigates the effect of the incidence angle on the aero-
dynamic characteristics of the flow around a Savonius wind rotor. Six configura-
tions with different incidence angles h = 0°, h = 30°, h = 60°, h = 90°, h = 120°
and h = 150° were studied. To this end, a numerical simulation was developed
using the Computational Fluid Dynamic (CFD) code “Fluent”. The considered
numerical model is based on the resolution of the Navier-Stokes equations together
with the k-e turbulence model. These equations were solved by a finite volume
discretization method. The results confirm that the variation of the incidence angle
has an effect on the local characteristics. The numerical results were compared to
those obtained by previous findings showing a good agreement and confirming the
numerical method efficiency.
The ninth chapter predicts numerically the flow effects of two coaxial jets with
different swirl numbers on the characteristics of the turbulent diffusion flame. The
study focused on the rotation influence of the secondary flow; that is to say two
configurations were processed and compared: co-swirl and counter swirl.
Obviously, the latter showed higher shear than the first. The calculation results were
validated by actual measurement of the same configuration for two cases: co and
counter swirl for reactant with combustion. The calculation results focus on the
characteristics of the average flow and its turbulence for the two cases cited above.
The obtained results confirm the swirl effects to stabilize the flame.
First and foremost, I would like to thank Dr. Nabil Khélifi, Springer Editor who
invited me to edit this new book after awarding the conference on CFD techniques
and Thermo-Mechanics Applications, which was held at the National School of
Engineers of Sfax (University of Sfax, Tunisia) in April 2016. All the ideas have
developed further with my co-editors and many reviewers; especially in the second
edition of the International Conferences on Mechanics and Energy (ICME’2016)
which was held in Hammamet (Tunisia) in December 2016 and the third edition
ICME’2017, held in Sousse (Tunisia) in December 2017.
I would like to thank all the authors who submitted chapters at our requests.
Especially, I wish to express my gratitude to all the reviewers who participated to
this book, provided support, talked things over, read, wrote, offered comments and
allowed us to quote their remarks.
Many colleagues have generously provided comments and material from their
past and current research. Particularly, I thank my co-editors Prof. Brahim Necib
from the University of Mentouri Constantine (Algeria) and Prof. Hao-Chun Zhang
from the Harbin Institute of Technology (China). Without them, this book would
never find its way to so many reserchers, engineers and Ph.D. students.
I would like to express my gratitude to all those who provided support and
assisted in the editing and proofreading. Particularly, I thank Prof. Abdelmajid
Dammak for the Linguistic improvements of all chapters in the book. In addition, I
would like to thank Reyhaneh Majidi, Shahid Mohammed, Kavitha Palanisamy and
Suganya Manoharan from Springer for helping me in the process of selection,
editing and design.
Last and not least: I beg forgiveness of all those who have been with me over the
course of the years and whose names I have failed to mention.
ix
Contents
xi
xii Contents
Nomenclature
A Area (m2)
Acoll Solar collector area (m2)
Dtube Tube diameter (m)
G Solar radiation (W/m2)
g Gravitational acceleration (ms−2)
h Heat transfer coefficient (Wm−2K−1)
m_ Mass flow rate (kg s−1)
ntube Tube number
R Collector radius (m)
T Temperature (K)
V Airflow velocity (ms−1)
Greek Symbols
k Thermal conductivity (W m−1 K−1)
l Dynamic viscosity (kg (s m)−1)
q Density (kg m−3)
sa Transmittance-absorbtance product
Subscript
c Solar collector cover
e Environment or external
f Fluid
i Internal
A. Dhahri A. Omri
Research Unit: Materials, Energy and Renewable Energies,
University of Gafsa, College of Sciences, 2112 Gafsa, Tunisia
J. Orfi (&)
Department of Mechanical Engineering, King Saud University,
P.O. Box 800, Riyadh 11421, Saudi Arabia
e-mail: orfij@ksu.edu.sa
m Average
r Storage reservoir
s Soil or solar
w Geothermal water
w, in Heat exchanger inlet
w, out Heat exchanger outlet
1 Solar collector inlet
2 Solar collector outlet
DT Temperature increase (K)
1 Introduction
Several solutions have, in fact, been proposed in order to enhance the perfor-
mance of solar chimney plants. One of the first enhancement methods was intro-
duced by the engineer, Yorg Schlaich. The idea was to place tight water-filled tubes
under the roof. The water heats up during the daytime and emits its heat at night.
These tubes are filled only once; no further water is needed, because the tubes
remain closed after the filling process (Schlaich 1995).
In order to ensure the solar chimney power system operation during the night, a
further publication by Zhou et al. investigated an alternative method of heat
extraction from a salt gradient solar pond to the solar collector (Zhou et al. 2009;
Dhahri and Omri 2013). A hybrid solar chimney power plant with a salinity gra-
dient solar pond was conducted by Akbarzadeh et al. (2009). Kreetz proposed the
use of additional water bags as a heat storage system (1997). This system absorbs
part of the radiated energy during the day and releases it into the collector at night.
Pretorius (2007) introduced the use of intermediate secondary roof under the first
collector roof to enhance heat storage in the soil. In order to ensure a continuous
and noninterrupted day and night electricity production operation, Al-Kayiem et al.
(2012) studied the case of an inclined solar chimney power plant integrated with
flue gas as a thermal backup. Another technique was suggested by Azeemuddin
et al. (2013) which goal was to use the waste heat energy in conduits within the
solar collector. Aja et al. (2011) introduced and analyzed a hybrid system with flue
gases waste heat. An additional thermal storage system such as closed water-filled
tanks on the natural soil was studied by Robert et al. (2012) and Barbier (2002).
Bernardes studied the possibility of using a water-filled tube on the collector floor
as a heat storage device after sunset (Al-Kayiem 2016). Other researchers carried
out numerical simulations to analyze the performance of the solar chimney power
plant system with an energy storage layer including ground solar radiation and heat
storage (Hurtado et al. 2012; Ming et al. 2009; Fanlong et al. 2011; Yan Zhou
2011). Most of the reported investigations showed improvements in the system
performance. But the proposed solutions remained insufficient to ensure a day and
night continuous operation of the plant.
A different concept was suggested by Alrobaei (2005). It involved the investi-
gation of methods to increase the efficiency of the collector zone. The new approach
focuses on the implementation of a solar chimney power generation system in the
southern region of Libya. The authors introduced a hybrid geothermal/PV/solar
chimney power plant. The plant would be able to deliver power during the night.
One of the rare techniques proposed for a 24-hour solar chimney operation is the
use of geothermal water tubes. In a recent publication about the solar chimney, Cao
et al. (2014) numerically studied the performance of a geothermal–solar chimney
power plant. The authors discussed the introduction of geothermal water on solar
chimneys. In their study, they considered a heating technique with water tubes
placed under the collector roof. The results reveal that this solution achieves a
greater contribution and can solve the continuous operation problem of the plant.
4 A. Dhahri et al.
The solar chimney is a solar thermal power plant that combines three well-known
components: the chimney (tower), the collector, and the turbine. During the day, the
sun radiation passes through the cover of the collector (raised a certain height above
the ground) and heats the air between the collector and the ground (greenhouse
effect). The warm air tends to escape through the solar chimney to the upper
atmosphere, while ambient air is pulled from the environment into the collector.
The warm air flows at high speed through the chimney and drives the turbine
installed at the base. The kinetic energy in the air is converted to electrical energy
using the generators (Fig. 1).
The solar chimney thermal plant needs to be able to operate reliably in all operation
modes. The primary goal of the combination concept is the enhancement of the
performance of conventional solar chimney. In the new system, the hot water is
pumped from the exploited geothermal well (located in the center of the chimney)
to the air in the collector. The proposed heating system ensures that the geothermal
energy is used only when the collector cannot provide enough heat, which improves
the system reliability and economy. This system is able to change the ground level
thermal conditions.
Theoretical Analysis of the Performance of a Solar Chimney … 5
In order to use the geothermal water heat, some design modifications have been
made to the classical solar chimney power plant. A heat exchanger consisting of
radial tubes (arranged horizontally) was introduced. The tubes proceed from the
collector outlet and end at the collector inlet. The tubes are separated from each
other by a fixed distance. Water flows in the same direction in all the pipes. The
cold water is released from the collector inlet to the injection well. The gained
energy is transferred to the indoor air.
The basic characteristic of the heat exchanger used in this study is that it supplies
a continuous and constant inflow of heat into the collector during the night what-
ever the outside meteorological changes.
The heat distribution is considered uniform. The hot water is distributed evenly
within each tube (Fig. 2). Our system is also fitted with a tank to ensure the storage
of hot water. A schematic of the hybrid solar chimney is shown in Fig. 3.
There are two distinct modes of operation of the hybrid system: For the first
mode, the geothermal exchanger acts as a backup system to overcome the deficits of
sunshine during the night (full geothermal mode operation). In the second mode,
the heat exchanger is used to provide additional energy during daytime periods
(solar–geothermal coupling).
The study focuses on modeling the night-time operation of the solar chimney and
assumes there is no solar irradiation on the system.
At sunset, the temperature inside the collector decreases. The heat exchange at
ground level is an important factor for conditioning and circulating air in the
6 A. Dhahri et al.
collector. In this case, maintaining the air temperature at a certain level requires
additional heating to compensate for energy losses. Since the Tunisian southern
regions have hot water energy sources (Table 1) with average temperatures ranging
between 65 and 80 °C (Ben Mohamed and Said 2008), the contribution of the
geothermal heat energy can remedy this problem.
The main advantage of the integration of the new component is the ability of
generating electricity without being limited to sunlight hours only.
Theoretical Analysis of the Performance of a Solar Chimney … 7
The solar and geothermal energies are two renewable energy sources often used for
the production of electricity. However, they are very often used independently
despite their obvious complementarity. The combination of these two sources has
the potential to consolidate energy production for solar chimneys. The goal of this
part is the coupling of these two renewable energy sources in daylight. The new
system allows a hybrid operation with both solar heat and low geothermal
temperature.
The proposed heating system uses the geothermal water as a source of heat. This
section focuses on the development of a thermal model of the heat exchanger. It
detailes the basic theory used for solving the energy balance equations leading to
the development of a full geothermal heat exchanger model. In the analytical
model, the heat exchange is considered to be steady. We consider an infinitesimal
element of the tube in the geothermal water flow direction (Fig. 4).
The analytical heat exchange rate dQ_ transferred to the indoor air as a function of
the soil surface temperature is given by Naili et al. (2016), and Saïd (1997):
dQ_ ¼ mC
_ p dTf ð xÞ ¼ Uð xÞðTs Tf ÞpDtube dx ð1Þ
Ts and Tf refer to the soil surface and fluid temperatures, respectively, while
x stands for the axial coordinate.
Supposing that the overall heat transfer coefficient (U) remains constant
throughout the exchanger, Eq. (1) becomes:
dTf ð xÞ UpDtube dx
¼ ð2Þ
ðTs Tf Þ _ p
mC
By integrating Eq. (2), the mean fluid temperature along the pipe can be found
analytically as:
ZTf Zx
dTf ð xÞ UpDtube
¼ dx ð3Þ
ðTs Tf Þ _ p
mC
Teau;e 0
UpDtube x
Tf ð xÞ ¼ Ts þ Tw;in Ts exp ð4Þ
_ p
mC
Equation 4 can be used to determine the water temperature at any point along the
tube. It demonstrates that the water temperature decreases exponentially along the
tube.
The outlet water temperature can be related to the average temperature of the soil
surface Ts by Eq. (5):
UpDtube L UStube
Tw;out ¼ Ts þ Tw;in Ts exp ¼ Ts þ Tw;in Ts exp
_ p
mC _ p
mC
ð5Þ
where
The total heat transfer rate in the heat exchanger may be expressed as:
Q_ tube ¼ ntube m_ w Cp;w Tw;in Tw;out ð7Þ
Theoretical Analysis of the Performance of a Solar Chimney … 9
The collector is a key component of the central solar chimney. It converts solar
energy into thermal energy that is subsequently converted into kinetic and electrical
energy. The study of the energy performance of the collector can be achieved
effectively through the establishment of adequate energy balances. In the following,
different mathematical models based on energy balance equations of the collector
are presented: one for the conventional solar chimney and the others for the
geothermal–solar system.
We consider the variation of the kinetic energy in the energy balance of the col-
lector. This introduces an additional term in the overall balance of the collector. In
this case, the following equation shall be used (Dhahri et al. 2014):
_ _ V12 V22
0 ¼ Qs Qp þ q1 A1 V1 Cp;a T1 þ q2 A2 V2 Cp;a T2 þ ð8Þ
2 2
The heat flow rate lost by convection between the collector and the environment
is given by:
where
p ¼ 2Cp T2 ð11Þ
2G 2hconv Acoll ðTc Tair Þ A1 V12
q¼ þ 2V1 Cp T1 þ ð12Þ
qcoll A2 qcoll A2 A2 2
Details on the development of the above equations can be found in Dhahri et al.
(2014).
10 A. Dhahri et al.
To establish the thermal balance of the collector, daytime and night-time periods
should be treated separately. Indeed, gross requirements depend directly on the
indoor temperature, generally different during the day and night, and the presence
of a geothermal exchanger over the night.
In the following section, the various equations for a fully operational geothermal–
solar chimney are presented.
hi is the heat transfer coefficient by convection between the indoor air and the
inner surface of the collector. In order to solve the various energy equations, hi
needs to be determined by the following equation (Ben Mohamed 2003; Verlodt
1983):
" 1=3 # " #1=3
qTm lTm
hi ¼ 0:2106 þ 0:0026v = ð14Þ
lgDT gDTcp k2 q2
The heat flux lost by convection between the collector (external wall) and the
environment may be expressed as follows:
The heat exchanger supplies energy to balance the heat losses occurring during
times when the collector air temperature exceeds the outside temperature. To
determine the maximum thermal capacity of the heating system, it is necessary to
consider that the losses to the collector cover are equal to zero. It is assumed that no
solar radiation is absorbed by the air. If losses are neglected, an energy balance for
the collector releasing no stored heat after sunset can be written as (Pretorius 2007):
ntube m_ w Cp;w A1 V3 V3
Tw;in Tw;out þ ðCp V1 T1 þ 1 Þ ¼ Cp V2 T2 þ 2 ð21Þ
qcoll A2 A2 2 2
2ntube m_ w Cp;w A1 V3
V23 þ 2Cp V2 T2 Tw;in Tw;out 2 ðCp V1 T1 þ 1 Þ ¼ 0 ð22Þ
qcoll A2 A2 2
By solving Eq. (22), one can obtain the expression of the collector exit velocity:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
3 q 1 27q2 þ 4p3 3 q 1 27q2 þ 4p3
V2 ¼ þ þ ð23Þ
2 2 27 2 2 27
where
p ¼ 2Cp T2 ð24Þ
2ntube m_ w Cp;w A1 V3
q¼ Tw;in Tw;out 2 ðCp V1 T1 þ 1 Þ ð25Þ
qcoll A2 A2 2
12 A. Dhahri et al.
In this case, the effect of collector losses is taken into account. All the terms of
Eq. (21) remain unchanged, and a new parameter is added.
ntube m_ w Cp;w A1 V3
Tw;in Tw;out þ ðCp V1 T1 þ 1 Þ
qcoll A2 A2 2
ð26Þ
he Acoll hi Acoll V3
ðTc Tair;e Þ ðTair;i Tc Þ ¼ Cp V2 T2 þ 2
qcoll A2 qcoll A2 2
The resolution of Eq. (27) leads to the determination of the air exit velocity
when the heat losses are taken into consideration.
When the hybrid plant is proposed by retrofitting an existing solar chimney plant
with a geothermal heat exchanger during daytime, Eq. (8) is no longer adequate.
For this reason, we developed a mathematical model to predict the thermal per-
formance of a collector involving three different interfaces (soil, air, and cover). In
this case, the heat is delivered by two energy sources, solar and geothermal energy.
The solar power received by the collector is expressed proportionally to the sunlight
collected by the transparent cover:
Q_ s ¼ ðsaÞAcoll G ð28Þ
heating tubes can receive a significant amount of solar radiation during daytime.
They reduce the amount of incoming solar radiation reaching the ground.
The energy stored in the soil depends on the heating method. Heating on the
ground competes with solar gain and reduces the storage capacity (Issanchou 1991;
Kittas 1987). The high value of the soil surface temperature prevents the recovery
of energy in the deeper layers.
This phenomenon of storage is not included in the thermal balance for two
reasons:
– Its influence is assumed to be relatively small in the whole year (Kittas 1987).
– It is practically removed in the presence of a heating system placed on the floor.
Thus, the energy balance can be established from the above equations as
follows:
Q_ s Q_ geo A1 V3 Q_ p V3
þ þ ðcp V1 T1 þ 1 Þ ¼ cp V2 T2 þ 2 ð31Þ
qcoll A2 qcoll A2 A2 2 qcoll A2 2
These expressions show that the airflow below the collector depends on several
parameters including the geometrical dimensions of the heat exchanger and the flow
rate of the circulating hot water. The resolution of the above set of equations is
performed through an iterative process knowing the input values of different surface
temperatures, water inlet temperature, and ambient temperature.
5 Numerical Solution
In this section, Manzanares prototype of solar chimney power plant was used as a
reference plant. For this prototype, a 195-m high and 10-m diameter chimney was
built, surrounded by a collector with 240 m in diameter (Schlaich 1995).
The geometrical structure of our computational domain is the same as in Fig. 1.
A model is constructed by GAMBIT software as shown in Fig. 5. The computa-
tional domain is discretized by an unstructured grid consisting of approximately
1,228,452 cells. To have a better control over the mesh quality, the complete
computational domain is subdivided into several sub-zones (collector, transition
zone, and chimney), each of which is meshed individually. The zone with extre-
mely high pressure gradients requires a fine mesh. For this reason, the transition
zone is meshed by the tetrahedral mesh. The other two sub-zones are meshed using
hexahedral mesh (Dhahri et al. 2014).
The numerical analysis was conducted using the commercial software FLUENT
relying on the aforementioned research findings. The standard k–ɛ was used in the
current study for the turbulence modeling, and the discrete ordinate (DO) model was
selected as the thermal radiation model. The PISO method was used for the pres-
sure–velocity coupling. A pressure inlet boundary condition was specified for the
14 A. Dhahri et al.
collector inlet. The momentum, continuity, and energy equations were solved using
the second-order upwind scheme. A pressure outlet was applied as a boundary
condition for the chimney outlet. Nonslip conditions were imposed on the walls. The
collector cover is considered as semitransparent wall (Dhahri et al. 2014).
The performances of the classical solar chimney and the hybrid system were ana-
lyzed using the various mathematical models presented in the previous sections.
6.1 Validation
Fig. 6 Updraft velocity distribution as function of the solar radiation intensity (Dhahri et al. 2014)
To achieve the objective of this work, the effect of the controlling parameters such
as the collector size, water inlet temperature, water flow rate, and collector losses on
the performance of the hybrid plant were systematically examined. Several simu-
lations have been conducted.
16 A. Dhahri et al.
The effect of the collector radius on the water outlet temperature, velocity, and
indoor air temperature was examined for a radius of 40, 60, 80, 100, 120, and
140 m. For a given water inlet temperature (70 °C), the effect of the radius on the
air outlet velocity is shown in Fig. 7. It is clear that during night, the air exit
velocity decreases as the collector radius increases. The results shown in Fig. 7 also
reveal that for a greater collector radius, the indoor air temperature is low. This is
because for larger collectors, the effect of the heating system becomes lower and the
temperature inside is quite close to the outside temperature.
Simulations were performed using the complete specifications of the reference plant
(Manzanares prototype). The temperature of the inlet water is assumed to be
constant (70 °C). This section studies the effect that the geothermal water flow has
on the air exit velocity. For this reason, we varied the inflow rate into each tube. It is
assumed that all the other conditions are fixed. Figure 8 illustrates clearly the
increased outlet velocity as a result of the increase in water flow rate for an
unchanged collector radius (120 m). This is due to the fact that the increased flow
rate allows more energy to pass to the indoor air flowing through the collector.
Figure 9 evaluates the influence of varying the geothermal water temperature at the
inlet of the heat exchanger from 55 to 75 °C. The water mass flow rate is set to
0.2 kg/s. The geometric and operating parameters were kept unchanged. A closer
examination of Fig. 9 reveals that the air velocity at the collector exit depends
strongly on the water inlet temperature. In fact, higher water inlet temperatures
means more energy is available to heat the air within the collector, which leads to
greater air velocities. Air velocity rises almost linearly as the water inlet temperature
increases.
During the night-time operation mode of the solar chimney power plant, large heat
losses may appear by means of convection from the collector cover to the
18 A. Dhahri et al.
environment. During the night, the heat losses through the cover to the environment
start to exceed the incoming quantity of energy (stored in the ground during the
day). The use of the heating tubes can overcome this problem and can reduce the
losses. However, as the collector radius increases (beyond 40 m), the thermal losses
increase and the incoming geothermal heat flow becomes insufficient. For a radius
of 120 m, the thermal losses are lower due to the decrease in the convective losses
through the collector roof to the environment. When the radius reaches 140 m, the
cover is cooler than the outside air and, therefore, it gains heat (Fig. 10). This
process is known as the inversion phenomenon.
Comparative simulations were performed using the Spanish prototype as a ref-
erence, with one model including thermal losses while the other does not consider
them.
Figure 11 investigates the collector thermal loss effect on the air velocity for
different collector radius values (40, 60, 80, 100, 120, and 140 m). For the radii of
40 and 140 m, there is no major difference in air exit velocity between the cases
with or without thermal losses. For intermediate radius values, Fig. 11 reveals that
the difference between the two curves can be significant.
As it can be deduced from Fig. 10, any increase in the collector size increases
the thermal losses during the night and leads to a concomitant decrease in the air
velocity at the exit of the collector. It follows from Fig. 10 that the inclusion of the
thermal losses is responsible for an average reduction of 3–7% in the air velocity.
Therefore, with less energy being lost to the environment, more heat is available to
warm the collector air, giving an updraft velocity through the plant.
To compare the performance of the hybrid plant to the conventional plant in terms
of air temperature increase (DT), a series of simulations were performed. Figure 12
shows the simulation results by varying the collector radius from 40 to 140 m. The
solar radiation (800 W/m2) and the water flow rate in each tube (0.2 kg/s) were kept
unchanged.
By comparing the two curves given in Fig. 12, it is found that incorporating the
heating system during the day greatly alters the profile of air temperature in the
collector. An overall value of 6.4 °C rise in the air temperature is obtained. For
the solar–geothermal plant, the temperature gradient is higher compared to the
conventional plant. This is due to the fact that as the air moves through the col-
lector, it is warmed and its density decreases. The hotter the air is, the more
efficiently the system operates. It can be concluded that it is possible to improve the
overall performance of the plant by increasing the collector size.
From Fig. 13, it is evident that enlarging the collector size leads to a significant
increase in the air velocity for both plants. Moreover, the comparison of the two
curves reveals a difference between the velocity values. For a constant geothermal
heat flow, the observed differences are significant for radii less than 100 m.
However, this difference diminishes for high collector radius values. This result can
be explained by the fact that for collectors of large sizes, the amount of heat
20 A. Dhahri et al.
Fig. 12 Temperature
increase versus radius
delivered to the interior air is higher and the role of the exchanger becomes limited.
As shown in Fig. 13, it seems that the implementation of a heating equipment
during daytime can increase significantly the air velocity.
Conversely to the night operation case, the increase in the collector radius is
found to be advantageous for both classical and hybrid solar chimney power plants.
Theoretical Analysis of the Performance of a Solar Chimney … 21
Figure 15 focuses on the effect of the variation of the geothermal water flow on the
air velocity at the outlet of the collector during the day. A water flow rate ranging
from 0.2 to 0.5 kg/s in a single tube was considered. The temperature at the inlet of
the tubes remains constant at 70 °C. Solar radiation intensity is set to 800 W/m2.
From Fig. 15, it is clear that at a high water flow rate, the air velocity at the exit of
the collector is important.
In this section, a conventional plant versus a plant equipped with a heating system
was studied. Calculations taking into account heat losses have been conducted. As
expected, the size of the collector greatly influences the amount of dissipated energy
for both plant configurations. When evaluating Fig. 16, it is clear that the thermal
losses for a plant employing a heating system (solar–geothermal plant) are larger
due to the higher interior temperature.
The reference location selected was an area of south Tunisia. This particular
location is dry and hot and characterized by clear sky days and nights. It is situated
in a part where a high amount of solar irradiation is received annually. All of these
factors contribute to making this zone an ideal location for the construction of a
large-scale solar chimney power plant.
In order to evaluate the effect of incorporating a heating system, a comparison
between the heat flow rates transferred by the heat exchanger (during day and night
periods) from January 1 to December 31 is illustrated in Fig. 17. It is clear that the
Theoretical Analysis of the Performance of a Solar Chimney … 23
heating rates are generally lower in the summer periods. This therefore shows
unequal performance of the heat exchanger during the year. For the period between
the months of April and September, an average heating capacity of 10.06 MW
(day) and 18.26 MW (night) was recorded. Furthermore, this value increases for the
second half of the year and a capacity of 15.14 MW (day) and 22.17 MW (night)
was reached. These results show that the heating capacity strongly depends on the
weather conditions and is only effective for a period of the year. It should be
pointed out that the heating system is interesting in the winter.
Two simulations were conducted, one during the daytime without integrating the
heating system and the other during the night with the heating system. When
examining Fig. 18, it is evident that in periods of strong sunshine, solar gains can
24 A. Dhahri et al.
cover a significant portion of the collector requirements. However, for low sunlight
periods, the internal temperature may drop. It is then necessary to provide energy
excess to heat the air. This explains the wide disparity between the velocities
calculated in the summer and those found in the winter.
A closer glance shows that the velocity increases on average by 20.6% in winter
and 9.6% in summer with the inclusion of the heating tubes. Overally, it can be
concluded that the performance of the solar chimney power plant in daytime could
be enhanced through the use of the geothermal water as a heating source.
The system can operate under three distinct modes. A comparative study was
conducted to evaluate the system performance under the possible operating sce-
narios. The first tests were carried out for a conventional installation during the day
without the inclusion of a heating system (case 1). The second case corresponds to a
fully geothermal operation. Case 3 concerns the combination of the solar and
geothermal energy during the day. Figures 19 and 20, respectively, show the
variation in the indoor air temperature and the temperature at the soil surface for the
three modes.
If we briefly analyze the thermal behavior of the collector, more energy is
subsequently transferred to the collector air, giving a greater power output during
the day. This explains the air temperature values obtained for mode 1. During the
night periods (mode 2), the heat exchange inside a conventional solar chimney
occurs mainly between the air and soil (Yu 2014). This means that the soil is the
only way to supply heat to the collector during the night. The use of an auxiliary
heater may be required to increase the air circulation. It is also interesting to note
that the plants equipped with hot water pipes enjoy an increase in the interior
temperature during the night and show significantly greater air temperatures for
most of the day. This is due to the fact that the hot ground continues to act as a
supplier of energy during the day.
Figure 20 shows the effect of the water pipe implementation on the temperature
at the ground level. A closer examination of the results reveals that the incorpo-
ration of the heating system during the day has a significant effect on the ground
surface temperature. Figure 20 shows that a plant with a heating equipment leads to
a higher average increase of temperature (about 14 °C) than for the case of a
conventional plant. The sizable increase in diurnal ground temperature may be
attributed to the fact that more energy is made available at the ground surface
received from both sun and geothermal water. The level of the nocturnal ground
temperature is explained by the lack of sunlight.
7 Conclusion
This chapter presented a method to enable a continuous day and night operation of a
solar chimney power plant. The combination concept of a solar chimney with a
geothermal heat exchanger has been assessed. The performances of the two dif-
ferent plants were investigated using theoretical and numerical models. CFD
modeling has been conducted with FLUENT. The following conclusions can be
drawn from the analyses.
26 A. Dhahri et al.
References
Aja OC, Alkayiem HH, Karim ZAA (2011) Thermal field study and analysis in hybrid solar flue
gas chimney power plant. In: National postgraduate conference (NPC). Tronoh, Perak:
Universiti Teknologi Petronas, p 1–6
Akbarzadeh A, Johnson P, Singh R (2009) Examining potential benefits of combining a chimney
with a salinity gradient solar pond for production of power in salt affected areas. Sol Energy
83:1345–1359
Al-Kayiem HH (2016) Hybrid techniques to enhance solar thermal: the way forward. Int J of
Energy Prod and Mgmt 1:50–60. https://doi.org/10.2495/eq-v1-n1-50-60
Al-Kayiem HH, Aja OC (2016) Historic and recent progress in solar chimney power plant
enhancing technologies. Renew Sustain Energy Rev 58:1269–1292
Al-Kayiem HH, Sing CY, Yin KY (2012) Numerical simulation of solar chimney integrated with
exhaust of thermal power plant, Chapter in the special session on enhanced heat transfer. In:
Advanced computational methods and experiments in heat transfer XII, WIT transaction of
engineering. WITpress, UK. (ISSN: 1743-3533)
Alrobaei H (2005) Hybrid geothermal/solar energy technology for power generation. http://
publications.solar-tower.org.uk
Aurélio M, Bernardes DS (2010) Solar chimney power plants—developments and advancements.
In: Rugescu RD (ed) Solar energy. InTech, Croatia, ISBN: 978-953-307-052-0
Azeemuddin, Al-Kayiem HH, Gilani SI (2013) Simulation of a collector using waste heat energy
in a solar chimney power plant system. The sustainable city VIII, vol 2, WIT Transactions on
Ecology and The Environment, vol 179, WIT Press
Barbier E (2002) Geothermal energy technology and current status: an overview. Renew Sustain
Energy Rev 6:3–65
Ben Mohamed M (2003) Geothermal resource development in agriculture in Kebili region,
Southern Tunisia. Geothermics 32:505–511
Ben Mohamed M, Saïd M (2008) Geothermal energy development in Tunisia: present status and
future outlook, proceedings, 30th anniversary workshop. UNU-GTP, Reykjavík, Iceland
Theoretical Analysis of the Performance of a Solar Chimney … 27
1 Introduction
The solar chimney power plants (SCPP) are innovative devices constructed for the
generation of solar energy. The SCPP makes a significant contribution to energy
generation in countries where sunlight is available through the year. The solar setup
uses the natural buoyancy of the heated air to harness the energy from the sun.
Then, the airflow is driven by a pressure difference along the chimney pipe. The
construction of an SCPP requires interesting investments. For this reason, the
optimization of the SCPP is essential. In recent years, the number of studies
interested in the numerical methods has remarkably increased aiming at optimizing
the SCPP performance (Ayadi et al. 2017a, b). Maia et al. (2009) evaluated the
impact of the geometrical parameters and the material on the behavior of the airflow
within the SCPP. In their work, they have noted that the dimensions of the chimney
pipe directly affect the design of the solar chimney. They also noted that an increase
in the chimney diameter and height causes an increase in the mass flow rate.
Chergui et al. (2010) analyzed the natural laminar convective heat transfer problem
taking place in the chimney. As a result, they found that the maximum values of the
air velocity were achieved at the chimney pipe entrance. Asnaghi et al. (2012)
carried out a numerical model to evaluate the efficiency of the solar chimney power
plant. In their work, they revealed that the generated power depends on the solar
irradiations, the sunshine duration, and the value of the ambient air temperature.
Indeed, they noted that the generated power is proportional to the rate of the mass
flow. Kasaeian et al. (2014) presented a mathematical model that describes the
airflow inside the prototype of the solar chimney. In their work, they presented the
temperature and the magnitude velocity profiles of the airflow while varying
the collector roof height from h = 60 to 120 mm, the chimney height from
H = 1000 to 4000 mm, and the chimney diameter from d = 50 to 300 mm. In
another work, Ghalamchi et al. (2016) varied the absorbent material and the geo-
metric dimensions using thermal and velocity data of the airflow. Peng-hua et al.
(2014) presented a CFD method including the solar radiation and turbine models.
As a result, they proved that radiation heat transfer is an important parameter in the
heat transfer process within the SCPP and should be considered in the numerical
simulations. Kalash et al. (2013) experimentally analyzed the solar updraft power
plant setup. In their study, they noted that the solar radiation and the ambient
temperature affect the temperature of the chimney inlet. Indeed, the chimney inlet
temperature increases with the increase of the radiation. However, they remarked
that the influence of the ambient air temperature is minimal during the noon period
since the intensity of the solar radiation starts to decrease. Hurtado et al. (2012)
analyzed the thermodynamic behavior of the airflow inside a solar chimney power
plant over a daily operation cycle while considering the soil as a heat storage
system. They have studied the impact of the soil thermal inertia and the soil
compaction degree on the generated power. The analysis has proven that the
generated power increases with the increase of the soil compaction. According to
these previous results, it is clear that the study of the SCPP using the CFD method
has evolved noticeably. In these conditions, the choice of the numerical parameters
such as the turbulence model is needed (Driss et al. 2014, 2016).
In this chapter, we were interested in the study of the turbulence model impact
on the local characteristics of the airflow such as the magnitude velocity, temper-
ature, static pressure, and turbulence characteristics. The choice of the adequate
turbulence model was based on experimental data.
2 Geometrical Parameters
Figure 1 depicts the geometrical design of the solar chimney prototype which is the
same as that of Kasaeian et al. (2014) application. In their work, they studied a
small solar chimney prototype which was constructed at the University of Tehran,
Study of the Turbulence Model Effect on the Airflow … 31
Iran. Tehran city is characterized by the geographical length equal to 51.4° and the
geographical width equal to 35.7°. The geometry of the solar setup is simple. It
consists of a chimney, a collector and an absorber. The chimney’s height and
diameter are H = 2000 mm and d = 200 mm, respectively. The collector diameter
and the roof height are D = 3000 mm and h = 60 mm, respectively. In this
application, a polycarbonate pipe was used as the chimney. It is characterized by a
thickness equal to ech = 4 mm. However, the absorber is made up of 17 pieces of
steel with a thickness equal to est = 2 mm and a wooden chipboard characterized by
a thickness equal to ew = 8 mm. The experimental data of the air temperature
distribution of Kasaeian et al. (2014) were recorded on the June 19–21, 2013, with
the same sunny climatic conditions. The ambient temperature was equal to
T = 306 K, and the global solar radiation was equal to G = 800 W m−2.
3 Numerical Model
The governing equations, describing the airflow inside the solar chimney, were
simulated using the commercial code ANSYS Fluent 17.0. The solar setup is a 2D
symmetric system. The chimney axis is assumed to be symmetric. The ambient
temperature is equal to T = 306 K. The wall boundary was applied for the chimney
with a heat flux value equal to zero since the chimney pipe is supposed to be an
adiabatic wall. Indeed, the convective heat transfer option was applied to the wall of
the absorber and the collector. An inlet pressure boundary was specified for the
collector inlet, and a pressure outlet boundary was applied to the chimney outlet.
32 A. Ayadi et al.
In fact, the pressure inlet is equal to Pin = 101,325 Pa and the pressure outlet is
equal to Pout = 101,325 Pa. On the other hand, the discrete ordinates method
(DO) was used to calculate the radiation heat transfer for the whole system of the
solar setup. These boundary conditions have been taken similar to the experimental
conditions of Kasaeian et al. (2014). The applied boundary conditions are illustrated
in Fig. 2.
3.2 Meshing
The meshing is an important step that precedes the CFD simulations. In fact, the
choice of the best meshing causes accurate computational results. In order to choose
an accurate meshing size with a moderate time, a meshing model with 326,105 cells
was chosen to minimize the resolution calculation time (Fig. 3).
Figure 4 shows a comparison between the numerical profiles of the air temperature
along the collector radius for the considered turbulence models and the experi-
mental data achieved by Kasaeian et al. (2014). From this comparison, it was noted
that all the considered configurations present the same profiles. A difference in
values has been recorded when moving from one case to another. In fact, the
maximum error of the values was reached for the second and third configurations.
However, all the k–ɛ models present a good agreement with the experimental
results. The most negligible error is recorded for the standard k–ɛ model.
Study of the Turbulence Model Effect on the Airflow … 33
Fig. 3 Meshing
5 Numerical Results
Five turbulence models have been tested to choose the accurate turbulence model
which can predict precise solutions of the airflow characteristics inside the solar
chimney power plant. The conditions and parameters of the numerical simulations,
such as the ambient temperature, the solar radiation, and the air pressure, are the
same as Kasaeian’s et al. (2014) application. Particularly, the standard k-e model,
the realizable k-e model, the RNG k-e model, the transition-k–kl–x model, and the
transition-SST model have been simulated.
5.1 Temperature
The distribution of the static temperature in the axisymmetric plane of the solar
chimney power plant for different turbulence models is shown in Fig. 5. From these
results, it can be noted that the variation of the turbulence model has a small impact
on the shape of the distribution of the air temperature inside the SCPP. In fact, the
location of the maximum and the minimum values of the air temperature is inde-
pendent of the turbulence model. For all the cases, the maximum values of the air
temperature are recorded in the absorber and the chimney near the axis. However, a
variation of the maximum values of the air temperature was also remarked for the
considered turbulence models. The comparison between these results showed that
the turbulence model directly affects the air temperature within the solar chimney
power plant. For example, the maximum value of the air temperature for the
standard k–ɛ model, the realizable k–ɛ model, and the RNG k–ɛ model is equal to
T = 331 K. However, the maximum value of the air temperature for the
transition-SST model and the transition-k–kl–x model is equal to T = 333 K.
Figure 6 shows the distribution of the velocity fields in the axisymmetric plane of
the solar chimney for the different considered models. According to these results,
the different turbulence models practically present similar distributions. In fact, an
acceleration zone located in the base of the chimney was observed difference in
values from one model to another. For example, the maximum air velocity value is
V = 2.02 m s−1 using the standard k–ɛ model, V = 2.28 m s−1 using the realizable
k–ɛ model, V = 1.84 m s−1 using the transition-SST model, V = 2.25 m s−1 using
the transition-k–kl–x model, and V = 2.18 m s−1 using the RNG k–ɛ model.
Study of the Turbulence Model Effect on the Airflow … 35
Fig. 5 (continued)
Comparing the air velocity profiles along the chimney axis (Fig. 7), a peak in
values was noticed at the chimney bottom defined by z = 150 mm for the con-
sidered turbulence model types. However, a difference in shape between the con-
sidered turbulence models was clear, especially for the k–kl–x model. The
comparison between these results confirm that the turbulence model type directly
affects the magnitude velocity of the airflow inside the solar setup.
The distribution of the static pressure in the axisymmetric plane of the solar
chimney power plant is illustrated in Fig. 8. According to these results, a com-
pression zone was observed to be located in the collector and in the top of the
chimney. However, the depression zone is located in the base of the chimney for all
configurations. The difference between the considered models appears essentially at
the depression zone recorded at the bottom of the chimney. The occupied area of
the depression zone varies from one turbulence model to another. These results are
confirmed by Fig. 9 which depicts the profiles of the static pressure along the
collector for the different turbulence models. The maximum depression values are
p = 101,321.9 Pa using the standard k–ɛ model, p = 101,321.6 Pa using the real-
izable k–ɛ model, p = 101,322.9 Pa using the transition-SST model,
p = 101,321.8 Pa using the transition-k–kl–x model, and p = 101,321.8 Pa using
the RNG k–ɛ model.
Study of the Turbulence Model Effect on the Airflow … 37
Figure 10 shows the profiles of the dynamic pressure along the chimney axis.
According to these results, the dynamic pressure was remarked to evolve in the
same way as the magnitude velocity for the considered models. This fact is
explained by the analytic expression of the dynamic pressure which is proportional
to the square of the magnitude velocity. The maximum value of the dynamic
Fig. 6 (continued)
pressure appears next to the base of the chimney in z = 180 mm for the considered
configurations. However, the zone, surrounding the maximum values of the
dynamic pressure, is noted to have the most extended along the chimney for the
realizable k–ɛ model. The comparison between these results confirms that
the dynamic pressure is dependent on the turbulence model. In fact, dynamic
pressure values vary from one case to another. For example, the dynamic pressure is
Study of the Turbulence Model Effect on the Airflow … 39
101325,0 101325,0
101324,9 101324,9
101324,7 101324,7
101324,6 101324,5
101324,4 101324,3
101324,2 101324,2
101324,1 101324,0
101323,9 101323,8
101323,5 101323,3
101323,3 101323,2
101323,2 101323,0
101323,0 101322,8
101322,7 101322,5
101322,5 101322,3
101322,4 101322,1
101322,2 101322,0
101321,9 101321,6
101325,0 101325,0
101324,9 101324,9
101324,8 101324,7
101324,7 101324,5
101324,6 101324,4
101324,4 101324,2
101324,3 101324,1
101324,2 101323,9
101323,9 101323,4
101323,7 101323,3
101323,6 101323,1
101323,5 101323,0
101323,3 101322,6
101323,2 101322,5
101323,0 101322,3
101322,9 101322,2
101322,7 101321,8
101325,0
101324,9
101324,7
101324,5
101324,4
101324,2
101324,0
101323,9
101323,4
101323,2
101323,1
101322,9
101322,6
101322,4
101322,3
101322,1
101321,8
Fig. 8 (continued)
pd = 2.92 Pa using the realizable k–ɛ model, but it is pd = 2.82 Pa using the
transition-SST model.
Figure 11 shows the distribution of the turbulent kinetic energy in the axisymmetric
plane of the solar chimney system. According to these results, it is clear that the
turbulent kinetic energy is very weak in the test section except at the chimney inlet.
Fig. 10 (continued)
From one turbulence model to another, no differences in shapes nor in values were
noted. The maximum value of the turbulent kinetic energy is k = 0.271 m2 s−2
using the standard k–ɛ model, k = 0.281 m2 s−2 using the realizable k–ɛ model,
k = 0.266 m2 s−2 using the transition-SST model, k = 0.411 m2 s−2 using the
transition-k–kl–x model, and k = 0.285 m2 s−2 using the RNG k–ɛ model. The
comparison between these results confirms that the distribution of the turbulent
kinetic energy depends on the turbulent model.
Figure 12 shows the dissipation rate distribution of the turbulent kinetic energy in
the axisymmetric plane of the solar chimney. According to these results, it is clear
that the variation of the turbulence model has a direct effect on the dissipation rate
Study of the Turbulence Model Effect on the Airflow … 43
Fig. 11 (continued)
distribution of the turbulent kinetic energy. Indeed, it has been observed that
the dissipation rate of the turbulent kinetic energy reaches its maximum value at the
chimney inlet for the transition-SST model and the transition-k–kl–x model. For the
other considered models, the recorded dissipation rate of the turbulent kinetic
energy dissipation is negligible. The comparison between these results confirms that
the dissipation rate distribution of the turbulent kinetic energy depends on the
turbulent model. In fact, the maximum value of the dissipation rate is
ɛ = 535 m2 s−3 using the standard k–ɛ model, ɛ = 565 m2 s−3 using the realizable
k–ɛ model, ɛ = 17.9 m2 s−3 using the transition-SST model, ɛ = 0.967 m2 s−3
using the transition-k–kl–x model, and ɛ = 551 m2 s−3 using the RNG k–ɛ model.
Study of the Turbulence Model Effect on the Airflow … 45
Fig. 12 (continued)
6 Conclusion
In this work, several numerical simulations of the airflow within the SCPP setup
were performed. The turbulence model effect on the computational results was
studied and discussed. The numerical results show that the turbulence model choice
is an interesting step that should be considered by CFD users. Otherwise, the results
confirm that the distribution of the airflow characteristics depends on the turbulence
model. In fact, the local airflow characteristics such as velocity magnitude, tem-
perature, pressure, and turbulence vary from one turbulence model to another. For
our future numerical simulations, we will be focusing on the standard k-e model
since it has shown a very good agreement with the experimental results.
References
Asnaghi A, Ladjevardi SM (2012) Solar chimney power plant performance in Iran. Renew Sustain
Energy Rev 16:3383–3390
Ayadi A, Driss Z, Bouabidi A, Abid MS (2017a) A computational and an experimental study on
the effect of the chimney height on the thermal characteristics of a solar chimney power plant.
Proc Inst Mech Eng—Part E:J Process Mech Eng. https://doi.org/10.1177/0954408917719776
Ayadi A, Driss Z, Bouabidi A, Abid MS (2017b) Experimental and numerical study of the impact
of the collector roof inclination on the performance of a solar chimney power plant. Energy
Build 139:263–276
Chergui T, Larbi S, Bouhdjar A (2010) Thermo-hydrodynamic aspect analysis of flows in solar
chimney power plants—a case study. Renew Sustain Energy Rev 14:1410–1418
Driss Z, Mlayeh O, Driss S, Driss D, Maaloul M, Abid MS (2014) Study of the bucket design
effect on the turbulent flow around unconventional Savonius wind rotors. Energy 89:708–729
Study of the Turbulence Model Effect on the Airflow … 47
Driss Z, Mlayah O, Driss S, Maaloul M, Abid MS (2016) Study of the incidence angle effect on
the aerodynamic structure characteristics of an incurved Savonius wind rotor placed in a wind
tunnel. Energy 113:894–908
Hurtado FJ, Kaiser AS, Zamora B (2012) Evaluation of the influence of soil thermal inertia on the
performance of a solar chimney power plant. Energy 47:213–224
Kalash S, Naimeh W, Ajib S (2013) Experimental investigation of the solar collector temperature
field of a sloped solar updraft power plant prototype. Sol Energy 98:70–77
Kasaeian A, Ghalamchi M, Ghalamchi M (2014) Simulation and optimization of geometric
parameters of a solar chimney in Tehran. Energy Convers Manag 83:28–34
Maia CB, Ferreira AG, Valle RM, Cortez MFB (2009) Theoretical evaluation of the influence of
geometric parameters and materials on the behavior of the air flow in a solar chimney. Comput
Fluids 38:625–636
Ghalamchi M, Kasaeian A, Ghalamchi M, Mirzahosseini A (2016) An experimental study on the
thermal performance of a solar chimney with different dimensional parameters. Renew Energy
91:477–483
Peng-hua G, Li J, Yuan W (2014) Numerical simulations of solar chimney power plant with
radiation model. Renew Energy 62:24–30
Numerical Study of the Fluid
Characteristics Effect on the Penetration
of a Negatively Buoyant Jet
1 Introduction
When a lighter fluid is injected vertically downwards into a denser ambient fluid, its
momentum is continually being decreased by buoyancy forces until the vertical
velocity becomes zero at some finite distance from the source. As the jet reaches its
maximum penetration length hmax, it reverses its direction and rises back to the top
of the tank. Such jets are called negatively buoyant jets or fountains, and the density
difference between the ambient and the injected fluids may be due to a variation in
either chemical composition or temperature. Negatively buoyant jets or fountains
are common in both engineering and natural science. An everyday example is the
ventilation of large open structures such as aircraft hangars, which are heated using
ceiling-mounted fans to drive hot air toward the floor. In nature, the geophysical
buoyant jets resulting from temperature (or salinity) differences can occur in magma
chambers and in the ocean (Campbell and Turner 1989; Turner and Campbell
1986). During the last 50 years, the behavior of negatively buoyant jets or fountains
has been widely explored theoretically and experimentally (Morton 1959;
Papanicolaou and Kokkalis 2008). Since the pioneering work of Morton (1959),
2 Geometrical Arrangement
The system investigated in this study consists of a cubic tank opened in the top as
presented in Fig. 1. It had a length L = 30 mm, a width l = 30 mm, and a height
H = 30 mm. Above the tank full of salt water and in the middle, a square nozzle for
injection with an inner length D = 1.37 mm was placed inside the tank about
h = 1 mm in height. Through this nozzle, pure water was injected in a laminar
regime to obtain a jet flow in the surrounding salt water.
3 Numerical Method
Using the CFD code Open Foam, the jet flow was described by the Navier–Stokes
equations with the volume of fluid model (VOF). The computational domain was
split into four cubes because of the problem symmetry. Therefore, the quarter of the
system was taken with symmetrical planes. A value U was taken for the inlet
velocity. This velocity value can change with the flow rate and the nozzle diameter.
For the outlet pressure, a value of p = 101325 Pa was considered. Figure 2 shows
Fig. 1 Geometrical
arrangement
52 O. Eleuch et al.
(a) 2-D view of the mesh (b) 3-D view of the mesh
Fig. 3 Meshing
the boundary conditions in the initial state of flow colored by volume fraction where
the red color corresponds to a = 1 (pure water with a density q and kinematic
viscosity m0) and the blue color corresponds to a ¼ 0 (salt water with a density
q + Δq with Δq ˃ 0 and kinematic viscosity m).
A three-dimensional, non-uniform hexahedral grid was used in this study, with a
high-density mesh in regions near the inlet, outlet, walls, and jet boundaries where
high gradients were expected. The numerical results were obtained with a number
of cells equal to N = 222337 as sketched in Fig. 3.
Numerical Study of the Fluid Characteristics Effect … 53
Throughout our study, the parameters were varied as follows: the flow rate Q with three
different values: Q = 0.047 cm3 s−1, Q = 0.065 cm3 s−1, and Q = 0.086 cm3 s−1, the
nozzle inner diameter D was also varied in two values: D = 0.838 mm and
D = 1.37 mm, the relative difference of density between inner and outer liquids are
−3 Dq −3 Dq −3 Dq −2
defined by Dq q = 2 10 , q = 4.36 10 , q = 6.98 10 , q = 1.31 10 ,
−2
and Dq q = 2.59 10 . The kinematic viscosity of the pure water is equal to
m0 = 10 m s , while that of the salt water m is equal to m = m0 = 10−6 m2 s−1 in the
−6 2 −1
first case. However, the second case is defined by a kinematic viscosity m = 3.7 m0.
Finally, to study the effect of diffusivity, we varied the molecular diffusivity in five
cases defined by km = 1.5 10−9 m2 s−1, km = 10−8 m2 s−1, km = 10−7 m2 s−1,
km = 10−6 m2 s−1, and km = 10−5 m2 s−1.
In this section, we were interested in the study of the effect of molecular diffusivity
between the inner and outer liquids. In this study, the conditions were defined by
the flow rate Q = 0.047 cm3 s−1, a nozzle diameter D = 1.37 mm, a relative dif-
−3
ference of density between the two liquids Dq q = 4.36 10 , and a kinematic
−6 2 −1
viscosity m = m0 = 10 m s .
Figures 4, 5, 6, 7, and 8 show the distribution of the volume fraction of pure water
injected in the tank full of salt water for the four cases of molecular diffusivity km
for the different water flow instances t = 0.25 s, t = 0.5 s, t = 1 s, t = 1.5 s, and
t = 2 s with the same conditions. According to these results, it has been observed
that the increase in the molecular diffusivity km causes the disappearance of the
body of the jet where the jet mixes with the salt water. This transformation appears
from the molecular diffusivity value km = 10−6 m2 s−1 which has the same value of
the kinematic viscosity m. The Schmidt number Sc related to the molecular diffu-
sivity with the kinematic viscosity affects the diffusion of the jet and is written as
follows:
m
Sc ¼ ð1Þ
km
Thus, when Sc > 1, the jet preserves its shape with a low interface between the jet and
the stable liquid as shown in the case of km = 10−8 m2 s−1 and km = 10−7 m2 s−1,
54 O. Eleuch et al.
(a) km=10-8 m2.s-1 (b) km=10-7 m2.s-1 (c) km=10-6 m2.s-1 (d) km=10-5 m2.s-1
(a) km=10-8 m2.s-1 (b) km=10-7 m2.s-1 (c) km=10-6 m2.s-1 (d) km=10-5 m2.s-1
whereas for Sc < 1 as presented in the case of km = 10−5 m2 s−1, the body of the jet
disappears completely. Besides, there is no boundary layer between the jet and the outer
liquid. Consequently, there is a total miscibility at all instances. For Sc = 1 and precisely
in the case of km = m = 10−6 m2 s−1, the interface between the jet and the salt water can
be distinguished, but the jet loses its head and there is more miscibility compared to
Sc > 1.
Numerical Study of the Fluid Characteristics Effect … 55
(a) km=10-8 m2.s-1 (b) km=10-7 m2.s-1 (c) km=10-6 m2.s-1 (d) km=10-5 m2.s-1
(a) km=10-8 m2.s-1 (b) km=10-7 m2.s-1 (c) km=10-6 m2.s-1 (d) km=10-5 m2.s-1
Figure 9 shows the profile of the penetration depth for many cases of molecular
diffusivity km varied from km = 1.5 10−9 m2 s−1 to km = 10−5 m2 s−1 over time.
According to these results, it was observed that the penetration depth increases with
km. For km = 10−5 m2 s−1, the penetration depth is very important and the jet
56 O. Eleuch et al.
(a) km=10-8 m2.s-1 (b) km=10-7 m2.s-1 (c) km=10-6 m2.s-1 (d) km=10-5 m2.s-1
continues to slow down without having a steady-state regime due to the total
miscibility between the pure water jet and the salt water.
4.1.3 Mixing Layer Between the Jet and the Salt Water
For different molecular diffusivity km, Fig. 10 illustrates the mixing layer curve as a
function of time for km = 1.5 10−9 m2 s−1, km = 10−8 m2 s−1, and km = 10−7
m2 s−1. According to these results, it was remarked that the mixing layer dm is
Numerical Study of the Fluid Characteristics Effect … 57
important for km = 10−7 m2 s−1 compared to the two first cases. For km = 10−6
m2 s−1 and km = 10−5 m2 s−1. dm is not presented in Fig. 10 since the miscibility is
very important and we cannot estimate the interface between the inner and the outer
liquids. That is why, Fig. 11 presents the profile of the volume fraction for km =
10−7 m2 s−1, km = 10−6 m2 s−1, and km = 10−5 m2 s−1 as a function of the width x
at t = 1 s. In these conditions, it has been observed that for km = 10−7 m2 s−1,
corresponding to the beginning of the width, the volume fraction curve contains a
small landing in a = 0. This fact indicates the presence of pure water only. This
slope indicates the mixing layer between the jet and the outer liquid and for a = 0
until the end of the tank width, containing only the salt water. For km = 10−6 m2
s−1, the curve starts with a slightly below 1. This fact can be explained by the body
of the jet which weakly mixes with the salt water. In these conditions, the slope is
larger than in the first case. Thus, there is more mixing layer. In the last case, where
km = 10−5 m2 s−1, there is a mix of the jet and salt water (a = 0.46) from the
beginning with a hyperbolic form of the slope presenting the dispersion of the
mixture in the tank.
To study the effect of the relative difference of the density between the outer and
3 −1
q , we have fixed the values of the flow rate to Q = 0.065 cm s , the
inner liquids Dq
molecular diffusivity to km = 1.5 10−9 m2 s−1, and the diameter of the nozzle to
−3
D = 0.838 mm. Different density differences were considered: Dq q = 4.36 10 ,
Dq
q = 6.98 10−3, and Dq −2
q = 2.59 10 . The density difference between two
liquids was obtained by adding a quantity of salt in the tank full of water.
58 O. Eleuch et al.
Figures 12, 13, 14, 15, and 16 show the evolution of the volume fraction of the jet
for three cases of Dq
q at different instances: t = 0.25 s, t = 0.5 s, t = 1 s, t = 1.5 s,
and t = 2 s. According to these results, it is clear that when the difference of the
density increases, the penetration of the flow decreases and there is no big differ-
ence in the form of the jet head. In these conditions, the decrease in the depth of the
jet results from the increase in the buoyancy B0 that is expressed as follows:
Dq
B0 ¼ g Q ð2Þ
q
The buoyancy and the relative difference of the density are proportional leading
the jet to rise back quickly.
The density difference effect on the jet penetration depth is shown in Fig. 17.
According to these results, it was observed that the penetration depth increases
Δρ Δρ Δρ
(a) = 4.36 10-3 (b) = 6.98 10-3 (c) = 2.59 10-2
ρ ρ ρ
Δρ Δρ Δρ
(a) = 4.36 10-3 (b) = 6.98 10-3 (c) = 2.59 10-2
ρ ρ ρ
when the difference of the density decreases. At t = 2 s, the jet reaches the maximal
−2 −3
depth Hm = 0.00113 m for Dq q = 2.59 10 although for Dq q = 6.98 10 , the
maximal penetration depth of the jet is equal to Hm = 0.0203 m at t = 2.25 s.
However, at t = 2.5 s the maximal depth Hm = 0.0255 m is obtained for
−3
q = 4.36 10 . Thus, the jet quickly reaches the steady-state regime at a large
Dq
Δρ Δρ Δρ
(a) = 4.36 10-3 (b) = 6.98 10-3 (c) = 2.59 10-2
ρ ρ ρ
Figure 18 shows the width of the jet for the different density differences
−3 Dq −3 −2
q = 4.36 10 , q = 6.98 10 , and q = 2.59 10 . According to these
Dq Dq
results, the width of the jet decreases when the difference of the density increases.
The large difference of density means that the jet has become lighter at the same
time the outer liquid has become denser. This difference allows obtaining less width
of the jet with a small penetration depth. Consequently, the variation of the dif-
ference of the density between the jet and the surrounding liquid presenting a direct
effect on the behavior of the jet inside the full tank of a denser liquid.
62 O. Eleuch et al.
Δρ Δρ Δρ
(a) = 4.36 10-3 (b) = 6.98 10-3 (c) = 2.59 10-2
ρ ρ ρ
In this part, we are interested on the study of the kinematic viscosity effect of the
denser water with a constant viscosity of the lighter one. In these conditions, the
variation of the kinematic viscosity depends on the temperate of the salt water. The
flow rate is equal to Q = 0.086 cm3 s−1, and the diameter of the nozzle is equal to
D = 1.37 mm with the same molecular diffusion equal to km = 1.5 10−9 m2 s−1.
Numerical Study of the Fluid Characteristics Effect … 63
Δρ Δρ Δρ
(a) = 4.36 10-3 (b) = 6.98 10-3 (c) = 2.59 10-2
ρ ρ ρ
Figures 19, 20, 21, 22, 23, and 24 present the jet volume fraction evolution for the
two kinematic viscosity values of salt water m = m0 and m = 3.7 m0 and for the
following different instances: t = 0.25 s, t = 0.5 s, t = 1 s, t = 1.5 s, t = 2 s, and
t = 3 s. In these conditions, the jet has a constant kinematic viscosity equal to
m0 = 10−6 m2 s−1. The rise of viscosity allows the flow to resist more the friction
between the two liquids which explains the decrease in the shape of the jet in depth
and width compared to the first case in each instance. From these results, it was
observed that the head of the jet contains more pure water (a = 1) and less mixing
64 O. Eleuch et al.
layer when the viscosity of the salt water is important. Thus, the miscibility in the
interface between the jet and the stable water decreases with the increase in the
kinematic viscosity.
66 O. Eleuch et al.
Figure 25 presents the variation of the transient penetration depth of the head of the
jet for m = m0 and m = 3.7 m0. According to these results, it is clear that the increase
in the kinematic viscosity of the salt water allows the jet penetration depth to
68 O. Eleuch et al.
decrease because of the friction and resistance of the jet in the outer liquid. For
m = 3.7 m0, the maximal penetration depth reaches Hm = 0.0116 m at t = 1.75 s.
However, for m = m0, the jet reaches the maximal depth Hm = 0.014 m at t = 1.5 s.
Indeed, just like the nozzle diameter variation case and the density difference, the jet
reaches the maximal depth and the steady-state regime in a short time. The greatest
penetration depth was observed for low kinematic viscosity of the outer liquid.
Figure 26 shows the penetration profile as function of the depth of the jet for
different kinematic viscosities equal to m = m0 and m = 3.7 m0. According to these
Numerical Study of the Fluid Characteristics Effect … 69
results, the head of the jet is more expelled radially when it reaches the steady-state
regime with an important mixing layer in the interface between the two liquids for a
low kinematic viscosity of salt water. Thus, the variation of proprieties of two
liquids in term of viscosity presents a direct effect on the evolution of the jet flow
during its penetration inside the stable liquid.
5 Conclusion
References
Addad Y, Benhamadouche S, Laurence D (2004) The negatively buoyant wall-jet: LES results.
Int J Heat Fluid 25:795–808
Bashitialshaaer R, Larson M, Persson KM (2012) An experimental investigation on inclined
negatively buoyant jets. Water 4:720–738
Campbell IH, Turner JS (1989) Fountains in magma chambers. J Petrol 30:885–923
Christodoulou GC, Papakonstantis IG (2010) Simplified estimates of trajectory of inclined
negatively buoyant jets. Environmental hydraulics. Taylor & Francis, London, UK, pp 165–
170
Ferrari S, Querzoli G (2010) Mixing and re-entrainment in a negatively buoyant jet. J Hydraul Res
48:632–640
Friedman P, Katz J (2000) Rise height for negatively buoyant fountains and depth of penetration
for negatively buoyant jets impinging an interface (trans: ASME I). J Fluids Eng 122:779–782
Inghilesi R, Stocca V, Roman F, Armenio V (2008) Dispersion of a vertical jet of buoyant particles
in a stably stratified wind-driven Ekman layer. Int J Heat Fluid 29:733–742
Kapoor K, Jaluria Y (1993) Penetrative convection of a plane turbulent wall jet in a two-layer
thermally stable environment: a problem in enclosure fires. Int J Heat Mass Transf 36:155–167
Morton BR (1959) Forced plumes. J Fluid Mech 5:151–163
Papanicolaou PN, Kokkalis TJ (2008) Vertical buoyancy preserving and non-preserving fountains
in a homogeneous calm ambient. Int J Heat Mass Transf 51:4109–4120
Papakonstantis IG, Christodoulou GC, Papanicolaou PN (2011) Inclined negatively buoyant jets 1:
geometrical characteristics. J Hydraul Res 49:3–12
70 O. Eleuch et al.
Shy S (1995) Mixing dynamics of jet interaction with a sharp density interface. Expl Therm Fluid
Sci 10:355–369
Torrecilla I MM, Geyer A, Phillips JC, Idelsohn I SR, Oñate E (2010) Numerical simulations of
negatively buoyant jets in an immiscible fluid using the particle finite element method. Int J
Numer Meth Fluids 69:1016–1030
Toyoshima M, Okawa S (2013) An effect of a horizontal buoyant jet on the temperature
distribution inside a hot water storage tank. J Fluid Mech 440:403–413
Turner JS, Campbell IH (1986) Convection and mixing in magma chambers. Earth Sci Rev
23:255–352
Computer Simulation of Liquid Motion
in a Container Subjected to Sinusoidal
Excitation with Different Turbulence
Models
1 Introduction
The sloshing phenomenon appears in containers partially filled with liquid and
subjected to an external excitation. These phenomena occur in different tank
geometries under different excitation types. Several works have been conducted to
study and analyze the liquid sloshing. For example, Arafa (2007) developed a
numerical model to study the sloshing phenomena based on the finite element
formulation and analyze the baffle effect. Belakroum et al. (2010) predicted the
damping effect of baffles on sloshing in tanks partially filled with liquid. They found
that the fluid sloshing increases significantly, when the frequency of the ground
excitation approaches the natural frequencies of the fluid tank. To reduce the liquid
sloshing mainly at resonance, they investigated the use of the baffles, a horizontal
baffle and a vertical one. They showed that the baffle normal to the free surface has
an important damping effect. Godderidge et al. (2009) investigated the liquid
sloshing using multiphase CFD technique in a partially filled container subjected to
a lateral excitation. Panigrahy et al. (2009) conducted a series of experiments to
study the baffle effect on reducing the pressure on the container walls and the
sloshing violence. Bouabidi et al. (2013, 2016a) studied numerically the efficiency
of the vertical baffle in reducing the liquid motion. Bouabidi et al. (2015) analyzed
the time step size effect on the numerical simulation of the liquid sloshing problem.
They validated the numerical model by comparing their numerical results with
those of previous studies. They found that the time step value affects the numerical
results, and the choice of the optimum time step value leads to a good agreement for
the comparison with the numerical and the experimental results. Bouabidi et al.
2 Geometry Configuration
As shown in Fig. 1, the 2-D geometry is defined by a length L = 0.6 m and a height
equal to H = 0.6 m. The tank is partially filled with liquid with the height
h = 0.1 m. The geometrical arrangement is similar to the geometry of Panigrahy
et al. (2009).
L
Computer Simulation of Liquid Motion in a Container … 73
3 Numerical Model
The commercial CFD code “Fluent” was used to develop the simulation of the
sloshing problem in the partially filled tank.
The mathematical model is given by the Navier–Stokes equations defined as
follows:
@q @
þ ðqui Þ ¼ 0 ð1Þ
@t @xi
@ @ @p @ @ui @uj
ð q ui Þ þ q ui uj ¼ þ l þ þ Fi ð2Þ
@t @xj @xi @xj @xj @xi
@2X
Fi ¼ qgj þ q ð3Þ
@t2
4 Numerical Results
The numerical results are presented over time at four different instants during the
time period of the external excitation. The considered instants are equal to t = T/
4 = 1 s, t = T/2 = 2 s, t = 3T/4 = 3 s, and t = T = 4 s.
These results confirm that the turbulence model has a great effect, and the choice of
the optimum turbulence model is essential to develop a validated numerical model.
Figures 6, 7, 8, and 9 depict the static pressure distribution in the whole volume of
the tank over time at t = 1 s, t = 2 s, t = 3 s, and t = 4 s, respectively. The liquid
motion is presented for the different considered turbulence models: standard k–e,
RNG k–e, Realizable k–e, and standard k–e. According to these results, it can be
noted that the sloshing loads generate an inhomogeneous distribution of the static
pressure in the container. The container bottom is characterized by the highest
pressure value for the different considered cases. The static pressure distribution
depends on the external excitation direction. At t = 1 s and t = 2 s, the compression
zones are very important at the left corner since the liquid motion occurs from the
Computer Simulation of Liquid Motion in a Container … 75
right to the left. However, the compression zones are very important at the right
corner at t = 3 s and t = 4 s since the liquid motion occurs from the right to the left.
On the other hand, the numerical results show that the static pressure distribution
varies from turbulence model to another. In fact, the maximum static pressure value
is different from one case to another over time. These results confirm that the choice
of the optimum turbulence model is essential to develop numerical simulation to
study the sloshing phenomena.
Figures 10, 11, 12, and 13 depict the magnitude velocity over time at t = 1 s,
t = 2 s, t = 3 s, and t = 4 s, respectively. The magnitude velocity is presented for
the different considered turbulence models: standard k–e, RNG k–e, Realizable k–e
and standard k–e. According to these results, it has been noted that the turbulence
76 A. Bouabidi et al.
model used in the numerical simulation affects significantly the magnitude velocity
appeared under the liquid sloshing in the partially filled container. In fact, the
results of the magnitude velocity are very different from turbulence model to
another. The location of the maximum value varies from one case to another for the
different considered instants equal to t = 1 s, t = 2 s, t = 3 s, and t = 4 s. At
t = 1 s, the maximum value of the velocity is located close to the left wall on the
free surface for the cases of the standard k–e and the RNG k–e turbulence models,
whereas it is located in the center of the container close to the bottom wall for the
cases of the realizable k–e and the standard k–e turbulence models. At t = 2 s, the
maximum velocity is located close to the right wall on the free surface for the cases
of the standard k–e, the RNG k–e, and the realizable k–e turbulence models.
However, it is located close to the left wall on the free surface for the cases of the
standard k–e turbulence model. At t = 3 s, the location of the maximum value is
similar for the different considered models. At t = 4 s, the maximum velocity value
is observed on the free surface close to the left and the right walls for the cases of
the standard k–e turbulence model, whereas it appears in the tank center for the case
Computer Simulation of Liquid Motion in a Container … 77
of the RNG k–e turbulence model. For the case of the realizable k–e turbulence
model, the highest velocity value appears near the right wall on the free surface.
However, it appears near the left wall for the standard k–e turbulence model. These
results confirm that the magnitude velocity significantly depends on the used tur-
bulence model.
Figures 14, 15, 16 and 17 depict the turbulent kinetic energy distribution over time,
respectively, at t = 1 s, t = 2 s, t = 3 s, and t = 4 s. The liquid motion is presented
for the different considered turbulence models: standard k–e, RNG k–e, Realizable
k–e, and standard k–e. According to these results, it can be noted that sloshing
generated the appearance of an inhomogeneous distribution of the turbulent kinetic
energy in the whole volume of the container. This distribution varies over time for
the different considered models. In addition, it was observed that the turbulence
78 A. Bouabidi et al.
model used in the numerical simulation affects significantly the turbulent kinetic
energy appearing in the partially filled container. At t = 1 s, the distribution of the
turbulent kinetic energy is very similar for the considered turbulence models. The
maximum value appears in the container center, whereas the minimum value
appears near the walls. At t = 2 s, the distribution of the turbulent kinetic energy is
very different from one turbulence model to another. In fact, the maximum value
appears in the container center, whereas the minimum value appears near the walls
for the cases of the standard k–e and the standard k–e turbulence models. However,
the maximum value appears on the free surface close to the right wall for the RNG
k–e and the Realizable k–e turbulence models. At t = 3 s, the maximum value
appears close to the right wall on the free surface for the cases of the RNG k–e and
the Realizable k–e turbulence models, whereas it appears in the container center for
the cases of the standard k–e and the standard k–e turbulence models. At t = 3 s, the
maximum turbulent kinetic energy value appears on the free surface close to the
right wall for the cases of the standard k–e, the Realizable k–e, and the standard k–e
turbulence models. However, it appears in the container center for the case of the
RNG k–e turbulence model. At t = 4 s, the highest value of the turbulent kinetic
Computer Simulation of Liquid Motion in a Container … 79
energy has been observed on the free surface close to the right wall for the standard
k–e turbulence model, whereas it appears close to the left wall for the case of the
Realizable k–e and the standard k–e. For the RNG k–e turbulence model, the
maximum value appears in the container center. The above results confirm that
the turbulent kinetic energy depends significantly on the turbulence model used in
the numerical simulations.
Figures 18, 19, 20, and 21 depict the dissipation rate of the turbulent kinetic energy
distribution over time at t = 1 s, t = 2 s, t = 3 s, and t = 4 s, respectively. The
liquid motion is presented for the different considered turbulence models: standard
k–e, RNG k–e, Realizable k–e, and standard k–e. According to these results, it has
been noted that sloshing generates the appearance of an inhomogeneous distribution
of the dissipation rate of the turbulent kinetic energy in the whole volume of the
80 A. Bouabidi et al.
container. In addition, it has been observed that the dissipation rate of the turbulent
kinetic energy distribution is affected by the turbulence model. At t = 1 s, the
dissipation rate of the turbulent kinetic energy is null in the whole volume of the
container, whereas the maximum value appears near the walls for the standard k–e
and the Realizable k–e turbulence models. For the RNG k–e turbulence model, the
maximum dissipation rate of the turbulent kinetic energy appears on the free surface
close to the left wall. However, it appears close to the right wall for the case of the
standard k–e turbulence model. At t = 2 s, the same observation of the previous
instant is noted for the turbulence model k–e standard. However, the maximum
dissipation rate of the turbulent kinetic energy value appears on the free surface
close to the right wall for the RNG k–e, the Realizable k–e, and the standard k–e
Computer Simulation of Liquid Motion in a Container … 81
In this section, the experimental data of Panigrahy et al. (2009) were compared to
the numerical results for the different turbulence models considered in this study:
standard k–e, RNG k–e, Realizable k–e, and standard k–e. The comparison is given
for the static pressure evolution in the point P1 (0, 0.05) as shown in Fig. 22.
According to these results, it can be noted that the static pressure varies peri-
odically over time for the different turbulence models (Fig. 23). The maximum
value of the static pressure was observed for the Realizable k–e turbulence model,
whereas the minimum value was noted for the standard k–e turbulence model. On
the other hand, it has been remarked that the static pressure behavior over time
varies with the turbulence model. The comparison with experimental results con-
firms that the standard k–e turbulence model gives a good agreement.
Computer Simulation of Liquid Motion in a Container … 83
P1 x
94 A. Bouabidi et al.
6 Conclusion
In this chapter, the effect of the turbulence model on the numerical simulation of the
liquid sloshing phenomenon in a partially filled container was studied and dis-
cussed. A series of numerical simulations were conducted to predict the liquid
motion for the different turbulence models: standard k–e, RNG k–e, Realizable k–e,
and standard k–e. The numerical results presented and analyzed over time at four
instants: t = 1 s, t = 2 s, t = 3 s, and t = 4 s show that the used turbulence model
affects significantly the results of the numerical simulation. The comparison with
the experimental results shows that the standard k–e turbulence model gives the
greatest satisfaction.
References
Arafa M (2007) Finite element analysis of sloshing in rectangular liquid-filled tanks. J Vib Control
13:883–903
Belakroum R, Kadja M, Mai TH, Maalouf C (2010) An efficient passive technique for reducing
sloshing in rectangular tank partially filled with liquid. Mech Res Commun 37:341–346
Bouabidi A, Driss Z, Abid MS (2013) Vertical baffles height effect on liquid sloshing in an
accelerating rectangular tank. Int J Mech Appl 3:105–116
Bouabidi A, Driss Z, Abid MS (2015) Time step size effect on the liquid sloshing phenomena. Int J
Fluid Mech Thermal Sci 1:8–13
Computer Simulation of Liquid Motion in a Container … 95
Bouabidi A, Driss Z, Abid MS (2016a) Numerical investigation of the baffle effect on the
hydrodynamic structure of a container partially filled with liquid. Am J Mech Eng 4:112–123
Bouabidi A, Driss Z, Kossentini M, Abid MS (2016b) Numerical and experimental investigation
of the hydrostatic pump in a battery cell with mixing element. Arab J Sci Eng 41:1595–1608
Bouabidi A, Driss Z, Abid MS (2016c) Study of hydrostatic pump created under liquid sloshing in
a rectangular tank subjected to external excitation. Int J Appl Mech 08:1–15
Bouabidi A, Driss Z, Cherif N, Abid MS (2016d) Computational investigation of the external
excitation frequency effect on liquid sloshing phenomenon. WSEAS Trans Fluid Mech 11:1–9
Bouabidi A, Driss Z, Abid MS (2016e) Numerical study on the liquid sloshing in a battery cell
equipped with new baffle design. Int J Mech Appl 6:31–38
Godderidge B, Turnock S, Tan M, Earl C (2009) An investigation of multiphase CFD modelling
of a lateral sloshing tank. Comput Fluids 38:183–193
Panigrahy PK, Saha UK, Maity D (2009) Experimental studies on sloshing behavior due to
horizontal movement of liquids in baffled tanks. Ocean Eng 36:213–222
Numerical Investigation for a Vanned
Mixed Flow Turbine Volute Under Steady
Conditions
Nomenclature
C Absolute flow velocity, m s−1
Cis Spouting velocity or isentropic velocity, m s−1
h Enthalpy per unit mass, J kg−1
k Turbulence kinetic energy, J kg−1
Kp Total pressure loss coefficient, dimensionless
m_ Mass flow, kg s−1
pffiffiffiffi
MFP Mass flow parameter, kg s−1 K Pa−1
P Pressure, Pa
PR Pressure ratio, dimensionless
r Radius, m
S Swirl coefficient, dimensionless
T Temperature, K
U Blade tip velocity, m s−1
V Velocity, m s−1
Greeks
a Absolute flow angle, (°)
e Turbulence dissipation rate, m2 s−3
f Loss coefficient, dimensionless
η Isentropic efficiency, dimensionless
w Azimuth angle, (°)
Subscripts
ex Exit
in Inlet
is Isentropic condition
ts Total to static
S Stator
h Tangential component
0 Stagnation condition
1 Volute inlet
2 Volute exit
3 Vane exit
4 Rotor exit
1 Introduction
2 Turbine Geometry
The present turbine consists of a mixed flow rotor and a nozzleless volute. Figure 1
gives a three-dimensional view of the full turbine stage geometry obtained by an
assembly of different designed turbine parts. Table 1 presents the overall rotor and
volute geometrical parameters adopted in this work. The mixed flow rotor is a
combination between radial and axial turbines and becomes more utilized in tur-
bocharger applications because of its advantage of working efficiently at lower
velocity ratios. This present rotor was cited as the rotor ‘A’ in several anterior works
and was designed previously by Abidat (1991) at the Imperial College Laboratory.
The blade shape of the rotor was designed with polynomial Bezier curves using
BladeGen and Blade Modeler packages. However, the volute is a commercial one
which was modified to acquire a nozzle ring allowing volute nozzleless and nozzled
configurations (Rajoo 2007). In this current work, we are interested in the volute
nozzleless configuration designed using the CFturbo software.
In fact, the volute cross sections have approximately non-symmetrical trape-
zoidal shapes which are drawn by means of the radial based method. Figure 2a
shows the distribution of the ratio of the cross-sectional area to its centroid radius as
a function of the volute azimuthal angle.
However, Fig. 2b illustrates an example of the cross-sectional shape for 0° of the
azimuth angle. In addition, an inlet duct having 400 m of length is used for the
volute flow entry. In order to visualize flow parameter contours, different transverse
and cross-sectional planes were defined as shown in Fig. 3.
3 Numerical Model
This section presents the numerical model solved using the CFX 17.0 package
which includes a finite volume method for the computational domain discretization.
The conducted simulations are based on Reynolds-averaged Navier–Stokes equa-
tions of mass, momentum, and energy conservation for a compressible flow. The
standard k–e turbulence model was used to model additional terms used in these
equations. This turbulence model showed good capabilities to compute the flow in
several anterior aerodynamic works (Driss et al. 2014, 2016a, b).
The meshing is an important step for a numerical since the numerical results’
accuracy is considerably related to the mesh quality, the cells number, and element
types. The computational domain of the present turbine was divided into tetrahedral
elements constituting the generated unstructured mesh using the ICEM-CFD 17.0
software. This software presents many available meshing features to better adjust
the numerical model to the real physical problem. Thus, some mesh zones of the
unstructured mesh such as the tongue, interfaces, and turbine blades were refined
more than others using the local meshing feature. Figure 4a and b respectively
shows the three-dimensional and the two-dimensional views of our computational
domain meshing. This obtained mesh consists of 329,619 cells and 106,070 nodes.
Numerical Investigation for a Vanned Mixed Flow Turbine Volute … 103
In fact, the rotating domain includes 164,651 cells and 36,145 nodes when the
volute is meshed with 164,968 cells and 69,925 nodes. The near wall zones were
discretized using prism layers, and an inflation which consists of ten layers is
applied by fixing the first cell high in order to get a suitable non-dimensional wall
distance in the range required by the turbulence model.
static pressure in order to avoid that a reverse exit flow occurs during simulations.
The value of the static pressure can be calculated by fixing the turbine expansion
ratio.
The different applied boundary conditions are shown in Fig. 5. A rotational
speed of 59,783.4 rpm was affected to the rotating domain. In addition, to simplify
the rotating model approach, the rotating and stationary domains were assembled
by means of Multiple Reference Frame (MRF) interface keeping an automatic pitch
change. Furthermore, all the wall boundaries such as the hub, blades, and volute
walls were defined with a non-slip wall function. However, a counter-rotating wall
option was imposed to the shroud wall in addition of the non-slip wall function. As
initial conditions for starting calculation, zero values were fixed for all the flow
parameters.
Numerical Investigation for a Vanned Mixed Flow Turbine Volute … 105
(a) 3D view
(b) 2D views
Frozen rotor
This section aims to verify our model accuracy by comparing the numerical results
to the experimental data, collected by Romagnoli and Martinez Botas (2011). The
total-to-static isentropic efficiency and the mass flow parameter are chosen as results
for the validation of our numerical model. The total-to-static efficiency and the mass
flow parameter (MFP) were calculated from Eqs. (1) and (2), respectively:
h01 h04
gts ¼ ð1Þ
h01 h4is
106 A. Ketata and Z. Driss
pffiffiffiffiffiffiffi
m_ T01
MFP ¼ ð2Þ
P01
where h01 , h04 , and h4is are the turbine inlet total enthalpy (J kg−1), the turbine exit
total enthalpy (J kg−1), and the turbine exit static isentropic enthalpy (J kg−1),
respectively. However, m, _ P01 , and T01 present the mass flow (kg s−1), the inlet total
pressure (Pa), and the inlet total temperature (K), respectively. Commonly, the mass
flow parameter is often plotted versus the pressure ratio ‘PR’ of the inlet volute total
pressure to the turbine exit static pressure. In addition, the distribution of the
isentropic efficiency is often given as a function of the ratio of the blade tip velocity,
cited as ‘U’, to the isentropic velocity, identified as ‘Cis’. Figure 6 shows the
dramatically. This peak of the exit absolute Mach number is a result of the flow
acceleration just behind the volute tongue. However, the decrease in the absolute
Mach number in front of the volute tongue can be explained by an occurring flow
deceleration. Figure 7c shows the distribution of the volute exit static pressure as a
function of the azimuth angle for different spanwise locations. From these results, it
is obvious that the magnitude of the static pressure at the mid-span is found to be
greater compared to its values at hub and shroud near zones. The static pressure
Numerical Investigation for a Vanned Mixed Flow Turbine Volute … 109
remains relatively at the same level throughout the exit circumference. Moreover,
its magnitude grows considerably in front of the tongue position to achieve its
highest point which is about 0.195 MPa at 25° of the azimuth angle for the
mid-span location. From these results, it can be deduced that the volute discharge
parameters are too sensitive to the volute tongue and their distribution changes
radically near this zone. Figure 8 shows the mass flow weighted discharge
parameters at the vane outlet as a function of the turbine mass flow parameter.
Figure 8a shows the distribution of the averaged vane discharge flow angle. It is
clear that the increase in the mass flow leads to an important deterioration of the flow
angle. In fact, the averaged flow angle is found to be about 70.57° at the lowest
computed mass flow parameter and then it drops to reach about 67.52° at the highest
computed mass flow parameter. Comparing the averaged values of the flow angle at
2.968 of the turbine expansion ratio, the flow angle seems to be considerably higher
at the vane exit compared to its value at the volute outlet. As a result, the flow angle
gets an upward trend when the flow passes through the volute vane. Figure 8b shows
the distribution of the mean absolute Mach number at the vane exit. It is obvious that
the outlet Mach number increases steeply with the surge of the mass flow. The mean
Mach number is still in the range of 0.4–0.7, and its maximum computed value is
about 0.66 at 6 10−5 of the mass flow parameter. Even with higher tested turbine
expansion levels, the flow at the volute vane exit remains subsonic. Figure 8c shows
the distribution of both total and static vane exit pressures. It is clear that both mean
exit total and static pressures show a downward trend with the surge of the mass flow
parameter. However, the variation of the mean static pressure as a function of the
mass flow is found to be more important than that computed of the mean total
pressure. In fact, the difference between extreme values of the magnitude of the mean
total pressure is about 11,190 Pa. However, this difference is found to be 39,480 Pa
for the mean static pressure. Therefore, the exit mean static pressure is more sensitive
to the mass flow variation than the exit mean total pressure.
5.2 Velocity
Figure 9a shows the velocity vectors at the vanned volute mid-span transverse
planes for 2.968 turbine expansion ratio and 59,783.4 rpm rotational speed. From
Fig. 9a, it is shown that the flow velocity magnitude and direction are uniform at
the volute inlet duct and then the direction of the flow changes gradually to follow
the volute scroll. At the vane exit, the flow rotates around the circumference and
tends to move toward the turbine rotating center. However, Fig. 9a does not show
any clear recirculation zone. Thus, it is obvious that the velocity magnitude remains
relatively constant throughout the inlet duct. In addition, the magnitude of the
velocity increases considerably throughout the volute vane and gets its highest
values at the volute vane exit. This can be explained by the reduction of the area
through which the flow crosses the volute vane. Figure 9b shows the distribution of
the tangential components of the velocity at the mid-span transverse plane for 2.968
110 A. Ketata and Z. Driss
(c) Pressure
turbine expansion ratio and 59,783.4 rpm rotational speed. From these results, the
velocity tangential component presents low values in the range of 16–32 m s−1 at
the volute inlet and increases slowly along the inlet duct. Near the volute first throat
area, the velocity tangential component increases rapidly to be about 100 m s−1.
From these findings, the velocity tangential component continues to rise slowly
Numerical Investigation for a Vanned Mixed Flow Turbine Volute … 111
with the radius decrease until it reaches the vane inlet with values between 113 and
129 m s−1 approximately. Through the vane, the flow tangential velocity increases
rapidly and gets its highest values in the range of 200–227 m s−1 at the zone close
to the rotor inlet.
The tangential component of the velocity gets an upward trend with the drop of
the radial distance, and thus, the flow is accelerated to reach the rotor inlet with a
higher kinetic energy. Figure 9c shows the distribution of the velocity radial
component at the mid-span transverse plane for 2.968 turbine expansion ratio and
59,783.4 rpm rotational speed. It should be noted that the radial velocity absolute
value remains relatively constant along the inlet duct and starts to decrease grad-
ually across the volute.
It can be seen that the radial velocity absolute value gets very low values at an
extended region near the volute wall and, sometimes, it becomes practically null.
However, the radial velocity absolute value gets higher values near the vane inlet
zone. Furthermore, the radial velocity absolute value increases considerably
through the vane and achieves higher values in the range of 80–95 m s−1 at the zone
near the vane exit. The increase in the radial velocity absolute value leads to a
strong rotational flow at the zone near the vane exit and, as a result, the creation of
certain forced vortex.
Figure 10 shows the velocity vectors at volute cross sections for different azi-
muth angles. These velocity vectors are obtained at 2.968 turbine expansion ratio
and 59,783.4 rpm rotational speed. According to these results, it can be seen that
the deviation of the flow direction is more important around corner zones, where the
flow follows the corner curvature, than other volute cross-sectional regions. The
flow direction deviation is greater at smaller than larger cross-sectional areas.
Furthermore, a secondary flow in a small zone is detected near the cross-sectional
corners defined by 45° of the azimuth angle. However, no secondary flow region
was observed for the rest of cross-sectional velocity vectors. Consequently, the
quasi-trapezoidal shape utilized in this volute cross-sectional design seems to be
able to avoid secondary flow generation which leads to the deterioration of the
turbine performance. From these results, it is obvious that the flow accelerated
when it moves from the volute cross-sectional centroid to the vane outlet. This
acceleration can be explained by the sudden surge of the radial component absolute
value of velocity. Besides, the smaller the cross-sectional area is, the more
important the velocity acceleration and the flow deviation are.
Figure 11 displays the distribution of the static pressure at the mid-span transverse
plane for 59,783.4 rpm rotational speed. It can be clearly seen that the static
pressure is totally uniform and keeps the same value which is about 0.205 MPa
along the volute inlet duct excluding the zone near the volute inlet cross section
which is defined by 0° of the azimuth angle.
Numerical Investigation for a Vanned Mixed Flow Turbine Volute … 113
At this zone, the static pressure drops slightly to be about 0.2 MPa at the lower
area of the cross section. However, the upper area of the volute inlet cross section
remains within the same values recorded along the inlet duct. In addition, the zone
near the upper volute wall keeps higher pressure ratios compared to the other zones.
114 A. Ketata and Z. Driss
Fig. 11 Static pressure contour at the mid-span transverse plane of the volute
Then, the static pressure drops gradually along the spiral casing until it reaches
the vane inlet where the static pressure is about 0.194 MPa. Downstream the vane
inlet, the static with a strong slope practically at the vane exit pressure continues to
decline significantly near zone where the flow nearly enters the rotor. Figure 12
illustrates the static pressure distribution at different volute cross sections for
59,783.4 rpm rotational speed and at 2.968 turbine pressure ratio. According to
these results, it is clear that the static pressure drops in a uniform way when the
radial distance decreases. Furthermore, the cross sections’ upper zone presents
higher static pressure values than those computed at lower zones. This zone which
presents higher pressure values becomes smaller with the surge of the azimuth
angle and, thus, with the decrease in the volute cross-sectional area until it disap-
pears practically at the tongue zone. The pressure drop at the tongue zone results in
a secondary flow loss which reduces the turbine overall performance. Then, it is
worth noting that the axial distribution of the static pressure at a fixed radial
distance is uniform and the value of the static pressure remains constant along the
axial distance excluding the zone near the volute vane exit. At this zone, the
magnitude of the static pressure is relatively higher at a lower radial distance (near
the hub) than at a higher radial distance (near the shroud).
This section presents the distribution of different losses occurring within both the
volute and its vane. The loss coefficient of the turbine stator can be calculated as
expressed by Eq. (3) (Abidat et al. 2008):
Vex
fS ¼ 1 ð3Þ
Vex;is
where Vex and Vex,is are the exit velocity (m s−1) and the isentropic exit velocity
(m s−1), respectively. The total pressure does not change for an ideal stator.
However, some total pressure loss occurs for a real stator. The total pressure loss
coefficient is defined as given by (4):
Numerical Investigation for a Vanned Mixed Flow Turbine Volute … 115
P0;in P0;ex
Kp ¼ ð4Þ
P0;ex Pex
where P0,in, P0,ex, and Pex are, respectively, the inlet total pressure (Pa), the exit
total pressure (Pa), and the exit static pressure (Pa).
The angular momentum in an ideal volute is preserved as shown by the free
vortex law. But, this is not the case in a real volute for which losses occur. Thus, a
swirl coefficient ‘S’ is introduced in the free vortex equation as given by Eq. (5).
where Ch,in andCh,ex are, respectively, the inlet absolute tangential velocity (m s−1)
and the exit absolute tangential velocity (m s−1). However, rin and rex are the exit
mean radius (m) and the inlet mean radius (m), respectively. Figure 13a shows the
distribution of the volute and the vane loss coefficients as a function of the mass
flow parameter. From 3.4 10−5 to 6 10−5 of the mass flow parameter, it can be
seen that the loss coefficient of the volute is considerably higher than that found at
the vane. However, the volute loss coefficient is found to be lower than its coun-
terpart for the vane for mass flow parameters less than 3.4 10−5. Furthermore, the
volute loss coefficient decreases with a strong slope with the surge of the mass flow
parameter.
On the other hand, the vane loss coefficient decreases with a slight slope when
the mass flow parameter grows until it reaches 5.89 10−5. From this point on, the
vane coefficient decreases rapidly to be about 0.054 at 6 10−5 of the mass flow
parameter. Figure 13b shows the distribution of volute and vane total pressure loss
coefficients as a function of the mass flow parameter. According to these results, it
can be seen that the total pressure loss coefficient presents approximately the same
trend as the loss coefficient but with different values. From these results, the volute
total pressure loss coefficient is lower than its counterpart of the volute vane for a
mass flow parameter less than 3.3 10−5. However, the volute presents a higher
total pressure loss coefficient than the vane for a mass flow parameter greater than
3.3 10−5. At this last point, the total pressure loss coefficient of the volute and its
vane are equal. In addition, the total pressure loss within the volute decreases with a
slight slope with the surge of the mass flow parameter. Oppositely, the total
pressure loss drops with a strong slope when the mass flow parameter rises. The
total pressure loss coefficient varies from 0.116 to 0.144 for the volute and from
0.043 to 0.155 for the vane.
Figure 14 shows the distribution of the volute and the vane computed swirl
coefficients as a function of the mass flow parameter for 59,783.4 rpm rotational
speed. It is obvious that the volute swirl coefficient is lower than that found of the
vane for mass flow parameter higher than 3.65 10−5. However, for the mass flow
parameters lower than 3.65 10−5, the volute swirl coefficient becomes higher
than that of the volute vane. From these results, it is clear that the volute swirl
coefficient decreases with the surge of the mass flow parameter. In fact, at
Numerical Investigation for a Vanned Mixed Flow Turbine Volute … 117
3.12 10−5 of the mass flow parameter, the volute swirl coefficient is about 1.005
and then drops to be about 0.982 at nearly 6 10−5 of the mass flow parameter.
However, the vane swirl coefficient increases with the rise of the mass flow. At
3.12 10−5 of the mass flow parameter, the vane swirl coefficient is about 0.956
and then rises to be 1.033 at about 6 10−5 of the mass flow parameter. All the
recorded values of the swirl coefficients for both the volute and its vane are rela-
tively very close to 1. Therefore, the angular momentum within this vanned volute
can be considered to be preserved and so the volute seems to behave relatively as an
ideal volute in which the free vortex law is respected.
To evaluate the volute losses, the distribution of the entropy is plotted at
cross-sectional and transverse planes as shown respectively in Figs. 15 and 16.
From these results, a considerable entropy generation appears near the walls which
corresponds equally to the occurring losses in these regions. Furthermore, the inlet
duct in which the entropy remains at the same level does not show any entropy
generation and thus any occurring losses. A significant entropy generation was
recorded at the tongue position, confirming that this zone is a seat of an
aerodynamic loss. Then, the entropy remains stable inside the volute spiral casing,
proving that there is no loss in this region. On the contrary, the entropy increases
within the volute vane and practically at the vane exit near the zone where it reaches
higher values. In addition, it seems that vane exhibits more entropy generation than
the volute where the entropy generation occurs essentially at the tongue zone. This
important generation of the entropy reduces the full turbine stage performance.
From Fig. 16, it is clear that the entropy distribution in the volute cross section
remains the same for various azimuth angles. In fact, no entropy change is recorded
within the cross sections excluding the near wall zones which present an entropy
production. Furthermore, an entropy generation is recorded inside the volute vane
cross section including near wall zones. In addition, Fig. 16 confirms, as previously
discussed, that the occurring losses within the vane are more important than those
existing within the volute.
Figure 17a shows the distribution of the turbulent kinetic energy at the volute
mid-span transverse plane. It can be seen that the turbulent kinetic energy remains
constant with a weak value along the volute inlet duct. Downstream the volute inlet
throat, the turbulent kinetic energy remains at the same level except for the zone
near the tongue where the volute cross section decreases considerably. At the
tongue upstream, the turbulent kinetic energy increases significantly and reaches
higher values at the tongue position. In addition, all the wall zones present higher
turbulent kinetic energy values compared to the volute internal zone.
Moreover, the turbulent kinetic energy recorded at the wall zones rises gradually
when moving toward the tongue position. However, the turbulent kinetic energy is
found to be higher at the vane than at the volute. From the vane inlet zone, the
turbulent kinetic energy increases considerably throughout the volute vane until the
vane exit. Indeed, the distribution of the turbulent kinetic energy is non-uniform in
the vane domain. From −70° to 135° of the azimuth angle, the turbulent kinetic
energy presents higher values compared to those found outside this zone and
particularly in the volute vane domain. This great turbulent kinetic energy value in
this zone may be explained by the presence of the volute tongue which seems able
to produce turbulent kinetic energy that propagates along the upper region of the
volute vane. Figure 17b, c, d, and e show the distribution of the turbulent kinetic
energy at different cross sections defined respectively by 0°, 90°, 180°, and 270° of
the azimuth angle. From these results, it can be observed that the upper zone of the
different cross sections shows an increase in the turbulent kinetic energy.
Furthermore, this upper zone becomes larger with the decrease in the
cross-sectional throat area. However, a weak value of the turbulent kinetic energy is
Numerical Investigation for a Vanned Mixed Flow Turbine Volute … 121
recorded in the cross-sectional core area and it shows a uniform distribution for the
different azimuth angles excluding the zone near the volute tongue. Besides, there is
a significant surge of the turbulent kinetic energy at the volute vane exit zone.
Furthermore, it is clear that the near-hub zone presents a wake zone, characteristic
of the maximum values of the turbulent kinetic energy. The expansion of this wake
decreases near the shroud zone.
122 A. Ketata and Z. Driss
Figure 18a shows the distribution of the dissipation rate of the turbulent kinetic
energy at the volute mid-span transverse plane for 59,783.4 rpm rotational speed
and at 2.968 turbine expansion ratio. According to these results, it is clear that the
dissipation rate of the turbulent kinetic energy is very weak at the entry of the volute
inlet duct. This fact is related to the occurring turbulent kinetic energy losses at the
volute inlet duct entry. Downstream the volute inlet, the dissipation rate of the
turbulent kinetic energy remains relatively at the same level and does not show any
great variation until the tongue upper region. However, a small increase in the
turbulent kinetic energy dissipation rate has been observed at the tongue zone.
Furthermore, a wake characteristic of the maximum value of the turbulent kinetic
Fig. 18 Turbulent kinetic energy dissipation rate distribution at different volute planes
Numerical Investigation for a Vanned Mixed Flow Turbine Volute … 123
energy dissipation rate is recorded at the volute vane domain. This wake appears in
a considerable zone of the volute vane domain, and it is limited by azimuth angles
equal to 0° and 135°. This considerable dissipation zone leads to a turbulent kinetic
energy loss and, then, to a reduction in the capacity of the flow to perform
mechanical work and, equally, the deterioration of the turbine stage performance.
Figure 18b, c, d, and e shows the dissipation rate distribution of the turbulent
kinetic energy for different cross sections defined respectively by 0°, 90°, 180°, and
270° of the azimuth angle. From these figures, it is shown that the turbulent kinetic
energy dissipation rate is uniform and remains steady inside the area of the different
volute cross sections. However, a growth of this kinetic energy dissipation rate has
been observed within the volute vane.
At the volute vane exit, the rise of the turbulent kinetic energy dissipation rate is
more significant at the hub than the shroud. Then, it is obvious that the value of the
turbulent kinetic energy dissipation rate within the volute vane is less important at a
higher azimuth angle compared to that found for lower azimuth positions.
Figure 19 shows the distribution of the turbulent viscosity at the volute mid-span
transverse plane of the for 59,783.4 rpm rotational speed and at 2.968 turbine
expansion ratio.
According to these results, it is clear that the weak values of the turbulent
viscosity appear along the volute inlet duct excluding the zone near the wall where
there is a wake characteristic of the turbulent viscosity maximum values.
Downstream of the volute inlet, the turbulent viscosity increases progressively
throughout the volute spiral casing. At the volute wall, the turbulent viscosity is
found to be higher than in its internal domain. Furthermore, the turbulent viscosity
at the volute wall increases considerably with the increase in the azimuth angle.
Another wake characteristic of the turbulent viscosity maximum values was
observed at the tongue position. This wake reduces the occurring momentum
transfer which is a result of turbulent viscosity and then gives rise to an internal
fluid friction at the volute tongue. Figure 19b, c, d, and e displays the turbulent
viscosity distribution at different cross sections defined respectively by 0°, 90°,
180°, and 270° of the azimuth angle. It is obvious that the cross-sectional distri-
bution of the turbulent viscosity is considerably sensitive to the azimuth angle. In
addition, it is clearly observed that the turbulent viscosity at volute cross sections’
upper regions increases when the volute throat area drops. However, the turbulent
viscosity within the volute vane is too sensitive to the azimuth angle and it exhibits
relatively higher values at positions close to the volute tongue.
124 A. Ketata and Z. Driss
6 Conclusion
This chapter discussed the overall performance and the distribution of several flow
parameters of a vanned mixed flow turbine volute. Three-dimensional numerical sim-
ulations based on Navier–Stokes equations were conducted under steady conditions.
Numerical Investigation for a Vanned Mixed Flow Turbine Volute … 125
The good agreement found between numerical and experimental results of the full
turbine stage swallowing capacity and efficiency ensures our numerical model
validation. The swallowing capacity and the total-to-static isentropic efficiency
seem to be significantly sensitive to the variation of the turbine expansion ratio and
the velocity ratio, respectively. Furthermore, the results confirm the uniformity of
the flow throughout the volute discharge circumference excepting the volute tongue
position where non-uniformity was recorded. The discharge parameters such as the
exit flow angle, the exit Mach number, and the exit pressure show a considerable
variation at the volute tongue near zone, and they are significantly influenced by the
variation of the mass flow rate. The results confirm that the quasi-trapezoidal
cross-sectional design leads to a less secondary flow generation and then better
turbine performance. In addition, the plotted velocity distributions indicate that the
velocity deviation becomes more significant when the volute throat decreases. The
variation of the tangential velocity within the volute scroll casing increases with the
decrease in the radial distance and contributes to the creation of a rotational flow at
the rotor inlet. On the other hand, our results indicate that the static pressure drops
gradually along the volute spiral casing. A drop of the static pressure was recorded
at the volute tongue which proves an occurring secondary flow at this region. The
loss analysis shows that the total pressure loss coefficient drops with the surge of the
mass flow rate for both the volute and its vane. However, the volute loss coefficients
seem to be higher than those found for the vane for higher mass flow rates. At lower
mass flow rates, the volute loss coefficients become lower than those of the vane.
The swirl coefficient values for both the volute and the vane are close to 1, and then,
the volute can be considered as relatively ideal in which the free vortex law is
respected. A significant entropy generation was recorded at the volute tongue
confirming the occurrence of an aerodynamic loss at this region. Within the volute
vane, an entropy generation which leads to the reduction in the turbine overall
performance was recorded especially at the vane exit. Our results show that the
volute tongue near zone and the volute vane upper region display an important
surge of the turbulent kinetic energy. Therefore, the volute tongue geometry seems
to be a source of turbulent kinetic energy production. Furthermore, the turbulent
kinetic energy dissipation rate within the volute is found to be more important from
0° to 135° of the azimuth angle than the other regions. Moreover, the turbulent
viscosity distribution depends on the azimuth angle and higher values are recorded
at the zone close to the volute tongue leading to an internal fluid friction. The volute
tongue and the volute vane seem to be a seat of aerodynamic losses, resulting in the
turbine performance deterioration. These results will be taken into account for the
design of a new turbocharger volute in order to further reduce the occurrence of
aerodynamic losses.
126 A. Ketata and Z. Driss
References
Abidat M (1991) Design and testing of a highly loaded mixed flow turbine, Ph.D. Thesis, Imperial
College, London
Abidat M, Hamidou MK, Hachemi M, Hamel M (2008) Design and flow analysis of radial and
mixed flow turbine volutes, ASME paper, GT2008-50503:2329-2338. https://doi.org/10.1115/
GT2008-50503
Ayder E, Van den Braembussche RA (1994) Numerical analysis of the threedimensional swirling
flow in centrifugal compressor volutes. J Turbomach 116:462–468. https://doi.org/10.1115/1.
2929435
Ayder E, Van den Braembussche RA, Brasz J (1993) Experimental and theoretical analysis of the
flow in a centrifugal compressor volute. J Turbomach 115:582–589. https://doi.org/10.1115/1.
2929293
Barnard MCS, Benson RS (1968) Radial gaz turbines. Proc IMechE 183:59–70. https://doi.org/10.
1243/PIME_CONF_1968_183_237_02
Chapple PM, Flynn PF, Mulloy JM (1980) Aerodynamic design of fixed and variable geometry
nozzle-less casings. ASME J Eng Power 102:141–147. https://doi.org/10.1115/1.3230212
Chen H (1996) Design methods of volute casings for turbocharger turbine applications. J Power
Energy 210:149–156. https://doi.org/10.1243/PIME_PROC_1996_210_022_02
Chen H (2009) A discussion on volute design method for radial inflow turbines. ASME paper,
GT2009-59110. https://doi.org/10.1115/GT2009-59110
Driss Z, Mlayeh O, Driss D, Maaloul M, Abid MS (2014) Numerical simulation and experimental
validation of the turbulent flow around a small incurved Savonius wind rotor. Energy 74:506–
17. https://doi.org/10.1016/j.energy.2014.07.016
Driss S, Driss Z, Kallel Kammoun I (2016a) Computational study and experimental validation of
the heat ventilation in a living room with a solar patio system. Energy Build 28:40–119. https://
doi.org/10.1016/j.enbuild.2016.03.016
Driss Z, Mlayeh O, Driss S, Driss D, Maaloul M, Abid MS (2016b) Study of the incidence angle
effect on the aerodynamic structure characteristics of an incurved Savonius wind rotor placed in
a wind tunnel. Energy 894:908–113. https://doi.org/10.1016/j.energy.2016.07.112
Gu F, Engeda A, Benisek E (2001) A comparative study of incompressible and compressible
design approaches of radial inflow turbine volutes. Proc Inst Mech Eng 215:475–486. https://
doi.org/10.1243/0957650011538730
Hakeem I, Su CC, Costall A et al (2007) Effect of volute geometry on the steady and unsteady
performance of mixed-flow turbines. Proc IMechE Part A J Power Energy 221:535–549.
https://doi.org/10.1243/09576509JPE314
Hara K, Furukawa M, Inoue M (1994) Behavior of three-dimensional boundary layers in a radial
inflow turbine scroll. J Turbomach. Trans ASME 116:446–452. https://doi.org/10.1115/1.
2929431
Lymberopoulos N, Baines NC, Watson N (1988) Flow in single and twin-entry radial turbine
volutes. ASME paper, 88-GT-59. https://doi.org/10.1115/88-gt-59
MacGregor SA, Whitefild A, Mohd Noor AB (1994) Design and performance of vaneless volutes
for radial inflow turbines Part 3: experimental investigation of the internal flow structure.
J Power Energy 208:295–302. https://doi.org/10.1243/PIME_PROC_1994_208_050_02
Rajoo S (2007) Steady and pulsating performance of a variable geometry mixed flow turbochager
turbine. Ph.D. Thesis, Imperial College, London
Romagnoli A, Martinez Botas RF (2011) Performance prediction of a nozzled and nozzleless
mixed-flow turbine in steady conditions. Int J Mech Sci 53:557–574. https://doi.org/10.1016/j.
ijmecsci.2011.05.003
Suhrmann JF, Peitsch D, Gugau M, Heuer T (2012) On the effect of volute tongue design on radial
turbine performance. ASME paper, GT2012-69525. https://doi.org/10.1115/GT2012-69525
Wei H, Zhu T, Shu G, Tan L, Wang Y (2012) Gasoline engine exhaust gas recirculation—a
review. Appl Energy 99:534–544. https://doi.org/10.1016/j.apenergy.2012.05.011
Numerical Investigation for a Vanned Mixed Flow Turbine Volute … 127
Whitefild A, Mohd Noor AB (1994) Design and performance of vaneless volutes for radial inflow
turbines Part 1: non-dimensional conceptual design considerations. J Power Energy 208:199–
111. https://doi.org/10.1243/PIME_PROC_1994_208_035_02
Whitefild A, MacGregor SA, Mohd Noor AB (1994) Design and performance of vaneless volutes
for radial inflow turbines Part 2: experimental investigation of the meanline performance—
assessment of empirical design parameters. J Power Energy 208:213–224. https://doi.org/10.
1243/PIME_PROC_1994_208_036_02
Yang M, Martinez Botas R, Rajoo S, Yokoyama T, Ibaraki S (2014) Influence of volute
cross-sectional shape of a nozzleless turbocharger turbine under pulsating flow conditions.
Proceedings of ASME Turbo Expo 2014: GT2014-26150. https://doi.org/10.1115/GT2014-
26150
Yang M, Martinez Botas R, Rajoo S, Yokoyama T, Ibaraki S (2015) An investigation of volute
cross-sectional shape on turbocharger turbine under pulsating conditions in internal combus-
tion engine. Energy Convers Manag 105:167–177. https://doi.org/10.1016/j.enconman.2015.
06.038
CFD Investigation of the Hydrodynamic
Structure Around a Modified Anchor
System
1 Introduction
Computational fluid dynamics (CFD) plays a key role in helping to understand the
flow inside stirred tanks. It is becoming a useful tool in the analysis of the highly
complex flow inside stirred vessels. The CFD has been used in the last two decades
to devise solutions and gain insight of the flow. Together with experimental vali-
dation, CFD has been able to improve the design of many reactor systems. Reaction
system with anchor impellers is used specially for highly viscous flows, with the
viscosity ranging from 10 to 100 Pa s, which is typical of polymer reactions and
some food industries processes. There are few works in the literature which study
anchor-type impellers. The great majority of works for stirred vessels refers to
turbine impellers, especially the pitched blades and the Rushton turbines, under
turbulent flow. A great deal of research focused on the optimization of the design of
the stirred tanks and impellers geometry. For example, Deglon and Meyer (2006)
investigated the effect of grid resolution and discretization scheme on the CFD
simulation of a fluid flow in a baffled mixing tank stirred by a Rushton turbine.
Murthy and Joshi (2008) tested five impeller designs namely disc turbine (DT), a
variety of pitched blade down flow turbine impellers varying in blade angle
(Standard PBTD60, 45 and 30) and hydrofoil (HF) impeller. Alvarez et al. (2002)
studied a stirred tank system with a single Rushton impeller mounted in a central
shaft. Using UV-visualization techniques, they illustrated the 3D mechanism by
which fluorescent dye is dispersed within the chaotic region of the tank. Also, they
compared a system of three Rushton impellers with a three-disk system at the same
locations. Montante et al. (2001) studied the flows recirculation zone in the tran-
sition model in a tank agitated by the laser Doppler anemometry (LDA) technique.
They studied the influence of the position of the turbine compared to the bottom and
the influence of the baffles on the flows hydrodynamics in stirred tanks with a
Rushton turbine in order to define an optimal position for a maximum axial
velocity. Stitt (2002) noted that multiphase reactor designs from larger scale and
non-catalytic processes are now being considered. These include trickle beds,
bubble columns, and jet or loop reactors. Alcamo et al. (2005) computed the
turbulent flow field generated in an unbaffled stirred tank by a Rushton turbine
using the large-eddy simulation (LES). Zalc et al. (2002) explored laminar flow in
an impeller stirred tank using CFD tools. They extended the analysis to include
short- and long-time mixing performance as a function of the impeller speed. The
simulated flow fields are validated extensively by particle image velocimetry (PIV).
Also, they used planar laser-induced fluorescence (PLIF) to compare the experi-
mental and computed mixing patterns. Guillard and Trägardh (2003) designed and
tested a new model for estimating mixing times in aerated stirred tanks with three
reactors equipped with two, three, and four Rushton impellers. The results showed
that the analogy model developed is independent of the scale, the geometry of the
tank, the number of used impellers, the distance between impellers, and the con-
sidered degree of homogeneity. Only the region in which the pulses were added was
found to affect the results. Brucato et al. (1998) studied the turbulent flow generated
by one and two Rushton turbines in different axial positions. They studied the effect
of the grid scaling on the assessment of the three velocity components: the turbulent
kinetic energy, the dissipation rate, and the evolution of the power number. Ammar
et al. (2011) analyzed numerically the effect of the baffles length on the turbulent
flows in stirred tanks equipped by a Rushton turbine. The numerical results from the
application of the CFD code Fluent with the MRF model were presented in the
vertical and horizontal planes in the impeller stream region. Chtourou et al. (2011)
were interested in providing predictions of turbulent flow in a stirred vessel and
assessing the ability to predict the dissipation rate of turbulent energy that consti-
tutes the most stringent test of prediction capability due to the small scales at which
dissipation takes place. The amplitude of local and overall dissipation rate is shown
to be strongly dependent on the choice of turbulence models. Driss et al. (2010)
developed a computational study of the pitched blade turbines design effect on the
stirred tank flow characteristics. Particularly, they studied the effects of different
inclined angles, equal to 45°, 60°, and 75°, on the local and global flow charac-
teristics. Kchaou et al. (2008) compared the effect of the flat blade turbine with 45°
and −45° pitched blade turbines on the hydrodynamic structure of the stirred tank.
The main aim of this chapter is to study the hydrodynamic characters generated
by a special agitator made through the superposition of two types of impellers: the
classical anchor and the pitched blade turbines. A comparison has been achieved
between our CFD results and experimental data extracted from the work of Wu and
Patterson (1989). The present work was carried out using the commercial software
CFD Investigation of the Hydrodynamic Structure … 131
ANSYS Fluent 17.0, for the steady-state simulation of the stirred tank reactors. The
simulation was solved in parallel calculations and under double precision option,
using the multiple reference frames impeller rotation model and the standard k–e
turbulence model.
2 Geometrical Arrangement
In this chapter, the system consists of a modified anchor turbine made by super-
position of the classical anchor and two stages of pitched blade turbine. The two
stages have two inclined blades with b = 45° (Fig. 1). Under these conditions, h1
and h2 are the axial positions of the pitched blades turbines relative to the bottom of
the tank. All the geometrical parameters are presented in Table 1.
3 CFD Technics
In this application, we have used ANSYS Fluent 17.0, to study the turbulent flow
for a Reynolds number equal to Re = 176.4 103. The calculation of the Reynolds
number is based on the impeller diameter. The discretized equations were solved
iteratively using the SIMPLE algorithm for pressure-velocity coupling, and under
the standard k–e turbulence model. The solution was considered converged when
the total residuals for the continuity equation dropped below 10−5.
Figure 2 shows the subdivision of the computational domain with the MRF
approach. The computational domain was split into two cylindrical zones, one of
which contains the agitator (modified anchor) and it is assumed to rotate with the
impeller angular velocity x = 2pN = 215 rpm. While the remaining space was
modeled with a stationary reference, the outer zone was stationary relative to the
tank walls. The shaft wall was also split into two zones. The inner zone which is
included in the rotating zone normally rotates with the same velocity as the
impeller. The outer zone which is adjacent to the stationary zones should also rotate
with the same angular velocity as the rotating space.
CFD Investigation of the Hydrodynamic Structure … 133
The integration domain was meshed using the commercial grid generator package
ANSYS MESH ICEM CFD which creates a hybrid three-dimensional grid. The
hybrid grid is actually an unstructured grid that contains different types of elements.
The choice of the unstructured grid was made due to the fact that in a complex flow,
details of the flow field everywhere in the tank and especially in the discharge area
of the impellers and behind the baffles must be captured. The grid was refined near
the impeller wall to resolve the large flow gradients. From the above study, the
control volume consists of 1532348 elements (Fig. 3).
4 Numerical Results
Our simulations results, such as the velocity fields, magnitude velocity, static
pressure, dynamic pressure, turbulent kinetic energy, dissipation rate of the tur-
bulent kinetic, offer local and global information about the hydrodynamic structure.
They give a more precise understanding of the hydrodynamic mechanism than
those obtained by experimental studies. The numerical results are presented, in
three r–z planes defined by the angular position h = 0°, h = 60°, and h = 90°
containing the blades of the anchor turbines (Fig. 4). Also, we have considered six
other planes r–h defined by the non-dimensional axis positions defined by
z/D = 0.06, z/D = 0.3, z/D = 0.41, z/D = 0.65, z/D = 0.73, and z/D = 0.86.
Figures 5 and 6 show the distribution of the velocity fields respectively in the
r–z planes defined by the angular positions equal to h = 0°, h = 60°, and h = 90°
and in the r–h planes defined by the non-dimensional axial positions equal to
z/D = 0.86, z/D = 0.73, z/D = 0.65, z/D = 0.41, z/D = 0.3, and z/D = 0.06.
According to these results, it was noted that the down-pitched blade generates a
radial jet developed from the blade and propagates in the upper part of the tank
(Fig. 5a). Also, it was observed that the down stage presents an up-pumping mode.
At the side wall of the tank, this jet was transformed into axial jets upward and
downward (Fig. 5b). Indeed, two recirculation loops appeared, with two different
directions, located in the upper and down zones limited by the radial direction of the
jet. In the other sides, the upper pitched blade generates a radial jet developed from
the blade and propagates in the down part of the tank (Fig. 5a). The upper stage of
the pitched blades turbine presents a down-pumping mode. Another recirculation
loop was observed in the upper zone of the tank (Fig. 5b), below the down hori-
zontal blade of the anchor and above the vertical blade (Fig. 5a, b).Also, we can see
a pumping flow that supplies the down stage of the pitched blade turbine, developed
by the down horizontal blade of the anchor (Fig. 5b). The recirculation zone above
the vertical blade of the anchor supplies the upper stage of the pitched blades
turbine. In the medium plane of the anchor, and inside the anchor, the velocity
vectors were noticed to display a centrifugal direction. However, at the level of the
gap, these vectors present a centrifugal direction on the side of the wall tank.
Indeed, the flow was noticed to be strongly dominated by the jet developed by the
vertical blade of the anchor. Also, it was noted that the flow pattern is developed by
the pitched blades turbine. Overall, it has been noted that the maximum values of
the velocity fields were remarked to be located near the vertical rotating blade of the
anchor. Elsewhere, the velocity field values decreased.
136 Z. Driss et al.
Figures 7 and 8 show the distribution of the mean velocity respectively in the
r–z planes for the angular positions h = 0°, h = 60°, and h = 90° and in the r–h
planes for the non-dimensional axial positions z/D = 0.86, z/D = 0.73, z/D = 0.65,
z/D = 0.41, z/D = 0.3, and z/D = 0.06. Globally, the wake characteristic of the
maximum values is developed in the area swept by the vertical blades of the anchor.
Around these blades, it is clear that the magnitude velocity decreases gradually
away from them. Indeed, the magnitude velocity increases near the pitched blades
and between the two stages of the impellers. This fact improves that the two stages
reduce the dead zones. Also, it has been noted that in the bottom of the tank, and
below the agitator for the non-dimensional axial position z/D = 0.06 (Fig. 8f), the
mean velocity reaches important values which decrease gradually with the radial
direction. This fact improves the pumping character in the bottom of the tank. For
the down stage of the pitched blades, the mean velocity is generated with limited
extended zone. The axial direction of this velocity is more dominated. For the upper
stage, it is clear that the mean velocity has a much more extended zone and the
CFD Investigation of the Hydrodynamic Structure … 137
radial component is more dominated. The maximum value of the mean velocity is
equal to 0.2691 m s−1. Near the baffles and the bottom of the tank, the mean
velocity decreases slightly.
Figures 9 and 10 show the static pressure distribution respectively in the r–z planes
defined by the angular positions h = 0°, h = 60°, and h = 90° and in the r–h planes
defined by the non-dimensional axial positions z/D = 0.86, z/D = 0.73, z/D = 0.65,
z/D = 0.41, z/D = 0.3, and z/D = 0.06. According to these results, the compression
zone characterized by the maximum values of the static pressure appears in the
vicinity of the tank wall developed by the agitator blades. For the two stages with
pitched blade turbine, a zone has been observed but with medium values of static
pressure. This fact improves the supply character generated by the down stage.
Indeed, a compression zone has been observed surrounding the vertical blade of the
anchor. The maximum value of this zone is located in the external edge of the blade.
It has also been noted that the static pressure increases near the walls of
the vessel and widely near the baffles. For z/D = 0.73 and z/D = 0.41 (Fig. 10b, d),
the static pressure developed by the upper and the down stages of the pitched blades
138 Z. Driss et al.
turbine seems to be the same but slightly extended for the upper stage. Around the
vertical rotating blade of the anchor and for one face of this blade, the static pressure
is low contrary to the other face. Near the baffles and the vessel wall, the static
pressure is very important and reaches a maximum value equal to P = 4.250 103 Pa.
Figures 11 and 12 show the distribution of the dynamic pressure respectively in the
r–z planes defined by the angular positions h = 0°, h = 60°, and h = 90° and in the
r–h planes defined by the non-dimensional axial positions z/D = 0.86, z/D = 0.73,
z/D = 0.65, z/D = 0.41, z/D = 0.3, and z/D = 0.06. According to these results, it is
clear that a compression zone characteristic of the maximum values of the dynamic
pressure in the vicinity of the upper stage of the pitched blade turbine developed
near the down face of the blade and around the vertical blade of the anchor. For the
CFD Investigation of the Hydrodynamic Structure … 139
down stage, this zone is rather limited. Also, a compression zone of dynamic
pressure that lied the two stages of impellers has been observed. This fact improves
the supply character generated by the down stage. Under these conditions, the
maximum value of this zone is located in the blade external edge. Indeed, it has
been noted that the dynamic pressure increases near the vessel walls and widely
near the baffles. The dynamic pressure reaches a maximum value equal to
P = 3.648 103 Pa. For z/D = 0.73 and z/D = 0.41 (Fig. 12b, d), the compression
zone characteristic of the maximum value of the dynamic pressure developed by the
upper pitched blade is more extended and more important than the compression
zone generated by the down stage. The maximum values are located around the
vertical blade of the anchor.
Figures 13 and 14 show the distribution of the turbulent kinetic energy respectively
in the r–z planes defined by the angular positions h = 0°, h = 60°, and h = 90° and
in the r–h planes defined by the non-dimensional axial positions z/D = 0.86,
140 Z. Driss et al.
z/D = 0.73, z/D = 0.65, z/D = 0.41, z/D = 0.3, and z/D = 0.06. According to these
results, it is clear that the wake characteristic of the maximum value of the turbulent
kinetic energy appears in the area swept by the vertical blade of the anchor, near the
baffles and the vessel wall level. Away from this area, the turbulent kinetic energy
decreases gradually. It has also been noted that these wakes are more extended near
the blades of the anchor than the two stages of the pitched blades turbines. In these
conditions, the maximum value of the turbulent kinetic energy is equal to
k = 9.884 10−1 J kg−1.
Figures 15 and 16 show the turbulent kinetic energy dissipation rate distribution
respectively in the r–z planes defined by the angular positions h = 0°, h = 60°, and
h = 90° and in the r–h planes defined by the non-dimensional axial positions
CFD Investigation of the Hydrodynamic Structure … 141
z/D = 0.86, z/D = 0.73, z/D = 0.65, z/D = 0.41, z/D = 0.3, and z/D = 0.06.
According to these results, the maximum value is noticed to be related to the
rotating vertical blade of the anchor and especially at the upper edges. This dis-
tribution is more strongly located around this blade. The maximum value is equal to
e = 1.983 103 m2 s−3. Outside this area, there is a quasi-uniform distribution
characterized by rather weak dissipation rate values.
turbulent viscosity near the baffles. Elsewhere, the turbulent viscosity decreases and
becomes very weak near the wall.
Figures 19 and 20 show the axial profiles of the normalized mean radial velocity for
the modified anchor system with “foam-breaker” for two areas defined by the radial
positions r/D = 0.185 and r/D = 0.38. Also, we have superposed the profile of
radial velocity for the Rushton impeller and the dual agitator system already studied
in the previous works (Driss et al. 2010). The comparison between the numerical
CFD Investigation of the Hydrodynamic Structure … 143
and experimental results was investigated through the results achieved in the works
of Wu and Patterson (1989). According to these results, it is clear that the radial
velocity component for the anchor system rises and shows a maximum value near
two zones. The first zone is between z/D = 0.2 and z/D = 0.4 and the second zone
is between z/D = 0.5 and z/D = 0.7. For the first zone, a good agreement was
noticed between the CFD results of the modified anchor system and those of the
dual agitator system. For the second zone, the radial velocity increases for
the anchor and decreases for the “foam-breaking” system. It can be concluded that
the two zones are strongly affected by the presence of the pitched blades stages
and the anchor blade. It has also been observed that the radial velocity decreases
between the two stages. However, the maximum value of the radial velocity appears
around the upper stage.
144 Z. Driss et al.
Fig. 15 Turbulent kinetic energy dissipation rate distribution in the r–z planes
146 Z. Driss et al.
Fig. 16 Turbulent kinetic energy dissipation rate distribution in the r–h planes
CFD Investigation of the Hydrodynamic Structure … 147
6 Conclusion
References
1 Introduction
Heat exchangers with annular geometry are often used in the heating process such as
food industries, chemical, pharmaceutical, plastic industries. In some situations, an
eccentricity occurs in the duct. Its effect is not negligible and has been the subject of
several studies (Feldman et al. 1982a, b; Manglik and Fang 2002). The heating fluid
consistency is temperature dependent (thermodependent), which generates unex-
pected phenomena that remain with very little consideration in literature till date.
The combination of the eccentricity effect and that of the dependence of con-
sistency to temperature is the subject of the present work.
2 Problem Description
A. Horimek (&)
Laboratoire de Développement en Mécanique et Matériaux, Mechanical Engineering
Department, Ziane Achour University, Djelfa, Algeria
e-mail: Horimek_aer@yahoo.fr
N. A. Messaoudene
Department of Mechanical Engineering, Faculty of Engineering, University of Hail,
Hail, Saudi Arabia
e-mail: n.messaoudene@uoh-edu.sa; naitmessaoudene@yahoo.com
N. A. Messaoudene
Laboratoire des Applications énergétiques de L’Hydrogène (LApEH),
University Blida1, Blida, Algeria
The annular eccentric geometry problems are often solved using the bipolar coor-
dinates system (a, b, z*). This system is related to the Cartesian coordinate system
by the following relations:
x ¼ h sin b ; y ¼ h sh a ; z ¼ z
1\a\ þ 1 ; 0 b 2p ; 0 z L
where h ¼ a =ðch a cos bÞ, where a is the location of the positive pole p. R1 and
R2 the radius of the interior and exterior cylinders, respectively; e the dimensionless
eccentricity defined as: e ¼ e=ðR2 R1 Þ, where e the distance between the cylin-
ders axis (Fig. 1).
Since the dynamic regime is assumed to be fully developed at the entrance, the
continuity Eq. (1) and the momentum one in z direction (2) are solved numerically
Laminar Flow for a Newtonian Thermodependent … 153
to obtain the axial velocity profile at the inlet section. Once this profile is obtained,
it will be injected into the energy Eq. (3) to determine the temperature field along
the heating zone.
Continuity equation:
1 @ @W
ðh U Þ þ ¼0 ð1Þ
h2 @a @z
Z momentum equation:
2
U @W @W @P 1 @
la @W @
la @W @ W @2W
þW ¼ þ þ a
þl þ
h @a @z @z Re h2 @a @a @b @b @a2 @b2
ð2Þ
Energy equation:
@h 1 1 @2h 1 @2h
W ¼ þ ð3Þ
@z Pe h2 @a2 h2 @b2
z h P W U ðT Te Þk
z¼ ; h¼ ;P ¼ ;W¼ ;U ¼ ;h ¼
Dh Dh qUd 2 Ud Ud UDh
where U; W are the velocity components following (a; z); Ud : the mean axial
velocity; Dh : the Hydraulic diameter; Te : inlet temperature; U ¼ ðU1 þ U2 Þ=2:
entered heat flux;
Boundary conditions:
1 @h
a ¼ a1 ; U ¼ W ¼ 0; h @a¼ 1
1 @h
a ¼ a2 ; U ¼ W ¼ 0; ¼ þ1
h @a ð4Þ
Z ¼ 0; W ¼ Ufd ða; bÞ; U ¼ 0; h ¼ 0
@U @W @h
b ¼ 0 or p; @b ¼ @b ¼ 0; @b ¼ 0
The governing equations along with their boundaries conditions are solved
numerically using the finite difference method with an implicit scheme (ADI).
A regular mesh in the three directions is used. To clarify better the results, we have
154 A. Horimek and N. A. Messaoudene
presented the axial position as Z = z/Pe. Calculations were conducted until Z = 0.5;
a distance that is largely sufficient to be sure that the fully developed thermal stage
was reached.
For the case of a concentric duct, the profile is characterized by concentric circles
with nil velocity at the walls (no-slip condition), and increasing magnitude until it
reaches its maximum at an intermediate position between two cylinders (in fact, this
position is slightly closer to the inner cylinder). When eccentricity is applied to the
duct, a new reorganization of the flow occurs. It is then characterized by an
acceleration in the wide part (following the widening of the flow section on this
side) and deceleration in the narrow part.
As an illustration, for e = 0.05 and r1 = 0.5, the maximum velocity in the wide
region is 9.2% larger than that for the concentric case, while it is 8.9% less in the
narrow part. These values are 33.8% and 38.2% for e = 0.2. However, they
become, respectively, 60% in the wide part and 89% in the narrow part for e = 0.6.
The flow tends to slow down in the narrow part and for eccentricities greater than
0.6. The validity of the assumption of negligible axial conduction should be con-
sidered with caution. (Feldman et al. 1982a, b), require a Pe > 50 while (Manglik
and Fang 2002) require a larger Pe.
Remark : These precautions are most important when the fluid is pseudoplastic
(Fang et al. 1999; Manglik and Fang 2002) or has a yield stress (Nouar 2005).
In Table 1, values of the maximum axial velocity (Wmax) in the widest part
(b = 0) and in the narrowest one (b = p) are reported for three geometrical radii
(r1). Further different values of these can be found in (Ait-Messaoudene et al.
2011). It should be recalled that for a bipolar coordinate system, the angle is always
counted from the large part to the narrow one. In addition, this system has a
singularity for e = 0.0; therefore, the concentric case is reproduced with e = 10−5.
Justifications for the selection of this value are discussed in [Ait-Messaoudene et al.
(2011, p. 7)].
It is evident from the table that the closeness of the two cylinders (by increasing
r1) associated with a large eccentricity favors a blocking of the flow in the narrow
part. In what follows, the study was limited to the case of r1 = 0.5.
In fact, the whole velocity field is affected by the degree of eccentricity, as
velocity profiles get more and more distorted with increasing eccentricity compared
to concentric circular profiles, encountered in the concentric case (Fig. 2).
As the two cylinders come closer to each other in the narrow part, a strong
heating occurs in the area; the reverse is observed in the wide part. This phe-
nomenon of thermal stratification in the temperature field increases with eccen-
tricity. Figure 3 presents this field for selected axial positions, from the inlet to the
fully developed zone. It is clearly seen that the high-temperature zone (red) is
present on both sides for b = p with maximum temperature increasing with
eccentricity (see Captions). In addition, it can be seen that for high eccentricities
(here e = 0.4 and e = 0.6), heat rapidly reaches the core zone on the narrow side
(See for Z = 5 10−2), while this happens farther downstream for smaller
eccentricities. The combination of the flow slowing effect and the strong heating in
the narrow part may make the heat transfer in the narrow conductive. Thus, the
assumption of neglecting the axial diffusion may become invalid; hence, large Pe
values are required for assuming negligible axial diffusion in such problems.
Remark : A thermal regime is considered to be fully developed either by invari-
ability of Nu number or the thermal profiles (Oosthuizen and Naylor 1999).
ε=0.1 Z=5.0 10-3 Z=5.0 10-2 Z=2.0 10-1 Z=3.75 10-1 Z=0.5
ε=0.2
ε=0.4
ε=0.6
Fig. 3 Evolution of the temperature along the duct for different e; r1 = 0.5; Re = 30.0; Pr = 10;
U1 = U2 = 6000 W/m²
ε=0.2 Z=5.0 10-3 Z=5.0 10-2 Z=2.0 10-1 Z=3.75 10-1 Z=0.5
ε=0.4
ε=0.6
ε=0.2 Z=5.0 10-3 Z=5.0 10-2 Z=2.0 10-1 Z=3.75 10-1 Z=0.5
ε=0.4
ε=0.6
Fig. 4 Evolution of the temperature along the duct for different e; effect of thermodependency.
Top: Pn = 1.0; bottom: Pn = 2.5; r1 = 0.5; Re = 30.0; Pr = 10.0; U1 = U2 = 2500 W/m2
The most important result of this work is summarized in Fig. 5, in which the
axial velocity profile for the same Z positions chosen for temperature is presented. It
clearly shows that the blocking phenomenon of the main flow in the narrow part
tends to decrease, or more correctly, a new acceleration in this part is observed far
158 A. Horimek and N. A. Messaoudene
ε=0.2 Z=5.0 10-3 Z=5.0 10-2 Z=2.0 10-1 Z=3.75 10-1 Z=0.5
ε=0.4
ε=0.6
ε=0.2 Z=5.0 10-3 Z=5.0 10-2 Z=2.0 10-1 Z=3.75 10-1 Z=0.5
ε=0.4
ε=0.6
Fig. 5 Effect of thermodependency on axial velocity profile for different e; top: Pn = 1.0; bottom:
Pn = 2.5; r1 = 0.5; Re = 30.0; Pr = 10.0; U1 = U2 = 2500 W/m2
from the entrance with decreasing consistency (viscosity) due to the strong heating
in this area. For the two Pn values considered, it is clear that the acceleration due to
the reduction of consistency is more pronounced as Pn increases. Moreover, this
phenomenon is accentuated with the increasing eccentricity. This is very clear for
e ¼ 0:4 and e ¼ 0:6 and Pn = 2.5, where there is a sharp acceleration in the narrow
part, with a velocity of the same order as in the wide side. This result is of great
practical importance because it indicates a decrease in the conductive heat transfer
Laminar Flow for a Newtonian Thermodependent … 159
4 Conclusion
References
Feldman EE, Hornbeck RW, Osterle JF (1982a) A numerical solution of laminar developing flow
in eccentric annular ducts. Int J Heat Mass Transf 25:23–241
Feldman EE, Hornbeck RW, Osterle JF (1982b) A numerical solution of developing temperature
for laminar developing flow in eccentric annular ducts. Int J Heat Mass Transf 25:243–253
Fang P, Manglik RM, Jog MA (1999) Characteristics of laminar viscous shear-thinning fluid flows
in eccentric annular channels. J Non-Newt Fluid Mech 84:1–17
Manglik RM, Fang P (1995) Effect of eccentricity and thermal boundary conditions on laminar
fully developed flow in annular ducts. Int J Heat Fluid Flow 16:298–306
160 A. Horimek and N. A. Messaoudene
Manglik RM, Fang P (2002) Thermal processing of viscous non-Newtonian fluids in annular
ducts: effects of power-law rheology, duct eccentricity, and thermal boundary conditions. Int J
Heat Mass Transf 45:803–814
Messaoudene AN, Horimek A, Nouar C, Benaouda-Zouaoui B (2011) Laminar mixed convection
in an eccentric annular horizontal duct for a thermodependent non-Newtonian fluid. Int J Heat
Mass Transf 54:4220–4234
Nouar C (2005) Thermal convection for a thermo-dependent yield stress fluid in an axisymmetric
horizontal duct. Int J Heat Mass Transf 48:5520–5535
Oosthuizen PH, Naylor D (1999) An introduction to convective heat transfer. Wcb/McGraw-Hill,
USA
Study of the Incidence Angle Effect
on a Savonius Wind Rotor Aerodynamic
Structure
Sobhi Frikha, Zied Driss, Hedi Kchaou and Mohamed Salah Abid
Nomenclature
Cp Coefficient of the power, dimensionless
C1e Constant of the k–e turbulence model, dimensionless
C2e Constant of the k–e turbulence model, dimensionless
Cl Constant of the k–e turbulence model, dimensionless
CMs Static torque coefficient
d Rotor diameter, m
e Bucket thickness, m
Fi Force components, N
Gk Production term of turbulence, kg m−1 s−3
k Turbulent kinetic energy, J kg−1
Ms Static torque
P Pressure, Pa
ui Velocity components, m s−1
ui ′ Fluctuating velocity components, m s−1
e Dissipation rate of the turbulent kinetic energy, W kg−1
l Dynamic viscosity, Pa s
lt Turbulent viscosity, Pa s
q Density, kg m−3
rk Constant of the k–e turbulence model
re Constant of the k–e turbulence model
1 Introduction
With the recent deficiency in fossil fuels, demands for renewable energy sources are
increasing and wind energy has become the most reliable technology for power
generation. Two major types of wind turbines exist based on their blade configu-
ration and operation: the horizontal axis wind turbine and the vertical axis wind
turbine. The vertical axis wind turbine rotates around an axis that is perpendicular to
the oncoming flow; hence, it can take wind from any direction. It consists of two
major types, the Darrieus rotor and Savonius rotor. The Darrieus wind turbine
rotates around a central axis due to the lift produced by the rotating airfoils, whereas
a Savonius rotor rotates due to the drag force created by the blades in blades. The
Savonius wind rotor has the advantage of being compact, economical, and aes-
thetic. In addition, it has good starting characteristics, operates at relatively low
operating speeds, and has the ability to accept the wind from any direction. For
several years, many studies have significantly improved the performance of
Savonius rotors. For example Kamoji et al. (2009) investigated the performance of
modified forms of conventional rotors with and without central shaft between the
end plates. Menet and Bourabaa (2004) tested different configurations of the
Savonius rotor and found that the best value of the static torque coefficient is
obtained for an incidence angle equal to h = 45° and a relative overlap equal to e/
d = 0.24. They compared their numerical results with those obtained by Blachwell
et al. (1978) and a good agreement was obtained. Aldos (1984) studied the power
increase of the Savonius rotor by allowing the rotor blades to swing back when on
the upwind side. They reported a power increase of the order of 11.25% with the
increase in Cp from 0.015 to 0.17. Ushiyama and Nagai (1988) tested several
parameters of the Savonius rotor including gap ratio, aspect ratio, number of
cylindrical buckets, number of stages, endplate effects, overlap ratio, and bucket
design. The highest efficiency of all configurations tested was 24% for a two-stage,
two-bucket rotor. Grinspan et al. (2001) developed a new blade shape with a twist
for the Savonius rotor. They obtained a maximum power coefficient of 0.5 with this
model. Saha and Rajkumar (2005) compared the performance of a metallic-bladed
Savonius rotor, a bladed metallic Savonius rotor, to a conventional semi-circular
blade with no twists. The twist produced good starting torque and larger rotational
speeds and gives an efficiency of 0.14. The best torque was obtained with blades
twisted at an angle a = 12.5°. Akwa et al. (2012) studied the influence of the
buckets overlap ratio of a Savonius wind rotor on the averaged torque and power
coefficients by changing the geometry of the rotor. They noticed that the maximum
device performance occurs for buckets overlap ratios with values close to 0.15.
Khan et al. (2009) tested different blade profiles of a Savonius rotor both in tunnel
and natural wind conditions and they varied the overlap. The highest Cp of 0.375
was obtained for a blade profile of S-section Savonius rotor at an optimum overlap
ratio of 30%. Rogowski and Maroński (2015) studied the aerodynamic efficiency of
the Savonius rotor using fluid dynamics computational methods. The obtained CFD
results are compared with the experimental ones. The study has demonstrated that
Study of the Incidence Angle Effect … 163
the CFD methods confirm the experimental results and can be used to optimize the
buckets shape of the Savonius rotor. Driss and Abid (2012) conducted a compu-
tational fluid dynamic study to present the local characteristics of the turbulent flow
around a Savonius wind rotor. They compared their numerical results with the
experimental results and a good agreement was obtained. Driss et al. (2014) made a
numerical simulation of the turbulent flow around a small incurved Savonius rotor
and compared the results with the findings of experiments conducted in an open
wind tunnel. Compared to a circular Savonius rotor, the flow circulation of this
rotor is enhanced. Driss et al. (2015) compared different rotor designs characterized
by the bucket angles equal to w = 60°, w = 75°, w = 90°, and w = 130°. It has been
noted that the depression zones increase with the increase of the bucket arc angle.
The acceleration zone, where the maximum velocity values were recorded, is
formed in the convex surface of the rotor bucket and gets greater at the bucket arc.
The wakes characteristics of the maximum turbulent values are more developed
with the increase of the bucket arc angle. Mohamed et al. (2010) considered an
improved design in order to increase the output power of a Savonius turbine with
either two or three blades. Other works [D’alessandro et al. (2011); Dobreva and
Massouh (2011); Kacprzak et al. (2013); Zhou and Rempfer (2013)] performed
unsteady simulations and compared the improved version of Savonius rotor to
contribute to the improvement of Savonius rotor. Roy and Saha (2013) reviewed the
numerical works. They have shown that with the selection of a proper computa-
tional methodology, the Savonius rotor design, performance, and efficiency can be
enhanced significantly.
In this context, we were interested in studying the effect of the incidence angle
on the aerodynamic characteristics of the flow around a Savonius wind rotor. To
this end, we developed numerical simulations of the turbulent flow using a CFD
code.
3 Numerical Results
Figure 3 presents the distribution of the velocity field in the considered computa-
tional domain with a zoom around the Savonius rotor for different incidence angles
equal to h = 0°, h = 30°, h = 60°, h = 90°, h = 120°, and h = 150°. According to
these results, the flow appears uniform in the entry of the computational domain.
After that, a slowdown of the flow has been noted in the two buckets. Two wakes
characteristics of the velocity maximum values appear on the two rotor sides.
A wake characteristic of the minimum value of the velocity occurs downstream of
the rotor. A clear asymmetry of the two wakes characteristics of the maximum
Study of the Incidence Angle Effect … 165
(a) θ=0°
(b) θ=30°
(c) θ=60°
(d) θ=90°
(e) θ=120°
(f) θ =150°
Figure 4 presents the distribution of the mean velocity in the computational domain
with a zoom around the Savonius rotor for different incidence angles. According to
these results, a mean velocity equal to V = 9.95 m s−1 was recorded in the entry of
the computational domain. This is the value already imposed by the boundary
conditions. At the Savonius rotor, this value decreases greatly and reaches low
values. Two maximum values wakes characteristics appear on the two sides of the
rotor. Downstream of the rotor, a wake characteristic of the minimum values is
observed. A dissymmetry of the two wakes characteristics of the maximum values
is shown for the incidence angles equal to h = 0°, h = 30°, and h = 150° (Fig. 4a,
b, f). With the increase of the incidence angle, the two wakes become clearly
symmetrical for h = 60°, h = 90°, and h = 120° (Fig. 4c, d, e). However, a for-
mation of a wake characteristic of the minimum values was noted behind the rotor.
This wake appears in a form of a band form which depends on the incidence angle.
In fact, the maximum width of the band is obtained for h = 30° and h = 90°. For
the different values of the incidence angle equal to h = 0°, h = 30°, h = 60°,
h = 90°, h = 120°, and h = 150°, the maximum values are equal, respectively, to
V = 14.5 m s−1 (Fig. 4a), V = 15.2 m s−1 (Fig. 4b), V = 14.8 m s−1 (Fig. 4c),
V = 14.9 m s−1 (Fig. 4d), V = 14.7 m s−1 (Fig. 4e), and V = 14.1 m s−1 (Fig. 4f).
Indeed, it has been noted that the maximum value is obtained for an incidence angle
equal to h = 30°.
Study of the Incidence Angle Effect … 167
(a) θ=0°
(b) θ=30°
(c) θ=60°
(d) θ=90°
(e) θ=120°
(f) θ =150°
Figure 5 presents the distribution of the static pressure in the computational domain
with a zoom around the Savonius rotor for different incidence angles. While
examining these results, the static pressure was noted to be at its maximum at the
rotor. In fact, a compressure zone appears upstream of the rotor and grows on the
concave surface of the upper bucket. Behind the rotor, a depression zone has been
observed. This zone extends up to the output of the computational domain. By
comparing these results to each other, the variation of the incidence angle was
remarked to affect the distribution of the static pressure. In fact, for h = 0°, a
depression zone has been observed near the Savonius rotor (Fig. 5a). However, for
h = 30°, h = 60°, and h = 90° (Fig. 5b, c, d), a compression zone was observed on
the concave surface of the upper bucket of the rotor. This depression appears again
for the incidence angles equal to h = 120° and h = 150° (Fig. 5e, f). For the dif-
ferent values of the incidence angle equal to h = 0°, h = 30°, h = 60°, h = 90°,
h = 120°, and h = 150°, the maximum values of the static pressure are, respec-
tively, equal to P = 73.9 Pa (Fig. 5a), P = 144 Pa (Fig. 5b), P = 92 Pa (Fig. 5c),
P = 20.5 Pa (Fig. 5d), P = 39.9 Pa (Fig. 5e), and P = 77.2 Pa (Fig. 5f). Indeed,
the maximum value of the static pressure was observed to be obtained for an
incidence angle equal to h = 30°.
(a) θ=0°
(b) θ=30°
(c) θ=60°
(d) θ=90°
(e) θ=120°
(f) θ =150°
(f) θ =150° (e) θ=120° (d) θ=90° (c) θ=60° (b) θ=30° (a) θ=0°
Figure 7 shows the distribution of the turbulent kinetic energy in the computational
domain with a zoom around the Savonius rotor for different incidence angles.
According to these results, the turbulent kinetic energy is noted to be very low at the
entry of the computational domain. On the rotor, the turbulent kinetic energy
increases, especially at the edges of the two buckets. In fact, the occurrence of a
wake zone characteristic of the maximum values of the turbulent kinetic energy was
remarked. For h = 0° (Fig. 7a), this region is located on the outer surface of the
upper bucket of the rotor. Whereas, for h = 30° (Fig. 7b), this region appears on the
outer surface of the lower bucket. By increasing the incidence angle h = 60° and
h = 90° (Fig. 7c, d), two wake characteristics of the maximum values were
observed. These wakes are located in the external attack side of the upper bucket
and on the convex surface of the lower bucket of the Savonius rotor. These two
wakes become more extended for an incidence angle equal to h = 120° (Fig. 7e).
Away from the rotor, the turbulent kinetic energy becomes very low. For the
different values of the incidence angle equal to h = 0°, h = 30°, h = 60°, h = 90°,
h = 120°, and h = 150°, the maximum values of the turbulent kinetic energy are,
respectively, equal to k = 15.9 m2 s−2 (Fig. 7a), k = 30 m2 s−2 (Fig. 7b),
k = 45.1 m2 s−2 (Fig. 7c), k = 41.9 m2 s−2 (Fig. 7d), k = 22.2 m2 s−2 (Fig. 7e), and
k = 18.2 m2 s−2 (Fig. 7f). The maximum value of the turbulent kinetic energy is
obtained for the incidence angle h = 60°.
Figure 8 presents the turbulent kinetic energy dissipation rate in the computational
domain with a zoom around the Savonius rotor for different incidence angles.
According to these results, the wake characteristic of the maximum values was
remarked to be located on the convex surface of the lower bucket of the Savonius
rotor as well as in the external attack zone of the upper bucket. Outside this area, a
very fast decrease of the dissipation rate has been observed. In addition, these
results show that the variation of the incidence angle has a direct effect on the
location of the wake zone characteristic of the maximum values. For example, for
the incidence angle h = 0°, the wake is located on the convex surface of the upper
bucket of the Savonius rotor (Fig. 8a). However, for h = 30°, an extension of the
wake zone in the external attack zone of the upper bucket (Fig. 8b) was observed.
Also, the occurrence of an extended wake on the convex surface of the lower bucket
was noticed. This wake disappears for h = 60° and h = 90° and it appears on the
external attack side of the upper bucket of the rotor. The same fact was remarked for
the incidence angles h = 120° and h = 150° with the emergence of a second wake
area on the convex surface of the lower bucket. For the different values of the
incidence angles equal to h = 0°, h = 30°, h = 60°, h = 90°, h = 120°, and
172 S. Frikha et al.
(a) θ=0°
(b) θ=30°
(c) θ=60°
(d) θ=90°
(e) θ=120°
(f) θ =150°
(a) θ=0°
(b) θ=30°
(c) θ=60°
(d) θ=90°
(e) θ=120°
(f) θ =150°
h = 150°, the maximum values of the dissipation rate of the turbulent kinetic energy
are, respectively, equal to e = 7330 m2 s−3 (Fig. 8a), e = 14,900 m2 s−3 (Fig. 8b),
e = 62,000 m2 s−3 (Fig. 8c), e = 43,500 m2 s−3 (Fig. 8d), e = 19,000 m2 s−3
(Fig. 8e), and e = 87,200 m2 s−3 (Fig. 8f). The maximum value of the dissipation
rate of the turbulent kinetic energy is obtained for the incidence angle h = 150°.
In this section, we were interested in the study of the influence of the incidence
angle h on the variation of the static torque coefficient CMs of the Savonius rotor.
Several incidence angles (h = 0°, h = 30°, h = 60°, h = 90°, h = 120°, and
h = 150°) have been examined. Table 1 summarizes the different values of the
static torque Ms as well as the static torque coefficient CMs. The variation of the CMs
as a function of the incidence angle is presented in Fig. 10 using the Cartesian and
polar schematization. According to these results, the value of the CMs is remarked to
be quite low for an incidence angle h = 0°. By increasing h, an increase of the static
torque coefficient CMs was noticed. Indeed, the static torque coefficient reaches a
maximum value equal to CMs = 0.75 for h = 60°. From this angle, the value of the
CMs was remarked to decrease, reaching a minimum value equal to CMs = 0.12 for
an incidence angle h = 120°. After that, the values of CMs were remarked to
increase and decrease again. The results are already expected due to the symmetry
presented by the Savonius rotor. The values of the static torque coefficients found
for different incidence angles are compared with those found by Menet and
Study of the Incidence Angle Effect … 175
(a) θ=0°
(b) θ=30°
(c) θ=60°
(d) θ=90°
(e) θ=120°
(f) θ =150°
Cottier (2003). The profiles of the static torques present the same evolution of the
curve. The good agreement confirms the validity of the numerical method.
5 Conclusion
A numerical simulation of the turbulent flow around a Savonius wind rotor was
investigated for different incidence angles. According to the obtained results, the
incidence angle of the Savonius wind rotor has a direct effect on the turbulent flow.
The local characteristics such as velocity field, mean velocity, static pressure,
dynamic pressure, turbulent kinetic energy, dissipation rate of the turbulent kinetic
energy, and turbulent viscosity are different from one configuration to another. The
variation of the coefficient of the static torque CMs of the Savonius rotor was also
studied, and the numerical results were compared with those obtained by previous
results. A good agreement was obtained and confirmed the efficiency of our
numerical method. In the future, we are planning to study the effect of the overlap
of the buckets on the turbulent flow around the Savonius wind rotor.
Study of the Incidence Angle Effect … 177
References
Akwa JV, Júnior GA, Petry AP (2012) Discussion on the verification of the overlap ratio influence
on performance coefficients of a Savonius wind rotor using computational fluid dynamics.
Renewable Energy 38:141–149
Aldos TK (1984) Savonius rotor using swinging blades as an augmentation system. Wind Eng
8:214–220
Blackwell BF, Sheldahl RE, Feltz LV (1978) Wind Tunnel performance data for two and
three-bucket Savonius rotor. J Energy 2–3:160–164
D’Alessandro V, Montelpare S, Ricci R, Secchiaroli A (2011) Unsteady Aerodynamics of a
Savonius wind rotor: a new computational approach for the simulation of energy performance.
Energy 35:3349–3363
Dobreva I, Massouh F (2011) CFD and PIV investigation of unsteady flow through Savonius wind
turbine. Energy Procedia 6:711–720
Driss Z, Abid MS (2012) Numerical investigation of the aerodynamic structure flow around
Savonius wind rotor. Sci Acad Trans Renewable Energy Syst Eng Technol 2(2):196–204
Driss Z, Mlayeh O, Driss D, Maaloul M, Abid MS (2014) Numerical simulation and experimental
validation of the turbulent flow around a small incurved Savonius wind rotor. Energy 74:506–
517
Driss Z, Mlayeh O, Driss S, Driss D, Maaloul M, Abid MS (2015) Study of the bucket design
effect on the turbulent flow around unconventional Savonius wind rotors. Energy 89:708–729
Grinspan AS, Kumar PS, Saha UK, Mahanta P, Ratnarao DV, Veda Bhanu G (2001) Design,
development and testing of Savonius wind turbine rotor with twisted blades. Proc Int Conf
Fuid Mech Fluid Power, India 28:28–31
Kacprzak K, Liskiewicz G, Sobczak K (2013) Numerical investigation of conventional and
modified Savonius wind turbines. Renewable Energy 60:578–585
Kamoji MA, Kedare SB, Prabhu SV (2009) Experimental investigations on single stage modified
Savonius rotor. Appl Energy 86:1064–1073
Khan N, Tariq IM, Hinchey M, Masek V (2009) Performance of Savonius rotor as water current
turbine. J Ocean Technol 4(2):27–29
Menet JL, Bourabaa N (2004) Increase in the Savonius rotors efficiency via a parametric
investigation. Eur Wind Energy Conf, London
Menet JL, Cottier F (2003) “Étude paramétrique du comportement aérodynamique d’une éolienne
lente à axe vertical de type Savonius”, 16è Congrès Français de Mécanique, Nice
Mohamed MH, Janiga G, Thévenin E, Pap D (2010) Optimization of Savonius turbines using an
obstacle shielding the returning blade. Renewable Energy 35:2618–2626
Rogowski K, Maronski R (2015) CFD computation of the Savonius rotor. J Theor Appl Mech 53
(1):37–45
Roy S, Saha UK (2013) Review on the numerical investigations into the design and development
of Savonius wind rotors. Renew Sustain Energy Rev 24:73–83
Saha UK, Rajkumar M (2005) On the performance analysis of Savonius rotor with twisted blades.
J Renew Energy 960–1481
Ushiyama I, Nagai H (1988) Optimum design configurations and performances of Savonius rotors.
Wind Eng 12–1:59–75
Zhou T, Rempfer D (2013) Numerical study of detailed flow field and performance of Savonius
wind turbines. Renewable Energy 51:373–381
Study of Swirl Contribution to
Stabilization Turbulent Diffusion Flame
1 Introduction
Turbulent reactive flows with swirl have always been an important topic of study in
the combustion community because they promote flame stabilization and are
commonly used to efficiently blend fuel into the air. Swirl is an essential element in
modern combustion chambers designed to function as a lean mixture to reduce the
formation of pollutants. Thus, the contribution of the swirl movement to the mixture
makes it possible to reduce the emissions of pollutants, on the one hand and to
increase the efficiency and facilitate the stabilization of combustion, on the other.
Rawe and Kremer state that in a swirled flow, the rotation of the fluid on itself
creates a depression at the axis. If this depression is sufficiently large enough, it can
create a recirculation zone on the axis of the chamber. The amount of tangential
movement makes it possible to stabilize and improve the mixture. Many experi-
mental studies on turbulent flows reactive with swirl in combustion chambers have
been conducted using the Anemometer Laser Doppler System (LDA). Detailed
mean fluctuation distributions of axial, radial, tangential velocity, and probability
density functions (PDFs) for instantaneous axial and tangential gas velocities are
obtained by Anacleto et al. (2003) under different circumstances. They provide
useful data for combustion chamber design and optimization and also validate some
D. Lalmi (&)
Faculty of Exact Sciences, Natural Sciences and Life,
University of L’Arbi Ben M’hidi, Oeb, Algeria
e-mail: eldjemoui@gmail.com
D. Lalmi
Unité de Recherche Appliquée en Energies Renouvelables,
URAER, Ghardaïa, Algeria
R. Hadef
Faculty of Sciences and Applied Sciences, University of L’Arbi Ben M’hidi,
Oum El Bouaghi, Algeria
combustion models. Most swirl reactive studies use high swirl values, usually
S > 0.6. Thus, the formation of a recirculation zone is ensured, which amplifies
the mixture dynamics and contributes to the flame stabilization. The effect of the
recirculation zone is studied by Chen and Driscoll (1988). They show that when the
number of swirls increases, the length of the flame can be reduced by a factor of 5.
Other studies by Tangirala et al. (1987) show that the reaction favors the formation
of the recirculation zone and that even a cold test case does not have a recirculation
zone. More recent experiments by Feikema et al. show that with a sufficiently small
number of swirls, the recirculation zone is not formed. These low swirl conditions
can have a beneficial effect on the stability and extinguishing limits of a flame
resulting from coaxial flows of fuel oil and air. In particular, poor flames are
generally more stable when the flow is weakly twisted because the speed of the
swirl may stretch the flame until it is extinguished. Beer and Chigier report that for
non-reactive swirled flows, S > 0.6 is necessary to establish an internal recirculation
zone. However, with the addition of heat release, recirculation zones may be
established in flows with substantially smaller swirl numbers. Chen and Driscoll
(1988), for example, assert the occurrence of the recirculation zone in a swirling
flame with S = 0.2.
Similarly, Tangirala et al. (1987) find that for a reactive flow, the swirl number
S = 0.7 is necessary for the appearance of the recirculation zone. The significant
conclusion is that the release of heat increases the susceptibility of a vortex flow;
this can be explained by the change of density effect.
According to Lartigue, new digital tools are now able to predict combustion
instabilities by introducing swirled flows. He describes a new formulation of a
large-scale simulation code (the AVBP code developed jointly by CERFACS and
IFP) to make it possible very precisely the thermodynamic and chemical phe-
nomena associated with combustion. A validation of this work was presented in a
complex geometry (PRECCINSTA focus).
The numerical results are compared successfully with the experimental mea-
surements carried out by DLR Stuttgart (Germany). Zhang and Nieh studied
numerically and experimentally the turbulent vortex flow and pulverized coal
combustion in the vortex combustor (VC) using the Algebraic Reynolds Model
(ARSM). They described in detail the flow characteristics and the combustion of gas
particles, relating to turbulence, temperature, species concentrations, particle density,
and trajectories. They found that the gas flow into the VC with a coaxial central tube
and multiple air injections is characterized by the appearance of a recirculation zone.
Other numerical simulations of highly swirled turbulent flows were performed in
a VC chamber by Ridluan et al. (2007).
A complete work was carried out on a three-dimensional isothermal flow in a
VC using three first-order turbulence models: the standard k–e turbulence model,
the Renormalized Group (RNG) k–e model, and the Shear Stress Transport model
K–x and a second-order turbulence model, the Reynolds Stress Model (RSM) as
well as a second-order difference scheme.
The calculation indicates that the RSM is superior to other turbulence models by
capturing the swirl effect in comparison with the measurements.
Study of Swirl Contribution to Stabilization … 181
The numerical results for the VC flow provide the flow characteristics in terms of
parameters suitable for the VC design, consisting of axial and tangential velocities,
a pressure field, and kinetic turbulence energy.
Large-scale simulations were carried out for two cases of reactive and
non-reactive flows in an industrial gas turbine burner using a compressible
unstructured solver by Selle et al. (2004).
The numerical results are compared with the experimental measurements in
terms of axial and tangential velocities (mean and RMS), the mean temperature and
the existence of natural instabilities such as precessing vortex core (PVC).
On the other hand, the LES is performed by Roux et al. (2005) with a two-stage
mechanism for combustion of air-methane and a thick flame pattern.
The combustion regime is partially premixed. For this very complex geometry,
the results demonstrate the ability of the LES to predict the mean flow, with and
without combustion, as well as these unstable modes: for example, the PVC mode
is very strong for cold flow but disappears with combustion. Martin et al. (2006)
also emphasize the need for well-defined boundary conditions: for example, the
calculation should include valves called swirled or inclined injectors. In order to
reduce nitrogen oxide (NOx) emissions in industrial combustion systems, the use of
injectors operating under lean and premixed combustion has become widespread in
recent years.
However, in this regime, strong combustion instabilities may occur and damage
the device or cause the flame to extinguish. An overview of recent studies of
combustion in swirl flows is presented by Syred and Beer (1974).
More recent studies have been carried out by Claypole and Syred, on com-
bustion chambers with swirled flows to analyze the effect of swirl on NOx for-
mation and stability limits by Rawe and Kremer (1986). Among the study cases, we
can cite the variable geometry combustion chamber installed at the University of
Maryland, to characterize the stability and the emission levels in swirled complex
flows by Gupta et al. A study by John and Samuelson uses active control techniques
to a flame stabilized by the swirl in a combustion chamber.
They vary the intensity of the swirl and the air flow to optimize the performance
of the burner in terms of NOx production and combustion efficiency. The power
generation industry largely uses “swirlés” swirling flow burners because they
provide flame stability meaning NOx control.
Intensive experimental and numerical investigations investigate the properties of
such burners by Gupta et al. (1991), Zhang and Nieh (1997), Widmann et al. (1999).
The ultimate aim of this whirling flow research with combustion must provide the
optimum burner to minimize NOx emissions with high combustion efficiency.
The introduction of a “swirl” rotational movement makes it possible to increase
the stability of the flame by creating a recirculation of exhaust gas from the injector
and also to limit NOx emissions by improving the combustible mixture (Syred
2006).
However, Ducruix et al. (2003) report that swirled flows often periodically
develop large vortex structures whose interaction with the flame and acoustic
modes of the burner can cause strong instabilities.
182 D. Lalmi and R. Hadef
The influence of turbulence models on the simulation of jets and flames has been
explored by several authors such as Kucukgokoglan et al. (1999) who have pre-
sented in their work a description an isothermal turbulent flow testing the perfor-
mance of three variants of the k–e turbulence model: k–e Standard, RNG k–e, and
k–e realizable.
They explained that it is at a distance equal to at least one and a half times the
diameter of this type of burner, downstream of the outlet, that the ignition of flame
generally occurs and that it is in this region too, that there is a production of a
substantial amount of all emissions of nitrogen oxides (NOx).
They found that the use of these three variants of the k–e model for the simu-
lation of isothermal turbulent flows with swirl gives very interesting results and is
closer to the experimental results. In our study, we presented the results obtained
with the calculation of co- and counter-swirl rotation effect on the characteristics
and stabilization of the turbulent diffusion flame.
The simulation contribution to improve the experimental methods is the only
reason for choosing this study. The use of two flows produces less NOx than a
single flow because of its capacity to homogenize temperature.
2 Mathematical Formulation
q~ui
@ @ @P @
þ q~ui ~uj þ ¼ u00i u00j
sij q ð2Þ
@t @xi @xi @xi
qY~k
@ @ ~ @
~_ k
þ ~ui Yk þ
q u00i Yk00 ¼ x
Vk;i Yk þ q ð3Þ
@t @xi @xi
¼q
P ~r T~ ð5Þ
where Ui and u0i are the average and fluctuating velocity components in the direction
xi , Yk is the methane mass fraction, P is the pressure, l is the dynamic viscosity, and
q is the density of the fluid. The hs is the sensible enthalpy, k is the thermal
conductivity; Dk is the species diffusivity; T is the temperature; x_ k is the secrecies
reaction rate. The viscous heating term and radiation term in Eq. (4) are neglected
as they are negligible compared to the combustion source term. The mass flux is
described by Fick’s law in Eq. (3).The Schmidt number ScK is assumed to be 0.7 in
this work.
The SST K–x model is based on the general model k–ɛ Magnussen and Hjertager
(1976), Claypole and Syred (1981), Gupta et al. (1984, 1991), Wegner et al. (2004a,
b) whose transported variables are the turbulent kinetic energy K and the turbulent
frequency x. Its equations are as follows Lalmi and Hadef (2015a, b, c, d):
@k @k @ @k ~ k b kx
þ Ul ¼ ð m þ rk m t Þ þP ð6Þ
@t @xl @xl @xl
@x @x @ @x x
þ Ul ¼ ð m þ rx m t Þ þ a2 P k
@t @xl @xl @xl k
ð7Þ
r x;2 @k @x
b2 x2 þ 2ð1 F1 Þ
x @xl @xl
where F1 is the mixture function of (equal to the unit in close wall and zero in the
remote area) defined by:
( pffiffiffi 4 )
k 500m 4rx;2 k
F1 ¼ tan h min max ; 2 ; ð8Þ
b xy y x CDkx y2
where y is the normal distance to the wall nearest and CDkx is the positive
equivalent portion of cross-diffusion in the Eq. (9). CDkx , is defined by:
1 @k @x 10
CDkx ¼ max 2rx;2 ; 10 ð9Þ
x @xj @xj
184 D. Lalmi and R. Hadef
The transition between the two formulations, k–e and k–x, is done through the
function F1 . When F1 0 far from the walls, the formulation k–e is activated and the
turbulent kinematic viscosity is given by:
a1 k
mt ¼ ð10Þ
maxða1 x; SF2 Þ
The model SST also contains a limiting device in order to avoid the artificial
construction of turbulence in the stagnation areas:
@Ui @Uj @Ui ~ k ¼ minðPk ; 10b kxÞ
Pk ¼ mt þ !P ð12Þ
@xj @xi @xj
3 Combustion Model
One of the crucial steps to modeling combustion in ANSYS Fluent 14.0 is the
choice of a combustion model. So we first have to choose a premixed combustion
or non-premixed combustion options. This determines whether or not the reactants
are initially mixed. From this stage, several choices of combustion model are
available. Here we would like to study a turbulent diffusion flame (non-premixed
combustion). This model needs to create the PDF table, where all parameters and its
information on thermochemistry interactions with turbulence are adopted. All
quantity values or all species such as fuel and oxide at the inlet have been induced
in their experimental values cited in fourth part in the PDF table. The PDF table
displays temperature and reacting species: methane, oxygen, and carbon dioxide
with mixture fraction figure blow. We can see that the maximum of temperature is
2200 K corresponding to 0.125 of mixture fraction Fig. 1.
All the calculations were performed using an atmospheric air blast atomizer in a
cylindrical combustion chamber Fig. 2a. The atomizer consists of a modular
arrangement of two radial swirl generators, an atomizer lip which separates the two
airstreams from each other within the nozzle, and an air diffuser with a throat
diameter of D0 = 2R = 25 mm.
Study of Swirl Contribution to Stabilization … 185
1,000
2000
0,800
CH4
Mole Fraction
Temperature
1500
0,600 O2
CO2
1000
0,400 Temperature
0,200 500
0,000 0
0 0,2 0,4 0,6 0,8 1
Z
(a)
Air
CH4
Air
CH4
Air
(b)
The air mass flow rate is adjusted to 64 kg/h (Mi/Mo = 0.37) and heated to
400 °C. Theoretical swirl numbers S0th of the inner as well as the outer airflow are
Si = 0.46 and So = 1, resulting in a global swirl number of 0.81 Fig. 2b. The
Reynolds number is calculated as the product of the axial average air velocity at
the nozzle exit (39.82 m/s) and the throat diameter of the diffuser divided by the
186 D. Lalmi and R. Hadef
kinematic viscosity of the air and yields approximately 60,000. For more detailed
specifications concerning the combustor, we refer to Merkle et al. (2003).
For the equations resolution is carried out numerically. The transport equations of
all the parameters are solved in a steady operation, by QUICK scheme.
Pressure–velocity is coupled with the SIMPLE algorithm. The model SST K–x
is used in a grid of 0.80 million hexahedral cells. In this work, we use three types of
boundary conditions. All parameters at the inlet are measurements, velocity species,
and turbulent kinetic energy with UDF algorithm profiles, except for dissipation
which is estimated by:
k 1:5
e ¼ Cl0:75 ð13Þ
0:7D
Adiabatic walls were used for all geometrical configuration combustion chamber
models. An outflow boundary condition was imposed for the exhaust of the model.
As validation, we define different axial sections to plot all the radial parameter
profiles as shown in Fig. 3.
6 Results
Figure 4 shows stream function contours for two cases, co- and counter-swirl. It is
seen that the two recirculation zones in both cases are predicted, the first is the corner
recirculation zone and the second is the toroidal central recirculation zone. We can
see that the central recirculation zone in the counter-swirl is greater than that in the
co-swirl, which explains the amount of recalculated mass detected by Merkle et al.
(2003) in their experimental study. They obtained 0.56 in co-swirl and 0.76 in
counter-swirl. However, the stream function is respectively −0.814 and −0.84
detected in the corner recirculation zone for co- and counter-swirl.
In Fig. 5, we presented the axial velocities contours for more information about
the shape and the size of the corner and central recirculation zones. The presence of
negative values of axial velocity in both cases is clear, indicating the existence of
the recirculation zone along the axial axis. This information validates the precedent
interpretation cited in the last figure. We can also see the detected length and
disposition of two recirculation zones in both cases.
We start our description of the mean flow by comparing the obtained results and
measurements (Merkle et al. 2003). Figures 6, 7, and 8, respectively, represent the
mean components: axial, radial, and tangential velocities. The comparisons between
calculations and measurements are satisfactory. It can be noticed that the com-
parisons are limited in space (near the injector); in fact, we note that shape and size
of the recirculation zone are predicted correctly. This first point is essential for
correctly predicting the flame stabilization area in both cases, co- and counter-swirl.
Near the injectors x/R = 0.2, the influence of the nozzles geometry (air injection) is
perfectly visible in the two cases.
The fuel injected between the two air flows opens into a shear zone, while taking
advantage of the ideal conditions to mix with the air. Moreover, the fuel being
injected axially tends to reduce the tangential component at the outer swirler and
thus tightens the jet.
The comparisons of the tangential velocities show a good prediction at the outer
swirler, which is crucial to predict the blast opening of the jet. The values gap near
the axis has little influence on the swirl number since the amount movement flow is
proportional to the square radius. It should be noted that the radial and tangential
velocities changed its sign on the symmetric axis.
Figure 8 shows the mean field temperature contour in the two cases. The
temperature distribution is uniforms and equal to the command value in the
expansion section, just downstream the expansion, the temperature begins to rise
sharply, indicating the start of combustion. The isothermal contours clearly illus-
trate the shape of the flame front. The maximum temperature “1800 K” is reached
on the axis at the central recirculation zone. The combustion is almost complete
with x = 0.20 m, and there is no more fresh gas beyond this. It should be noted that
190 D. Lalmi and R. Hadef
we do not observe any thermal boundary layers along the combustion chamber
walls because the walls are supposed adiabatic. As can be seen from Fig. 9, the
difference between co- and counter-swirl is only at the corner zone. There is a heat
loss through the combustion chamber walls; it is about 7% between the two cases.
Otherwise, this region is dominated by the jet itself; this can be explained by the
existence of two injection diameters. The maximum temperature is seen in this zone
is 1800 K in the first case.
The flame is not attached to the injector lips because the flow velocity is too
great. The rich zones are delimited by the stoichiometric line. Inside this line, a rich
premix pocket is formed, and while on the outside, the premix is poor.
The rich zone burns in premix (corresponding to the pink contours of the
reaction rate in Fig. 9). Behind the front flame, air coming from the inner swirling
flow (central swirler) is heated by the burnt gases retained in the central recircu-
lation zone. This diluted air will burn in diffusion with the excess fuel coming from
the rich premix flame (blue contours in Fig. 9). The flame closes at the level of
outer swirler. The burned gas returns to the recirculation zone of the step and warms
Study of Swirl Contribution to Stabilization … 191
the fresh mixture. On the whole, it can be seen that the areas where the flame burns
in diffusion are very small compared with the zones where the flame burns is
premixed. The average fields reveal that the flame is compact and stabilized near the
injectors for both cases.
The mass fraction of methane is represented in radial profiles form at different
positions: 5, 10, and 25 mm as shown in Fig. 10.
The fuel consumption is clearly illustrated, because the mass fraction decreases
from the initial value at the outlet jet and is totally burned just in the front flame to
give the carbon dioxide and the water vapor.
It is noted that the excess of air is translated by an amount of oxygen which
remains in the products of the combustion (rate of reaction).
Among the species produced, the CO2 mass fraction is also expressed as radial
profiles in the last positions.
The behavior of H2O is similar to CO2. By crossing the front flame, the mass
fraction is increased to reach maximum values. Downstream the front flame, the
concentration of these species decreases since they mix with the surrounding air.
Figure 11 shows the reaction rate normalized by its maximum. The reaction rate
is similar in both cases; however, there are only a small difference in the central
toroidal recirculation zone for the first configuration.
The intermediate species CO as defined by the reduce reaction mechanism in
two steps by Peters and Williams is illustrated in Fig. 12.
The mass fraction of this component is zero in the burned gases and becomes
maximal in the front the flame. In this simple mechanism, the CO recombines to
give the CO2; when at the outlet, there are both CO2 and H2O.
192 D. Lalmi and R. Hadef
Fig. 12 Profiles of CO at
different sections of both
cases co- and counter-swirl
7 Conclusion
The work undertaken in this chapter is a numerical study of the flow with chemical
reaction in two geometric configurations concerning the direction of rotation of
swirl: co- and counter-swirl similar to a combustion chamber. Particular attention
has been paid to the effect of the swirling rotation imposed at the entry on the
aerodynamic and thermochemical behavior. The calculation code was used for the
study of the combustion taking into account the influence of the thermochemical
properties of the mixture on the flow. Efforts to achieve these objectives have
enabled the acquisition of good experience in the fields of the use of a commercial
code, on the one hand, and a good initiation to research in the field of combustion,
on the other. Comparisons have thus been made, on both reactive flows with two
different configurations. In each of these two cases, the comparison of the numerical
calculations and the experiments proved to be encouraging.
The exploration of the results allowed a better understanding of the phe-
nomenology of the flow in question, in particular:
• The complex behavior of the temperature field influenced essentially by the
swirl and the section changes.
• Appearance of a recirculation zone was detected in both cases.
• The results showed a toroidal recirculation central zone which was formed just
downstream of the injector and the flame stabilized in a compact form. This
makes it possible to reduce the dimensions of the chamber.
Finally, comparing the two rotation directions, it is found that the quantity of the
mass flow in the central zone and at the corners is greater in the counter-swirl even
at high temperatures. This confirms the results previously reported in the literature.
It is suggested to test the performance of other combustion models for these two
configurations. It would be very interesting to consider the study of pollutant
emissions in this counter-swirl configuration.
194 D. Lalmi and R. Hadef
References
Anacleto PM, Femandes EC, Heitor MV, Shtork SI (2003) Swirl flow structure and flame
characteristics in a model lean premixed combustor. Combust Sci Tech 175:1369–1388
Chen RH, Driscoll JF (1988) The role of the recirculation vortex in improving fuel air mixing
within swirling flames. Proc Combust Institute 22:531–540
Claypole TC, Syred N (1981) The effect of swirl burner aerodynamics on NOx formation. Proc
Combust Inst 18:81–89
Ducruix S, Schuller T, Durox D, Candel S (2003) J Propul Power 19:722–734
Gupta AK, Lilley DG, Syred N (1984) Swirl flows. Abacus Press, Tunbridge Wells
Gupta AK, Ramavajjala M, Chomiak J, Marchionna N (1991) Burner geometry effects on
combustion and emission characteristics using variable geometry swirl combustor. J Propul
Power 7:473–480
Kucukgokoglan S, Aroussi A, Pickering SJ (1999) Prediction of interaction between burners in
multi-burner systems. University of Nottingham, Nottingham, NG7 2RD, UK, p 1
Lalmi D, Hadef R (2015a) 3D, numerical study of turbulent mixing (fuel/air) in confined swirling
flow. Congrèe Français de Mécanique 24–28
Lalmi D, Hadef R (2015b) Evaluation of the statistical approach for the simulation of a swirling
turbulent flow. Am J Mech Eng AJME 03(N°3A27-31)
Lalmi D, Hadef R (2015c) evaluation of the performance of two turbulent models in the prediction
of a swirling flow. Int J Mech Energy (IJME) 3(1) (ISSN: 2286-584)
Lalmi D, Hadef R (2015d) Numerical simulation of co- and counter- swirls on the isothermal flow
and mixture field in a combustion chamber. Adv Appl Fluid Mech AAFM 18(2):199–212
Magnussen BF, Hjertager BH (1976) On mathematical models of turbulent combustion with
special emphasis on soot formation and combustion. Proc Combust Inst 16:719–729
Martin C, Benoit L, Sommerer Y, Nicoud F, Poinsot T (2006) Large-eddy simulation and acoustic
analysis of a swirled staged turbulent combustor. AIAA J 44:741–749
Merkle K, Haessler H, Büchner H, Zarzalis N (2003) Effect of co- and counter-swirl on the isothermal
flow- and mixture-field of an airblast atomizer nozzle. Int J Heat Fluid Flow 24:529–537
Ridluan A, Eiamsa-ard S, Promvonge P (2007) Numerical simulation of 3D turbulent isothermal
flow in a vortex combustor. Int Commun Heat Mass Transfer 34:860–869
Rawe R, Kremer H (1981) Stability limits of natural gas diffusion flames with swirl. Proc Combust
Institute 18:667–677
Roux S, Lartigue G, Poinsot T, Meier U, Berat C (2005) Studies of mean and unsteady flow in a
swirled combustor using experiments, acoustic analysis, and large eddy simulations. Combust
Flame 141:40–54
Selle L, Lartigue G, Poinsot T, Koch R, Schildmacher KU, Krebs W, Kaufman P, Veynante D
(2004) Compressible large eddy simulation of turbulent combustion in complex geometry on
unstructured meshes. Combust Flame 137:489–505
Syred N (2006) Prog Energy Combust Sci 32:93–161
Syred N, Beer JM (1974) Combustion on swirling flows. Combust Flame 23:143–181
Tangirala V, Chen RH, Driscoll JF (1987) Combust Sci Technol 51–75
Wegner B, Maltsev A, Schneider C, Sadiki A, Dreizler A, Janicka J (2004a) Assessment of
unsteady RANS in predicting swirl flow instability based on LES and experiments. Int J Heat
Fluid Flow 25:528–536
Wegner B, Kempf A, Schneider C, Sadiki A, Dreizler A, Janicka J, Schäfer M (2004b) large eddy
simulation of combustion processes under gas turbine conditions. Prog Comput Fluid Dyn
4:257–263
Widmann JF, Charagundla SR, Presser C (1999) Report No. NISTIR 6370, National Institute of
Standards and Technology
Zhang J, Nieh S (1997) Comprehensive modelling of pulverized coal combustion in a vortex
combustor. Fuel 76:123–131