air turbulent flames

Energy Conversion and Management 140 (2017) 121–132

Contents lists available at ScienceDirect

Energy Conversion and Management journal homepage: www.elsevier.com/locate/enconman

A parametric study of microjet assisted methane/air turbulent flames Sirine Chouaieb a,⇑, Wassim Kriaa a, Hatem Mhiri a, Philippe Bournot b a b

UTTPI, National Engineering School of Monastir, Monastir, Tunisia Institut de Mécanique de Marseille, 60 rue Juliot Curie Technopôle de Château-Gombert, 13453 Marseille Cedex 13, France

a r t i c l e

i n f o

Article history: Received 5 December 2016 Received in revised form 24 February 2017 Accepted 25 February 2017

Keywords: Methane/air flames K-epsilon Standard model with Pope Correction Microjet Velocity Diameter Soot

a b s t r a c t A parametric study of microjet assisted methane/air turbulent flames characteristics is numerically investigated. The Presumed Probability Density Function model and the Discrete Ordinates model are respectively considered for combustion and radiation modeling. The k-epsilon Standard model with Pope Correction is adopted as a turbulence closure model. The two step Tesner model is used to quantify the soot particle production in the flame configuration. Comparison with our previous work using the kepsilon Realizable model shows that the k-epsilon Standard model with Pope Correction ensures better predictions. The microjet velocity and diameter effects on thermal field, mixing process and soot emission are then discussed. Numerical findings show that the microjet can be used as an efficient tool controlling methane/air turbulent flames. On the one hand, it is shown that the microjet creates an inner flame in the vicinity of the central nozzle exit but does not globally alter the methane/air flame shape. On the other hand, mixing process can be enhanced for high microjet Reynolds number either by increasing the microjet velocity or by decreasing its nozzle diameter for a constant microjet mass flow rate. Soot production can be consequently reduced for low microjet diameter and high velocity values. Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction Reactive mixture control constitutes the major objective of researchers and industrials dealing with combustion equipments. Indeed, an efficient mixture guarantees flame stability, ensures better performance and reduces consequently pollutants’ emission. Several techniques were applied to control reactive mixing process. Kang et al. [1] have underlined the effect of moderate or intense low-oxygen dilution (MILD) regime in reducing unburned hydrocarbons fraction at the furnace outlet. Pulse combustion has been used as an efficient tool able to maximize thermal efficiency and minimize pollutant emission [2]. Zhuang et al. [3] have controlled the intake air swirl motion in a spark-ignition directinjection engine in order to improve fuel/air mixing process under low load conditions. Yu et al. [4] have highlighted the repetitive laser-induced plasma efficiency in the stabilization of a premixed methane/air flame. Jeon et al. [5] have presented the homogeneous charge compression ignition method as an effective strategy in controlling combustion performance, flame, and soot in a compression-ignition engine. The use of hydrogen as an alternative to fossil fuels has been elucidated by Nicoletti et al. [6]. By defining a global environmental index for different fuels, the authors have shown that industrial devices performance is the most increased ⇑ Corresponding author. E-mail address: [email protected] (S. Chouaieb). http://dx.doi.org/10.1016/j.enconman.2017.02.079 0196-8904/Ó 2017 Elsevier Ltd. All rights reserved.

when hydrogen is used [6]. Zhang et al. [7] have proven the effectiveness of hydrogen addition in improving a butanol engine performance and controlling nitric oxides (NOx) for a lean combustion regime. Amrouche et al. [8] have shown that hydrogen enrichment improves combustion process through shortening the flame length. Kashir et al. [9] have studied propane enriched flames developed in a bluff-body burner. Local heat release rate was found to be increased with hydrogen addition due to the inflow carbon atoms decrease [9]. Ilbasß et al. [10] have studied the effect of the swirl number on combustion characteristics and pollutants emission of hydrogen enriched flames developed in a gas-fired combustor. Chouaieb et al. [11] and Tabet et al. [12] have presented hydrogen enrichment as a promising technique able to enhance mixing process. The effect of microjet association to hydrogen enriched flames in reducing soot emission and promoting reactive mixture have been highlighted by Chouaieb et al. [11]. This work constitutes a contribution to underline the microjet technique as a promising tool in controlling turbulent flames. Actually, the concept of microjet in reactive coaxial configurations was introduced by the experimental works of Ganguly et al. [13] and Sinha et al. [14]. Indeed, a microjet can be simply defined as a jet of air or inert species ejected from a low diameter nozzle. Ganguly et al. [13] have equipped an unconfined coaxial burner by a central air microjet of diameter Dj equal to 1 mm. Their goal was to show experimentally the capacity of the added microjet to control methane/air flame shape, luminosity and emissions for

122

S. Chouaieb et al. / Energy Conversion and Management 140 (2017) 121–132

Nomenclature D Dh Dk E f 0 f g H h k L P T ui, uj u0i ; v0j V X

diameter, m hydraulic diameter, m molecular diffusion coefficient, m2
s1 total mass energy, J
kg1 mixture fraction fluctuation of the mixture fraction gravity acceleration, m
s2 enthalpy, J specific enthalpy, J
kg1 kinetic energy of turbulence, m2
s2 length of the cylindrical volume, m static pressure, Pa temperature, K mean velocity components along x and y directions, m
s1 fluctuating velocity components, m
s1 velocity magnitude, m
s1 axial position, m

different microjet rates. Experimental results demonstrate the microjet ability in reducing flame’s length. It was also found that the microjet assisted flames were characterized by a lower brightness reflecting less soot production. The experiments of Ganguly et al. [13] have been prosecuted by those of Sinha et al. [14]. These authors have been interested in microjet assisted methane/air flames developed in a confined volume. The microjet has been presented as a hydrodynamic tool to control methane/air flames by improving the air training in the near field of the central nozzle exit area [14]. Another experimental study was conducted by Yuchun et al. [15] using several lateral microjets of diameters equal to 1 mm. In their work, the authors have been interested in jet flames issued from a burner equipped with six lateral microjets. This study has shown that, for the same air and fuel rates, the lateral microjets were able to modify flames structure and reduce their lengths by improving mixture. Certainly, the above studies reflect the interest of the microjet technique and its efficiency in controlling methane/air diffusion flames characteristics and pollutants especially because a microjet can be easily handled in industrial combustion engines. Nevertheless, the number of researches dealing with this axis is limited and few numerical studies are carried out. Kanchi et al. [16] has explored, numerically, the feasibility of microjets’ addition to improve round and flat burner performance. Using the k-epsilon Standard model, simulations showed that the presence of microjets with a flow rate not exceeding 1% of the main flow rate implies an increase in heat release by 22% and 140% respectively for round and flat burners. A more recent numerical study was conducted by Chouaieb et al. [11]. In this work, hydrogen enrichment was applied to reduce soot emission and enhance mixing process in methane/air confined flames under the addition of an air microjet. Numerical findings have shown that the use of a central microjet associated to hydrogen enriched flames succeeds in ensuring better mixing process and less soot production. Certainly, hydrogen enrichment can be judged as a promising technique in improving reactive mixture and decreasing pollutants emissions [17]. Nevertheless, the use of hydrogen in industrial devices is limited by the high cost related to its storage and production. In this context, contrary to our previous work [11], our objective is to underline, presently, the microjet efficiency in controlling

Y Yk

radial position, m mass fraction of species k

Greek symbols q density, kg
m3 l dynamic molecular viscosity, kg
s1
m1 m kinematic viscosity, m2
s1 e dissipation rate of the turbulent kinetic energy, m2
s3 £i thermo-chemical scalar (mass fraction, density and temperature) sij stress tensor, kg
s2
m1 Subscript a g j t

air fuel (methane) microjet turbulent

methane/air flames characteristics without implying a composite fuel (methane-hydrogen mixture). A parametric study is then carried out to optimize mixing process and soot emission for microjet assisted methane/air flames developed in the Brookes and Moss confined coaxial configuration [18]. Particularly, in the first part of this paper, previous numerical validation is enhanced in terms of temperature, mixture fraction and soot profiles. The use of the k-epsilon Standard turbulence model with Pope Correction instead of the Realizable model [11] shows better agreement with the experimental data of Brookes and Moss [18]. In the second part, the effect of the microjet velocity and diameter on methane/air flames characteristics is investigated.

2. Presentation of the computational domain The Brookes and Moss configuration [18] is used to study numerically a confined turbulent reactive flow issuing from two coaxial jets. These authors have studied experimentally [18] then numerically [19] the development of a turbulent methane/air diffusion flame in a confined volume. Given the availability of several measurements in terms of temperature, mixture fraction and soot particles in both axial and radial positions, this configuration was previously studied by many authors. Kronenburg et al. [20] were the first who have studied numerically this configuration. The importance of soot particles’ differential diffusion in predicting soot profiles has been demonstrated by Kronenburg et al. [20] and Navarro-Martinez et al. [21] for methane/air flames. The same aspect has been discussed by Wolley et al. [22] for both methane/air and propane/air flames. Tabet et al. [12] have shown the efficiency of hydrogen addition to methane/air flames in enhancing mixing process. Saqr et al. [23] have proven that pollutants production can be reduced by increasing the air free stream turbulence intensity. The effect of turbulence model and radiation model in capturing thermal and dynamic fields has been studied respectively by Saqr et al. [24] and Kassem et al. [25]. Furthermore, Kassem et al. [26] have shown the capacity of the Eddy Dissipation model implementation in OpenFoam code in capturing thermal and species fields. It should be noted that the Brookes and Moss configuration is equipped with an internal nozzle, of diameter Dg equal to 4 mm ejecting methane. An annular nozzle of diameter Da equal to 155 mm is used to hold the external air jet. The resulting turbulent

123

S. Chouaieb et al. / Energy Conversion and Management 140 (2017) 121–132

diffusion flame is developed in a cylindrical confinement characterized by a length equal to 1000 mm. In this work, a central microjet of diameter Dj varying from 1 mm to 3 mm is added as shown in Fig. 1. Details of the tested microjet assisted methane/air flames configuration are illustrated in Fig. 1. 3. Numerical approach An incompressible multispecies turbulent reactive flow is numerically investigated. The governing equations are the reactive Navier–Stokes equations (NSE). The steady-state of the conservation equations for mass, momentum, species and energy are numerically solved. Density variation in the flow field is taken into account using the Favre decomposition. The fluid is assumed to be incompressible with a variable density and the reaction mixture species obey to the ideal gas law. 3.1. Governing equations Based on the Favre decomposition, the conservation equations of mass and momentum are written as follows:

@
 q
u~j ¼ 0 @xj

ð1Þ

  
u
~ @u ~ 2 @u ~ ~i u ~j Þ @ðq @p @ @u ¼ þ l i þ j  dij i @xj @xi @xj @xj @xj 3 @xj @ 
gi qu0i u0j þ q þ @xj

ð2Þ

Reynolds stresses are related to the average velocity gradients according to the following expression:

    @ uei @ ueJ 2 @ ue  qu0i u0j ¼ lt þ q
k þ lt i dij 3 @xJ @xi @xi

ð3Þ

lt represents the turbulent dynamic viscosity and k is the turbulence kinetic energy.

The k-epsilon Standard model with Pope Correction [27] was adopted as a turbulence closure model. It should be noted that this model was previously used by several researchers studying the Brookes and Moss configuration [18]. Indeed, it has been shown that the Pope correction previously used by Tabet et al. [12], Brookes and Moss [19] and Kronenburg et al. [20] succeeds in predicting the reactive coaxial jet spread and as a consequence, the experimental data has been well reproduced. 3.3. Radiation modeling As used in our previous work [11], the Discrete Ordinates model (DO) [28] was adopted to consider thermal radiation in numerical simulations. Indeed, the DO model is suitable for optically thin problems (a
L < 1) [28]. In combustion chambers, L and a correspond respectively to the combustor diameter and the absorption coefficient varying between 0.01 and 0.1 m1. In the current study, the problem can be assumed as optically thin since L is equal to 0.155 m. The absorption coefficient is calculated according to the Weighted Sum of Gray Gases Model (WSGGM) [29]. The radiation intensity transport equation is presented in Ref. [30]. 3.4. Combustion modeling The Presumed PDF model was used for combustion modeling. In our previous work [11], this model was compared to the Eddy Dissipation model. Numerical results have shown the superiority of the Presumed PDF model based on the mixture fraction scalar f to predict temperature and species field [11]. The conservation equation of the mixture fraction f and its vari0 ance f are written in the following form [31]:

@  @ q
ueJ ef ¼ @xj @xj

lt @~f rt @xj

!

ð4Þ

0 1 !2   f 02 @ @ @lt @ f A @~f e 02 f 02
ff þ C g lt q
ueJ f ¼  Cdq @xj @xj rt @xj @xj k

ð5Þ

0 With f ¼ f  ~f . The default values of constants rt , Cg and Cd are respectively equal to 0.85, 2.86, and 2.0. For non-adiabatic reactive systems, an additional transport equation related to the average enthalpy H is considered:

Wall (D)

Axis (E)

Outflow (F)

3.2. Turbulence modeling

~ @  @ lt @ H q
u~j H~ ¼ @xj @xj Prt @xj

!

þ SE

ð6Þ

Thermo-chemical scalars (species fractions, density and temperature) are calculated using the following expression:

f £i ¼

Z

1

 e df pðf Þ£i ~f ; H

ð7Þ

0

 e is the instantaneous value of the thermo-chemical scalars £i ~f ; H (B) CH4 (Dg/2)

(C) Microjet air (Dj)

(A) Air (Da/2)

X Y (C) (B) (A) « Mass flow inlet »

Fig. 1. Grid mesh distribution and boundary conditions of the Brookes and Moss configuration [18].

and p(f) represents the probability density function denoting turbulence and chemistry interaction. The probability density function p(f) is calculated using the mathematical beta-PDF function given in Ref. [30]:

f pðf Þ ¼ R 1 R 0

a1

f

ð1  f Þ

a1

b1 b1

ð1  f Þ

df

" # " # ~f ð1~f Þ ~f ð1~f Þ ~ ~ With: a ¼ f  1 and b ¼ ð1  f Þ 1 . e e f 02 f 02

ð8Þ

124

S. Chouaieb et al. / Energy Conversion and Management 140 (2017) 121–132

The combustion equation of methane (CH4) in air is written in the following form:

CH4 þ 2O2 ! CO2 þ 2H2 O

ð9Þ

3.5. Soot modeling

Table 2 Details of the tested grid densities. Test case No.

Grid density

Computational domain

1 2 3 4

10,500 30,400 60,800 90,620

Half-plane (x, y) avec 0 6 x 6 L

The Tesner model [32] was used to quantify soot formation in the flow field by the intermediate of radical nuclei particles. More details about soot particles calculation can be found in Ref. [30] and in our previous work [11].

2000

1600

The injection flux distribution and boundary conditions illustrated in Fig. 1 are summarized in Table 1. 4. Numerical method The grid mesh used in this work is composed of 60,680 cells (Fig. 1). The simulated domain was meshed using an unstructured triangular grid. The considered mesh was sufficiently dense in the nozzles vicinity and the reaction zone (Dy = 0.25 ⁄ 103 m and Dx = 104 m). Far from the fuel nozzle, the mesh was less tight (Dy  Dx  103 m). It should be noted that several meshes have been tested where grid densities were varied from 10,400 to 90,620 cells. Details of the tested mesh densities are summarized in Table 2. In order to ensure numerical results independency from grid density, temperature profiles are illustrated in Fig. 2 for the different tested meshes. In fact, mesh quality affects directly the accuracy of results and computational time. From Fig. 2, it is seen that the mesh No. 1 (10,500 cells) provides a temperature profile far from the rest of the three meshes. As well as the grid becomes finer, numerical results become closer. It is shown that numerical temperature profiles are found to be less dependent for a grid density counting more than 60,800 (mesh No. 3). Indeed, the temperature difference does not exceed 2% when the grid density is increased by 49% (mesh No. 4). As a consequence, the grid No. 3 is chosen. A second order UPWIND scheme was used for the discretization of the conservative governing equations. The algorithm SIMPLE

Table 1 Boundary conditions and injection flux distribution. Domain

Boundary conditions type

Boundary conditions

(A) Air nozzle Da = 155 mm

Mass flow inlet

(B)

Mass flow inlet

Wall

m_ a = 118 * 104 kg
s1 f=0 Dh = 0. 151 m It = 5.5% m_ g = 1.72 * 104 kg
s1 f=1 Dh = 0.003 m It = 5.5% m_ j varying from 9.42 * 108 kg
s1 to 2.355 * 105 kg
s1 f=0 Dh varying from 0.001 m to 0.003 m It = 6.3% –

Axis

–

Outflow

–

(C)

Mass flow inlet Microjet nozzle Dj = 1 mm to Dj = 3 mm

(D) Confinement wall (E) Symmetry axis (F) Outlet

Temperature (K)

3.6. Boundary conditions and flux distribution

Fuel nozzle Dg = 4 mm

et 0 6 y 6 D2a

1200 Mesh density N°1: 10500 cells Mesh density N°2: 30400 cells Mesh density N°3: 60800 cells Mesh density N°4: 90620 cells

800

400

0 0

0.2

0.4

0.6

0.8

1

Axial position (m) Fig. 2. Numerical temperature profiles for different tested grid meshes.

was adopted for the coupling between velocity and pressure. The scaled residual was used as a convergence indicator. Convergence criterion of residuals was taken equal to 106 for energy and soot particles calculation and 103 for all other equations. 5. Results and discussion In this section, the effect of the microjet velocity and diameter on methane/air flames is investigated. Before developing the parametric study, numerical results should be validated. The Brookes and Moss configuration is then used as a reference case [18]. Indeed, this configuration consists on a prototype of a coaxial burner. This apparatus was chosen since coaxial burners are widely encountered in several engineering applications including ceramic and chemical industries. In addition, the choice of this apparatus is promoted by the availability of several experimental measurements (temperature, mixture fraction and soot emission) for different positions in the confinement. In the following section, numerical results are validated in terms of temperature, mixture fraction and soot profiles. 5.1. Numerical validation In the present work, previous validation [11] is improved. Fig. 3 illustrates axial numerical mixture fraction and temperature profiles in comparison to Brookes and Moss measurements [18] and previous numerical results [11]. Compared to our previous work [11], present results show better agreement with experimental mixture fraction and temperature evolutions [18]. Actually, the difference can be attributed to turbulence modeling. Indeed, the k-epsilon Standard model with Pope Correction [27] presently used seems to be more adequate in capturing thermal and species fields. Indeed, the use of this model in several works dealing with the same configuration was appropriate to take into account coaxial jets spreading [27].

125

S. Chouaieb et al. / Energy Conversion and Management 140 (2017) 121–132

1

Experimental data [18] Numerical data [11]

Experimental data [18] Numerical data [11] Numerical data: Present study

2000

Numerical data: Present study 0.8

Temperature (K)

Mixture fraction

1600 0.6

0.4

0.2

1200

800

400

0

0 0

0.2

0.4

0.6

0.8

0

1

0.2

Axial position (m)

0.4

0.6

0.8

1

Axial position (m)

Fig. 3. Axial profiles of mixture fraction and temperature: comparison with literature [11,18].

Mixture fraction and temperature radial profiles at x = 0.2 m are presented in Fig. 4. It is noticed that the used models reproduce properly the experimental evolutions. Indeed, the peak temperature values and positions are well predicted. Besides, it is found that the mixture fraction peak value was presently improved by 15% compared to previous results [11]. The enhancement in calculating the developed temperature and species fields could be also observed in soot particles emissions since their quantification depends strongly on the flame characteristics. Indeed, species and temperature fields strongly affect soot particles production. In fact, soot emission is directly linked to the unburned carbon atoms which characterize the combustion process. Axial evolutions of soot volume fraction are depicted in Fig. 5. Previously [11], it was demonstrated that the Tesner model succeeds in capturing soot emissions for the flame configuration. Fig. 5 elucidates the k-epsilon Standard model with Pope Correction capacity [27] in giving closer results to the experimental measurements [18] than those using the k-epsilon Realizable [33]. As expected, soot calculation enhancement is accorded to the

5.2. Effect of Vj on the microjet assisted methane/air flames characteristics In this section, the effect of Vj on the microjet assisted methane/ air flames characteristics is underlined in terms of dynamic, thermal and species fields. A special emphasis on mixing process and soot production is presented. To the authors’ knowledge, these aspects have not been discussed for methane/air microjet assisted flames. Nevertheless, it should be noted that the effect of Vj was previously investigated, experimentally, for high momentum microjets to control methane/air flames’ shape, luminosity and pollutants emissions [14]. In this work, several simulations were carried out in which the microjet velocities were varied from low (0.1 m
s1) to high values (25 m
s1). The different test cases are detailed in Table 3 (Test cases No. 1 to 7).

Experimental data [18] Numerical data [11] Numerical data: Present study

2000

0.25

Experimental data [18] Numerical data [11] Numerical data: Present study

1600

Temperature (K)

0.2

Mixture fraction

improvement in predicting thermal and species fields of the considered flame (Figs. 3 and 4).

x=0.2m

0.15

0.1

800

0.05

400

0

0 0

0.02

0.04

Radial position (m)

0.06

0.08

x=0.2m

1200

0

0.02

0.04

Radial position (m)

Fig. 4. Radial profiles of mixture fraction and temperature at x = 0.2 m.

0.06

126

S. Chouaieb et al. / Energy Conversion and Management 140 (2017) 121–132

Experimental data [18] Numerical data [11] Numerical data: Present study

2.5E-007

Soot volume fraction

2E-007

1.5E-007

1E-007

5E-008

0 0

0.2

0.4

0.6

0.8

1

Axial position (m) Fig. 5. Numerical soot profile: comparison with experimental data [14] and previous numerical results [11].

The axial velocity contours and the streamlines of the microjet assisted methane/air flames for different microjet velocities (Vj = 0.1 m
s1, Vj = 0.5 m
s1, Vj = 0.8 m
s1 et Vj = 1 m
s1) are illustrated in Fig. 6. A global similarity is generally shown for the velocity contours in the nozzles exit near field. However, streamlines mark the presence of recirculation region in the central microjet nozzle vicinity for low microjet velocities (Vj < 1 m
s1). Indeed, as indicated by Mahmoud et al. [34], the development of recirculation bubbles is induced by air training [34]. For the low microjet velocities (Vj < 1 m
s1), the central air microjet is driven by the secondary flow of methane. By increasing the microjet velocity (Vj < 1 m
s1) the recirculation bubble moves downstream the central nozzle exit until disappearing for a microjet velocity equal to 1 m
s1. Temperature contours of methane/air flames for different microjet velocities (Vj = 0 m
s1, Vj = 1 m
s1, Vj = 5 m
s1, Vj = 10 m
s1, Vj = 15 m
s1 and Vj = 25 m
s1) are illustrated in Fig. 7. As reported in Ref. [11], the same contour is globally observed for the different flames with and without microjet (Vj = 0 m
s1). Nevertheless, a zoom view in the central nozzle vicinity shows the presence of an inner flame for the different microjet velocities. This inner flame was previously identified experimentally by Ganguly et al. [13] and Sinha et al. [14] and numerically for methane/air flames [35] and methane-hydrogen/air microjet assisted flames [11]. Moreover, it could be noted that the increase in the microjet mass flow rate elongates the inner flames. This can be due to the microjet mass flow rate augmentation which promotes combustion reaction between the microjet air and the fuel (methane). Axial temperature evolutions of the microjet assisted flames for different microjet velocities are depicted in Fig. 8. Compared to the reference case (without microjet), the addition of the microjet does

not alter the maximum temperature value. Indeed, this result is in conformity with the experimental observation of Sinha et al. [14]. Actually, it has been previously observed that microjet addition has no determinant cooling effect to modify the overall heat transfer [14]. In our case, this can be interpreted by the low value of the air mass flow ejected from the microjet which does not exceed 0.16% of the overall air mass flow for m_ j equal to 2.355 ⁄ 105 kg
s1 (test case No. 7). Moreover, Fig. 8 shows that axial temperature profiles of the microjet assisted flames are characterized with the same trends with two peaks: a primary peak and a secondary peak located in the central nozzle vicinity (x < 0.015 m). As explained for methane-hydrogen/air flames [11], it can be clearly seen that the microjet is responsible in creating the secondary peak. Indeed, as shown in Fig. 8, contrary to the microjet assisted flames, the conventional flame (without microjet) centerline temperature profile is marked uniquely by the presence of the primary peak. Indeed, this secondary peak can be considered as an indicator of faster combustion process. It should be noted also that the secondary peak position and value depend strongly on the microjet mass flow rate: When the air mass flow rate ejected from the microjet is less than 0.094 ⁄ 105 kg
s1 (Vj = 1 m
s1), the secondary peak is attached to the central axis. As depicetd in Fig. 9, it can be noted that the secondary peak position moves away from the central microjet nozzle and its value becomes higher by increasing the microjet velocity. This can be explained by the dynamic field shown in Fig. 6, particulary by the recirculation zones developed in the central nozzle vicinity. In fact, several researchers have shown the importance of recirculation regions in enhancing mixing process [36]. As observed in Fig. 6, for the lowest microjet velocities (Vj < 1 m
s1), the central recirculation region developed in the microjet nozzle vicinity promotes mixing process between the carburant (methane) and the comburant (microjet air). This means a faster combustion reaction and causes consequently the high temperature value (870 K) reached immediately at the microjet nozzle outlet (Fig. 8). In this case, mixing enhancement in the nozzle vicinity is due to the interaction between the two coaxial streams of the air microjet and the fuel jet [37]. According to Fig. 8, it is recommended to avoid low microjet velocities (Vj < 1 m
s1) since high temperatures reached in the nozzle exit near field can be responsible in destroying the burner construction material and present a risk of flame flashback. For these reasons, the microjet velocity Vj equal to 25 m
s1 can be judged as the best test case. The centerline mixture fraction profiles of the methane/air flames for different microjet velocities (Vj = 0 m
s1, Vj = 1 m
s1, Vj = 5 m
s1, and Vj = 25 m
s1) are depicted in Fig. 10. For the microjet assisted flames, it is clear that the mixture fraction evolutions decrease significantly with the microjet mass flow rate augmentation. Indeed, the reduction of the mixture fraction peak value is about 29% for the test case No. 7 (Vj = 25 m
s1) compared to the test case No. 4 (Vj = 1 m
s1). Nevertheless, the decrease of the mixture fraction value from unity to 0.71 is caused essentially

Table 3 Details of the microjet assisted methane/air flames test cases. Test case No. 1 2 3 4 5 6 7 8 9 10

Vj (m/s) 0.1 0.5 0.8 1 5 15 25 11.10 6.24 2.77

Dj (mm) 1 1 1 1 1 1 1 1.5 2 3

Rej 7 33 53 66 335 1006 1677 1116 837 558

_ j (kg
s1) m

_ g (kg
s1) m

8

4

9.42 * 10 47.1 * 108 75.36 * 108 0.094 * 105 0.471 * 105 1.413 * 105 2.355 * 105 2.355 * 105 2.355 * 105 2.355 * 105

1.72 * 10

_ a (kg
s1) m 118 * 104

127

S. Chouaieb et al. / Energy Conversion and Management 140 (2017) 121–132

Vj=0.1m/s

Vj=0.5m/s

0.005

0.005 X Velocity: -2 0 2 4 6 8 10 12 14 16 18 20 22

X Velocity: -2 0 2 4 6 8 10 12 14 16 18 20 22

0.004

0.004

0.003

X

X

0.003

0.002

0.002

0.001

0.001

0

0 0

0.001

0.002

0.003

0.004

0.005

0

0.001

0.002

Y

0.003

0.004

0.005

Y

Vj=0.8m/s

Vj=1m/s

0.005

0.005

X Velocity: -2 0 2 4 6 8 10 12 14 16 18 20 22

X Velocity: -2 0 2 4 6 8 10 12 14 16 18 20 22

0.004

0.003

0.003

X

X

0.004

0.002

0.002

0.001

0.001

0 0

0.001

0.002

0.003

0.004

0.005

Y

0 0

0.001

0.002

0.003

0.004

0.005

Y

Fig. 6. Effect of Vj on the axial velocity contours and the streamlines of the microjet assisted methane/air flames.

by the introduction of the microjet nozzle. On the one hand, the presence of the microjet makes the methane jet able to react not only with the surrounding coflow air jet but also with that issued from the central added microjet. Consequently, the destruction rate of methane molecules is found to be more important for the microjet assisted flames. On the other hand, the reduction from unity to 0.71 (test case No. 4) can be simply explained by the definition of the mixture fraction. Indeed, a fuel boundary (configuration without microjet) is characterized by a mixture fraction centerline value equal to unity. In the presence of the air microjet, the reduction in the mixture fraction value explains simply that the central boundary exit does not hold pure methane for the microjet assisted flames case. For these reasons, the mixture fraction could not be taken as a mixing quality indicator. At this stage, in order to evaluate the microjet velocity effect on the reactive mixture, the methane concentration core is rather considered. This method allows us to follow the methane distribution at both axial and

radial positions in the nozzle vicinity. It should be noted that a smallest methane core indicates that the fuel is the most consumed and reflects as a consequence better mixture between methane and the surrounding air. Methane concentration cores of the conventional (without microjet) and the microjet assisted flames for different microjet velocities (Vj = 0 m
s1, Vj = 1 m
s1, Vj = 5 m
s1, Vj = 10 m
s1, Vj = 15 m
s1 and Vj = 25 m
s1) are illustrated in Fig. 11. It should be noticed that the introduction of the microjet reduces significantly the length of the methane concentration core. However, the microjet flames are less sensitive than the conventional flame to the microjet velocity variation. Indeed, the addition of the central air microjet with a mass flow not exceeding 0.094 ⁄ 105 kg/s causes a reduction of 62% in the methane core peak value. However, by multiplying the microjet velocity by a factor of five (Vj passes from 1 m
s1 to 5 m
s1), the reduction of the methane core length does not exceed 17%. These results emphasizes the microjet

S. Chouaieb et al. / Energy Conversion and Management 140 (2017) 121–132

X

128

T(K) Fig. 7. Effect of Vj on methane/air flames temperature contours.

efficiency in promoting mixture and proves that higher microjet velocities provides better mixing. Indeed, a higher methane destruction rate corresponds to better mixture between microjet air and fuel jet. As a conclusion, the microjet velocity equal to 25 m
s1 (test case No. 7) can be considered as the best test case. Soot volume fraction at the confinment outlet of the microjet assisted methane/air flames for different microjet velocities are depicted in Fig. 12. The soot values at the confinement outlet depend linearly on the microjet velocities as described by Eq. (10):

V f ;soot;exit ¼ 103  V j þ 0:043

ð10Þ

It should be noted that the introduction of the microjet, even for a low velocity (Vj = 1 m
s1), implies a decrease in soot particles emission by 94% compared to the initial case (without microjet). For the microjet assisted flames, reduction occurs more intensively with the rise of Vj. This is due to mixing enhancement [37]. In this case, the added microjet reduces significantly soot emissions by enhancing mixing in the near field central nozzle region (Fig. 11). Indeed, the presence of soot particles is essentially caused by unburned carbon atoms accompanying incomplete combustion process. In fact, the reduction of soot particles can be attributed to the decrease of inflow carbon atoms [9].

129

S. Chouaieb et al. / Energy Conversion and Management 140 (2017) 121–132

2000

0.025

without microjet V j=0.1m/s

without microjet Without microjet V j=1m/s Vj=1m/s V j=5m/s Vj=5m/s V j=10m/s Vj=10m/s V j=15m/s Vj=15m/s V j=25m/s Vj=25m/s

Primary peak

V j=1m/s V j=5m/s

0.02

V j=10m/s

1600

V j=15m/s

Axial position (m)

Temperature (K)

V j=25m/s

Secondary peak

1200

800

0.015

0.01

0.005 400 0 0

0 0.0001

0.001

0.01

0.1

0.0005

0.0015

0.002

Radial position (m)

1

Axial position (m)

0.001

Fig. 11. Effect of Vj on methane/air flames methane concentration core.

Secondary peak temperature position downstream the central nozzle (mm)

16

12

1398K 1379K

8 1361K

1298K

4

1217K

0 0

5

10

15

20

25

Soot volume fraction at the confinement outlet (*10-10)

Fig. 8. Effect of Vj on methane/air flames centerline temperature profiles.

0.05

0.04

0.03

0.02

0.01 0

V j (m/s)

5

10

15

20

25

Vj (m/s)

Fig. 9. Effect of Vj on the secondary peak positions and values. Fig. 12. Effect of Vj on soot volume fraction at the confinement outlet.

1

numerical foundings show that mixing process and soot emission can be controlled for higher microjet velocities (Vj) corresponding to higher Rej.

Mixture fraction

0.8 V j=0m/s V j=1m/s V j=5m/s V j=25m/s

0.6

0.4

5.3. Effect of Dj on the microjet assisted methane/air flames characteristics

0.2

0 0.001

0.01

0.1

1

Axial position (m) Fig. 10. Effect of Vj on methane/air flames mixture fraction.

According to Figs. 11 and 12, the microjet technique can be judged as an efficient tool controlling methane/air flames. In fact,

In this section, the microjet diameter effect on the microjet assisted flames characteristics is numerically investigated in terms of thermal field, mixing and soot emission for different microjet diameters varying from 1 mm to 3 mm. In fact, the main goal of this section consists on optimizing the microjet diameter for the same mass flow rate corresponding to the test case No. 7 (Table 3). Indeed, it is important to discuss, for the best test case (2.355 ⁄ 105 kg
s1), whether the microjet is more efficient for lower or higher diameter. The main purpose of this section consists on evaluating the microjet diameter effect for the same burner injection conditions. As a result, the air mirojet mass flow is fixed _ j = 2.355 ⁄ 105 kg
s1 and the same burner power (8.6 kW) is at m used for the different test cases. Details of the test cases are summarized in Table 3.

130

S. Chouaieb et al. / Energy Conversion and Management 140 (2017) 121–132

Dj =1mm

without microjet

Dj =2mm

Dj =3mm

T(K) Fig. 13. Effect of Dj on methane/air flames temperature contours.

Fig. 13 presents temperature contours of the methane/air flames with and without central air microjet for different diameters. Similarly to Fig. 7 results, it is clear that the presence of the air microjet does not modify the general shape of the methane/air flames. Nevertheless, the microjet creates an inner flame in the central nozzle vicinity for the different tested diameters. The centerline temperature evolutions of the microjet assisted flames for Dj=1 mm, Dj=2 mm and Dj=3 mm are presented in Fig. 14. For the different test cases, the effect of the microjet diameter on the flame length is negligeable. Indeed, the primary peak position is quite similar for the different microjet assisted flames equiped by various diameters values. However, the secondary peak position moves upstream the microjet exit as the diameter value is reduced. In fact, as shown in Table 3, for the same microjet mass 2000

flow rate, this is caused by higher inertia forces (higher Rej) characterizing smaller diameters. Moreover, the secondary peak value decreases slightly with reducing the microjet diameter. Nevertheless, this reduction does not exceed 5% for the microjet equiped by 3 mm nozzle diameter compared to that of 1 mm diameter. These two constatations may be explained by the fact that the microjet enhances mixing process between the fuel and the added central air in the near field region [37]. Axial mixture fraction trends for different microjet diameters (Dj=1 mm, Dj=2 mm and Dj=3 mm) are depicted in Fig. 15. As expected from Fig. 14, for smaller microjet diameters, the mixture fraction peak is lower and the reduction reaches 8% for the 1 mm microjet assisted flame compared to the 3 mm one. This indicates better mixture for the lowest microjet diameters.

D j =1mm

0.8

D j =2mm

D j =1mm D j =2mm D j =3mm

0.6 1200

Mixture fraction

Temperature (K)

1600

D j =3mm

800

400

0.4

0.2

0 0.0001

0.001

0.01

0.1

1

Axial position (m) Fig. 14. Effect of Dj on axial temperature profiles of the microjet assisted methane/ air flames.

0 0.0001

0.001

0.01

0.1

1

Axial position (m) Fig. 15. Effect of Dj on methane/air flames centerline mixture fraction profiles.

S. Chouaieb et al. / Energy Conversion and Management 140 (2017) 121–132

increasing the microjet velocity or decreasing the microjet diameter for the same microjet mass flow rate.

10

Soot volume fraction at the confinement outlet (*1010)

131

References 1

0.1

0.01 0.5

1

1.5

2

2.5

3

Dj (mm) Fig. 16. Effect of Dj on soot volume fraction at the methane/air flames confinement outlet.

Soot volume fraction of the microjet assisted methane/air flames at the confinment outlet for microjet diameters varying from Dj=1 mm to Dj=3 mm are illustrated in Fig. 16. It is clearly seen that soot production is lower for the lowest diameter (test case No. 7). The difference is directly linked to the mixture enhancement for the microjet nozzle characterized by higher microjet inertia forces. Based on Figs. 15 and 16, the configuration equiped with a 1 mm microjet diameter can be judged as the best test case providing better mixing process and lower soot production. In fact, it is found that the microjet is usefull in controlling methane/air flames for higher Rej. Indeed, mixing process enhancement and soot production reduction can be ensured either by increasing the microjet velocity or reducing the microjet diameter (Dj) for a constant microjet mass flow rate.

6. Conclusion A parametric study was carried out to investigate numerically methane/air microjet assisted flames characteristics. The Presumed PDF model, the DO model and the k-epsilon Standard model with Pope Correction were adopted respectively for combustion, radiation and turbulence modeling. In the first part, our previous predictions were enhanced. Indeed, it has been shown that the use of the k-epsilon model with Pope Correction instead of the kepsilon Realizable model is more suitable in describing the reactive coaxial jet behavior. Simulations were conducted to evaluate the effect of the microjet on methane/air flames characteristics. Two parameters were discussed: the microjet velocity and diameter. Numerical simulations have shown that the microjet technique can be used to promote the reactive mixture and reduce soot emission. On the one hand, the parametric study showed that the mixture is improved for high microjet velocities (Vj = 25 m
s1). It was demonstrated that even for lower microjet velocity (Vj = 1 m
s1), mixing process was increased by 62% and soot production was reduced by 94%. Nevertheless, it was found that higher microjet velocities are rather recommended to avoid high temperatures reached at the central nozzle outlet. On the other hand, by varying the diameter of the microjet for a constant mass flow rate, it was found that the best test case corresponds to the 1 mm diameter nozzle. As a conclusion, microjet assisted methane/air flames mixing process and soot emission can be controlled for higher Rej either by

[1] Kang Y, Lu T, Lu Xi, Wang Qu, Huang Xi, Peng Sh, et al. Study on combustion characteristics of dimethyl ether under the moderate or intense low-oxygen dilution condition. Energy Convers Manage 2016;108:549–65. [2] Meng Xi, de Jong W, Kudra T. A state-of-the-art review of pulse combustion: principles, modeling, applications and R&D issues. Renew Sustain Energy Rev 2016;55:73–114. [3] Zhuang H, Hung DLS. Characterization of the effect of intake air swirl motion on time-resolved in-cylinder flow field using quadruple proper orthogonal decomposition. Energy Convers Manage 2016;10:366–76. [4] Yu Y, Li X, An X, Yu X, Fan R, Chen D, et al. Stabilization of a premixed methaneair flame with a high repetition nanosecond laser-induced plasma. Opt Laser Technol 2017;92:24–31. [5] Jeon J, Tae Lee J, Kwon S, Park S. Combustion performance, flame, and soot characteristics of gasoline–diesel pre-blended fuel in an optical compressionignition engine. Energy Convers Manage 2016;116:174–83. [6] Nicoletti G, Arcuri N, Nicoletti G, Bruno R. A technical and environmental comparison between hydrogen and some fossil fuels. Energy Convers Manage 2015;89:205–13. [7] Zhang B, Ji Ch, Wang Sh. Investigation on the lean combustion performance of a hydrogenenriched n-butanol engine. Energy Convers Manage 2017;136:36–43. [8] Amrouche F, Erickson PA, Varnhagen S, Park JW. An experimental study of a hydrogen-enriched ethanol fueled Wankel rotary engine at ultra lean and full load conditions. Energy Convers Manage 2016;123:174–84. [9] Kashir B, Tabejamaat S, Jalalatian N. The impact of hydrogen enrichment and bluff-body lip thickness on characteristics of blended propane/hydrogen bluffbody stabilized turbulent diffusion flames. Energy Convers Manage 2015;103:1–13. [10] Ilbasß M, Karyeyen S, Yilmaz I. Effect of swirl number on combustion characteristics of hydrogen-containing fuels in a combustor. Int J Hydrogen Energy 2016;41:7185–91. [11] Chouaieb S, Kriaa W, Mhiri H, Bournot Ph. Presumed PDF modeling of microjet assisted CH4–H2/air turbulent flames. Energy Convers Manage 2016;120:412–21. [12] Tabet F, Sarh B, Gokalp I. Hydrogen–hydrocarbon turbulent non-premixed flame structure. Int J Hydrogen Energy 2009;34:5040–7. [13] Ganguly R, Puri IK. Nonpremixed flame control with microjets. Exp Fluids 2004;36:635–41. http://dx.doi.org/10.1007/s00348-003-0749-z. [14] Sinha A, Ganguly R, Puri IK. Control of confined nonpremixed flames using a microjet. Int J Heat Fluid Flow 2005;26:431–9. [15] Yuchun C, Jinxing W, Jianchun M, Yu Zh. Flame structure of a jet flame with penetration of side micro-jets. Chin J Chem Eng 2008;16(6):861–6. [16] Kanchi H, Russell K, Anderson MJ, Beard SP, Strykowski PJ, Mashayek F. Fluidic control with microjets in dump combustors. Int J Heat Mass Transf 2011;54:5395–405. [17] Kumar Maurya R, Akhil N. Development of a new reduced hydrogen combustion mechanism with NOx and parametric study of hydrogen HCCI combustion using stochastic reactor model. Energy Convers Manage 2017;132:65–81. [18] Brookes SJ, Moss JB. Measurements of soot production and thermal radiation from confined turbulent jet diffusion flames of methane. Combust Flame 1999;116:49–61. [19] Brookes SJ, Moss JB. Predictions of soot and thermal radiation properties in confined turbulent jet diffusion flames. Combust Flame 1999;116:486–503. [20] Kronenburg A, Bilger RW, Kent JH. Modeling soot formation in turbulent methane–air jet Diffusion Flames. Combust Flame 2000;121:24–40. [21] Navarro-Martinez S, Rigopoulos S. Differential diffusion modelling in LES with RCCE-reduced chemistry. Flow Turbul Combust 2012;89:311–28. [22] Woolley RM, Fairweather M. Conditional moment closure modelling of soot formation in turbulent, non-premixed methane and propane flames. Fuel 2009;88:393–407. [23] Saqr KM, Aly HS, Sies MM, Wahid MA. Effect of free stream turbulence on NOx and soot formation in turbulent diffusion CH4-air flames. Int Commun Heat Mass Transfer 2010;37:611–7. [24] Saqr KM, Wahid MA. Comparison of four eddy-viscosity turbulence models in the eddy dissipation modeling of turbulent diffusion flames. Int J Appl Math Mech 2011;7(19):1–18. [25] Kassem HI, Saqr KM, Aly HS, Sies MM, Abdul Wahid M. Integrating a simplified P-N radiation model with EdmFoam1.5: model assessment and validation. Int Commun Heat Mass Transf 2012;39:697–704. [26] Kassem HI, Saqr KM, Aly HS, Sies MM, Abdul M. Wahid. Implementation of the eddy dissipation model of turbulent non-premixed combustion in OpenFOAM. Int Commun Heat Mass Transf 2011;38:363–7. [27] Pope SB. An explanation of the round jet/plane jet anomaly. AIAA J 1978;16 (3):279–81. [28] Chui EH, Raithby GD. Computation of radiant heat transfer on a non orthogonal mesh using the finite-volume method. Numer Heat Transfer, Part B 1993;23:269–88. [29] C Hottel H, Sarofim AF. Radiative transfer. New-York: McGraw-Hill; 1967. [30] FLUENT 6.3, Tutorial Guide Fluent 6.3. Inc; 2006.

132

S. Chouaieb et al. / Energy Conversion and Management 140 (2017) 121–132

[31] Sivathanu YR, Faeth GM. Generalized state relationships for scalar properties in non-premixed hydrocarbon/air flames. Combust Flame 1990;82:211–30. [32] Tesner PA, Snegiriova TD, Knorre VG. Kinetics of dispersed carbon formation. Combust Flame 1971;17:253–60. [33] Shih TH, Liou WW, Shabbir A, Yang Z, Zhu J. A new k- eddyviscosity model for high reynolds number turbulent flows – model development and validation. Comput Fluids 1995;24(3):227–38. [34] Mahmoud H, Kriaa W, Mhiri H, Le Palec G, Bournot Ph. A numerical study of a turbulent axisymmetric jet emerging in a co-flowing stream. Energy Convers Manage 2010;51:2117–26.

[35] Chouaieb S, Kriaa W, Mhiri H, Bournot Ph.. Étude numérique des champs thermique et dynamique d’un brûleur coaxial avec un microjet d’air central. Congrès Français de Thermique; Gerardmer du 28 au 31 Mai 2013. [36] Li S, Zheng Y, Zhu M, Martinez DM, Jiang X. Large-eddy simulation of flow and combustion dynamics in a lean partially-premixed swirling combustor. Energy Procedia 2015;66:333–6. [37] Ditaranto M, Hals J. Flame length above a coaxial burner with oxygen enrichment. In: Joint meeting of the Scandinavian-Nordic and Italian Sections of the Combustion Institute; 2003.