Analysis of Turbulent Lean Premixed Methane–Air Flame Statistics at

Oct 20, 2017 - The computations are performed initially for the cold flow field and then for the reactive flow statistics using the Fluent software. F...
0 downloads 6 Views 5MB Size
Subscriber access provided by AUSTRALIAN NATIONAL UNIV

Article

Analysis of Turbulent Lean Premixed MethaneAir Flame Statistics at Elevated Pressures Baris Yilmaz, and Iskender Gökalp Energy Fuels, Just Accepted Manuscript • DOI: 10.1021/acs.energyfuels.7b02097 • Publication Date (Web): 20 Oct 2017 Downloaded from http://pubs.acs.org on October 23, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Energy & Fuels is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Analysis of Turbulent Lean Premixed Methane-Air Flame Statistics at Elevated Pressures Baris Yilmaz*1, Iskender Gökalp2 1

Marmara University, Faculty of Engineering, Mechanical Eng. Dept, 34722 Istanbul, Turkey

2

ICARE – CNRS, 1c, Av. de la Recherche Scientifique, 45071, Orleans Cedex 2, France.

KEYWORDS: Turbulent Premixed Flames, High Pressures effect, Methane, Modeling

ABSTRACT

Combustion takes place within turbulent environments and under pressurized conditions in most of industrial applications, for instance; gas turbines and internal combustion engines.

The

mixing of fuel and oxidizer enhances by turbulence so that combustion will be more efficient. In addition, during combustion heat is released so it causes flow instability due to gas expansion and high gradients of temperature. Interaction of turbulence and combustion under pressurized conditions may lead to modifications, even disruption, of the flame. There are still ongoing researches on the prediction of flame and flow statistics which is crucial for the stable operation of high pressure combustion systems. In the present study, the aim is to investigate the influence of elevated pressures on the structure of turbulent premixed methane-air flames, within the range

ACS Paragon Plus Environment

1

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 33

of atmospheric to 0.9 MPa, by numerical modeling and experimental validation. The computations are performed initially for the cold flow field and then for the reactive flow statistics using the Fluent software. For the cold flow case results, it is found that the velocity potential core region extends until 3 diameters downstream from the burner exit and the decay rate decreases with increasing pressure. However, in the reactive case simulations, the decay in the velocity profiles is not observed due to gas expansion through the flame front, similar to the experiments. The turbulent flame front statistics in terms of the flame front location, the flame brush thickness and the flame surface density (FSD) have been computed. The flame tip location and the flame brush thickness are found more downstream and thicker at elevated pressures, respectively. Moreover, both the profiles and the peak values of FSD are well captured for various pressure conditions in computations as compared to the measurements.

1.

Introduction

Turbulent premixed combustion systems are widely operated under high-pressure conditions in many practical applications such as automotive engines (SI engines) and gas turbines in order to attain high power output. In addition, the laminar premixed flame structure varies significantly at elevated pressures compared to atmospheric flames. Finally, at high pressure conditions small scale turbulent structures dominate the turbulent wrinkling of the flame surface so that the flame surface density and the turbulent burning rate are modified. All these features justify the recent studies on the pressure effects on turbulent premixed flames. Turbulent combustion involves a wide range of turbulence and time scales. For turbulent premixed type flames, the flame sheet is both wrinkled and stretched. Therefore, the flame surface area increases. It also enhances the burning rate or in other words the overall flame

ACS Paragon Plus Environment

2

Page 3 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

speed. Large eddies cannot penetrate in but wrinkle and corrugate the flame sheet. On the other hand, small eddies, comparable to the flame thickness, may modify the laminar flame by penetrating into the flame. Furthermore, this process enhances the transport of heat and mass, mainly in the preheating zone of flame1-3. The interactions between the turbulent flow field and the flame front are described by means of characteristic dimensionless numbers. The most commonly used are the turbulent Reynolds (ReT), Damköhler (Da) and Karlovitz (Ka) numbers defined in equations (1-3). The relations have been derived based on the assumption that the thermal diffusivity is equal to the mass diffusivity, i.e., Schmidt number (Sc ≡1).

ReT =

u ′lT v

(1)

Da =

lT s L u ′δ L

(2)

l Ka =  T δL

  

−1 / 2

 u′     sL 

3/ 2

(3)

The Reynolds number given in equation (1) corresponds to turbulent Reynolds number obtained based on the integral length scale lT; u’ and v represent the characteristic turbulent fluctuation velocity and the kinematic viscosity, respectively. The Damköhler number (Da), Equation 2, relates the turbulent mixing time lT /u’ and the chemical reaction time δL/SL, where δL represents the laminar flame thickness and SL corresponds to the laminar flame speed. Large Damköhler numbers (Da >> 1) refer to a flame where the chemistry is fast and thus the combustion is controlled by mixing processes. Flames at small Damköhler numbers (Da < 1) are characterized by intense mixing. The combustion is then controlled by chemical kinetics and flames are located at the well-stirred reactor region on the flame regimes diagram. The Karlovitz number

ACS Paragon Plus Environment

3

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 33

represents the ratio of the reaction zone thickness, represented by δL, to the Kolmogorov length scale η, representing the smallest turbulence length scale. Therefore, since the Karlovitz number relates the flame sheet thickness and turbulence scales it may be used for flame sheet stretch. Using these three characteristic numbers, namely ReT, Da and Ka numbers, it is possible to categorize the turbulent premixed flames on a flame regime diagram, first proposed by Borghi4 and then extended by Peters5 and other authors1. Several experimental and theoretical studies dealing with high pressure turbulent lean premixed flames are published. Kobayashi et al.6 measured the burning velocities of turbulent premixed methane/air mixtures, stabilized on a Bunsen type burner, for pressures varying from atmospheric conditions to 1.0 MPa. Griebel et al.7 studied experimentally the flow field characterization of turbulent premixed flames under pressurized conditions. They examined the influence of turbulent Reynolds, Damkohler and Karlovitz numbers on the flame front location and its structure. They concluded that increasing both the pressure and the turbulence intensity affects the flame tip position slightly but increases the flame front corrugation level. In their work, Soika et al.

8

have conducted series of experiments for methane/air flames

stabilized on a bluff-body at lean and stoichiometric conditions. It is concluded that the flame front curvatures are strongly affected by changing the pressure. Fragner9 et al. have conducted experiments on turbulent premixed methane/air flames where the pressure ranges between 0.1 to 0.4 MPa. They reported that increasing the pressure while keeping the laminar flame thickness constant, there is a strong interaction between the small scale turbulent structures and the flame front. They also demonstrated that albeit the integral length scales are not much affected by the pressure increase, the small scales strongly decrease with pressure and also they become relatively more energetic.

ACS Paragon Plus Environment

4

Page 5 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Liu et al.10 performed measurements for the premixed methane/air flames interacting with intense isotropic turbulence. They concluded that when the turbulent Re number changes from 6700 to 14,200 both the laminar and turbulent flame speeds decrease with pressure. The precise calculation of the laminar flame speed is crucial for the turbulent flame combustion models since it will be a key input for such models. Several experimental studies11-13 have been devoted to the investigation of the laminar flame speed at elevated pressures. Vagelopoulos et al.11 studied experimentally the laminar flame speeds of ultra-lean hydrogen flames by controlling strain rates on a twin flame test setup. They extended the study for methane/air and propane/air mixtures for various equivalence ratio conditions. They found slightly lower flame speeds compared to the previously reported data. Ogami and Kobayashi12 measured flame speeds of premixed CH4/air mixtures at high temperature and pressure conditions. They compared measurements to the modeling results obtained using different chemical kinetics models. Hassan et al.13 conducted experiments in order to examine the influence of pressure on laminar flame speed of premixed methane/air flames. A freely propagating flame is studied at various equivalence ratios and pressures. They observed that, at elevated pressures, the flame becomes unstable due to preferential-diffusion effects. At ICARE, the influence of pressure on the premixed flame front statistics of various mixtures has been studied extensively in a high pressure combustion chamber facility. Different premixed mixtures such as methane-air, hydrogen-methane-air and CO2-methane-air, are considered for various operating conditions by Lachaux14, Halter15 and Cohé16, respectively. Several numerical studies have been realized to examine the influence of pressure on premixed methane flames17-21. Ratzke et al.17 examined the pressure effect for the internal combustion

ACS Paragon Plus Environment

5

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 33

engines both experimentally and numerically. They have tested several turbulent flame speed correlations at high pressures, up to 7 MPa. Goswami et al.18 studied numerically on a 2D axisymmetric flat burners at high pressure conditions. They examined the laminar premixed flames at elevated pressures of 1.5 MPa using one-step and detailed chemical reaction mechanisms. A nonlinear increase of the flame curvature at elevated pressure conditions was observed. Dinkelacker et al.19 modeled turbulent lean premixed flames and compared results with two different experimental results. They focused mainly on the effect of Lewis number and reported good agreement with experiments. Hu et al.20 have studied numerically the laminar methane-air premixed flames and compared the computations with experimental measurements. They reported that the flame structures were highly affected by varying the equivalence ratio and the pressure. Kumar et al.21 have also studied numerically the effect of the pressure and premixture temperature on the laminar flame speed for hydrogen enriched methane-air premixed combustion. They observed that the flame speed decreases with increasing pressure. The objective of this work is to contribute to the understanding of the premixed flame front statistics at elevated pressure conditions by numerical modeling. These flames are computed using sub-models for two complex processes, namely combustion and turbulence, and their interactions. The flame stability and efficiency depend mainly on the interaction of these processes at different scales. Therefore, the prediction of the flame statistics is crucial for stable and efficient operation of turbulent combustion systems.

ACS Paragon Plus Environment

6

Page 7 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

In the present study, the influence of pressure on the flame statistics the effect of pressure on the turbulent lean premixed methane-air flames has been investigated computationally and the results are compared to available measurements conducted in ICARE. The computational conditions are defined as in the experimental studies14-16. The burner exit velocity is kept constant and therefore mass flow rates are increased at elevated pressures up to 0.9 MPa. Furthermore, flame front statistics in terms of progress variable distributions and flame surface density (FSD) have been examined. 2.

The High Pressure Experimental Setup

Experiments are conducted on a two-block cylindrical combustion chamber with inner diameter of 300 mm. Pressure can be varied from atmospheric to 1.0 MPa conditions. Laser diagnostics are utilized for cold and reactive flow measurements, as shown in Figure 1. The pressure level is set using a pressure gauge located at the exit of the chamber. In addition, a thermocouple is used to measure exhaust gas temperature. The premixed Bunsen burner is located at the bottom of the combustion chamber. The inner diameter of the burner is 25 mm. Turbulence is generated using a perforated plate located at 50 mm upstream location, as seen in Figure 1. A stoichiometric CH4/air premixed pilot flame is used to stabilize the main flame. The velocity field statistics such as mean axial velocity and its fluctuation are obtained using LDA measurements. In addition, the premixed flame measurements are conducted using laser induced Mie scattering technique. In order to simulate an operation of a gas turbine at various operating conditions the burner exit velocity is kept constant for various pressure conditions. The details of the measurement techniques can be found in Refs.14-16

ACS Paragon Plus Environment

7

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

3.

Page 8 of 33

Numerical Studies

Two dimensional and axisymmetric model of the combustion chamber geometry is developed at first step. Then, the flow domain is meshed using structured meshes. The regions in the domain where both the flow/turbulence and temperature gradients are high meshed finely. After mesh generation, at the beginning the cold flow and then the reactive flow field have been computed using well known Fluent software22. The operating conditions are chosen as in the experiments in which the average burner exit velocity is assumed to be constant about 2 m/s. On the other hand, the radial mean velocity and turbulence profiles measured 5 mm downstream from the burner exit are used to specify initial conditions in modeling studies. The inlet mean flow conditions for different pressures are tabulated in Table 2. As seen in the table, increasing the pressure increases the density and thereby increases the turbulent Reynolds number via the reduction of the kinematic viscosity. At the beginning of the numerical work, the cold flow field of the chamber has been computed. The turbulent flow field is simulated using the k-ε turbulence model with Pope correction23. Several versions of this turbulence model have been evaluated for cold flow24. It was concluded that the turbulence model with Pope correction results in better agreement with experiments comparing in-chamber mean and turbulent flow field statistics. The reactive flow studies are performed in two steps. Firstly, the laminar flame properties such as the flame speed and the Zeldovich flame thickness are computed for different pressure conditions using the Chemkin software25. The well-known GRI Mech3.0 chemical kinetics model consisting of 325 reactions and 53 species has been utilized in these calculations. As mentioned before, the correct calculation of the laminar flame properties, specifically, the

ACS Paragon Plus Environment

8

Page 9 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

laminar flame speed, is crucial for the turbulent flame combustion models since the calculated value will be an input for implementing combustion models, especially for Turbulent Flame Closure (TFC) type models. Finally, based on the obtained data from Chemkin computations, the turbulent flame front statistics are computed using Fluent 6.2 software. The reactive turbulent flow field is obtained by solving the required transport equations of momentum and energy. The results are compared to the available measured data. Besides solving the coupled continuity, momentum and energy conservation equations, the combustion phenomena is modelled using different methods in Fluent.

In the present

computations, the Zimont turbulent premixed combustion model based on algebraic turbulent flame closure23,27 is utilized to investigate the turbulent premixed flame-front statistics. This model solves a scalar transport equation for the progress variable besides the flow field transport equations, given in equation (4). In addition, the model assumes infinitely fast chemistry. The progress variable is defined to be the ratio of the normalized summation of the mass fractions of the product species by the summation of the mass fractions at equilibrium conditions as given in equation (5). It equals to 0 in fresh gases whereas it is 1 the burned gases.

  ∂ρ c ∂ (ρ uc ) = ∂  µ t ∂c  + ρS c + ∂ t ∂x i ∂xi  Sct ∂xi  n

c = ∑ Yi i =1

(4)

n

∑Y

i ,eq

(5)

i =1

the source term in equation (4) is closed using an algebraic equation, deveoped by Zimont et al.26 as,

ρS c = ρ uU t ∇c

(6)

ACS Paragon Plus Environment

9

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 33

and Ut represents the turbulent flame speed, is given in Equation 7 using the following relation22:

U t= A (u ′)3 / 4 S L1/ 2α −1/ 4lt1/ 4

(7)

where A is the model constant proposed to be 0.52 for methane/air mixtures22. SL represents the laminar flame speed, α refers to the heat diffusivity of the unburnt mixture and lT is the turbulent integral length scale. The stretch effects on the flame is considered by a multiplication factor in the source term for the progress variable equation22,26. As it is seen in Equation 7, the laminar flame speed, SL, is the key parameter for the closure of the progress variable transport equation. 4. Results The computational studies are realized in two steps. Firstly, the cold flow field developed in the chamber is computed. The mean flow field parameters are examined at different pressure conditions in terms of normalized axial velocity, turbulent kinetic energy and integral length scale distributions along axial direction. In the second stage, the reactive flow is solved based on cold flow simulation results. Moreover, the flame parameters, needed for reactive flow simulations, are obtained using the Chemkin software. The flame front statistics, the progress variable and the flame surface density profiles are derived from reactive case studies. 4.1. Cold Flow Results. The simulations have been started from solving cold flow field inside the high pressure chamber setup. The influence of elevated pressures on both the characteristic scales of turbulent flow and the mean flow field has been examined. The modeling results are compared to experiments for pressures, starting from atmospheric conditions, 0.1 MPa, 0.3 MPa, 0.5 MPa, 0.7 MPa and 0.9 MPa. The radial velocity measurements at 5 mm

ACS Paragon Plus Environment

10

Page 11 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

downstream have been used as the initial velocity conditions in terms of inlet profiles for each case. The simulations have been performed with the same mesh configuration as in the atmospheric conditions. Moreover, the same numerical algorithms and second order differentiation schemes have been applied. The k- ε turbulence model modified with Pope correction, added to the Fluent using user defined functions (UDF), is utilized for the turbulence modeling. The operating conditions are set as given in Table 2. The average burner exit velocity is kept constant, at about 2 m/s, as in the experiments. Therefore, the Reynolds number increases by increasing the pressure. The influence of pressure on the cold flow statistics is investigated by comparing the mean flow field statistics to the experimental measurements. The comparison of normalized axial velocity profiles is shown in Figure 2. It is observed that increasing the pressure decreases the decay rate of the mean axial velocity profile in the axial direction. This is due to the fact that the jet penetrates in a medium at higher density and its mixing rate is accordingly decreased for higher pressures. At the highest pressure case the mean velocity profile is almost not affected by the pressure. The differences are highest between 0.1 MPa and 0.3 MPa; however, the velocity profiles are almost similar for 0.3, 0.5, 0.7 and 0.9 MPa cases. The simulation results are in good agreement with experiments. The normalized turbulence kinetic energy axial profiles are depicted in Figure 3. It is observed that the increase in the pressure results in the decrease in normalized turbulent kinetic energy profiles. This is related to the above mentioned result of decreased mixing due to high medium density at high pressures where the entrainment rate is reduced generating less turbulent shear stresses, namely less lateral velocity fluctuations. The computations follow relatively well the experimental results and capture the above phenomena. The computed kinetic energies are

ACS Paragon Plus Environment

11

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 33

divided by ko calculated from the velocity measurements at 5 mm downstream for the normalization. The influence of pressure on the turbulent integral length scale is plotted in Figure 4. Good agreement has been found between experimental data and simulations. As also observed in previous works6,14 , the pressure effect on this large scale turbulence characteristics is not visible. As known from the turbulence literature, the integral length scale increases with distance downstream from the turbulence grid within the potential core of the jet. As is evident from the above comparisons and discussions, the computations simulate well the behavior of the cold flow turbulence structure. The next paragraph will discuss the results of the reacting flow computations and their comparisons with the experimental data on flame statistics. 4.2 Reactive Flow Results. The reactive flow field characteristics are presented in this second part of the study. The influence of both pressure and the presence of the flame on normalized axial mean velocity profiles are shown in Figure 5. The computed results are in reasonable agreement with the experiments for the available data range as seen in Figure 5. Compared to Figure 2, the mean axial velocity decay rates are much lower for the reactive case computations due to the heat release effect and the axial velocity profiles are almost homogeneous along the axial distance. Therefore, the flow turbulence in the reactive case is well predicted by the computations. Concerning combustion modelling, first, the laminar flame speeds at different pressure conditions are computed as explained in the paragraphs above. Next, with this input, the turbulent flame statistics, such as the average progress variable and flame surface density distributions, are calculated. The results are compared to available experimental data.

ACS Paragon Plus Environment

12

Page 13 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Laminar Flame Computations. The methane-air laminar flame propagation speeds are computed at various pressure conditions using Chemkin with GRI Mech 3.0 chemical mechanism for the stoichiometric equivalence ratio. The results are compared to various experimental studies11-13, plotted in Figure 6. The laminar flame speeds decrease with increasing pressure. The results show good agreement with available experimental observations. As u’ is rather insensitive to the pressure, this result indicates that the ratio u’/SL increases with pressure when all other parameters are kept constant as in this study. The Zeldovich flame thicknesses, formulated as in Equation 8, are examined at elevated pressures as well in Figure 7.

δ=

λ (8)

ρC p S L

where λ, ρ, Cp and SL represent conductivity, density, specific heat and laminar flame speed, respectively. It is observed that the flame thickness decreases at elevated pressures since it is inversely proportional to the laminar flame speed. This indicates that small scale flame wrinkling should increase but that smallest scales will have more difficulty to penetrate in the flame thickness. On the other hand, as shown in9, with increasing pressure smaller scales become smaller and more energetic, possible counteracting the previous effect. It is therefore clear that capturing the pressure effects on turbulent premixed flames is not an easy task. Turbulent Flame Computations. The turbulent flame computations are performed using Zimont combustion model for the methane-air premixed mixture under lean equivalence ratio (equal to 0.6) conditions.

ACS Paragon Plus Environment

13

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 33

The average progress variable axial distributions are depicted in Figure 8. The computed profiles are found strongly downstream, about 1.5 diameters, than the measurements. This shift is observed for all pressures. On the other hand, the computed profiles and thereby the flame brush thicknesses are observed similar to the experimental measurements.

It is also found that

increasing the pressure leads to a downstream shift in the flame front locations both experimentally and computationally. In the simulations, the influence of the pressure on the progress variable distribution lessens at higher pressures, i.e., the progress variable distributions are getting closer at higher pressures as seen in Figure 8. This discrepancy between the experimental and computational average flame front locations is puzzling and seems to indicate that the turbulent flame propagation velocity is under-estimated by the Zimont model (Equation 7 above). There are several issues that should be discussed with these results. Equation 7 can be rearranged in the following form:  



= ( ) /  /

(9)



where ReT is the turbulent Reynolds number based on lT and where the unity Schmidt number hypothesis is made. As shown on Table 4, both u’/SL and ReT strongly increases with pressure, therefore Ut/SL should increase with pressure and logically higher pressure flames should stabilize closer to the burner exit and not further downstream compared to the atmospheric pressure case. However, the experimental results also show the same trend as the computations. Therefore there is a subtle interaction between the cold flow turbulence structures at high pressures (see Fig. 3) where the

ACS Paragon Plus Environment

14

Page 15 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

turbulence kinetic energy on the centerline is reduced at high pressures and the turbulent flame speed or the response of the turbulent flame to the cold flow turbulence. The second issue is the downstream positioning of the computed flames indicating lower turbulent propagation velocities compared to the experimental data. This seems indeed to indicate that Equation 7 under-estimates the turbulent propagation velocity for all pressures. The influence of the pressure on the flame surface density (FSD) distributions has been examined as well. In order to obtain FSD profiles, the relation proposed by the Bray-Moss-Libby (BML), formulated in Equation 9, has been applied using progress variable results from Zimont model solutions. The orientation factor y, and the Ly/g ratio have been experimentally reported by Halter15 as in Table 3. Based on these data the flame surface density profiles versus progress variable have been obtained and plotted in Figure 9.

Σ=

g c (1 − c ) σ y Ly

(9)

where y is the flamelet orientation factor, Ly refers to the wrinkling integral length scale of flame front and g is a model constant of order of unity1,15. In Figure 9, it has been observed that increasing the pressure results in higher flame surface densities. This behavior is expected since the FSD and the turbulent length scale are inversely proportional. On the other hand, increase in the pressure results in decrease in the turbulent length scales. Thus, the flame surface density increases with increasing the pressure as expected. The maximum flame surface densities are computed to be 280, 326 and 407 m-1 for 0.1, 0.5 and

ACS Paragon Plus Environment

15

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 33

0.9 MPa cases, respectively. The predictions are compared to measurements of Cohé16 where the maximum flame surface densities are reported to be 270, 317 and 365 m-1 for studied 0.1, 0.5 and 0.9 MPa cases, respectively. The influence of the pressure is observed more effective on the flame surface density at higher pressure conditions. The characteristic numbers, namely ReT, Da and Ka have been also computed based on the flow and flame results. In derivations, the molecular viscosity assumed to be constant. The estimated turbulent flow and flame characteristics at different pressure conditions are summarized in Table 4. It is observed that increasing the pressure increases the level of u’/SL since SL decreases with increasing the pressure, however, u’ is computed almost same at higher pressures. Furthermore, the Kolmogorov scale decreases significantly, that is expected since higher Re numbers lead to decrease in small scale turbulence characteristics. Therefore, the ratio of η/δL decreases at elevated pressures since the Kolmogorov length scale at 0.9 MPa is almost one-fifth of that of 0.1 MPa case (Table 4). Turbulent Reynolds number (ReT) increases due to the increase in density as tabulated in Table 4. Damköhler number decreases at higher pressures while Karlovitz number increases. In their experimental works, both Halter15 and Cohé16 reported also the same tendency for Da and Ka numbers with pressure. Once the simulation results are plotted on the flame regime diagram, Figure 10, it is observed that the premixed flames are located at the wrinkled flame region at lower pressure conditions. Moreover, the flame is shifted to the thickened-wrinkled regime boundary at 0.9 MPa. On the other hand, in the work of Driscoll27, it is discussed that the boundaries of the diagram is questionable and may be extended to Ka = 10. It is concluded that

ACS Paragon Plus Environment

16

Page 17 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

the increase in pressure changes the location of premixed flames towards thickened wrinkled flame regime. Therefore fast chemistry assumption, of which the Zimont model based on, may no longer be applicable for this flame regime. 5. Conclusions The complete understanding of the influence of pressure on the premixed flame front statistics is necessary since most of the industrial combustion systems operate under pressurized conditions. Therefore, the prediction of flame statistics is crucial for the stable and efficient operation of such systems such as gas turbines. The aim of this work is to contribute to reveal the effects of elevated pressure conditions on the premixed flame front statistics by numerical modeling. The numerical studies are performed in two steps. Initially, the cold flow field developed in the chamber at various pressures from 0.1 to 0.9 MPa, has been computed and then the reactive case studies have been performed for the same pressures. The results are compared with the available measurements for both cases. The computed cold flow field statistics such as the mean axial velocity, the turbulent kinetic energy and turbulent length scale profiles have been compared with the experiments and good agreement has been found. The velocity potential core region resists until 3-diameter downstream from the burner exit. Moreover, the decay rate of the normalized mean axial velocity decreases with increasing the pressure since the jet penetrates in a medium at higher density and its mixing rate is therefore reduced at higher chamber pressures. However, in the reactive case simulations, the decay in the velocity profiles are observed much lower due to hot gas expansion through the flame front similar to the experiments. The laminar flame front statistics in terms of the laminar flame speed and Zeldovich flame thickness have been computed

ACS Paragon Plus Environment

17

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 33

using the Chemkin software in the reactive flow cases. The flame speed decreases with pressure by the combined pressure effects on the chemical kinetics and the thermodynamic properties. The turbulent flame structure is investigated using Zimont premixed combustion model based on the solution of the progress variable transport equation with a turbulent flame closure model. The progress variable distributions along the symmetry axis are shifted downstream as in the experiments however the location is computed at more downstream locations, about one burner exit diameter, compared to the measurements. In addition, thicker flame brush thicknesses are found at higher pressure conditions as in experiments. The flame surface density distributions are computed and good agreement with experimental results has been observed. Pressure increases both the flame surface density distribution and the magnitude of the maximum FSD. A better agreement with experiments is obtained at lower pressures. The interaction between turbulence and the combustion scales are represented in terms of Da and Ka numbers. Da number decreases whereas Ka number increases with increasing pressure. When the studied cases are plotted on the Borghi diagram, it is observed that the studied flames are in the thin wrinkled flame regime at the atmospheric pressure but move towards the thickenedwrinkled flame regime zone because of the flame front thickening effect of smaller scale turbulent eddies generated by increased pressure. Therefore, fast chemistry assumption may no longer be valid for high pressure conditions since the chemical and the turbulence scales become comparable.

ACS Paragon Plus Environment

18

Page 19 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

REFERENCES 1.

T. Poinsot, D. Veynante, Theoretical and Numerical Combustion, R.T. Edwards Publications, 2001.

2.

N. Peters, Turbulent Combustion, 2000, Cambridge University Press, UK.

3.

D. Veynante, L. Vervisch, “Turbulent Combustion”, Von Karman Institute for Fluid Dynamics Lecture Series Notes 2006-2007, , J.P.A.J., 2007.

4.

R. Borghi, Rec. Adv. Aerosp. Sci.,1985, 117-138.

5.

N. Peters, J. Fluid Mech., 1999,384, 107-132.

6.

H. Kobayashi, K. Seyama, H. Hagiwara, Y. Ogami, Proc. of the Comb. Ins., 2005, 30, 827–834.

7.

P. Griebel, R. Schären, P. Siewert, R. Bombach, A. Inauen, W. Kreutner, Proc. of ASME Turbo Expo 2003 Power for Land, Sea, and Air, 2003.

8.

A. Soika, F. Dinkelacker, A. Leipertz, Combustion and Flame, 2003,132, 451–462.

9.

R. Fragner, F. Halter, N. Mazellier, C. Chauveau, I. Gökalp, Proc. Combust. Inst., 2015, 35-2, 1527– 1535.

10. C.C. Liu, S. S. Shy, M.W. Peng, C.W. Chiu, Y.C. Dong, Combustion and Flame, 2012, 159, 2608– 2619. 11. C.M. Vagelopoulos, F.N. Egolfopoulos, C.K. Law, Proc. Combust. Inst., 1994, 25, 1341-1347. 12. Y. Ogami and Hideaki Kobayashi, JSME International Journal, 2005, 48B-3, 603-609. 13. M.I. Hassan, K.T. Aung, G.M. Faeth, Combustion and Flame, 1998, 115, 539-550. 14. T. Lachaux, PhD Thesis, 2004, Universite d’Orleans, Orleans, France. 15. F. Halter, PhD Thesis, 2005, Universite d’Orleans, Orleans, France. 16. C. Cohé, PhD Thesis, 2007, Universite d’Orleans, Orleans, France. 17. A. Ratzke, T. Schöffler, K. Kuppa, F. Dinkelacker, Combustion and Flame, 2015, 162-7, 2778-2787. 18. M. Goswami, K. Coumans, R.J.M. Bastiaans, A.A. Konnov, L.P.H. De Goey, Combustion Sci. and Technology, 2014, 186, 10-11, 1447-1459. 19. F. Dinkelacker, B. Manickam, , S.P.R. Muppala, Combustion and Flame, 2011, 158, 9, 1742-1749. 20. S. Hu, J. Gao, Y. Zhou, C. Gong, X.S. Bai, Z. Li, M. Alden, Energy Procedia, 2017,105, 4970-4975. 21. P. Kumar, P. A. Kishan, A. Dhar, Int. J. of Hydrogen Energy, 2016, 41, 22, 9644-9652. 22. Fluent 6.2 Users Manual, Fluent Inc., 2005. 23. S.B. Pope, AIAA Journal, 1978, 16(3), 279-281. 24. B. Yilmaz, S. Özdogan, I. Gökalp, Energy and Fuels, 2009, 23,4, 1843-1848. 25. R.J. Kee, F.M. Rupley, J.A. Miller, Report SAND89-8009, 1989, Sandia National Laboratories, USA. 26. V. Zimont, W. Polifke, M. Bettelini, W. Weisenstein, J. of Gas Turbines Power, 1998, 120, 526-532.

ACS Paragon Plus Environment

19

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 33

27. J.F. Driscoll, Prog. in Energy and Combust. Sci., 2008, 34, 91-134.

ACS Paragon Plus Environment

20

Page 21 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

LIST OF FIGURES 1. High pressure combustion chamber and schematic view of Bunsen burner on the left and right, respectively. 2. The normalized mean axial velocity profiles along normalized axial direction for several pressure conditions. (Symbols: experiments, lines: simulations) 3. The normalized turbulent kinetic energy distribution along normalized axial direction at several pressure conditions. (Symbols: experiments, lines: simulations) 4. The distribution of integral length scale along the normalized axial direction for several pressure conditions. (Symbols: experiments, lines: simulations) 5. The reactive normalized mean axial velocity profiles are compared with experiments. Experimental data are measured by Lachaux.14 6. The calculated laminar flame speeds versus pressure and comparison with experiments. 7. The computed Zeldovich flame thicknesses at elevated pressures. 8. The progress variable distributions computed at various pressures and comparison with experiments.15 9. The FSD distributions versus progress variable and comparison with experiments.16 10. The positions of the flames on the flame regime diagram (symbols ▲, ■ and ♦ represents 0.1, 0.5 and 0.9 MPa conditions, respectively).

ACS Paragon Plus Environment

21

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 33

Main Flow Pilot Flame Flow

Figure 1. High pressure combustion chamber and schematic view of bunsen burner on the left and right, respectively.

ACS Paragon Plus Environment

22

Page 23 of 33

1 ++++++

0.8

U / U0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

0.6

+ 0.4

0.2

0

1

++++++

++++

+++

+++

0.1 MPa 0.3 MPa 0.5 MPa 0.7 MPa 0.9 MPa 0.1 MPa Exp 0.3 MPa Exp 0.5 MPa Exp 0.7 MPa Exp 0.9 MPa Exp

2

3

+++

4

++

5

z/D Figure 2. The normalized mean axial velocity distributions along the normalized axial direction for several pressure conditions. (Symbols: experiments, lines: simulations)

ACS Paragon Plus Environment

23

Energy & Fuels

0.1 MPa 0.3 MPa 0.5 MPa 0.7 MPa 0.9 MPa 0.1 MPa Exp 0.3 MPa Exp 0.5 MPa Exp 0.7 MPa Exp 0.9 MPa Exp

3.5 3 2.5

k / k0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 33

+

2

+

+

+ ++

+

+ + ++

1.5 1 ++

+++++++++++

+

+

++

0.5 0

0

1

2

3

4

5

6

z/D Figure 3. The normalized turbulent kinetic energy distribution along the normalized axial direction for several pressure conditions. (Symbols: experiments, lines: simulations)

ACS Paragon Plus Environment

24

Page 25 of 33

10

+ 8

lT [mm]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

+

+

+ +

6

+ +

4

+

+

+

+

+

+

2

0

0

1

2

3

4

5

6

z/D Figure 4. The distribution of integral length scale along the normalized axial direction for several pressure conditions. (Symbols: experiments, lines: simulations)

ACS Paragon Plus Environment

25

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 33

Figure 5. The reactive normalized mean axial velocity profiles are compared with experiments. Experimental data are measured by Lachaux.14

ACS Paragon Plus Environment

26

Page 27 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Ogami et al. [12] Vagelopoulos et al. [11] Hassan et al. [13] Present work

Figure 6. The calculated laminar flame speeds versus pressure and comparison with experiments.

ACS Paragon Plus Environment

27

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 33

Figure 7. The computed Zeldovich flame thicknesses at elevated pressures.

ACS Paragon Plus Environment

28

Page 29 of 33

< C > Sim 0.1 < C > Sim 0.3 < C > Sim 0.5 < C > Sim 0.7 < C > Sim 0.9 Exp 0.1 M Pa Exp 0.3 M Pa Exp 0.5 M Pa Exp 0.7 M Pa Exp 0.9 M Pa

1

0.8

0.6

M Pa M Pa M Pa M Pa M Pa



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

0.4

0.2

0

0

1

2

3

z/D

4

5

6

Figure 8. The progress variable distributions computed at various pressures and comparison with experiments15.

ACS Paragon Plus Environment

29

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 30 of 33

Figure 9. The FSD distributions versus progress variable and comparison with experiments16.

ACS Paragon Plus Environment

30

Page 31 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Figure 10. The positions of the flames on the flame regime diagram (symbols ▲ , ■ and ♦ represents 0.1, 0.5 and 0.9 MPa conditions, respectively.)

ACS Paragon Plus Environment

31

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 32 of 33

TABLES

Table 1. Turbulent premixed flame regimes and their characteristics.

Flame regime

ReT

Da

Ka

u’/SL

δL / η

Wrinkled flame

>1

>1

>1

>1

1

>1

>1

>>1

Distributed flame >1

Table 2. The inlet conditions for the pressure case simulations P (MPa)

0.1

0.3

0.5

0.7

0.9

 x10-3(kg/s)

1.122

3.5645

5.9474

8.3276

1.0583

Umean (m/s)

1.90

2.01

2.01

2.01

2.01

ρ (kg/m3)

1.201

3.607

6.017

8.425

10.709

µx10-5 (kg/m2s)

1.85

1.85

1.85

1.85

1.85

Re

3090

9810

16400

22900

29100

Table 3. The values of parameters used in Equation 4 for various pressure conditions15. Pressure(MPa) σy

Ly / g (mm)

0.1

0.59

1.50

0.5

0.61

1.25

0.9

0.61

1.00

ACS Paragon Plus Environment

32

Page 33 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Table 4. The influence of the pressure on the turbulent flow and flame characteristics at =0.05. P[MPa]

’[m/s]

’/SL

η[mm]

η/δL

ReT

Da

Ka

0.1

0.144

1.27

0.348

1.77

36.65

20.71

0.292

0.5

0.133

3.15

0.109

1.03

156.87

14.48

0.865

0.9

0.133

4.46

0.069

0.83

297.5

13.24

1.303

ACS Paragon Plus Environment

33