You are on page 1of 18

Chinese Journal of Aeronautics, (2020), 33(1): 116–133

Chinese Society of Aeronautics and Astronautics


& Beihang University
Chinese Journal of Aeronautics
cja@buaa.edu.cn
www.sciencedirect.com

Fluid Mechanics and Flight Mechanics–Research Article

LES of the Sandia flame series D-F using the


Eulerian stochastic field method coupled with
tabulated chemistry
Yifan DUAN, Zhixun XIA, Likun MA *, Zhenbing LUO, Xu HUANG,
Xiong DENG

College of Aerospace Science and Technology, National University of Defense Technology, Changsha 410073, China

Received 2 September 2018; revised 12 November 2018; accepted 31 March 2019


Available online 11 October 2019

KEYWORDS Abstract In this paper, the Eulerian Stochastic Field (ESF) model in the Transported Probability
Eulerian Stochastic Field Density Function (TPDF) class model is combined with the Flamelet Generated Manifolds (FGM)
(ESF); model. This method solves the joint probability density function transport equation by ESF method
Flamelet Generated Mani- that considers the interaction mechanism between flame and turbulence with high precision. At the
folds (FGM); same time, by making use of the advantage of the FGM model, this model is able to incorporate the
Large Eddy Simulation detailed chemical reaction mechanism (GRI 3.0) with acceptable computational cost. The new
(LES); model has been implemented in the open source CFD suite-OpenFOAM. Validation of the model
OpenFOAM; has been carried out by simulating the Sandia flame series (three turbulent piloted methane jet
Turbulent combustion flames) issued by the National Laboratory of the United States. The accuracy and advancement
of the ESF/FGM turbulent combustion model are verified by comparing the LES results of the
new model with the rich experimental data as well as the RANS results. The results demonstrate
that the model has a strong ability in capturing combustion phenomena such as extinction and
re-ignition in turbulent flame, which is essential in the accurate prediction of the combustion pro-
cess in real combustion devices, for example, aircraft engines.
Ó 2019 Chinese Society of Aeronautics and Astronautics. Production and hosting by Elsevier Ltd. This is
an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

As the most common and main combustion mode closely


* Corresponding author. related to human production and life, turbulent combustion
E-mail addresses: zxxia@nudt.edu.cn (Z. XIA), malikun@nudt.edu. has always been a hot topic and important issue for researchers
cn (L. MA). in many countries. With the rapid advancement in the com-
Peer review under responsibility of Editorial Committee of CJA. puter technology, numerical simulation technology is becom-
ing an increasingly important research tool for studying
turbulent combustion problems. The turbulent combustion
Production and hosting by Elsevier process is an extremely complex physical and chemical process

https://doi.org/10.1016/j.cja.2019.09.022
1000-9361 Ó 2019 Chinese Society of Aeronautics and Astronautics. Production and hosting by Elsevier Ltd.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Eulerian stochastic field method coupled with tabulated chemistry 117

involving turbulent flow, heat transfer, mass transfer, and ables are independent of each other, but this is not the case
chemical reactions. The fundamental mechanism of turbulent under many circumstances. Chen et al.11 found in their study
flow has not been fully understood yet, and it is one of the of partial premixed flames that the Beta PDF (a widely used
remaining unsolved issues in classical physics. The turbulent P-PDF method) distribution of the independence variables is
flow has a strong nonlinear coupling relationship with the inconsistent with the DNS data, proving that the statistically
simultaneous chemical reaction,1–4 which makes the numerical independence assumption of control variables is questionable.
simulation of turbulent combustion very difficult. Secondly, with the increase of the number of control variables
At present, the Large Eddy Simulation (LES) method has used in the FGM model or the increase of the reaction mech-
become a commonly used numerical simulation method for anism, the size of the look-up table increases exponentially and
studying complex turbulent combustion problems due to the the memory requirements are huge. In addition, the assumed
continuous increase of computer resources. In this method, shape coefficients in the P-PDF method are obtained from
the turbulent flow associated with large-scale eddy is directly empirical data, and the PDF distribution required for different
numerically simulated. By establishing a corresponding model types of flames cannot be universally described by these coef-
to simulate the influence of small-scale vortices on large-scale ficients with good accuracy. Bray et al.12 found in the study
vortex motion, the method achieves the best compromise of turbulent diffusion flames, that the assumed shape coeffi-
between computational accuracy and computational complex- cients of the three most widely used models in the P-PDF
ity compared to Direct Numerical Simulation (DNS) and method have a great influence on the prediction of chemical
Reynolds-Averaged Navier-Stokes (RANS) simulation.5 The reaction rate. Finally, the P-PDF method is derived from the
LES method is capable of capturing unsteady processes asso- empirical data of the gas phase, which limits the further appli-
ciated with turbulent combustion, but this method does not cation of the FGM model in multiphase combustion, e.g. spray
fully solve all turbulence and scalar structures. For the simula- combustion, where the controlling variables are no longer con-
tion of the turbulent combustion process, the LES method served variables. Ge and Gutheil13 found in the study of a tur-
needs to establish a corresponding model to describe the bulent spray with TPDF method that the actual probability
turbulent-flame interaction at the sub-grid scale. For turbulent density of the mixture fraction, the gas temperature and the
combustion processes involving complex chemical reactions, enthalpy has a large difference from the distribution obtained
the LES method requires appropriate models and a large by the Beta model.
amount of computational resources for accurate numerical The coupled ESF/FGM model for turbulent combustion
simulation. At present, for LES, the mainstream turbulent developed in the current study still decouples the chemical
combustion model can be divided into Transported Probabil- reactions and turbulent flow by means of tabulated chemistry
ity Density Function (TPDF) method, flamelet-based model, method, i.e. FGM model, but directly solves the PDF of the
moment closure model and other combined models.6 The control variables by the ESF model. It maximizes the advan-
TPDF model group consists of Eulerian Stochastic Field tages of the two types of models, accomplishing a class of tur-
(ESF) model and Monte Carlo TPDF model. The most promi- bulent combustion model that can achieve high precision at
nent advantage of this model is that the chemical reaction low computational cost, and is able to consider the detailed
source term can be directly closed by the joint Probability Den- chemical reaction mechanism, facilitating the accurate simula-
sity Function (PDF) without the need to introduce a closure tion of, for example, pollutant emission as well as ignition/
model, which makes the calculation accuracy very high and extinction processes. The model treats the turbulent flow field
has the advantage of being unrestricted by the combustion in a random way. Through multiple real-time solutions to the
mode. However, for the single-point-single-time PDF, it is dif- PDF transport equation of the control variables under local
ficult to account for the molecular diffusion process, and the flow field conditions, the statistical average gets the local con-
computational cost of the simulation is enormous. The flame- trol variables under the influence of turbulent fluctuation, and
let model considers that, under certain conditions, the turbu- then the other scalar fields of the simulated flame are obtained
lent combustion field consists of a series of flamelets and a by retrieving the FGM look-up table. Similar application has
non-reactive turbulent flow field surrounding these flamelets, been applied in Ref. 14. In this paper, we report the implemen-
which realizes the decoupling calculation of turbulent flow tation of the new model in the open source CFD suite-
and chemical reaction process, has high computational effi- OpenFOAM and its validation. We select the Sandia flame
ciency and can take into account the advantages of detailed D-F15 as target cases to validate the accuracy and advance-
chemical reaction mechanism. ment of the new model.
In recent years, the Flamelet Generated Manifolds (FGM)
proposed by van Oijen and de Goey7 has been widely used in
numerical simulation of various turbulent flames, and the cal- 2. Theoretical calculation model
culation results that agree well with the experimental data have
been obtained.8–10 The FGM model establishes a multi- 2.1. FGM model
dimensional flamelet look-up table by calculating a series of
one-dimensional laminar flames that consider the detailed In the FGM model, there is no need to solve the transport
chemical reaction mechanism. The traditional FGM model equations for all components and energy, and the chemical
expands the laminar flamelet look-up table into a turbulent fla- reactions in turbulent combustion are thought to occur in
melet look-up table by considering the interaction between tur- low-dimensional manifolds, which means that only a few inde-
bulence and flame by a Presumed-Probability Density pendent variables are required in the entire component space
Function (P-PDF) method. However, P-PDF itself has many to characterize chemical reaction in turbulent combustion. In
limitations. Firstly, the model assumes that the control vari- the model, the ‘‘mixture fraction”, Z, that characterizes the
118 Y. DUAN et al.

mixing state of fuel and oxidizer, and the ‘‘progress variable”, The influence of turbulence on the local flame structure is
C, that characterizes the progress of chemical reaction, are considered by the joint PDF of the control variables (Z and
usually selected as independent variables. Of course, depend- C) in the FGM model. The Favre average of the scalar u in
ing on the physical model being simulated, variables such as the turbulent flow field (e.g. temperature, mass fraction of each
pressure and enthalpy loss can be added as supplementary components) can be calculated as
independent variables.8,10,16,17 The FGM turbulent combus- Z 1Z 1
 
tion model under the LES context can be expressed as18 u¼ uðZ; CÞ P ðZ; CÞdZdC ð9Þ
0 0
 
@ q @ q uj  
þ ¼ Sq ð1Þ where P ðZ; CÞ is the joint PDF of the mixture fraction and the
@t @xj
progress variable. In the original FGM model, it is usually
    assumed that the control variables are statistically independent
 @ q ui uj   
@ q uj @p @  D  of each other and the PDF obeys a certain hypothesis distribu-
þ ¼ þ 2 l Sij  sij þ Sui ð2Þ
@t @xj @xi @xj tion obtained from the first moment and the second moment of
the control variable. In order to avoid the many defects of the
   
" #
@   @Z method described in the foregoing, the ESF model is used for
@ q Z @ q uj Z 
þ ¼ q D þDt þ SZ ð3Þ the direct solution of the joint PDF.
@t @xj @xj @xj

"  #
2.2. ESF model
   
@ q YC @ q uj YC @    @Y
C

þ ¼ q D þDt þ x_ YC ð4Þ
@t @xj @xj @xj The basic idea of the ESF model is: solving the joint PDF of
the control variables using a set of N random fields and solving
  that by using a set of Stochastic Differential Equations (SDE).
1 @ui @uj
Sij ¼ þ ð5Þ In this paper, according to the Eulerian random field formula
2 @xj @xi
adopted in Ref. 20, the N random fields
 
1 na ðxi ; tÞ ¼ n1a ; n2a ; . . . ; nN
a for the control variables Z and C
ij ¼ Sij  dij Skk
SD ð6Þ are based on the stochastic partial differential equation derived
3
from the joint PDF transport equation. These stochastic par-
where the subscript ‘‘i”or ‘‘j”stand for directions i or j; the tial differential equations can be expressed as14
notations of ‘‘” and ‘‘” are different, they stand for time
  
or ensemble average in RANS, and spatial average at the  @   n  @ l l @nna
d q nna ¼  q uj na dt þ þ t dt
Sub-Grid Scale (SGS) in LES; q, u and p represent density, @xj @xj Sc Sct @xj
 1=2 n
q  n  
velocity and pressure; Sq , Sui and SZ represent the source term 
2 l @na
for density, velocity and mixture fraction due to the non- þ x_ na dt  na  ua dt þ 2q t dWnj
2ssgs Sct @xj
combustion caused source term respectively, for example body
force or radiation; x_ YC is the source term for progress variable n ¼ 1; 2;    N ð10Þ
due to reaction; SD ij is the deviatoric part of the strain rate ten- where l is the dynamic viscosity, and lt is the apparent turbulent
sor component Sij and sij is SGS stress, closed with standard viscosity; Sc is the Schmidt number, and the subscript of Sct

Smagorinsky model in LES model; dij is Kronecker delta, stand for turbulence term; ua denotes the controlling variable.
and Skk is linear strain; l is the molecular viscosity; D is the dif- Hence, in case that N stochastic fields are utilized, one needs
fusivity, and the subscript of Dt stand for turbulence term; Z is to solve N  2 scalar transport equations. Herein, x_ na denotes
the mixture fraction, and is defined according to Bilger’s the chemical source term describing the production or the con-
method19 of element definition with the same diffusion coeffi- sumption of the controlling variable due to the chemical reac-
cient for all components; YC is an un-normalized progress vari- tion. The second last term in the right-hand side of the
able, and its definition in this study is as follows: equation represents the molecular diffusion (micro-mixing)
YCO2 YH2 O YH2 term. Since the contribution of the molecular diffusion term in
YC ¼ þ þ ð7Þ the LES method is not as large as in the RANS method, the clo-
WCO2 WH2 O WH2
sure is performed using a relatively simple and computationally
where W is molar mass and Y is mass fraction. The normalized cheap Linear Mean Square Estimation (LMSE) model21. The
progress variable C is defined as turbulent mixing time ssgs is determined according to the mixing
time model proposed in Ref. 22 and reads
YC  YuC
C¼ ð8Þ 
YbC  YuC 1 m þmt
¼ CX 2 ð11Þ
ssgs D
where superscripts ‘‘b” and ‘‘u” represent the burned and
unburned states respectively. The variables Z and C are the where the micro-mixing constant is selected according to other
independent variables of the look-up table. The equations of related work,23 CX ¼ 2. D is the grid size. m denotes the kine-
the unnormalized progress variable YC are directly solved in matic viscosity, and the subscript of mt stand for turbulence
the simulation because there are additional source terms that term. The last term in Eq. (10) represents the random term
need to be modeled in the transport equation of the normal- caused by turbulence, which is subject to an independent ran-
ized progress variable C, which increases the complexity of dom Wiener process for each random field. The Wiener pro-
the model. cess varies over time, but is independent of space and
Eulerian stochastic field method coupled with tabulated chemistry 119

position. In a given random field, all scalars use the same dWnj as validation cases. The fuel of this series of flames is a mixture
value, and the dWnj value represents the increment of the vec- of methane and air. The fuel is 25% methane diluted in 75%
tor Wiener process, which is expressed as air by volume, and the diameter of the jet inlet is 7.2 mm.
pffiffiffiffiffi The average velocity of the fuel jets of the Sandia D, E and
dWnj ¼ fð0; 1Þ Dt ð12Þ F flames are 49.9, 74.4 and 99.2 m/s respectively. The pilot
flame of this series of flames is high-temperature combustion
where fð0; 1Þ represents a Gaussian random variable with a
product. The equivalent ratio is set to 0.77 and the diameter
mean of 0 and a variance of 1, and Dt is the time step size.
is 18.9 mm. The average speed of the pilot flames of Sandia
Eq. (10) is a general formula for the joint PDF transport equa-
D, E and F are 11.4, 17.1 and 22.8 m/s respectively. The tem-
tions of arbitrary control variables, which provides a good
perature of fuel jet and pilot flame are 290 K and 1880 K
basis for the new model to add more control variables.
respectively. Other flow field parameters can be found in
Ref. 15.
3. Validation cases and numerical setup
This series of flames, especially the Sandia flame D, has
been widely used as validation cases for many new mod-
This paper uses a series Sandia flame D-F released by the San- els.24–28 Among them, Mustata et al.21 applied the ESF
dia National Laboratory, which is rich in measurement data, method to simulation of the Sandia flame D, Renzo et al.29

Fig. 1 Transient and time-averaged simulation for Sandia flame D, E and F.

Fig. 2 Instantaneous scatter plots of temperature vs mixture fraction for Sandia flame D, E and F (x/d = 7.5).
120 Y. DUAN et al.

Fig. 3 Central profiles of mean and RMS temperature and main components mass fraction for Sandia flame D.

Fig. 4 Central profiles of mean and RMS temperature and main components mass fraction for Sandia flame E.

applied the Flamelet/Progress Variable (FPV) model to the method combined with the simplified mechanism of the
Sandia flame D, and Jones and Prasad30 applied a system- Augmented Reduced Mechanism (ARM). The authors31
atic study of the Sandia flames series D-F using the ESF used the ESF/FGM model studied in this paper to study
Eulerian stochastic field method coupled with tabulated chemistry 121

Fig. 5 Central profiles of mean and RMS temperature and main components mass fraction for Sandia flame F.

Fig. 6 Radial profiles of mean and RMS temperature at six axial locations for Sandia flame D.

the Sandia flame D under the RANS context, and achieved The velocity inlet boundary condition of the fuel and pilot
good results. flames of Sandia flames D-F adopt the measure profiles in the
The open source computational fluid dynamics software current study. The solution region extends 17 jet diameters (d)
(OpenFOAM) is used to complete the program development. in the radial direction and 60 jet diameters in the axial direction.
122 Y. DUAN et al.

Fig. 7 Radial profiles of mean and RMS temperature at six axial locations for Sandia flame E.

Fig. 8 Radial profiles of mean and RMS temperature at six axial locations for Sandia flame F.

The grid used is refined in the direction of the fuel jet, with a total effect of turbulent flow on the flamelet. The simulation was per-
mesh size of about 3.5 million, and all lateral boundaries are formed on the Tianhe-1 supercomputer. One simulation of the
free-slip. This paper uses eight stochastic fields to describe the case requires 132 CPUs to run for about 4 days.
Eulerian stochastic field method coupled with tabulated chemistry 123

Fig. 9 Radial profiles of mean and RMS CH4 mass fraction at six axial locations for Sandia flame D.

Fig. 10 Radial profiles of mean and RMS CH4 mass fraction at six axial locations for Sandia flame E.

The flamelet databases (FGM table) of flame were gener- 4. Simulation results and analysis
ated by solving the non-premixed counterflow flame with
Chem1D software. The boundary conditions for generating In this section, by analyzing the scatter plot of the temperature,
the FGM table were given according to the boundary condi- the distribution of species mass fraction along the central axis,
tions of the Sandia flame series. and the radial profiles LES results for each typical location,
124 Y. DUAN et al.

Fig. 11 Radial profiles of mean and RMS CH4 mass fraction at six axial locations for Sandia flame F.

Fig. 12 Radial profiles of mean and RMS O2 mass fraction at six axial locations for Sandia flame D.

and compared with the experimental results32 and RANS given literature, it will not be described in detail in this paper. Based
in Ref. 31, the characteristics of the LES model of ESF/FGM on the average velocity of the jet flow, the results given in this
are validated and analyzed. Since the results of the ESF/FGM section are extracted from the average of the 10–15th flow-
model and the original FGM model have been analyzed in the through cycles after the steady flow of the jet is achieved. Six typ-
Eulerian stochastic field method coupled with tabulated chemistry 125

Fig. 13 Radial profiles of mean and RMS O2 mass fraction at six axial locations for Sandia flame E.

Fig. 14 Radial profiles of mean and RMS O2 mass fraction at six axial locations for Sandia flame F.

ical axial positions are selected to analyze the radial profile ing occurred is: x/d = 7.5, and there is a position of re-ignition
results, which are the ignition positions near the jet inlet: x/ combustion phenomenon: x/d = 15, in accordance with the typ-
d = 1, x/d = 2, x/d = 3, the position where partial extinguish- ical characteristics of the diffusion flame: x/d = 30.
126 Y. DUAN et al.

Fig. 15 Radial profiles of mean and RMS CO2 mass fraction at six axial locations for Sandia flame D.

Fig. 16 Radial profiles of mean and RMS CO2 mass fraction at six axial locations for Sandia flame E.

As shown in Fig. 1, it is an instantaneous result of the tem- fields. It can be clearly seen that the greater the velocity in
perature (T) field and the average result of 10–15 time periods the three flames is, the thicker the thickness of the simulated
after the full development of the Sandia series of flame flow flame is and the higher the height is, which is in line with
Eulerian stochastic field method coupled with tabulated chemistry 127

Fig. 17 Radial profiles of mean and RMS CO2 mass fraction at six axial locations for Sandia flame F.

Fig. 18 Radial profiles of mean and RMS CO mass fraction at six axial locations for Sandia flame D.

theoretical analysis. As shown in Fig. 2, it is the position of the d = 7.5. We can clearly see that from the Sandia flame D to
partial flameout phenomenon of the Sandia series flame, and the Sandia flame F, the phenomenon of partial flame flameout
the scatter plot of the experimental value and the simulated continues to increase, the Sandia flame E and Sandia flame F
value of the temperature and the mixing fraction at x/ are more obvious, while the Sandia flame D has almost no par-
128 Y. DUAN et al.

Fig. 19 Radial profiles of mean and RMS CO mass fraction at six axial locations for Sandia flame E.

Fig. 20 Radial profiles of mean and RMS CO mass fraction at six axial locations for Sandia flame F.

tial flameout, which is a typical diffused flame characteristic. In As shown in Figs. 3–5, the results of the LES simulation of
general, the simulation results are basically consistent with the the temperature, CH4, CO, CO2, H2O, and O2 mass fractions
experimental results. of the Sandia series flame at the central axis positions agree
Eulerian stochastic field method coupled with tabulated chemistry 129

Fig. 21 Radial profiles of mean and RMS H2O mass fraction at six axial locations for Sandia flame D.

Fig. 22 Radial profiles of mean and RMS H2O mass fraction at six axial locations for Sandia flame E.

well with the experimental values (Exp) and are better than simulation average results and Root Mean Square (RMS) val-
that the best 48 stochastic field results obtained by the authors ues of the LES of the Sandia series flames are in good agree-
in the previous stage.31 It can be seen that, in general, the ment with the experimental results, especially the LES
130 Y. DUAN et al.

Fig. 23 Radial profiles of mean and RMS H2O mass fraction at six axial locations for Sandia flame F.

Fig. 24 Radial profiles of mean and RMS OH mass fraction at six axial locations for Sandia flame D.

simulation results of the Sandia flame D are in good agreement especially the prediction of CO concentration. However, it
with the experimental results. Compared with the simulated has also been found that the accuracy of the simulation
average results of LES and RANS of Sandia flame D, LES decreases with the increase of the velocity of the flame jet, such
results effectively improved the problems of overestimation as the mass fraction of CH4 (fuel) and O2 (antioxidant) con-
of chemical reaction rate, high peak value and ignition delay, sumption position and the mass fraction of CO2 and CO (com-
Eulerian stochastic field method coupled with tabulated chemistry 131

Fig. 25 Radial profiles of mean and RMS OH mass fraction at six axial locations for Sandia flame E.

bustion products) production position are relatively greater in Refs. 28,30,33,34. This is supposed that the detailed chemi-
than the experimental results, because the simulated ignition cal reaction mechanism and the direct solution of the joint
process is slightly delayed compared with the actual one. PDF of the control variables adopted in the current study
As shown in Figs. 6–8, the LES simulation results of the can achieve high-precision simulation of turbulence/flame
temperature of Sandia series flame at six axial positions were interaction. For the prediction of CO2 mass fraction, the
compared with the experimental values and the best 48 LES model results effectively improved the peak prediction
stochastic field results of the RANS obtained by the author and solved the problems caused by the use of simplified chem-
in the previous stage. We can see that on the whole the simu- ical reaction mechanism in Ref. 21. However, in the compar-
lated average and RMS values of the LES of the Sandia series ison of the simulation results of the three flames, the model
flames are generally consistent with the experimental results. also has the common problem with Refs. 28,30,33,34, namely,
As the distance from the inlet position increases, the simula- the result of Sandia flame F is not accurate enough. The reason
tion accuracy of RANS becomes lower and lower. LES, by vir- for this is uncertain though it has to be acknowledged that
tue of the advantages of the model, can effectively improve the Sandia flame F is very close to blow-off and it represents a sev-
prediction of the impact of turbulent flow on chemical reac- ere test of all aspects of the simulation.
tion, making the simulation results of LES higher. Compared Compared with Sandia flame E and F, the result of the
with Sandia flame D and E, the simulation accuracy of Sandia RANS model is unstable. With the increase of fuel jet speed,
flame F is relatively low, especially for the prediction of tem- affected by the influence of turbulence intensity, the simulation
perature peak and flame thickness in the downstream of flame. result of RANS model is larger and larger than the actual
This may be because the jet velocity of Sandia flame F is high, value. LES model simulation result is stable, and the model
and the peak speed will reach 120 m/s, which makes the influ- can more accurately simulate the Sandia flame E and Sandia
ences of the small scales on the flame more significant and the flame F ignition location. From the mass fraction evolution
standard Smagorinsky method used in the current study may rates of fuel CH4, oxidant O2, combustion products H2O,
not be sufficient for the accurate description of this process. CO2 and CO, it can be found that LES model effectively
It is suggested that more advanced SGS models should be used improves the overestimation of chemical reaction rate by
in future studies. RANS model, because LES model effectively improves the
Figs. 9–26 show that the LES simulation results for the mixing effect of fuel and air.
mass fraction of each component for the Sandia series of As mentioned in Ref. 35 for studying of the Sandia flame D
flames at six axial positions, including CH4, O2, CO2, CO, by the Euler random field method alone, the results of 16 and 8
H2O and OH. It can be seen that, in general, the simulation random fields are similar. Therefore, we can think that the
average results and RMS values of LES of Sandia series flames model adopts more random number and more accurate solution
are in good agreement with the experimental results. In partic- of PDF, and cannot increase the simulation accuracy of Sandia
ular, the prediction results of minor species such as OH are flame D. However, Ref. 35 does not use the same method to
surprisingly good, and the results are better than those given study the Sandia flame F. Due to the complex turbulence char-
132 Y. DUAN et al.

Fig. 26 Radial profiles of mean and RMS OH mass fraction at six axial locations for Sandia flame F.

acteristics of the Sandia flame F, the results of this flame simu- accurately predict the shape of the flame and capture the phe-
lation which uses more random fields may be more accurate. It nomenon of extinction and re-ignition under different Rey-
is also possible to influence the simulation results with the selec- nolds numbers. At present, this combined model can
tion of micro-mixing constants. Ref. 36 shows that when solving accurately simulate the evolution of temperature, the mass
the joint probability-density function transport equation, the fractions of methane, carbon dioxide and hydroxyl groups of
selection of this constant is important for the micro-mixing Sandia flame D and E. However, for the Sandia flame F which
term. In the next work, a similar approach will be used to study contains severe extinction and blow-off state, the accuracy of
the role of micromixing constants in the LES method. Above the model simulation is not satisfactory without changing the
are the possible effects of selecting the simulation model. Simi- parameter settings of the model and example. In the future
larly, the error of the measured data released by Sandia study, we will investigate the effects of the number of random
National Laboratory can also lead to the deviation of the sim- fields, micro-mixing constants, and inlet temperatures on the
ulation results. As shown in Ref. 14, the temperature measure- simulation results of Sandia flame F.
ment error for the pilot flame is plus or minus 50 K, and the
temperature used in this paper is the standard value in the ref-
erence, 1880 K. The selection of the inlet boundary temperature Acknowledgement
has a direct impact on the process of building the flame surface
model and the sensitivity of the chemical reaction mechanism. This study was supported by the National Natural Science
These parameters will be checked in further study. Foundation of China (No. 51706241).

5. Conclusions References

1. Wang F, Liu R, Dou L, Liu D, Jin J. A dual timescale model for


In this paper, the Eulerian stochastic field method is combined micro-mixing and its application in LES-TPDF simulations of
with the flamelet generated manifolds model for large eddy turbulent nonpremixed flames. Chin J Aeronaut 2019;32
simulation modeling of turbulent combustion. The combined (4):875–87.
model enables the consideration of detailed chemical reaction 2. Ren Z, Wang B, Zheng L, Zhao D. Numerical studies on
mechanisms and high-precision simulation of the interaction supersonic spray combustion in high-temperature shear flows in a
between flame and turbulence under low computational cost, scramjet combustor. Chin J Aeronaut 2018;31(9):1870–9.
and the program implementation of the combined turbulent 3. Balakrishnan G, Williams FA. Turbulent combustion regimes for
hypersonic propulsion employing hydrogen-air diffusion flames. J
combustion model was based on OpenFOAM software. In
Propul Power 2012;10(3):434–7.
the numerical simulation of the Sandia series of turbulent jet
4. Durbin PA. Some recent developments in turbulence closure
flames using this model, we found that the LES model can modeling. Annu Rev Fluid Mech 2017;50:77–103.
Eulerian stochastic field method coupled with tabulated chemistry 133

5. Gonzalez-Juez ED, Kerstein AR, Ranjan R, Menon S. Advances methane/air diffusion flame (Sandia D). Combust Flame 2006;145
and challenges in modeling high-speed turbulent combustion in (1–2):88–104.
propulsion systems. Prog Energy Combust Sci 2017;60:26–67. 22. Jones WP, Navarro-Martinez S, Röhl O. Large eddy simulation of
6. Pope SB. Small scales, many species and the manifold challenges hydrogen auto-ignition with a probability density function
of turbulent combustion. Proc Combust Inst 2013;34(1):1–31. method. Proc Combust Inst 2007;31(2):1765–71.
7. Van Oijen JA, De Goey LPH. Modelling of premixed laminar 23. Consalvi JL, Nmira F, Burot D. Simulations of sooting turbulent
flames using flamelet-generated manifolds. Combust Sci Technol jet flames using a hybrid flamelet/stochastic eulerian field method.
2000;161(1):113–37. Combust Theor Model 2017;20(2):1–37.
8. Bekdemir C, Somers LMT, De Goey LPH. Modeling diesel engine 24. Nik MB, Yilmaz SL, Givi P, Sheikhi MRH, Pope SB. Simulation
combustion using pressure dependent flamelet generated mani- of sandia flame d using velocity-scalar filtered density function.
folds. Proc Combust Inst 2011;33(2):2887–94. AIAA J 2010;48(7):1513–22.
9. Donini A, Bastiaans R, Van Oijen JA, De Goey LPH. Numerical 25. Xu X, Chen YL, Wang H. Detailed numerical simulation of
simulations of a turbulent high-pressure premixed cooled jet flame thermal radiation influence in Sandia flame D. Int J Heat Mass
with the flamelet generated manifolds technique. J Eng Gas Transf 2006;49(13–14):2347–55.
Turbines Power 2015;137(7):071501. 26. Sheikhi MRH, Drozda TG, Givi P, Jaberi FA, Pope SB. Large
10. Ma L, Roekaerts D. Modeling of spray jet flame under mild eddy simulation of a turbulent nonpremixed piloted methane jet
condition with non-adiabatic fgm and a new conditional droplet flame (Sandia flame D). Proc Combust Inst 2004;30:549–56.
injection model. Combust Flame 2016;165:402–23. 27. Lysenko DA, Ertesvåg IS, Rian KE. Numerical simulations of the
11. Chen ZX, Doan NAK, Ruan S, Langella I, Swaminathan N. A Sandia flame D using the eddy dissipation concept. Flow Turbul
priori investigation of subgrid correlation of mixture fraction and Combust 2014;93(4):665–87.
progress variable in partially premixed flames. Combust Theor 28. Elbahloul S, Rigopoulos S. Rate-controlled constrained equilib-
Model 2018;22(5):862–82. rium (RCCE) simulations of turbulent partially premixed flames
12. Bray KNC, Champion M, Libby PA, Swaminathan N. Finite rate (Sandia D/E/F) and comparison with detailed chemistry. Combust
chemistry and presumed PDF models for premixed turbulent Flame 2015;162(5):2256–71.
combustion. Combust Flame 2006;146(4):665–73. 29. Renzo MD, Coclite A, De Tullio MD, Palma PD, Pascazio G.
13. Ge HW, Gutheil E. Simulation of a turbulent spray flame using LES of the Sandia flame D using an FPV combustion model.
coupled pdf gas phase and spray flamelet modeling. Combust Energy Procedia 2015;82:402–9.
Flame 2008;153(1):173–85. 30. Jones WP, Prasad VN. Large eddy simulation of the Sandia flame
14. Avdić A, Kuenne G, Janicka J. Flow physics of a bluff-body swirl series (D-F) using the eulerian stochastic field method. Combust
stabilized flame and their prediction by means of a joint eulerian Flame 2010;157(9):1621–36.
stochastic field and tabulated chemistry approach. Flow Turbul 31. Duan Y, Xia Z, Ma L, Luo Z. Numerical simulation of the Sandia
Combust 2016;97(4):1–26. flame d using the ESF method coupled with FGM model. Cluster
15. Barlow RS, Frank JH. Effects of turbulence on species mass Comput [Internet]. 2018 March 30 [cited 2018 August 20];
fractions in methane/air jet flames. Symp Combust 1998;27 Available from: https://doi.org/10.1007/s10586-018-2501-8.
(1):1087–95. 32. Sandia.gov [Internet]. New Mexico: Sandia National Laboratory;
16. Ma L, Roekaerts D. Structure of spray in hot-diluted coflow [updated 2003 January 16; cited 2018 August 16]. Available from:
flames under different coflow conditions: a numerical study. https://www.sandia.gov/TNF/DataArch/FlameD.html.
Combust Flame 2016;172:20–37. 33. Ge Y, Cleary MJ, Klimenko AY. A comparative study of Sandia
17. Ma L, Roekaerts D. Numerical study of the multi-flame structure flame series (D-F) using sparse-lagrangian mmc modelling. Proc
in spray combustion. Proc Combust Inst 2017;36(2):2603–13. Combust Inst 2013;34(1):1325–32.
18. Ma L. Computational modeling of turbulent spray combustion 34. Vreman AW, Albrecht BA, Van Oijen JA, De Goey LPH,
539 [dissertation]. Delft: Delft University of Technology; 2016. Bastiaans RJM. Premixed and nonpremixed generated manifolds
19. Bilger RW. The structure of turbulent nonpremixed flames. Symp in large-eddy simulation of sandia flame d and f. Combust Flame
Combust 1989;22(1):475–88. 2008;153(3):394–416.
20. Jones WP, Navarro-Martinez S. Numerical study of n-heptane 35. Jones WP, Navarro-Martinez S. Study of hydrogen auto-ignition
auto-ignition using les-pdf methods. Flow Turbul Combust 2009;83 in a turbulent air co-flow using a large eddy simulation approach.
(3):407–23. Comput Fluids 2008;37(7):802–8.
21. Mustata R, Valiño L, Jiménez C, Jones WP, Bondi S. A 36. Xu J, Pope SB. PDF calculations of turbulent nonpremixed flames
probability density function eulerian monte carlo field method with local extinction. Combust Flame 2000;123(3):281–307.
for large eddy simulations: application to a turbulent piloted

You might also like