You are on page 1of 74

PERGAMON Progress in Energy and Combustion Science 28 (2002) 193±266

www.elsevier.com/locate/pecs

Turbulent combustion modeling


Denis Veynante a,*, Luc Vervisch b
a
Laboratoire E.M2.C., CNRS et Ecole Centrale Paris, Grande Voie des Vignes, 92295 ChaÃtenay-Malabry Cedex, France
b
Institut National des Sciences AppliqueÂes de Rouen, UMR CNRS 6614/CORIA, Campus du Madrillet,
Avenue de l'Universite ÐBP 8, 76801 Saint Etienne du Rouvray Cedex, France
Received 4 November 2000; accepted 12 October 2001

Abstract
Numerical simulation of ¯ames is a growing ®eld bringing important improvements to our understanding of combustion. The
main issues and related closures of turbulent combustion modeling are reviewed. Combustion problems involve strong coupling
between chemistry, transport and ¯uid dynamics. The basic properties of laminar ¯ames are ®rst presented along with the major
tools developed for modeling turbulent combustion. The links between the available closures are illuminated from a generic
description of modeling tools. Then, examples of numerical models for mean burning rates are discussed for premixed turbulent
combustion. The use of direct numerical simulation (DNS) as a research instrument is illustrated for turbulent transport
occurring in premixed combustion, gradient and counter-gradient modeling of turbulent ¯uxes is addressed. Finally, a review
of the models for non-premixed turbulent ¯ames is given. q 2002 Published by Elsevier Science Ltd.
Keywords: Premixed ¯ames; Non-premixed ¯ames; Turbulent combustion; Scalar turbulent transport; Direct numerical simulation; Reynolds
averaged Navier±Stokes modeling

Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
2. Balance equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
2.1. Instantaneous balance equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
2.2. Reynolds and Favre averaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
2.3. Favre averaged balance equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
2.4. Filtering and Large Eddy Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
3. Major properties of premixed, non-premixed and partially premixed ¯ames . . . . . . . . . . . . . . . . . . . . 200
3.1. Laminar premixed ¯ames . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
3.2. Laminar diffusion ¯ames . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
3.3. Partially premixed ¯ames . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
4. A direct analysis: Taylor's expansion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
5. Scales and diagrams for turbulent combustion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
5.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
5.2. Turbulent premixed combustion diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
5.2.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
5.2.2. Combustion regimes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
5.2.3. Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
5.3. Non-premixed turbulent combustion diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
5.3.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210

* Corresponding author. Tel.: 133-1-41-13-10-80; fax: 133-1-47-02-80-35.


E-mail address: denis@em2c.ecp.fr (D. Veynante).

0360-1285/02/$ - see front matter q 2002 Published by Elsevier Science Ltd.


PII: S 0360-128 5(01)00017-X
194 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

6. Tools for turbulent combustion modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212


6.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
6.2. Scalar dissipation rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
6.3. Geometrical description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
6.3.1. G-®eld equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
6.3.2. Flame surface density description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
6.3.3. Flame wrinkling description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
6.4. Statistical approaches: probability density function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
6.4.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
6.4.2. Presumed probability density functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
6.4.3. Pdf balance equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
6.4.4. Joint velocity/concentrations pdf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
6.4.5. Conditional moment closure (CMC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
6.5. Similarities and links between the tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
7. Reynolds-averaged models for turbulent premixed combustion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
7.1. Turbulent ¯ame speed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
7.2. Eddy-break-up model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
7.3. Bray±Moss±Libby model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
7.3.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
7.3.2. BML model analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
7.3.3. Recovering mean reaction rate from tools relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
7.3.4. Reynolds and Favre averaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
7.3.5. Conditional averagingÐcounter-gradient turbulent transport . . . . . . . . . . . . . . . . . . . . . 227
7.3.6. Extensions to partially premixed combustion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
7.4. Models based on the ¯ame surface area estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
7.4.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
7.4.2. Algebraic expressions for the ¯ame surface density S . . . . . . . . . . . . . . . . . . . . . . . . . . 229
7.4.3. Flame surface density balance equation closures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
7.4.4. Analysis of the ¯ame surface density balance equation . . . . . . . . . . . . . . . . . . . . . . . . . 234
7.4.4.1. Turbulent transport. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
7.4.4.2. Strain rate induced by the mean ¯ow ®eld, AT. . . . . . . . . . . . . . . . . . . . . . . . 234
7.4.4.3. Strain rate aT due to turbulent motions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
7.4.4.4. Propagation and curvature terms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
7.4.5. Flame stabilization modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
7.4.6. A related approach: G-equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
8. Turbulent transport in premixed combustion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
8.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
8.2. Direct numerical simulation analysis of turbulent transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
8.2.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
8.2.2. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
8.3. Physical analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
8.3.1. Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
8.4. External pressure gradient effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
8.5. Counter-gradient transportÐexperimental results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
8.6. To include counter-gradient turbulent transport in modeling . . . . . . . . . . . . . . . . . . . . . 243
8.7. Towards a conditional turbulence modeling? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
9. Reynolds averaged models for non-premixed turbulent combustion . . . . . . . . . . . . . . . . . . . . . . 245
9.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
9.2. Fuel/air mixing modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
9.2.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
9.2.2. Balance equation and simple relaxation model for x~ . . . . . . . . . . . . . . . . . . . 246
9.3. Models assuming in®nitely fast chemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
9.3.1. Eddy dissipation model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
9.3.2. Presumed pdf: in®nitely fast chemistry model . . . . . . . . . . . . . . . . . . . . . . . . 247
9.4. Flamelet modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
9.4.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 195

9.4.2. Flame structure in composition space, Y iSLFM …Z p ; xp † ................... 249


9.4.3. Mixing modeling in SLFM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
9.4.4. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
9.5. Flame surface density modeling, coherent ¯ame model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
9.6. MIL model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
9.7. Conditional moment closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
9.8. Pdf modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
9.8.1. Turbulent micromixing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
9.8.2. Linear relaxation model, IEM/LMSE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
9.8.3. GIEM model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
9.8.4. Stochastic micromixing closures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
9.8.5. Interlinks PDF/¯ame surface modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
9.8.6. Joint velocity/concentrations pdf modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
10. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260

1. Introduction between the breakdown of the liquid sheets, the vapor-


ization of the liquid, turbulent mixing, and droplet
The number of combustion systems used in power combustion.
generation and transportation industries is growing ² Radiative heat transfer is generated within the ¯ame by
rapidly. This induces pollution and environmental problems some species and carbon particles resulting from soot
to become critical factors in our societies. The accurate formation and transported by the ¯ow motion. In
control of turbulent ¯ames therefore appears as a real furnaces, walls also interact with combustion through
challenge. radiative transfer.
Computing is now truly on par with experiment and
theory as a research tool to produce multi-scale information Turbulent combustion modeling is therefore a very
that is not available by using any other technique. Computa- broad subject. All the aspects of the problem are not
tional ¯uid dynamics (CFD) is ef®ciently used to improve addressed in the present review. We will only focus on
the design of aerodynamical systems, and today no real the closure schemes developed and used to understand
progress in design can be made without using CFD. With and calculate turbulent transport and mean burning rates
the same objectives, much work has been devoted to turbu- in turbulent ¯ames. The detail of chemistry, its reduc-
lent combustion modeling, following a variety of tion, tabulation, etc. are not considered. However, the
approaches and distinct modeling strategies. This paper is links existing between the models are identi®ed, showing
intended to provide a generic review of these numerical similarities which are sometimes much stronger than is
models. usually thought.
A wide range of coupled problems are involved in turbu- Numerical modeling of ¯ames is developed from the
lent ¯ames: following steps (Fig. 1):

² The ¯uid mechanical properties of the combustion system ² Under assumptions such as the high activation energy
must be well known to carefully describe the mixing limit, asymptotic analysis [1±4] allows the analytical
between reactants and, more generally, all transfer determination of ¯ame properties in well-de®ned model
phenomena occurring in turbulent ¯ames (heat transfer, problems (ignition, propagation of ¯ame front, instabil-
molecular diffusion, convection, turbulent transport, ities and acoustics, etc.). This approach, limited to simpli-
etc.). ®ed situations, leads to analytical results exhibiting
² Detailed chemical reaction schemes are necessary to esti- helpful scale factors (dimensionless numbers) and
mate the consumption rate of the fuel, the formation of major ¯ame behaviors. Asymptotic analysis is particu-
combustion products and pollutant species. A precise larly well suited to perform quantitative comparison
knowledge of the chemistry is absolutely required to between various phenomena.
predict ignition, stabilization or extinction of reaction ² Simpli®ed experiments are useful to understand the basic
zones together with pollution. properties of combustion (laminar ¯ames, ¯ame/vortex
² Two (liquid fuel) and three (solid fuel) phase systems may interactions, etc.) [5,6]. These experiments are accom-
be encountered. Liquid fuel injection is a common pro- panied by numerical simulations of laminar ¯ames incor-
cedure and the three-dimensional spatial distribution of porating complex chemistry and multi-species transports
gaseous reactants depends on complex interactions along with radiative heat losses [7].
196 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

Fig. 1. Combustion modeling steps.

² For given chemistry and transport model, in direct expansion for closing the mean burning rate (Section
numerical simulation (DNS) all the scales of the turbu- 4), the physical analysis leading to turbulent combustion
lence (time and length) are calculated without resorting to diagrams is developed (Section 5). Then, modeling tools
closures for turbulent ¯uxes and mean burning rate. available to derive turbulent combustion models are
Turbulent ¯ames are analyzed in simple con®gurations described and the relations between a priori quite differ-
to extract data impossible to measure in experiments, and ent formalisms are established (Section 6). The next
to isolate some speci®c phenomena (heat release, Lewis three sections are devoted to combustion modeling in
number¼) [8±12]. the context of Reynolds Averaged Navier±Stokes
(RANS) equations. For premixed turbulent combustion,
Because of the large number of degrees of freedom involved we review the available closures for the mean reaction
in turbulent combustion, a full DNS of a practical system rate (Section 7) and turbulent transport (Section 8). In a
cannot be performed and averaging techniques leading to subsequent section (Section 9), the modeling of the
unclosed equations are necessary. Models for turbulent ¯ames mean burning rate in non-premixed turbulent ¯ames is
are then developed: closure techniques are proposed for addressed.
unknown terms found in exact averaged balance equations.
Once the models have been implemented in numerical
codes, validation procedures are required. The numerical 2. Balance equations
modeling is validated against measurements obtained from
experiments. Con®gurations as close as possible to actual 2.1. Instantaneous balance equations
industrial systems are chosen for these tests. Then, the ultimate
The basic set of balance equations comprises the classical
step is the simulation of a real combustion device.
Navier±Stokes, species and energy transport equations.
The decomposition discussed in Fig. 1 is quite formal.
These instantaneous local balance equations are, using the
Turbulent combustion modeling is actually a continuous
classical lettering [13±15]:
ring between theoretical studies to analyze combustion,
understand ¯ames and improve models, implementation of
² Mass:
these models into CFD, experimental measurements
and comparison between these experimental data and the 2r 2ruj
1 ˆ0 …1†
numerical results. 2t 2xj
Following a short presentation of the balance equa- ² Momentum (i ˆ 1,2,3):
tions for reactive ¯ows (Section 2), a ®rst part is
devoted to a brief description of laminar ¯ames (Section 2rui 2ruj ui 2p 2tij
1 ˆ2 1 1 Fi …2†
3). After a presentation of the unsuccessful Taylor's 2t 2xj 2xi 2xj
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 197

where t ij denotes the viscous force tensor and Fi a body Then, the Lewis number Lek of the species k, comparing
force. thermal and mass diffusivities is introduced:
² Species (N species with k ˆ 1; ¼; N):   !
Sck l
Lek ˆ ˆ …10†
2rYk 2ruj Yk 2Jkj Pr rCp Dk
1 ˆ2 1 v_ k …3†
2t 2xj 2xj
Under the assumption of unity Lewis number, the enthalpy
where Jkj is the molecular diffusive ¯ux of the species k diffusive ¯ux (Eq. (8)) is simpli®ed and mass fraction and
and v_ k the mass reaction rate of this species per unit enthalpy balance equations are formally identical if 2P=2t;
volume. ui tij and ui Fj are negligible (low Mach number assumption)
² Total enthalpy ht ˆ h 1 ui ui =2 : [16]. This assumption is generally made to simplify tur-
bulent ¯ame modeling, especially in premixed ¯ames
2rht 2ruj ht 2p 2
1 ˆ 1 …Jh 1 ui tij † 1 uj Fj …4† when species mass fractions and temperature are assumed
2t 2xj 2t 2xj j to be equivalent variables. Nevertheless, thermo-diffusive
where ui tij and ui Fj denote respectively the power due to instabilities occur in premixed systems when the Lewis
number is lower than unity (for example for hydrogen).
viscous and body forces.
One direct consequence of these instabilities is an increase
of the premixed ¯ame area and of the global reaction rate
These equations are closed by expressions for the species [14,17].
molecular ¯uxes and the viscous forces. In practical situa-
tions, all ¯uids are assumed to be Newtonian, i.e. the viscous
2.2. Reynolds and Favre averaging
tensor is given by the Newton law:
!  
2ui 2uj 2 2uk Unfortunately, the full numerical solution of the instanta-
tij ˆ ml 1 2 ml dij …5† neous balance equations is limited to very simpli®ed cases
2xj 2xi 3 2xk
(DNS [9,11,12]), where the number of time and length
where the molecular viscosity ml depending on the ¯uid scales present in the ¯ow is not too great. To overcome
properties is introduced. d ij is the Kronecker symbol. this dif®culty, an additional step is introduced by averaging
Species molecular diffusivities are generally described the balance equations to describe only the mean ¯ow ®eld
using the Fick law, assuming a major species: (local ¯uctuations and turbulent structures are integrated in
mean quantities and these structures have no longer to be
ml 2Yk
Jkj ˆ 2 …6† described in the simulation). Each quantity Q is split into a
Sck 2xj
mean Q and a deviation from the mean denoted by Q 0 :
Sck is the Schmidt number of the species k, de®ned as:
Q ˆ Q 1 Q 0 with Q 0 ˆ 0 …11†
ml
Sck ˆ …7†
rDk Then, the previous instantaneous balance equations may be
ensemble averaged to derive transport equations for the
Dk is the molecular diffusivity of the species k relative to the  This classical Reynolds averaging tech-
mean quantity Q:
major species.
nique, widely used in non-reacting ¯uid mechanics, brings
More complex expressions may be used to describe multi-
unclosed correlations such as u 0 Q 0 that are unknown and
species molecular diffusion. Soret effect (species diffusion
must be modeled. The numerical procedure is called
under temperature gradients) and molecular transport due to
Reynolds Averaged Navier±Stokes (RANS) modeling.
pressure gradients are usually neglected. Enthalpy diffusion
In turbulent ¯ames, ¯uctuations of density are observed
is described according to the Fourier law:
" # because of the thermal heat release, and Reynolds averaging
XN  
induces some additional dif®culties. Averaging the mass
h ml 2h Pr 2Yk
Jj ˆ 2 1 2 1 hk …8† balance equation leads to:
Pr 2xj kˆ1
Sck 2xj

The Prandtl number Pr compares the diffusive transport 2r 2  


1 rui 1 r 0 u 0i ˆ 0 …12†
of momentum (viscous forces) and temperature. In the 2t 2xi
previous expressions, radiative heat transfer and Dufour
where the velocity/density ¯uctuations correlation r 0 u 0i
effect (enthalpy diffusion under mass fraction gradients)
appears. To avoid the explicit modeling of such correlations,
are neglected. The Prandtl number is written as a function
a Favre (mass weighted) [18] average Q~ is introduced and
of the thermal diffusivity l and the constant pressure
any quantity is then decomposed into Q ˆ Q~ 1 Q 00 :
speci®c heat Cp :
  
mi Cp rQ f r Q 2 Q~
Pr ˆ …9† Q~ ˆ ; 00
Q ˆ ˆ0 …13†
l r r
198 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

The Favre averaged continuity equation: an approximation for this term. The closure may be done
2r 2r u~i directly or by deriving balance equations for these
1 ˆ0 …14† Reynolds stresses. However, most combustion works
2t 2xi
are based on turbulence modeling developed for non-
is then formally identical to the Reynolds averaged conti- reacting ¯ows, such as k±1; simply rewritten in terms
nuity equation for constant density ¯ows. This result is true of Favre averaging, and heat release effects on the
for any balance equations (momentum, energy, mass frac- Reynolds stresses are generally not explicitly included.
tions, etc.). Nevertheless, Favre averaging is only a math- ² Species …ug
00 Y 00 † and temperature …ug
00 T† turbulent ¯uxes.
j k j
ematical formalism: These ¯uxes are usually closed using a gradient transport
hypothesis:
² There is no simple relation between Favre, Q~ and
Reynolds, Q;  averages. A relation between Q~ and Q mt 2Y~ k
requires the knowledge of density ¯uctuations correla-
r ug
00 Y 00 ˆ 2
j k …20†
Sckt 2xj
tions r 0 Q 0 remaining hidden in Favre averaging (see
Section 7.3.4): where m t is the turbulent viscosity, estimated from the
turbulence model, and Sckt a turbulent Schmidt number
r Q~ ˆ r Q 1 r 0 Q 0 …15† for the species k.
Nonetheless, theoretical and experimental works have
² Comparisons between numerical simulations, providing demonstrated that this assumption may be wrong in
Favre averaged quantities Q;~ with experimental results some premixed turbulent ¯ames and counter-gradient
are not obvious. Most experimental techniques determine turbulent transport may be observed [19,20] (i.e. in an
Reynolds averaged data Q and differences between Q~ and opposite direction compared to the one predicted by Eq.
Q may be signi®cant (Section 7.3.4 and Fig. 17). (20), see Sections 7.3.5 and 8).
² Laminar diffusive ¯uxes Jkj ; Jhj ; etc. are usually small
compared to turbulent transport, assuming a suf®ciently
2.3. Favre averaged balance equations large turbulence level (large Reynolds numbers limit).
² Species chemical reaction rates v_ k : Turbulent combus-
Averaging instantaneous balance equations yields:
tion modeling generally focuses on the closure of these
² Mass: mean burning rates.

2r 2r u~j These equations, closed with appropriate models, allow
1 ˆ0 …16†
2t 2xj only for the determination of mean quantities, that may
differ from the instantaneous ones. Strong unsteady mixing
² Momentum (i ˆ 1,2,3): effects, resulting from the rolling up of shear layers, are
observed in turbulent ¯ames, and the knowledge of steady
2r u~i 2r u~j u~i 2r ug
00 u 00
i j 2p 2t ij statistical means is indeed not always suf®cient to describe
1 ˆ2 2 1 1 F i …17† turbulent combustion. An alternative is to use large eddy
2t 2xj 2xj 2xi 2xj
simulation (LES).

² Chemical species (for N species, k ˆ 1; ¼; N): 2.4. Filtering and Large Eddy Simulation

2r Y~ k 2r u~j Y~ k 2r ug


00 Y 00
j k 2Jkj The objective of Large Eddy Simulation (LES) is to ex-
1 ˆ2 2 1 v_ k …18† plicitly compute the largest structures of the ¯ow (typically
2t 2xj 2xj 2xj
the structures larger than the computational mesh size),
² Total enthalpy h~t : while the effects of the smaller one are modeled. LES is
widely studied in the context of non-reactive ¯ows [21±
2r h~t 2r u~j h~t 24], its application to combustion modeling is still at an
1
2t 2xj early stage [25]. As in RANS, the complex coupling
…19† between micromixing and chemical reactions occurring at
2r ug
00 h 00
j t 2p 2  h  unresolved scales needs models, however, LES possesses
ˆ2 1 1 Jj 1 ui tij 1 uj Fj
2xj 2t 2xj some attractive properties:

The objective of turbulent combustion modeling is to ² Large structures in turbulent ¯ows generally depend on
propose closures for the unknown quantities appearing in the geometry of the system. On the contrary, smaller
the averaged balance equations, such as: scales feature more universal properties. Accordingly,
turbulence models may be more ef®cient when they
² Reynolds stresses ug
00 u 00 : The turbulence model provides
i j have to describe only the smallest structures.
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 199

² Turbulent mixing controls most of the global ¯ame prop- All these ®lters are normalized:
erties. In LES, unsteady large scale mixing (between Z1 1 Z1 1 Z1 1
fresh and burnt gases in premixed ¯ames or between F…x1 ; x2 ; x3 † dx1 ; dx2 ; dx3 ˆ 1 …25†
21 21 21
fuel and oxidizer in non-premixed burners) is simulated,
instead of being averaged. In combusting ¯ows, a mass-weighted, Favre ®ltering, is
² Most reacting ¯ows exhibit large scale coherent struc- introduced as:
tures [26], which are also especially observed when Z
combustion instabilities occur. These instabilities result ~
r Q…x† ˆ rQ…xp †F…x 2 xp †dxp …26†
from the coupling between heat release, the hydrody-
namic ¯ow ®eld and acoustic waves. They need to be Instantaneous balance equations (Section 2) may be ®ltered
avoided because they induce noise variations of the to derived balance equations for the ®ltered quantities Q or
main properties of the system, large heat transfers ~ This derivation should be carefully conducted:
Q:
and, even in some extreme cases, the destruction of Any quantity Q may be decomposed into a ®ltered
the device.LES may be a powerful tool to predict the component Q and a `¯uctuating' component Q 0 ; according
occurrence of such instabilities [27] and consequently to: Q ˆ Q 1 Q 0 : But, in disagreement with classical
improve passive or active control systems. Reynolds averaging (ensemble average), Q 0 may be non-
² With LES, large structures are explicitly computed and zero:
instantaneous fresh and burnt gases zones, with different Z 
turbulence characteristics (Section 8.7) are clearly iden-  p † F…x 2 xp †dxp
Q 0 …x† ˆ Q…xp † 2 Q…x
ti®ed. This may help to describe some properties of the
¯ame/turbulence interaction. Z Z
ˆ Q…xp †F…x 2 xp †dxp 2  p †F…x 2 xp †dxp
Q…x
In LES, the relevant quantities Q are ®ltered in the spec-
tral space (components greater than a given cut-off 
ˆ Q…x† 2 Q…x† …27†
frequency are suppressed) or in the physical space (weighted
where
averaging in a given volume). The ®ltered operation is
de®ned by: ZZ 
Q…x† ˆ Q…x1 †F…xp 2 x1 †dx1 F…x 2 xp †dxp
Z

Q…x† ˆ Q…xp †F…x 2 xp † dxp …21† ZZ
ˆ 
Q…x1 †F…xp 2 x1 †F…x 2 xp †dx1 dxp ± Q…x†
…28†
where F is the LES ®lter. Standard ®lters are:
To summarize:
² A cut-off ®lter in the spectral space:

Q ± Q; Q 0 ± 0; Q~~ ± Q;
~ f00 ± 0
Q …29†
( ~
1 if k # p=D  0 ~ ~
The relations used in RANS Q ˆ Q; Q ˆ 0; Q ˆ Q; Q ˆ 0 f 00

F…k† ˆ …22†
0 otherwise are true when a cut-off ®lter in the spectral space is chosen
(Eq. (22)). Then, all the frequency components greater than
a cut-off wave number kc ˆ p=D vanish.
where k is the spatial wave number. This ®lter preserves
The derivation of balance equations for the ®ltered quan-
the length scales greater than the cut-off length scale 2D:
tities Q or Q~ requires the exchange of ®ltering and differ-
² A box ®lter in the physical space:
entiation operators. This exchange is theoretically valid only
( under restrictive assumptions and is wrong, for example,
1=D3 if uxi u # D=2; i ˆ 1; 2; 3
F…x† ˆ F…x1 ; x2 ; x3 † ˆ when the ®lter size varies (®lter size corresponding to the
0 otherwise mesh size, depending on the spatial location). This point has
…23† been carefully investigated [28]. In most simulations, the
uncertainties due to this operator exchange are neglected
where …x1 ; x2 ; x3 † are the spatial coordinates of the loca- and assumed to be incorporated in subgrid scale modeling.
tion x. This ®lter corresponds to an averaging of the Filtering the instantaneous balance equations leads to
quantity Q over a box of size D . equations formally similar to the Reynolds averaged balance
² A Gaussian ®lter in the physical space: equations given in Section 2.3:

F…x† ˆ F…x1 ; x2 ; x3 † ² mass:


 3=2  
6 6  2r
1
2r u~ j
ˆ0 …30†
ˆ exp 2 2 x21 1 x22 1 x23 …24†
pD2 D 2t 2xj
200 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

behavior of the ®ltered ®elds. Compared to direct numerical


simulations (DNS), part of the information contained in the
unresolved scales is lost (and should be modeled).
Compared to RANS, LES provides valuable information
on the large resolved motions.
Either using RANS or LES, combustion occurs at the
unresolved scales of the computations. Then, the basic
tools and formalism of turbulent combustion modeling are
somehow the same for both techniques. Most of the RANS
combustion models can be modi®ed and adapted to LES
modeling. The scope of this review is limited to RANS.

3. Major properties of premixed, non-premixed and


partially premixed ¯ames

3.1. Laminar premixed ¯ames

The structure of a laminar premixed ¯ame is displayed in


Fig. 2. Structure of a laminar plane premixed ¯ame. Fig. 2. Fresh gases (fuel and oxidizer mixed at the molecular
level) and burnt gases (combustion products) are separated
by a thin reaction zone (typical thermal ¯ame thicknesses,
² momentum (for i ˆ 1; 2; 3): d l, are about 0.1±1 mm). A strong temperature gradient is
observed (typical ratios between burnt and fresh gases
2r u~i 2r u~j u~i
1 temperatures are about 5±7). Another characteristic of a
2t 2xj premixed ¯ame is its ability to propagate towards the
…31†
2 h  i 2p 2t ij fresh gases. Because of the temperature gradient and the
ˆ2 r ug ~ i u~ j 2
i uj 2 u 1 1 F i corresponding thermal ¯uxes, fresh gases are preheated
2xj 2xi 2xj
and then start to burn. The local imbalance between diffu-
sion of heat and chemical consumption leads to the propa-
² Chemical species (N species, k ˆ 1; ¼; N): gation of the front. The propagation speed SL of a laminar
¯ame depends on various parameters (fuel and oxidizer
2r Y~ j 2r u~j Y~ k 2 h g i
1 ˆ2 r uj Yk 2 u~j Y~ k 1 v_ k …32† compositions, fresh gases temperature, etc.) and is about
2t 2xj 2xj 0.1±1 m/s. There is an interesting relation between the ther-
mal ¯ame thickness, d l, the laminar ¯ame speed, SL and the
² Total enthalpy ht ˆ h 1 ui ui =2 kinematic viscosity of the fresh gases, n :

2r h~t 2r u~j h~t 2 h g i 2p dl SL


1 ˆ2 r uj ht 2 u~j h~t 1 Ref ˆ
n
<4 …34†
2t 2xj 2xj 2t
2  h  where the thermal thickness d l corresponds to a temperature
1 Jj 1 ui tij 1 uj Fj …33† jump of 98% of the temperature difference between fresh
2xj
and fully burnt products. The ¯ame Reynolds number, Ref,
where Q and Q~ denote LES ®ltered quantities instead of is then almost constant. This relation, derived, for example,
ensemble means. from the Zeldovich/Frank-Kamenetskii (ZFK) theory
[14,15] is often implicitly used in theoretical derivation of
The unknown quantities are: models for premixed turbulent combustion.
For a simple one-step irreversible chemical scheme:

² Unresolved Reynolds stresses ug i uj 2 u ~i u~j ; requiring a
reactants ! products
subgrid scale turbulence model. 
² Unresolved species ¯uxes ug j Yk 2 u ~j Y~ k and enthalpy the ¯ame is described using a progress variable c, such as

¯uxes ug ~j h~t :
j ht 2 u c ˆ 0 in the fresh gases and c ˆ 1 in the fully burnt ones.
² Filtered laminar diffusion ¯uxes Jkj ; Jhj : This progress variable may be de®ned as a reduced tempera-
² Filtered chemical reaction rate v_ k : ture or a reduced mass fraction:

These ®ltered balance equations, coupled to subgrid scale T 2 Tu YF 2 YFu


cˆ or cˆ …35†
models may be numerically solved to simulate the unsteady Tb 2 Tu YFb 2 YFu
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 201

progress variable:
2rc  
1 7´ ruc ˆ 7´ rD7c 1 v_ …36†
2t
The previous Eq. (36) may be recast in a propagative form,
introducing the displacement speed v of the iso-c surface:
  
2c 1 7´ rD7c 1 v_
1 u´7c ˆ j7cj ˆ wu7cu …37†
2t r u7cu
|‚‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚‚}
displacement speed

Eq. (36) then describes the displacement of an iso-c surface


with the displacement speed w measured relative to the ¯ow.
Introducing the vector n normal to the iso-c surface and
pointing towards fresh gases …n ˆ 27c=u7cu†; the displace-
ment speed may be split into three contributions:
1  1
wˆ nn : 7 rD7c 2 D7´n 1 v_ …38†
rj7cj rj7cj

1 2  1
Fig. 3. Generic structure of a laminar diffusion ¯ame. wˆ2 rDu7cu 2 D7´n 1 v_
ru7cu 2n ru7cu
|‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚} |‚{z‚} |‚‚{z‚‚}
wn wc wr
where T, Tu and Tb are respectively the local, the unburnt gas
where 2=2n ˆ n´7 denotes a normal derivative. wn corre-
and the burnt gas temperatures. YF, YFu and YFb are respec-
sponds to molecular diffusion normal to the iso-c surface,
tively the local, unburnt gas and burnt gas fuel mass frac-
wc is related to the curvature 7´n of this surface and corre-
tions. YFb is non-zero for a rich combustion (fuel in excess).
sponds to tangential diffusion. wr is due to the reaction rate
For a unity Lewis number (same molecular and thermal
v_ : In a ®rst approximation, w n 1 wr may be modeled with
diffusivities), without heat losses (adiabatic combustion)
the laminar ¯ame speed, SL, whereas wc incorporates wrink-
and compressibility effects, the two de®nitions (35) are
ling surface effects and may be expressed using Markstein
equivalent and mass and low Mach number energy balance
lengths [29].
equations reduce to a single balance equation for the
The propagation of reactive fronts has been the subject of
various developments and more discussion may be found in
Ref. [2] and references therein.

3.2. Laminar diffusion ¯ames

In laminar diffusion ¯ames, fuel and oxidizer are on both


sides of a reaction zone where the heat is released. The
burning rate is controlled by the molecular diffusion of the
reactants toward the reaction zone (Figs. 3 and 4). In a
counter-¯owing fuel and oxidizer ¯ame (Fig. 4), the amount
of heat transported away from the reaction zone is exactly
balanced by the heat released by combustion. A steady
planar diffusion ¯ame with determined thickness may be
observed in the vicinity of the stagnation point. Increasing
the jet velocity, quenching occurs when the heat ¯uxes
leaving the reaction zone are greater than the chemical
heat production. The structure of a steady diffusion ¯ame
therefore depends on ratios between characteristic times
representative of molecular diffusion and chemistry [30].
The thicknesses of the mixing zone and of the reaction
zone vary with these characteristic times. In opposition
with premixed ¯ames:

Fig. 4. Sketch of a counter-¯owing fuel and oxidizer diffusion ² Diffusion ¯ames do not bene®t from a self-induced propa-
¯ame. gation mechanism, but are mainly mixing controlled.
202 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

² The thickness of a diffusion ¯ame is not constant, but Other Schwab±Zeldovitch variables w (YF, T ) and w (YO, T )
depends on the local ¯ow properties. (conserved scalars) may be derived by combining the
variables (YF, T ) and (YO, T ). The mixture fraction and these
Let us consider the irreversible single step chemical reac- additional conserved scalars are linearly related and one
tion between fuel and oxidizer: may write:
F 1 sO ! …1 1 s†P
Y O …x; t† ˆ YO;o …1 2 Z…x; t††
where s is the mass stoichiometric coef®cient. |‚‚‚‚‚‚{z‚‚‚‚‚‚}
Mixing
In term of mass fraction, this chemical reaction may be
 
written: nO WO Cp
1 ‰Z…x; t†…TF;o 2 TO;o † 1 …TO;o 2 T…x; t††Š
nP dYP ˆ nF dYF 1 no dYo nF WF Q
|‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚}
Combustion
where d YF, d YO and d YP are the variations of fuel, oxidizer
…42†
and product mass fractions. n i are the stoichiometric molar
coef®cients of the reaction, Wi denotes the species molar
weight and v_ is the reaction rate. The balance equations
YF …x; t† ˆ YF;o Z…x; t†
for mass fractions and temperature are necessary to identify |‚‚‚{z‚‚‚}
Mixing
the properties of the ¯ame:
Cp …43†
2rYF
1 7´…ruY F † ˆ 7´…rD F 7Y F † 2 nF WF v_ 1 ‰Z…x; t†…TF;o 2 TO;o † 1 …TO;o 2 T…x; t††Š
2t Q
|‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚}
2rYO Combustion
1 7´…ruY O † ˆ 7´…rD O 7Y O † 2 nO WO v_
2t
! where TO,o and TF,o are the temperatures of the fuel and
 
2rT l Q oxidizer streams respectively. Using these algebraic re-
1 7´…ruT† ˆ 7´ 7T 1 n F WF v_ lations, the diffusion ¯ame is fully determined when the
2t CP Cp
mixture fraction Z and any one of T, YF, or YO is known.
The molecular diffusion is expressed using the Fick law, The conserved scalar approach may still be useful when
the chemical rate of fuel and oxidizer are respectively v_ F ˆ fuel and oxidizer molecular diffusivities differ, but an ad-
nF WF v_ and v_ O ˆ nO WO v_ : Q is the amount of heat released ditional mixture fraction:
by the combustion of an unit mass of fuel.
The internal structure of diffusion ¯ames is usually YF Y
F 2 O 11
discussed using the extent of mixing between fuel and oxidi- YF;o YO;o
zer. It is ®rst assumed that fuel and oxidizer molecular ZL ˆ …44†
F11
diffusivities are equal (i.e. DF ˆ DO ˆ D). Combining the
transport equation for YF and YO, a conserved scalar (quan- should be introduced, satisfying [31]:
tity that is not in¯uenced by the chemical reaction, a !
Schwab±Zeldovitch variable) w…YF ; YO † ˆ YF 2 YO =s is DZ 1 l
r ˆ 7´ 7Z L …45†
introduced, with the mass stoichiometric coef®cient s ˆ Dt L Cp
…nO WO =nF WF †: The mixture fraction Z is then de®ned by
normalizing w using values in the fuel and oxidizer streams. where:
Z evolves through the diffusive layer from zero (oxidizer) to
unity (fuel): L ˆ LeO …1 1 f†=…1 1 F† with F ˆ …Le O =LeF †f
Y Y …46†
f F 2 O 11
YF;o YO;o
Zˆ …39† where Lei is the Lewis number of the species i. The relations
f11
between Z and ZL are given in Table 1. When LeO ˆ LeF ;
YF,o is the fuel mass fraction in the fuel feeding stream. ZL ˆ Z: In experiments or in simulations involving complex
Similarly, YO,o is the oxidizer mass fraction in the oxidizer chemistry, the mixture fraction is de®ned from mass frac-
stream (for instance, in air, YO,o < 0.23), f is the chemical tions of atomic elements [32].
equivalence ratio: Mass fractions and temperature balance equations may be
sYF;o reorganized into a new frame where Z is one of the coordi-
fˆ …40† nates (see for instance Ref. [14] or Ref. [33]). A local or-
YO;o
thogonal coordinate system attached to the surface of
The mixture fraction follows the balance equation: stoichiometric mixture is introduced and the derivatives in
2rZ the stoichiometric plane are denoted '. For unity Lewis
1 7´…ruZ† ˆ 7´…rD7Z† …41† number and using Eq. (41), the species transport equation
2t
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 203

Table 1
Piecewise relations for in®nitely fast chemistry including non-unity Lewis number. Zst ˆ 1=…1 1 f† and ZLst ˆ 1=…1 1 F† (see Eq. (46)). The
subscript o denotes a quantity measured in pure fuel or oxidizer, Tf is the ¯ame temperature

Oxidizer side Fuel side


Z , Z st and ZL , ZLst Z . Zst and ZL . ZLst

Z ˆ ZL …1 1 F†=…1 1 f† Z ˆ …f…Z L …1 1 F† 2 1†=F 1 1†=…1 1 f†


YF ˆ 0 YO ˆ 0
YO ˆ YO;o …1 2 Z…1 1 f†† YF ˆ YF;o …Z…1 1 f† 2 1†=f
T ˆ …Tf 2 TO;o †Z…f 1 1† 1 TO;o T ˆ …TF;o 2 Tf †…Z…f 1 1† 2 1†=f 1 Tf

may be written: in the direction j through the ¯ame, using Eq. (50):
2Yi 22 Y Z1 1 Z1 v_ …x; Z†
r 1 ru' ´7 ' Yi ˆ rx 2i 1 7 ' ´…rD7 ' Yi † V_ i ˆ v_ i …j†dj ˆ i
dZ
2t 2Z 21 0 u7Zu
2 rD7 ' …lnu7Zu†´7 ' Yi 1 v_ i …47† Z1 2 2 Yi
ˆ2 rDu7Zu dZ …52†
In Eq. (47), x is the scalar dissipation rate of the mixture 0 2Z 2
fraction Z: Assuming that r , D and u7Zu do not vary across the ¯ame
! (this is typical of a ¯amelet assumption used in turbulent
2Z 2Z
xˆD ˆ Du7Zu 2 …48† combustion modeling where the ¯ame is assumed very
2xj 2xj
thin), V_ i becomes:
measuring the inverse of a diffusive time tx ˆ x21 : As this    
2Y i Zˆ1 2Y Zˆ1
time decreases, mass and heat transfers through the stoichio- V_ i < 2rDu7Zu ˆ 2 rD i …53†
2Z Zˆ0 2j Zˆ0
metric surface are enhanced.
When iso-Z surface curvatures are not too strong, the This last relation illustrates how the integrated reaction
gradients measured along the stoichiometric surface are rate of a species i is directly related to the molecular diffu-
smaller than the gradients in the direction Z perpendicular sion ¯ux of that species through the ¯ame.
to the stoichiometric surface, the balance equation for the Diffusion combustion is limited by two regimes corre-
mass fractions reduces to: sponding to pure mixing of the reactants and in®nitely fast
chemistry (Fig. 5). When the chemistry is in®nitely fast, the
2Yi 22 Y temperature depends on mixing through Z, but not on the
r ˆ rx 2i 1 v_ i …49†
2t 2Z rate of mixing x [35]. Then, piecewise relationships exist
Neglecting unsteady effects, the time derivative vanishes between Z, ZL, species mass fractions and temperature,
and for unity Lewis numbers, the ¯ame structure is fully summarized in Table 1. Eq. (43) provides the maximum
described by: ¯ame temperature Tf obtained when YF ˆ YO ˆ 0 and Z ˆ
Zst ˆ 1=…1 1 f†
22 Yi 22 T
rx 1 v_ i ˆ 0 and rx 1 v_ T ˆ 0 …50† Q
2Z 2 2Z 2 TF;o 1 TO;of 1 YF;o
Cp
showing that the chemical reaction rate is directly related to Tf ˆ
11f
the function T(Z, x ). Under these hypothesis, the diffusion
¯ame is completely determined as a function of the mixture In many combustion systems, the in®nitely fast chemistry
fraction Z and the scalar dissipation rate x (or 7Z): hypothesis cannot be invoked everywhere. For example in
ignition problems or in the vicinity of stabilization zones,
Y i ˆ Yi …Z; x†; T ˆ T…Z; x†
and more generally when large velocity gradients are found.
An expression for x (Z,t) and full solutions for various lami- The characterization of diffusion ¯ames from the in®nitely
nar ¯ames may be derived from asymptotic developments fast chemistry situation to the quenching limit is therefore of
[30,34], or solving Eq. (50) leading to Fig. 5. fundamental interest for turbulent combustion. The counter-
A coordinate j is de®ned across the one-dimensional ¯ow diffusion ¯ame (Fig. 4) is a generic con®guration well
¯amelet such as: suited to reproduce and to understand the structure and the
dZ extinction of laminar diffusion ¯ames. These extinction
ˆ 2n Z ´7Z ˆ u7Zu …51† phenomena have been theoretically described using asymp-
dj
totic developments [30,34,36]. A diffusive time tx < x21 st ˆ
where nZ denotes the normal vector to the iso-Z surfaces, …Du7Zu 2 †21
ZˆZst and a chemical time t c are combined to build a
pointing towards Z ˆ 0: The reaction rate may be integrated DamkoÈhler number Dap ˆ …tx =tc † < …tc xst †21 : The response
204 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

Fig. 5. Inner structure of non-premixed ¯ames. The distribution in mixture fraction space of fuel, oxidizer and temperature lies between the
in®nitely fast chemistry limit and the pure mixing case. The thickness of the diffusive zone ld is estimated from the scalar dissipation rate x at
the stoichiometric surface, whereas the characteristic thickness of the reaction zone lr depends on both ld and the DamkoÈhler number [12].

of the burning rate to variations of Da p leads to the so-called Moreover, for a given location within a diffusion ¯ame, by
`S' curve (Fig. 6) [14]. Starting from a situation where the traveling along the normal to the stoichiometric surface,
chemistry is fast, decreasing Da p (increasing x ) makes the T(Z) can be constructed and characterizes the combustion
burning rate and transport through the stoichiometric regime (i.e. fast or slow chemistry, Fig. 5). Many turbulent
surface greater, until chemistry cannot keep up with the combustion models are based on this description of diffusion
large heat ¯uxes. Then, extinction develops. The value of ¯ame; when the ¯ow is turbulent, T(Z) is replaced by the
the DamkoÈhler Dapq at the extinction point may be estimated mean temperature calculated for a given value of Z, i.e. for a
by quantifying the leakage of fuel (or oxidizer) through the given state in the mixing between fuel and oxidizer.
stoichiometric surface [37].
Two limit cases are thus important for non-premixed 3.3. Partially premixed ¯ames
turbulent combustion modeling: pure mixing without
combustion (Da p ! 0) and in®nitely fast chemistry In non-premixed combustion, some partial premixing of
(Da p ! 1). These cases delineate the domain where ¯ames the reactants may exist before the reaction zone develops.
may develop in planes (Z, YF), (Z, YO) and (Z, T ) (Fig. 5). Then, the pure diffusive/reactive layer, as observed in a
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 205

Fig. 6. Generic response of the heat released by a one-dimensional strained diffusion ¯ame versus DamkhoÈler number. The dashed line denotes
in®nitely fast chemistry. Dapq and Dapi are the critical values of Dap ˆ …tc xst †21 at quenching and ignition, respectively. t cis a given chemical
time and xst ˆ Du7Zu2ZˆZst is the scalar dissipation rate under stoichiometric conditions.

laminar diffusion ¯ame, may not be the unique relevant premixed ¯ame develops on the air side (Fig. 7). These
model problem. Furthermore, many ¯ames in burners are two premixed ¯ames are curved because their respective
stabilized by the recirculation of burnt gases, leading to propagation velocities decrease when moving away from
stabilization mechanisms controlled by the mixing between the stoichiometric condition. The overall structure,
fuel, oxidizer, and burnt gases. The mixtures feeding the composed of two premixed ¯ames and of a diffusion
reaction zone are then not always pure fuel and pure ¯ame, is usually called `triple ¯ame'. Such triple ¯ames
oxidizer. There are situations where partial premixing is have been ®rstly experimentally observed by Phillips [40].
clearly important: Since this pioneer work, more recent experiments have
con®rmed the existence of triple ¯ames in laminar ¯ows
² Auto-ignition in a non-homogeneous distribution of fuel [41±43]. Theoretical studies [44±48] and numerical simula-
and oxidizer, where the reactants can be mixed before tions [49±53] have been devoted to triple ¯ames. The propa-
auto-ignition occurs. gation speed of triple ¯ames is controlled by two
² Laminar or turbulent ¯ame stabilization, when combus- parameters: the curvature of the partially premixed front,
tion does not start at the very ®rst interface between fuel increasing with the scalar dissipation rate imposed in front
and oxidizer in the vicinity of burner exit, so that fuel and of the ¯ame, and the amount of heat release. The effect of
oxidizer may mix without burning. heat release is to de¯ect the ¯ow upstream of the triple
² After quenching of the reaction zone, the reactants may ¯ame, making the triple ¯ame speed greater than the propa-
mix leading to possibility of re-ignition and combustion gation speed of a planar stoichiometric ¯ame. This de¯ec-
in a partially premixed regime [38]. tion also induces a decrease of the mixture fraction gradient
in the trailing diffusion ¯ame. The triple ¯ame velocity
The triple ¯ame is an interesting model problem to decreases when increasing the scalar dissipation rate at the
approach partially premixed combustion. In a laminar ¯ame tip. Triple ¯ame velocity response to variations of
shear layer where the mixing between cold fuel and oxidizer scalar dissipation rate may be derived by approximating
develops, a diffusion ¯ame may be stabilized at the splitter the ¯ame tip by a parabolic pro®le and using results from
plate by the combination of heat losses with viscous ¯ow expansions in parabolic-cylinder coordinates. This analysis
effects, or, further downstream [39]. In this latter case, was used by Ghosal and Vervisch to include small but ®nite
combustion starts in a region where fuel and oxidizer have heat release and gas expansion, the triple ¯ame velocity UTF
been mixed in stoichiometric proportion. The resulting may be written [47]:
premixed kernel tends to propagate towards fresh gases s
and contributes to the stabilization of the trailing diffusion b l
UTF < SL …1 1 a† 2 p x …54†
¯ame. In a mixing layer con®guration, the stoichiometric Zst …1 1 a† 4nF 2 2 rCp st
premixed kernel evolves to a rich partially premixed ¯ame
in the direction of the fuel stream, while a lean partially where a ˆ …Tburnt 2 Tfresh †=Tburnt is de®ned from the
206 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

Fig. 7. Schematic of a freely propagation triple ¯ame.

temperatures on both sides of a stoichiometric premixed Arrhenius law as:


¯ame for the same mixture, b ˆ a…TA =Tburnt † is the Zeldo-
vitch number [14], TA is the activation temperature, n F  
T
the stoichiometric coef®cient of the fuel and x st is measured v_ F ˆ 2Ar2 T b YF YO exp 2 A …55†
T
far upstream in the mixing layer where the triple ¯ame
propagates. The value of the scalar dissipation rate at the
triple point is of the order of xst =…1 1 a†2 [47]. These re- where A is the pre-exponential constant, and TA is the acti-
lations are valid for small values of a and moderate, but vation temperature.
non-zero, values of x st. The triple ¯ame velocity given by As the reaction rate is highly non-linear, the averaged
Eq. (54) may be combined with Landau±Squire solution for reaction rate v_ F cannot be easily expressed as a function
non-reacting laminar round jet to construct a stability of the mean mass fractions Y~ F and Y ~ O the mean density r
diagram for lift-off and blowout of jet laminar diffusion and the mean temperature T: ~ The ®rst simple idea is to
¯ames [54]. expand the mean reaction rate v_ F as a Taylor series:
A variety of studies suggest that ®nite rate chemistry and
quenching in non-premixed combustion are somehow     !
T TA X
1 1
T 00n
linked to partially premixed combustion [55]. exp 2 A ˆ exp 2 11 Pn n ;
T~ T~ T~nˆ1
…56†
!
X
1 1
T 00n
4. A direct analysis: Taylor's expansion T b ˆ T~ b 1 1 Qn
nˆ1 T~ n
A direct approach to describe turbulent combustion is ®rst
discussed in this section. This simple formalism, based on
series expansion, illustrates the dif®culties arising from the where Pn and Qn are given by:
non-linear character of chemical sources.
X
n  k
Consider a simple irreversible reaction between fuel (F) …n 2 1†! TA
and oxidizer (O): Pn ˆ …21†n2k ;
kˆ1 …n 2 k†!‰…k 2 1†!Š2 k T~
…57†
F 1 sO ! …1 1 s†P
b…b 1 1†¼…b 1 n 2 1†
where the fuel mass reaction rate v_ F is expressed from the Qn ˆ
n!
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 207

The mean reaction rate, v_ F becomes [56]: Eq. (58) leads to various dif®culties. First, new quantities
! such as Yg00 T 00n have to be closed using algebraic expressions
k
 2 ~b ~ ~ TA or transport equations. Because of non-linearities, large
v_ F ˆ 2Ar T Y F Y O exp 2
T~ errors exist when only few terms of the series expansion
" ! are retained. Expression (58) is quite complicated, but is
Yg 00 00
Y Yg 00 00
FT Yg 00 00
OT only valid for a simple irreversible reaction and cannot be
 1 1 F O 1 …P1 1 Q1 † 1
Y F Y~ O
~ Y~ F T~ Y~ O T~ easily extended to realistic chemical schemes (at least 9
! # species and 19 reactions for hydrogen combustion, several
Yg 00 00 2
FT Yg 00 00 2
T hundred species and several thousand reactions for hydro-
1 …P2 1 Q2 1 P1 Q1 † 1 O 2 1¼
Y~ F T~ 2 Y~ O T~ carbon combustion, etc.). For these reasons, reaction rate
…58† closures in turbulent combustion are not based on Eq. (58).

Fig. 8. Turbulent premixed combustion regimes as identi®ed by Borghi and Destriau [62]: (a) ¯amelet (thin wrinkled ¯ame), (b) thickened
wrinkled ¯ame regime, and (c) thickened ¯ame regime.
208 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

Models are rather derived from physical analysis as discuss mixed by turbulent structures before reaction. In this
below. perfectly stirred reactor limit, the mean reaction rate may
Nevertheless, this approach is used in some simulations be expressed from Arrhenius laws using mean mass frac-
of supersonic reacting ¯ows [57] or to describe reaction in tions and temperature, corresponding to the ®rst term of the
atmospheric boundary layer where the temperature T may Taylor's expansion (58).
be roughly assumed to be constant [58]. In these situations, In turbulent ¯ames, as long as quenching does not occur,
only the ®rst two terms in the series expansion are kept. most practical situations correspond to high or medium
A segregation factor, a FO, is then introduced: values of the DamkoÈhler numbers. It is worth noting that
! various chemical time scales may be encountered: fuel
Yg 00 Y 00
F O Yg F YO oxidation generally corresponds to short chemical time
aFO ˆ 2 ˆ2 12 …59†
Y~ F Y~ O Y~ F Y~ O scales …Da q 1† whereas pollutant production or destruc-
tion such as CO oxidation or NO formation are slower.
to characterize the mixing between the reactants F and O. If
they are perfectly separated Yg F YO ˆ 0 and aFO ˆ 21: On
5.2. Turbulent premixed combustion diagram
the other hand, a perfect mixing …Yg 00 Y 00 ˆ 0† leads to a
F O FO ˆ
0: This segregation factor may be either postulated or 5.2.1. Introduction
provided by a balance equation (see Ref. [59] in a large The objective is to analyze premixed turbulent combus-
eddy simulation context). Then, the mean reaction rate tion regimes by comparing turbulence and chemical char-
becomes: acteristic length and time scales. This analysis leads to
  combustion diagrams where various regimes are presented
T
v_ ˆ 2A…1 1 aFO †r 2 T~ b Y~ F Y~ O exp 2 A …60† as function of various dimensionless numbers [14,23,29,60±
T~
62]. These diagrams could be a support to the selection and
development of the relevant combustion model for a given
situation. A formalism combining recent analysis [29,62] is
5. Scales and diagrams for turbulent combustion
retained here.
5.1. Introduction For turbulent premixed ¯ames, the chemical time scale,
t c, may be estimated as the ratio of the thickness d l and the
As the mean burning rate v_ cannot be found from an propagation speed SL of the laminar ¯ame. 1 Estimating the
averaging of Arrhenius laws, a physical approach is required turbulent time from turbulent integral scale characteristics
to derive models for turbulent combustion. Turbulent …tt ˆ lt =u 0 †; the DamkoÈhler number becomes:
combustion involves various lengths, velocity and time
scales describing turbulent ¯ow ®eld and chemical reac- tt l S
Da ˆ ˆ t L0 …63†
tions. The physical analysis is mainly based on comparison tc dl u
between these scales.
The turbulent ¯ow is characterized by a Reynolds number where a velocity ratio (u 0 /SL) and a length scale ratio (lt/d l)
comparing turbulent transport to viscous forces: are evidenced.
u 0 lt
Re ˆ …61† 5.2.2. Combustion regimes
v
For large values of the DamkoÈhler number …Da q 1†; the
where u 0 is the velocity rms (related to the square root of the ¯ame front is thin and its inner structure is not affected by
turbulent kinetic energy k), lt is the turbulence integral turbulence motions which only wrinkle the ¯ame surface.
length scale and n the kinematic viscosity of the ¯ow. This ¯amelet regime or thin wrinkled ¯ame regime (Fig. 8a)
The DamkoÈhler number compares the turbulent (t t) and occurs when the smallest turbulence scales (i.e. the Kolmo-
the chemical (t c) time scales: gorov scales), have a turbulent time t k larger than t c
t (turbulent motions are too slow to affect the ¯ame structure).
Da ˆ t …62†
tc This transition is described in term of the Karlovitz
number Ka:
In the limit of high DamkoÈhler numbers …Da q 1†; the
chemical time is short compared to the turbulent one, corre-
sponding to a thin reaction zone distorted and convected by tc d u
Ka ˆ ˆ l k …64†
the ¯ow ®eld. The internal structure of the ¯ame is not tk lk SL
strongly affected by turbulence and may be described as a
laminar ¯ame element called a `¯amelet'. The turbulent 1
This chemical time t c corresponds to the time required for the
structures wrinkle and strain the ¯ame surface. On the ¯ame to propagate over a distance equal to its own thickness. This
other hand, a low DamkoÈhler number …Da p 1† corresponds time may also be viewed as a diffusive time scale, using Eq. (34)
to a slow chemical reaction. Reactants and products are tc ˆ dl =SL ˆ …1=Ref †…d2l =n†:
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 209
0
The size lk and the velocity uk of Kolmogorov structures Ð(u /SL) . 1: wrinkled ¯ame with pockets (`corrugated
are given by [64]: ¯ames'). In this situation, larger structures become able
!1=4 to induce ¯ame front interactions leading to pockets.
v3 ² 1 , Ka # 100 …Kar , 1† : Thickened wrinkled ¯ame
lk ˆ ; uk ˆ …v1†1=4 …65†
1 regime or thin reaction zone. In this case, turbulent
motions are able to affect and to thicken the ¯ame preheat
where 1 is the dissipation rate of the turbulent kinetic energy
zone, but cannot modify the reaction zone which remains
k. The integral length scale lt may be written:
thin and close to a laminar reaction zone (Fig. 8b).
!
u 03 ² Ka . 100 (Kar . 1): Thickened ¯ame regime or well-
lt ˆ …66† stirred reactor. In this situation, preheat and reaction
1
zones are strongly affected by turbulent motions and no
using v ˆ dl SL ; corresponding to a unity ¯ame Reynolds laminar ¯ame structure may be identi®ed (Fig. 8c).
number Ref (Eq. (34)), yields
!3=2   These various regimes are generally displayed on a log-
u0 lt 21=2 arithmic diagram (u 0 /SL; lt/d l), similar to the one presented
Ka ˆ …67†
SL dl on Fig. 9.

Reynolds, Re, DamkoÈhler, Da, and Karlovitz, Ka, numbers


5.2.3. Comments
are related as:
This analysis, leading to a rough classi®cation of combus-
Re ˆ Da2 Ka2 …68† tion regimes as a function of characteristic numbers, has
been developed as a support to derive and choose turbulent
and a set of two parameters (Re, Da), (Re, Ka) or (Da, Ka) are combustion models. Following this classi®cation, most
necessary to discuss regimes in the case of premixed reactants. practical applications correspond to ¯amelet or thickened
The Karlovitz number also compares the ¯ame and the wrinkled ¯ame regimes. Nevertheless, such analyses are
Kolmogorov length scales according to: only qualitative and should be used with great care. A
 2 diagram such as the one displayed on Fig. 9 cannot be
dl
Ka ˆ …69† readily used to determine the combustion regime of a prac-
lk
tical system from (u 0 /SL) and (d l/lt) ratios:
The Karlovitz number is used to de®ne the Klimov±
Williams criterion, corresponding to Ka ˆ 1; delineating ² The analysis is based on the assumption of a homo-
between two combustion regimes. This criterion was ®rst geneous and isotropic turbulence unaffected by heat
interpreted as the transition between the ¯amelet regime release, which is not the case in combustion systems.
…Ka , 1†; previously described, and the distributed combus- ² Some quantities used are not clearly de®ned. For
tion regime where the ¯ame inner structure is strongly modi- example, the ¯ame thickness d l may be based on the
®ed by turbulence motions. A recent analysis [29] has thermal thickness or on the diffusive thickness. Accord-
shown that, for Karlovitz numbers larger than unity …Ka . ingly, the limits between the various regimes may notice-
1†; turbulent motions become able to affect the ¯ame inner ably change.
structure but not necessarily the reaction zone. This reaction ² All regime limits are based on order of magnitude esti-
zone, where heat is released, has a thickness d r much lower mations and not on precise derivations. For example, the
that the thermal thickness d l of the ¯ame …dr < 0:1dl †: The ¯amelet regime limit could correspond to a Karlovitz
Karlovitz number based on this reaction thickness is: number Ka ˆ 0:1 or Ka ˆ 10; rather than Ka ˆ 1:
 2  2   2   ² Various effects are not taken into account. Unsteady and
dr dr dl 1 dl 2 Ka
Kar ˆ ˆ < < …70† curvature effects play an important role neglected here.
lk dl lk 100 lk 100
Turbulent premixed combustion diagrams were analyzed
Then, the following turbulent premixed ¯ame regimes are using direct numerical simulations of ¯ame/vortex inter-
proposed [29]: actions [65]. Results show that the ¯amelet regime seems
to extend above the Klimov±Williams criterion (see Fig.
² Ka , 1: Flamelet regime or thin wrinkled ¯ame regime 9). DNS has revealed that small turbulent scales, which
(Fig. 8a). Two subdivisions may be proposed depending are supposed in classical theories to have the strongest
on the velocity ratio u 0 /SL: effects on ¯ames, have small lifetimes because of viscous
Ð(u 0 /SL) , 1: wrinkled ¯ame. As u 0 may be viewed as dissipation and therefore only limited effects on combus-
the rotation speed of the larger turbulent motions, tion, results recovered experimentally [66]. Peters [29]
turbulent structures are unable to wrinkle the ¯ame shows that the criterion Ka ˆ 100 (i.e. Kar ˆ 1† is in
surface up to ¯ame front interactions. The laminar quite good agreement with the transition proposed in
propagation is predominant and turbulence/combus- Ref. [65], at least when the length scale ratio, lt/d l, is
tion interactions remain limited. suf®ciently large.
210 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

Fig. 9. Turbulent premixed combustion diagram [29,62]. Combustion regimes are identi®ed using the length scale …lt =dl † and the velocity
…u 0 =SL † ratios. The Klimov±Williams criterion …Ka ˆ 1† corresponds to a ¯ame thickness d l equal to the Kolmogorov scale lk. Below this line,
the ¯ame is thinner than any turbulent scale. Below the line delineating the Peters criterion (Ka ˆ 100 or Kar ˆ 1), the reaction zone thickness,
d r, is thinner than any turbulent scale and is not affected by turbulent motions (the criterion is plotted assuming dr < 0:1dl ). The ¯amelet regime
limit devised by Poinsot et al. [65] from direct numerical simulations is also displayed. The criterion proposed in Ref. [140] to delineate
between gradient (above) and counter-gradient (below) turbulent transport is displayed assuming a heat release factor t ˆ Tb =Tu 2 1 ˆ 6 where
Tu and Tb are, respectively, the fresh and the burnt gases temperature (see Section 8).

² Additional length scales have been introduced in the dynamics controlling the thickness of the local mixing
literature. For instance the Gibson scale lG, to character- layers developing between fuel and oxidizer (Section 3.2)
ize the size of the smaller vortex able to affect the ¯ame and no ®xed reference length scale can be easily identi®ed
front was used [63]. This length was de®ned as the size of for diffusion ¯ames. This dif®culty is well illustrated in the
the vortex having the same velocity as the laminar ¯ame literature, where various characteristic scales have been
speed SL. retained depending on the authors [33,67±71]. These
² All these analyses are implicitly based on a single step classi®cations of non-premixed turbulent ¯ames may be
irreversible reaction. In actual turbulent combustion, a organized in three major groups:
large number of chemical species and reactions are
involved (several hundred species and several thousand ² The turbulent ¯ow regime is characterized by a Reynolds
reactions for propane burning in air). These reactions number, whereas a DamkoÈhler number is chosen for the
may correspond to a large range of chemical time scales. reaction zone [72].
For example, the propane oxidation may assumed to be ² The mixture fraction ®eld is retained to describe the
fast compared to turbulent time scale. On the other hand, turbulent mixing using Zf 002 and a Damko È hler number
the CO2 formation from carbon monoxide (CO) and OH (ratio of Kolmogorov to chemical time) characterizes
radical in the burnt gases is much slower with a chemical the ¯ame [33].
time of the same order as turbulent times. ² A velocity ratio (turbulence intensity to premixed lami-
nar ¯ame speed) and a length ratio (integral scale to
5.3. Non-premixed turbulent combustion diagram premixed laminar ¯ame thickness) may be constructed
[68] to delineate between regimes.
5.3.1. Introduction
Two numbers, a length ratio and a velocity ratio, have Additional lengths have also been introduced, using for
been used to identify premixed turbulent combustion instance thicknesses of pro®les in mixture fraction space
regimes. The problem is more dif®cult in non-premixed [67].
turbulent combustion because diffusion ¯ames do not propa- A laminar diffusion ¯ame is fully determined from a
gate and, therefore, exhibit no intrinsic characteristic speed. DamkoÈhler number Dap ˆ …tc xst †21 ; where the value of
In addition, the thickness of the ¯ame depends on the aero- the chemical time t c depends on the fuel chemistry [30]
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 211

Fig. 10. Non-premixed ¯ame/vortex interaction regimes by Cuenot and Poinsot [70]. This diagram delineates the steady laminar ¯amelet
assumption (LFA) validity regions, the quenching limits and the zone where unsteady and curvature effects are important during ¯ame vortex
interaction. u 0 is the level of velocity ¯uctuations, di ˆ ld is the ¯ame thickness (ld < u7Zu 21 ), t is a chemical time and r the characteristic size
of the vortices.

(Section 3.2). In this number, the scalar dissipation rate associated to a time varying strain rate was obtained by
under stoichiometric condition …Z ˆ Zst †; xst ˆ Du7Zu st2 ; Cuenot and Poinsot from DNS results of ¯ame/vortex inter-
measures at the same time a mechanical time, tx ˆ x21 st ; action [70]. In this diagram presented on Fig. 10, the ¯ame
and, a characteristic mixing length, ld ˆ …D=xst †1=2 : Accord- thickness is d i < ld, whereas r and u 0 denote respectively the
ing to asymptotic developments [30], the reaction zone characteristic size and velocity of the vortex pair. This
thickness is of the order of lr < ld …Dap †21=…a11† ; where a is analysis identi®es two limiting DamkoÈhler numbers,
the order of a global one-step reaction. DaLFA and Daext. When Da p is larger than DaLFA, the ¯ame
Because diffusion ¯ames do not feature a ®xed reference front may be viewed as a steady laminar ¯ame element and
length, a main dif®culty arises when effects of unsteadiness its inner structure is not affected by vortices. On the other
need to be quanti®ed. In a steady laminar ¯ame the local rate hand, when Dap # Daext ; ¯ame extinction occurs. In the
of strain is directly related to x st (and to a ¯ame thickness), intermediate DamkoÈhler number range (i.e. Daext , Da p ,
however, when the velocity ®eld ¯uctuates, unsteadiness in DaLFA), strong unsteadiness effects are observed.
diffusion ¯ames develops at two levels [73]: In a non-premixed turbulent ¯ame, the reaction zones
develop within a mean mixing zone whose thickness lz is
² The mixture fraction ®eld Z does not immediately of the order of the turbulent integral length scale lt (Fig. 11):
respond to velocity ¯uctuations, leading to a distribution !
of x st for given rates of strain. Because a strong correla- ~ 21 k3=2
lz < u7Zu < lt < …72†
tion exists between x st and velocity gradients taken along 1
the stoichiometric line [74], this effect is not the dominant
one when ®nite rate chemistry occurs. Turbulent small scale mixing mainly depends both on
² For ®nite rate chemistry, the burning rate does not im- velocity ¯uctuations, transporting the iso-Z surfaces (stir-
mediately follow variations of x st, leading to a second ring), and diffusion between these iso-surfaces that compose
level of unsteadiness, modifying the burning rate (Eq. the mixing layer of thickness ld, with
(71)).  
D 1=2
ld < …73†
u 0 ! ‰Unsteadiness in mixingŠ ! xs x~ st

! ‰Burning rate unsteadiness…for Dap , 1†Š ! v_ i where x~ st denotes the conditional mean value of the scalar
…71† dissipation rate x for Z ˆ Zst :
When transport of species and heat by velocity ¯uctua-
Summarizing these effects in a generic diagram is an tions is faster than transfer in the diffusion ¯ame, a departure
arduous task. from laminar ¯amelet is expected. Also, when the Kolmo-
A diagram for laminar ¯ames submitted to curvature gorov scale lk is of the order of the ¯ame thickness, the inner
212 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

Fig. 11. Sketch of a non-premixed turbulent ¯ame. Z is the mixture Fig. 12. Schematic of non-premixed turbulent combustion regimes
fraction, ld the diffusive thickness, lr the reaction zone thickness, lt as function of the DamkoÈhler number Da ˆ tt =tc (constructed from
the turbulence integral length scale. the turbulent integral time scale t t and chemical time t c) and Re the
turbulent Reynolds number.
structure of the reaction zone may be modi®ed by the turbu-
lence. As diffusion ¯ame scales strongly depend on the local
¯ow motions, one may write: practical interest that should be approximated by turbulent
a combustion models. The simplest and most direct approach
ld < a1 lk and x~ st < 2 …74† is to develop the chemical rate in Taylor series as a function
tk
of species mass fractions and temperature (Eq. (58)). This
where a1 $ 1 and a2 # 1 (the maximum local strain rate analysis is limited by its low accuracy and by the rapidly
would correspond to ld pˆlk ). growing complexity of the chemistry (Section 4). It is then
Then using tt =tk ˆ Re; the DamkoÈhler number com- concluded that the non-linear character of the problem
paring turbulent ¯ame scale and chemical ¯ame scale is requires the introduction of new tools. These new tools
recast as: must be designed to describe turbulent ¯ames and have to
t t t t a2 p provide an estimation of mean production or consumption
Da ˆ t ˆ t k < t < a2 Re Dap …75†
tc tk tc tk x~ st tc rates of chemical species. They also need to be based on
known quantities (mean ¯ow characteristics, for example)
Constant DamkoÈhler numbers Da p correspond to lines of
or on quantities that may be easily modeled or obtained from
slope 1/2 in a log±log (Da, Re) diagram. When the chemistry
closed balance equations. In this section, a generic descrip-
is suf®ciently fast (large Da values), the ¯ame is expected to
tion of the main concepts used to model turbulent combus-
have a laminar ¯ame structure. This condition may be
tion is proposed. Relations between the various approaches
simply expressed as Dap $ DaLFA On the other hand, for
are also emphasized, but the discussion of the closure
large chemical times (i.e. when Dap # Daext ), extinction
strategy is postponed to subsequent sections.
occurs. Laminar ¯ames are encountered for low Reynolds
The basic ingredients to describe turbulent ¯ames remain
numbers …Re , 1†: Results are summarized in Fig. 12.
the quantities introduced for laminar ¯ame analysis: the
In a practical combustion devices, a 1 and a 2 would
progress variable c for premixed combustion (c ˆ 0 in
evolve in space and time according to ¯ow ¯uctuations,
fresh gases and c ˆ 1 in burnt gases, see Section 3.2),
velocity and scalar energetic spectra. In a given burner, it
and, the mixture fraction Z for non-premixed ¯ames (Z is
is likely that one may observe at different locations ¯amelet
a passive scalar, with Z ˆ 0 in pure oxidizer and Z ˆ 1 in
behavior and strong unsteadiness, or even quenching.
pure fuel, see Section 3.2). The ¯ame position would corre-
As the classi®cation of premixed turbulent ¯ames, these
spond to values of the progress variable c lying between 0
considerations are limited by the numerous hypothesis
and 1, or, to Z taking on values in the vicinity of Z ˆ Zst :
necessary to derive the regimes.
Three main types of approaches are summarized in
Fig. 13:
6. Tools for turbulent combustion modeling The burning rate may be quanti®ed in terms of turbulent
mixing. When the DamkoÈhler number Da ˆ tt =tc ; com-
6.1. Introduction paring turbulent (t t) and chemical (t c) characteristic times,
is large (a common assumption in combustion modeling),
The mean heat release rate is one of the main quantities of the reaction rate is limited by turbulent mixing, described in
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 213

Fig. 13. Three types of analysis of premixed or non-premixed turbulent ¯ames.

terms of scalar dissipation rates [75]. The small scale dissi- value surfaces de®ned as ¯ame surfaces (iso-c p or iso-Zst).
pation rate of species controls the mixing of the reactants The ¯ame is then envisioned as an interface between fuel
and, accordingly, plays a dominant role in combustion and oxidizer (non-premixed) or between fresh and burnt
modeling, even for ®nite rate chemistry. gases (premixed). A ¯ame normal analysis is derived by
In the geometrical analysis, the ¯ame is described as a focusing the attention on the structure of the reacting ¯ow
geometrical surface, this approach is usually linked to a along the normal to the ¯ame surface. This leads to ¯amelet
¯amelet assumption (the ¯ame is thin compared to all modeling when this structure is compared to one-dimen-
¯ow scales). Following this view, scalar ®elds (c or Z) are sional laminar ¯ames. The density of ¯ame surface area
studied in terms of dynamics and physical properties of iso- per unit volume is also useful to estimate the burning rate.
214 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

The statistical properties of scalar ®elds may be collected gradient), the time evolution of the scalar variances are
and analyzed for any location within the ¯ow. Mean values governed by:
and correlations are then extracted via the knowledge of
Premixed combustion:
one-point probability density functions (pdf). The determi-
nation of these pdfs leads to pdf modeling. A one-point
statistical analysis restricted to a particular value of the 2r cg
00 2
ˆ 22rD7c 00 ´7c 00 1 2v_ c 00
scalar ®eld is related to the study of conditional statistics. dt
Conditional statistics which are obviously linked to the Non-premixed combustion:
geometrical analysis and to ¯ame surfaces when the con-
ditioning value is c p or Zst. 2r Zg00 2
ˆ 22rD7Z 00 ´7Z 00
dt
6.2. Scalar dissipation rate
These equations have important implications:
In a ®rst step, the transport equation for cf 002
or Zf
002
is The scalar dissipation rate directly measures the decaying
derived, these ¯uctuations characterize non-homogeneities speed of ¯uctuations via turbulent micromixing. Since the
and intermittencies. In the case of the progress variable, the burning rate depends on the contact between the reactants,
variance cf002
is de®ned as: in any models, the scalar dissipation rate enters directly or
indirectly the expression for the mean burning rate. For
r cf ~ 2 ˆ r …cf
002 ˆ r …c 2 c† 002 2 c~2 † ˆ rc2 2 r c~2 …76† instance, when assuming very fast chemistry and a combus-
tion limited by mixing, the mean burning rate is proportional
Starting from the balance equation for the progress vari- to the scalar dissipation rate of Z or c.
able (Eq. (36)), c is decomposed into c ˆ c~ 1 c 00 ; then the Within a premixed system, turbulent mixing occurs
new equation is multiplied by c 00 and averaged. After between fresh and burnt gases. One may then expect a
straightforward manipulations, the exact transport equation very strong coupling between mixing phenomena and
for cf002
reads: chemical reaction. This is observed in the equation for cg 00 2

where, at the same time, x~ and the chemical source v_ c 00 are


2r cf
002
1 7…r u~ cg
00 2
† 1 7…r ug
00 00 2
c †ˆ …77† involved.
2t In a non-premixed ¯ame, fresh fuel and fresh oxidizer
have to be mixed at the molecular level for reacting
7´…rD7c 00 2 † 1 2c 00 7´…rD7c†
~ and the ¯ame is mainly controlled by turbulent mixing
occurring between the fresh gases. In consequence, there
22r ug
00 00
c ´7c~ 2 2rD7c 00 ´7c 00 1 |‚{z‚}
2v_ c 00 is no chemical source acting on the evolution of Zf 002 :
|‚‚‚‚{z‚‚‚‚} |‚‚‚‚{z‚‚‚‚} The mixture fraction Z is sensitive to chemistry only via
Production Dissipation Source
density change, making the coupling between chemistry
In addition to the two diffusive terms 7´…rD7c 00 2 † and and mixing different than in the case of premixed
2c 00 7´…rD7c†;
~ which are non-zero, but expected small for combustion.
This preliminary analysis shows that dissipation rate of
large Reynolds number ¯ows (especially the second one),
scalars is a very key concept of turbulent combustion and,
two important terms are found: The ¯uctuating part of the directly or indirectly, x appears in any tools used to model
scalar dissipation rate 2rD7c 00 ´7c 00 and a correlation v_ c 00 ¯ames. The main stumbling block in turbulent combustion
involving the chemical source. modeling and bridges between the various modeling
In the literature, various expressions have been associated concepts emerge through the scalar dissipation rate.
to the terminology scalar dissipation rate (in laminar ¯ame
theory, it actually quanti®es a diffusion speed, Section 3.2). 6.3. Geometrical description
It may include the density r , a factor 2 and be written in
term of instantaneous (c) or ¯uctuating (c 00 ) values of the The ¯ame front is here described as a geometrical entity.
concentration species. Hereafter: This analysis is generally linked to the assumption of a
suf®ciently thin ¯ame, viewed as an interface between
~ c~ 1 2rD7c 00 ´7c~ 1 rD7c 00 ´7c 00
r x~ ˆ rD7c´7c ˆ rD7c´7
fresh and burnt gases in premixed combustion or as an inter-
leading to, when mean gradients are neglected: face between fuel and oxidizer in non-premixed situations.
Two formalisms have been proposed: ®eld equation or ¯ame
r x~ < rD7c 00 ´7c 00 …78† surface density concept.
Then, r x~ is the dissipation rate of the ¯uctuations of the
scalar ®eld. 6.3.1. G-®eld equation
In the simpli®ed case of homogeneous ¯ames (no c~ or Z~ The balance equation for the progress variable may
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 215

be written: w and Sc are related but may be quite different, especially in


2c 1 high ¯ame front curvature zones [78]. The displacement
1 u´7c ˆ …7´…rD7c† 1 v_ † ˆ wu7cu speed w may also be quite different from the laminar
2t r
…79† ¯ame speed SL.
2c The coupling between the consumption speed Sc and the
or 1 u c ´7c ˆ 0
2t displacement speed w is a very key point in G-®eld model-
ing. Three approaches have been proposed to overcome this
where uc is the absolute propagation velocity of the progress
dif®culty:
variable ®eld. This velocity is easily decomposed into the
¯ow velocity u and w the relative velocity of the iso-c
² Flame front tracking technique: The displacement of the
surface measured with respect to the ¯ow, leading to u c ˆ
¯ame front is evaluated from the displacement speed w,
u 1 wn; where n ˆ 2…7c=u7cu† is the local normal vector to
leading to an estimation of the volume of burnt gases
the iso-c surface, pointing toward the fresh gases. The rela-
produced along with the thermal heat release [79,80].
tive displacement speed w is given by Eq. (37).
Based on a purely geometric approach, this technique is
In laminar ¯ames, a G-®eld whose level G ˆ G 0 repre-
well suited to two dimensional simulations, but its exten-
sents the ¯ame surface, was introduced to simulate the
sion to 3D cases may not be straightforward.
propagation of fronts [76]. The `G-equation' may be written:
² Temperature (or energy) reconstruction: The temperature
2G 2G ®eld is directly estimated from the G-®eld as [81]:
1 uc ´7G ˆ 0 or 1 u´7G ˆ wu7Gu …80†
2t 2t Q
T ˆ Tu 1 H…G 2 Gp † …83†
This kinematic description of premixed combustion Cp
possesses some attractive aspects:
where Tu is the temperature of unburnt gases, Q the reac-
² The internal ¯ame front structure does not need to be tion heat release and H denotes the Heaviside function
resolved on the computational mesh. Only the G-®eld, (the Heaviside function is smoothed on the mesh of the
generally much thicker than the ¯ame front, needs to be simulation). This approach does not require a balance
resolved. Usually, G…x† is de®ned as the distance of the equation for the energy but is not applicable when heat
given location x to the ¯ame front [77]. losses or compressibility effects (for example in an in-
² Under the assumption of constant density (thermodiffu- ternal combustion engines) occur.
sive assumption), the G-equation may be used for low ² Estimation of the heat release rate from the G-®eld: The
cost direct numerical simulations. Each G iso-surface is G-®eld is used to estimate the heat release rate to be
then related to a ¯ame front, and one may argue that a incorporated in the balance energy equation from a
single simulation corresponds to the computation of formulation similar to Eq. (83) [82]. Accordingly, any
several ¯ames. other effects (heat losses, compressibility) may be
included.
In non-constant density ¯ows where thermal heat release
is included, the displacement speed w of the G-iso-surface is
This formalism may be extended to turbulent ¯ames.
affected by thermal expansion and should be corrected for
Averaging the progress variable Eq. (79) leads to:
density variations, even in the case of a steady laminar ¯ame 2 3
propagating at the constant laminar ¯ame speed SL, de®ned 2c~ ru 4 27´…r ug
00 c 00 † 1 rwu7cu
relatively to the unburnt gases. This correction is: ~ c~ ˆ
1 u´7 5 u7cu
~ …84†
2t r u7cu ~
r |‚‚‚‚‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚‚‚‚‚}
w ˆ u wu …81† ST
r
where wu is the ¯ame displacement speed relative to the where the turbulent ¯ame speed, ST, has to be modeled. The
unburnt gases of density r u. G-equation becomes:
A more dif®cult point is the coupling required between 2G~ ru
the G-equation and the species or energy balance equations. ~ G~ ˆ
1 u´7 ~
S u7Gu
2t r T
The G-equation provides a kinematic description of the
¯ame front and involves its displacement speed w. The The mean turbulent ¯ame brush is then located at the
reactant consumption and the heat release rate are controlled points where G~ ˆ G 0 : The G-equation does not required
by the consumption speed Sc: thin ¯ame elements per se. The overall turbulent ¯ame is
Z1 only viewed as a propagating surface without solving for the
ru Sc ˆ v_ dj …82† internal ¯ame structure. This formalism is therefore a good
21
candidate for the numerical simulation of large systems
where r u is the fresh gases density and j the spatial coordi- where the knowledge of the internal structure of the ¯ame
nate along the normal direction to the ¯ame front. Of course, brush is not required [79]. Nevertheless, a model has to be
216 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

Fig. 14. Flame surface density modeling.

supplied for ST, when doing so, turbulent transport and balance equation. The derivation using statistical tools is
propagation may be separated to carefully model the effect now brie¯y summarized.
of turbulence on w, the instantaneous displacement speed of In premixed combustion, the ¯ame surface density S of
the premixed front [29,33]. the iso-c p surface is estimated from the conditional gradient
of the progress variable c [86]:
 
6.3.2. Flame surface density description  p†
S…c p † ˆ u7cud…c 2 c p † ˆ u7cuuc ˆ c p P…c …86†
The ¯ame is identi®ed as a surface and the ¯ame surface
where d…c 2 cp † is a local measure
 of the probability (see
density S is introduced, S measures the available ¯ame area
Section 6.4.3), u7cuuc ˆ c p is the conditional average of
d A per unit volume d V. The mean burning rate of a species  p † is the probability to ®nd c ˆ cp at
u7cu for c ˆ c p and P…c
i is then modeled as:
the given location. From this de®nition and the balance
v_ i ˆ V_ i S …85† equation for the progress variable c, an exact equation for
the ¯ame surface density S may be derived according to the
where VÇ i is the mean local burning rate per unit of ¯ame area following steps [87]:
integrated along the normal direction to the ¯ame surface.
VÇ i is related to the properties of the local ¯ame front and is 1. Derivation of an equation for u7cu from the equation for
generally estimated from a prototype laminar ¯ame, incor- the instantaneous progress variable c.
porating more or less complexity. For instance, one may 2. Derivation of an equation for u7cu d…c 2 c† p by condi-
consider a planar laminar ¯ame, submitted or not to a steady tioning the previous u7cu balance equation.
strain, a laminar ¯ame where curvature effects have been 3. Averaging the u7cu d …c 2 c† p balance equation leading
introduced, or even a laminar unsteady strained and curved to an exact equation for the ¯ame surface density S ˆ
¯ame. The main advantage of this formulation, summarized u7cud…c 2 c p †:
on Fig. 14, is to decouple the chemical description (VÇ i) from
This derivation is valid for any iso-scalar surface (c p can
the ¯ame/turbulence interaction (S ). The ¯ame surface is
take any values between zero and unity) and S…c p ; x; t† is also
convected, diffused, curved and strained by the velocity
called a surface density function, the derivation of its balance
®eld [17,83].
equation is quite tedious [87] and similar to the derivation of
The ¯ame surface density S may be estimated either from
a balance equation for the probability density function
algebraic relations (see Section 7.4.2) or as a solution of a
(Section 6.4.3), details are not given here. Two equivalent
balance equation. Using a phenomenological analysis, this
forms of the progress variable equation may be used:
balance equation was ®rst proposed by Marble and
Broadwell [84] for non-premixed turbulent combustion. ² A classical reaction/diffusion formulation:
More rigorous derivations were obtained from geo-
metrical considerations [17,85] and from a statistical 2c 1
1 u´7c ˆ ‰7´…rD7c† 1 v_ Š …87†
description [86±88] leading to an exact, but unclosed, 2t r
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 217

² A propagative form: along the ¯ame normal direction and ¯uxes in the sample
2c space c p. The last two terms in the RHS of the propagative
1 u´7c ˆ wu7cu …88† Eq. (90) correspond to front convection due to a normal
2t
propagation and combined propagation/curvature effects.
where the displacement speed w of the ¯ame front rela- These two formulations of the S balance equation induce
tively to the ¯ow was introduced (Eq. (37)). This equa- the following comments:
tion corresponds to the G-equation [77] (Section 6.3.1).
² The two balance Eqs. (89) and (90) are mathematically
strictly equivalent but the problem is not expressed in the
Two different types of equation for S are associated to
same way. In the propagative form (Eq. (90)), many
these two formulations of the progress variable balance
effects are incorporated in the ¯ame front propagation
equation.
speed, w, that may differ from the laminar ¯ame speed
² A reaction/diffusion formulation: SL [78]. On the other hand, a ¯ux term in the sample space
is found in the reaction/diffusion derivation. In Eq. (90)
2S the imbalance between diffusion and reaction is cast in
1 7´…kul s S† ˆ k7´u 2 nn : 7ul s S
2t the form of the propagation velocity w, Eq. (89) recalls
* !+ that transfer phenomena between iso-surfaces are
1 2 1
2 ‰7´…rD7c† 1 v_ Š S …89† involved in this propagation.
u7cu 2n r s ² The derivation of the balance Eqs. (89) and (90) impli-
"* + # citly forgets some mathematical singularities that may
2 1
2 ‰7´…rD7c† 1 v_ Š S become important in particular situations. For example,
2c p r s the normal vector n is assumed to be well de®ned and
having a ®nite derivative. This may not be the case when
² A propagative formulation: two close ¯ame fronts interacts.
² In combustion modeling, a single iso-c p surface is
2S assumed to correspond to the ¯ame front. This is a priori
1 7´…kul s S†
2t …90† true when the ¯ame is in®nitely thin, assuming that the
  local reaction rate per unit ¯ame area, VÇ i, describes
ˆ k7´u 2 nn : 7ul s S 2 7´ kwnl s S 1 kw7´nl s S
whether the ¯ame is actually burning or not. However
in real turbulent ¯ames, the local burning zone is not
where n is the unit vector normal to the c ˆ c p surface and in®nitely thin and the ¯ame front, identi®ed as the
pointing towards the fresh gases …n ˆ 27c=u7cu†: 7´n corre- location of the maximum reaction, may differ from the
sponds to the ¯ame front curvature. 2=2n ˆ n´7 is a deriva- c p iso-surface. This may be for instance the case when
tive normal to the ¯ame front. 2/2c p is a derivative in the analyzing data of direct numerical simulations where the
sample space c p. The surface average of Q, kQls, is de®ned ¯ame front has to be resolved on the computational grid.
as: ² Using the Favre decomposition …u ˆ u~ 1 u 00 †; the
  convection ¯ux term may be decomposed into mean
Qu7cud…c 2 c p † Qu7cujc ˆ c p and turbulent components:
kQl s ˆ ˆ   …91†
u7cud…c 2 c p † u7cujc ˆ c p kul s S ˆ u~ S 1 ku 00 l s S …93†

Here the notation kQls implicitly indicates that the mean ² The strain rate term is also split into a contribution due to
is taken for c ˆ cp ; and kQls is a function of c p. the mean ¯ow and a contribution due to turbulent velo-
The propagative formulation is often written in term of city ¯uctuations:
total stretch Acp of the iso-c p surface
  k7´u 2 nn : 7ul s S ˆ …7´u~ 2 knnl s : 7ul
~ S
1 2S |‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚}s
1 7´…ku 1 wnl s S† ˆ A c p AT
S 2t
1 k7´u 00 2 nn : 7u†
~ S …94†
z‚}|‚{
Curvature |‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚}
aT
ˆ k |‚‚‚‚
7´u 2 nn : 7u
‚{z‚‚‚‚‚} 1 w7´n l s …92†
tangential strain rate
All these de®nitions may be extended to ¯ame fronts which
The LHS terms in the two balance Eqs. (89) and (90) corre-
are not in®nitely thin. Integrating Eq. (86) across iso-surface
spond to unsteady effects and to the ¯ame surface convec-
levels leads to:
tion. The ®rst term in the RHS expresses the action of the
tangential strain rate on the ¯ame surface. The last two terms Z1 Z1  
S…c p †dcp ˆ  p †dcp ˆ u7cu
u7cuuc ˆ c p P…c …95†
in Eq. (89) describe respectively reaction/diffusion effects 0 0
218 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

that is a direct estimation of the inverse of the mean local closely related. Balance equations may also be derived
¯ame thickness d …d ˆ …1=u7cu† 21 † and may be viewed as a and closed for J (c p) or J : These equations are considerably
`generalized ¯ame surface density', u7cu follows a balance more complicated than ¯ame surface density balance equa-
equation similar to the S (c p) balance Eq. (90), where tions, but the wrinkling factor may be more convenient for
surface averages kQl s are replaced by `generalized surface initial and/or boundary conditions (J $ 1 everywhere).
averages' kQls which do not depend on c p: This approach has been explored by Weller et al. [90,91].
1 Z1 6.4. Statistical approaches: probability density function
kQls ˆ kQl s S…c p †dcp …96†
u7cu 0
6.4.1. Introduction
In ¯ame surface density models, under a ¯amelet assump- Predictions of radicals and intermediate species such as
tion, for any values of c p, the ¯ame surface is assumed to OH, or pollutants like CO, require the description of the
behave such as S…c p † < u7cu ˆ S: ¯ame front internal structure, for intermediate states
All balance equations remain formally the same but have between fresh and burnt gases in premixed ¯ames or
to be closed. The same modeling issues emerge using the G-
between fuel and oxidizer in non-premixed ¯ames. Even
equation formalism [89]. One needs to develop closures for though G-®eld and density of ¯ame surface S need some
the turbulent ¯ux of ¯ame surface, for the propagation vel- statistical treatments, they are initially based on a geo-
ocity of the surface, as well as for the effects of curvature
metrical view describing the ¯ame as a thin interface. In
and strain rate. probability density function methods, one wishes to relax
this assumption by focussing on the statistical properties of
6.3.3. Flame wrinkling description intermediate states within the ¯ame front.
The previous formalism may be recast in term of ¯ame The probability density function (pdf) P…Y  p ; x; t† quanti-
surface wrinkling. The basic idea is to introduce the ratio ®es the probability to ®nd, for a given location x and a time t,
J (c p) of the ¯ame surface to its projection in the direction of a variable Y (mass fraction, temperature, velocity, etc.)
propagation np: within the range ‰Y p 2 DY=2; Y p 1 DY=2Š: This probability
   p ; x; t†DY [92±96]. The pdf satis®es the
 p† is equal to P…Y
S…c p † u7cuuc ˆ c p P…c
p
J…c † ˆ ˆ   following simple relations:
np ´…nu7cud…c 2 c p †† np ´ nu7cud…c 2 c p † Z Z
 p ; x; t†dY p ˆ 1
P…Y  p ; x; t†dY p ˆ Y…x;
P…Y  t†
  Y Y
u7cuuc ˆ c p 1 Z
ˆ   ˆ …97†  2 P…Y
…Y p 2 Y†  p ; x; t†dY p ˆ Y 02 …x; t†
np ´ nu7cuuc ˆ c p np ´knls
Y

where n and np are the unit vectors normal to the instan- where Y p is the sample space variable corresponding to the
taneous ¯ame front and to the mean propagating direction random variable Y. When more than one variable is required
respectively. These vectors are chosen pointing towards the to capture the ¯ame structure, a joint probability density
 1p ; ¼; YNp ; x; t† is introduced [97]. The mean
function P…Y
fresh gases and are given by:
  burning rate (or any mean quantity) is then estimated as:
7c 7cuc ˆ c p
nˆ2 ; n p ˆ 2   …98† v_ Y1 …x; t†
u7cu
7cuc ˆ c p Z Z
ˆ ¼  1p ; ¼; YNp ; x; t†dY1p ¼dYNp
v_ Y1 …Y1p ; ¼; YNp †P…Y
Then: Y1 YN
  …101†
u7cuuc ˆ c p S…c p †
J…c p † ˆ   ˆ   …99† Conditional statistics have been used to de®ned the ¯ame
 p
7cuc ˆ c p 7cuc ˆ c p P…c † surface density (Section 6.3.2), these conditional means are
also useful in a pdf context. Consider a non-premixed ¯ame
As in Section 6.3.2, a generalized ¯ame surface wrinkling, where the chemistry is reduced to a single step reaction, and
J ; is introduced: where radiative heat losses are neglected. Laminar combus-
Z1 tion would be parameterized with two variables, for
S…c p †dcp example, fuel mass fraction YF and mixture fraction Z (see
 0 u7cu S
J ˆ Z1   ˆ ˆ …100† Section 3.2). The turbulent ¯ame is then fully described by
 p p u7cu
 u7cu

7cuc ˆ c p P…c †dc the joint pdf of mixture fraction and fuel mass fraction,
0  Fp ; Z p ; x; t†: For such ¯ames, it is interesting to focus on
P…Y
For an in®nitely thin ¯ame front, J…c p † ˆ J for any c p the statistical properties of the fuel mass fraction YF for a
value. given value of the mixture fraction Z (Section 3.2 and
Flame surface density and ¯ame wrinkling factor are Fig. 5).The conditional pdf P c …YFp uZ p ; x; t† is introduced
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 219
p
and de®ned for a given value of Z as: variable c. The time evolution of a probability density func-
 Fp ; Z p ; x; t† ˆ P c …YFp uZ p ; x; t†P…Z
 p ; x; t† tion is easily derived by expressing the pdf as the average of
P…Y
a function d…cp 2 c…x; t†† de®ned as [100]:
The conditional mean is readily de®ned from: 
d cp 2 c…x; t† ˆ 1=Dcp if cp 2 Dcp =2 , c , cp 1 Dcp =2
  Z1
YF uZ…x; t† ˆ Z p ˆ YFp P c …YFp uZ p ; x; t†dYFp d…cp 2 c…x; t†† ˆ 0 otherwise
0
By de®nition:
 p ; x; t† describes the statistical
In this decomposition, P…Z
properties of fuel/air
 mixing, whereas P c …YFp uZ p ; x; t† and  p ; x; t† ˆ d…cp 2 c…x; t††
P…c …102†
p
YF uZ…x; t† ˆ Z are linked to the internal structure of the
¯ame front. The time evolution of the reactive species c…x; t† is given by:
Two main directions may be chosen to built numerical 2c 1
models from pdf: ˆ 2u´7c 1 ‰7´…rD7c…x; t†† 1 v_ Š …103†
2t r

² To presume the pdf shape from available mean quantities. and d…c p 2 c…x; t†† satis®es:
² To solve a balance equation for the pdf. 2 2 2c…x; t†
‰d…c p 2 c…x; t††Š ˆ ‰d…cp 2 c…x; t††Š
2C 2c 2C
6.4.2. Presumed probability density functions
The simplest approach to the mixing between fuel and 2 2c…x; t†
ˆ2 ‰d…cp 2 c…x; t††Š …104†
oxidizer in a non-premixed ¯ame is to presume the shape 2c p 2C
of the pdf P…Z ~ p ; x; t†; usually with a Beta function [72]:
b21 22
 …Z p †a21 1 2 Z p ‰d…cp 2 c…x; t††Š
P~ Z p ; x; t ˆ Z1 2C 2
b21 1
…Z 1 †a21 1 2 Z 1 dZ 2 22 c…x; t† 22
p
0 ˆ2 p ‰d…c 2 c…x; t††Š 2
1 p2 ‰d…cp 2 c…x; t††Š
2c 2C 2c
This presumed pdf should reproduce the mean of the !2
mixture fraction Z~ and its variance Zf
002 : 2c…x; t†
 (105)
Z1  2C
Z~ ˆ Z p P~ Z p ; x; t dZ p ;
0
where C can either be time or any spatial coordinates. These
Z1 2  relations (104) and (105) and Eq. (102) are key relations,
Zf
002
ˆ Z p 2 Z~ P~ Z p ; x; t dZ p useful to obtain all equations discussed in this section.
0
After simple manipulations combining Eqs. (102) and
and using the relation [98]: (103) with Eq. (104) the transport equation for the pdf
Z1  may be written:
…Z p †n P~ Z p ; x; t dZ p " ! #
0 2  p 2 2c…x; t† p  p
‰P…c ; x; t†Š ˆ 2 p uc…x; t† ˆ c P…c ; x; t†
a…a 1 1†¼…a 1 …n 2 1†† 2t 2c 2t
ˆ
…a 1 b†…a 1 b 1 1†¼…a 1 b 1 …n 2 1†† …106†
The two-parameters a and b are determined as: where conditional averaging
!  
~ 2 Z†
Z…1 ~ 1  
a ˆ Z~ 2 1 $ 0; bˆa 21 $0  p ; x; t†
Q…c†d…cp 2 c…x; t†† ˆ Q…c†uc…x; t† ˆ cp P…c
Zf002 Z~

This technique requires the solving of a balance equation for is introduced.


the mean and the variance of Z (Section 9.2). Zf 002
vanishes Eq. (106) shows that the time evolution of the pdf is
when reactants are perfectly mixed and reaches its maxi- controlled by a ¯ux in the sample space c p. This ¯ux is
 driven by a velocity equal to the conditional mean of the
mum value, Z~ 1 2 Z~ ; when fuel and oxidizer are comple-
tely segregated (Section 5.3), then the pdf takes the form of a time evolution of the progress variable c. In other words,
double peak function, with peaks located at Z ˆ 0 and Z ˆ 1: when the mean of …2c…x; t†=2t† is non-zero for the value c ˆ
This approach is also used in premixed ¯ames replacing the cp ; the probability of ®nding the occurrence c ˆ cp is modi-
mixture fraction Z by the progress variable c [99]. ®ed. Because the probability to ®nd all the possible values is
constant and equal to unity:
6.4.3. Pdf balance equation Z1
 p ; x; t†dcp ˆ 1
P…c
The pdf balance equation is ®rst written for the progress 0
220 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

an increase or a decrease of the density of probability for ² In terms of molecular diffusion:


c ˆ cp ; implies a modi®cation of this density for other 2  p
values of c p, justifying the convective term in sample ‰P…c ; x; t†Š
2t
space. This term insures that the density of probability is " !
transported from point to point in the sample space c p in a 2 1
ˆ2 p 2u´7c 1 7´…rD7c†uc…x; t† ˆ c p
conservative manner. The ¯ux depends on the conditional 2c r
mean of the time evolution of the progress variable given by ! #
Eq. (103):  p ; x; t†
1 v_ …cp † P…c (109)
!
2c
uc…x; t† ˆ cp ² In terms of scalar dissipation rate:
2t
  2  p
‰P…c ; x; t†Š
ˆ 2u´7cuc…x; t† ˆ c p 2t
2 h  
 p ; x; t†
i
! ˆ2 p 2u´7cuc…x; t† ˆ c p 1 v_ …cp † P…c
1   2c
1 ‰7´…rD7c…x; t††Šuc…x; t† ˆ c 1 v_ uc…x; t† ˆ cp
p
r 22 h 
 p ; x; t†
i
2 p2 xuc…x; t† ˆ cp P…c
…107† 2c

In Eq. (107), the conditional value of the source term is a The convective term may be split into mean,
function of c de®ned in one-point and is exactly known  p ; x; t†
 P…c
u´7
since:
and ¯uctuating components. Using an eddy viscosity model,
  the ¯uctuating part becomes:
v_ …c†uc…x; t† ˆ cp ˆ v_ …cp †
 p ; x; t†
2nt 7P…c
The main advantage of pdfs in turbulent combustion lies in leading to the pdf balance equation:
this availability to deal with chemistry, any term de®ned in
2  p  p ; x; t†
one-point (as chemical source) is closed. Nonetheless, this  P…c
‰P…c ; x; t†Š 1 u´7
advantage is offset by the fact that reactants are brought to 2t
the reaction zone by diffusion, and the conditional mean of  p ; x; t† 2 2 ‰v_ …cp †P…c
ˆ nt 7P…c  p ; x; t†Š
2c…x; t†=2t also includes a conditional diffusive term: 2cp
! 22 h 
 p ; x; t†
i
1 2 p2 xuc…x; t† ˆ cp P…c …110†
…7´…rD7c††uc…x; t† ˆ c p
…108† 2c
r Weighted, or Favre, averages are also introduced in pdfs, for
instance when r ˆ r…c† :
named the micromixing term and remaining unclosed (as
any term involving spatial derivatives). This micromixing r P…c  p ; x; t†
~ p ; x; t† ˆ r…cp †d…c…x; t† 2 cp † ˆ r…cp †P…c
term may be rewritten with the scalar dissipation rate,
When more than one species is taken into account, the pdf
x ˆ Du7cu 2 ; using Eq. (102) with Eq. (105) and assuming
balance equation is derived using the same formalism with
rD < cst :
~ 1p ; ¼; YNp † ˆ r…Y1 ; ¼; YNp †d…Y1 2 Y1p †¼d…YN 2 YNp † ‰93Š:
r P…Y
2 h 2

p  p
i
D7 cuc…x; t† ˆ c P…c ; x; t†
2cp
2 ~ p
r P…Y1 ; ¼YNp ; x; t†
h
2  i 2t
 p ; x; t† 1 2
ˆ 2D7 2 P…c  p ; x; t†
xuc…x; t† ˆ cp P…c
2c p2 XN
2 h  
p p
ˆ 2r 2u´7Y i uY…x; t† ˆ Y 1 v
_ i
The ®rst term in the RHS, D7 2 P…c  p ; x; t† is usually neg- iˆ1
2Yip
ligible compared to the transport due to velocity ¯uctua-  i
tions. The total dissipation rate, x is recovered as:  P~ Y1p ; ¼; YNp ; x; t
" !
Z1   XN X N
2 2 2Yi 2Yj
x ˆ  p ; x; t†dcp
xuc…x; t† ˆ cp P…c 2 r D uY…x; t† ˆ Y p
2Y p p
0 iˆ1 jˆ1 i 2Yj 2xk 2xk

Using diffusion or scalar dissipation rate, the probability #


 
density function balance equation may be organized in  P~ Y1p ; ¼; YNp ; x; t (111)
two different forms:
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 221

The same terms are identi®ed on the RHS: convection by the In premixed ¯ames, the mean value, Y~ i of Yi may be esti-
conditional velocity and by the chemical source, and micro- mated as:
mixing. As was done in Eq. (110), the convective term Z1  
in physical space can be decomposed into mean and r Y~ i ˆ  p ; x; t†dcp
rYi uc ˆ cp P…c …113†
0
¯uctuating parts.
One may solve a balance equation for the conditional quan-
6.4.4. Joint velocity/concentrations pdf tities Qi(c p) de®ned as:
To avoid the gradient transport assumption for the  
rYi uc ˆ cp
turbulent ¯ux, …u 0 ´7cuc…x; t† ˆ c p †; the joint velocity/ Qi …cp † ˆ  
concentration pdf is introduced. Once this joint pdf is ruc ˆ cp
known, turbulence models, such as k± 1 , are, a priori, no
longer required for the mean ¯ow. However, an equation This balance equation is [104,105]:
for 1 is still needed to estimate a characteristic mixing time,   2Q   2Q   22 Q
i i i
except when the frequency of mixing is also included in the ruc ˆ cp ˆ 2 rui uc ˆ cp 1 rxuc ˆ cp
2t 2xi 2cp2
joint pdf [101]. The transport equation for this joint compo-
 
sition/velocity pdf is given below for the progress variable c, 1 v_ i uc ˆ cp 1 EQi 1 EYi (114)
where the RHS contains the unclosed terms.
2  p p  p ; cp ; x; t† The two last terms of Eq. (114) are usually neglected,
‰P…u ; c ; x; t†Š 1 up ´7P…u
2t EQi appears from molecular diffusion along with differ-
  ential diffusion effects across the iso-c surface, EYi
1 2p 2  p p
1 n7 2 ui 2 ‰P…u ; c ; x; t†Š represents the effects of turbulent ¯uctuations on the
r 2xi 2upi
deviation from the conditional mean. The three ®rst
2  p ; cp ; x; t†Š terms on the right hand side are unclosed, they describe
1 ‰v_ …cp †P…u
2cp convective transport, micromixing (x enters this term) and
" ! # chemical source. Closures for the conditional values of the
2 1 2p 0  p p scalar dissipation rate, but also for the conditional value of
ˆ p u ˆ u ; c ˆ c P…u ; c ; x; t†
p p
2ui r 2xi the chemical source of Yi, calculated for a given value of the
|‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚}
Pressure fluctuations progress variable c p or the mixture fraction Z, are required
[106,107]. One equation has to be solved for each value of
2 h  i
 p ; cp ; x; t†
2 p n7 2 u 0i uu ˆ up ; c ˆ cp P…u c p retained. The number of these values is determined from
2ui the accuracy required to estimate both the mean from Eq.
|‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚ ‚{z‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚}
Viscous dissipation (113) and the second order derivative in the sample space
(i.e. 2 2/2c p2) found in Eq. (114). On the other hand, the
2 h 
 p ; cp ; x; t†
i
2 p D7 2 cuu ˆ up ; c ˆ cp P…u …112† probability density function entering expression (113),
2c
|‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚}  p ; x; t† is presumed.
P…c
Molecular diffusion
CMC may also be viewed as a multi-surface
 description,
The LHS terms are closed and represent respectively accu- any conditional quantity rYi uc ˆ cp corresponding to the
mulation, convection in physical space by the random vel- conditional average of Yi along the iso-surface c ˆ cp :
ocity ®eld (incorporating turbulent transport), convection of
the pdf in velocity space, here the convective velocity is the 6.5. Similarities and links between the tools
mean of the viscous dissipation and the mean pressure gradi-
ent, and ®nally the closed chemical source. The unclosed Major links between the tools used in turbulent combus-
terms (RHS) are the pressure ¯uctuations, the ¯uctuating tion modeling are now developed. Without loss of the
part of the viscous dissipation and micromixing. All these generic character of the discussion, we consider the case
phenomena remain to be closed. of a turbulent premixed ¯ame represented with a progress
variable c. To describe this turbulent ¯ame, three quantities
are useful:
6.4.5. Conditional moment closure (CMC)
Conditional moment closure modeling was ®rst proposed ² The scalar dissipation rate of the progress variable
in Refs. [102,103]. As with the pdf, the idea is to focus on r x~ ˆ rD7c´7c.
particular states between fresh gases and fully burnt product  p ; x; t†:
² The pdf of the progress variable P…c
in premixed ¯ames, or, between fuel and oxidizer in non- ² The ¯ame surface density S or the mean ®eld G:
premixed combustion. However, here only conditional
moments …rYi uc ˆ cp † are considered. In premixed ¯ames, Simple links exist between these quantities:
the conditional quantity is the progress variable c, whereas
for non-premixed combustion, the mixture fraction is used. ² The conditional value of the scalar dissipation rate enters
222 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

Table 2
 p †, or surface density, S (c p), of iso-surface c ˆ c p and
Exact relations between mean scalar dissipation rate, x~ , probability density function, P…c
generalized ¯ame surface density S ˆ u7cu: Connections with the G-equation are readily obtained with Eq. (117). kQl s is a surface average (Eq.
(91)), kQls corresponds to generalized surface average and is de®ned in Eq. (96). c is the conditional variable (progress variable in premixed
¯ames, mixture fraction in non-premixed combustion)

Scalar dissipation rate r x~ Probability density function Flame surface density S…c p †
 p†
P…c

Z1   Z1
r x~ ±  p †dcp
rDu7cu 2 uc ˆ cp P…c krDu7cul s S…c p †dcp
  0 0  
 p†
P…c xuc ˆ cp via PDF Eq: ± S…c p †= u7cuuc ˆ c p
r x~ Z1   Z1
S ˆ u7cu  p †dcp
u7cuuc p P…c S…c p †dcp
krDu7cul s 0 0

the pdf transport equation (Eq. (110)), therefore there is a where the conditional mean Qi …cp † may be written:
direct connection between P…c p ; x; t† and x~ :  
² The surface density function S…c p ; x; t†; or ¯ame surface, rYi uc ˆ cp krYi =u7cul s
p
Qi …c † ˆ   ˆ …119†
is  pdf via the conditional value of u7cu; S ˆ
 related to the ruc ˆ cp kr=u7cul s
 p † (Eq. (86)).
u7cuuc ˆ c p P…c
and appears as directly related to c ˆ cp surface averaged
Since x is proportional to u7cu; the ¯ame surface S , the quantities.
pdf P…c p † and the dissipation rate x~ are very strongly The links between the combustion modeling tools are
related. Using the joint pdf of c and x , the ¯ame surface summarized in Table 2. The mean burning rate is given by:
density may be written:
! Z1
r  p †dcp
x v_ ˆ v_ …cp †P…c …120†
S…c p ; x; t† ˆ  p ; x; t†
uc ˆ cp P…c 0
D
s Using relations (86) and (91) and Eq. (116), this expression
Z xp  p p becomes:
ˆ P…c ; x ; x; t†dxp …115†
x D Z1   Z1  v_ 
p v_ ˆ  p †dcp ˆ
v_ uc ˆ cp P…c S…c p †dcp
and characteristic length scales dc ˆ D=x of the iso-c 0 0 u7cu s
p
distribution are embedded within S (c ).      
Combining Eq. (86) and Eq. (91) with Eq. (96), the mean v_ v_ 1 v_
ˆ u7cu ˆ p S ˆ r x~
c-scalar dissipation rate is also a function of surface u7cu s x=D s krDu7cul s u7cu s
densities: …121†
Z1
r x~ ˆ rDu7cu 2 ˆ krDu7cul s S…c p †dcp ˆ krDu7cul s u7cu Models based on probability density functions, conditional
0 means, ¯ame surface density function and generalized ¯ame
…116† surface density are then related via the scalar dissipation rate.
where the generalized surface average kQl s is de®ned by These expressions are extended to burning rate depending
Eq. (96). Relations between ¯ame surface densities and on many species using multi-dimensional pdfs and con-
scalar dissipation rates were anticipated by Borghi [108]. ditional averaging. When the local reaction rate is a function
In premixed combustion, using a G-®eld equation, the of various quantities (species mass fractions, temperature,
¯ame front is identi®ed to a given level G ˆ G0 [29]. etc.), Yi and a sampling scalar c:
The ¯ame surface density is then: Z Z Z1
  v_ ˆ ¼  1p ; ¼; YNp ; cp †dY1p ; ¼; dYNp dcp
v_ i …Y1p ; ¼; YNp ; cp †P…Y
S…G 0 † ˆ u7GuuG ˆ G 0 P…G  0† …117† Y1 YN 0

Z1  Z Z 
Hence, previous relations may be recast in terms of the G- ˆ ¼  p †dcp
v_ i …Y1p ; ¼; YNp ; cp †P c …Y1p ; ¼; YNp ucp †dY1p ; ¼; dYNp P…c
0 Y1 YN
equation. |‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚}

The CMC formalism (see Section 6.4.5) may be re- v_ ;…Y1 ;¼;YN ;c†ucˆcp

organized in term of ¯ame surface density: …122†


Z1   Z1  rY   1p ; ¼; YNp ; cp † ˆ
r Y~ i ˆ  p ; x; t†dcp ˆ
rYi uc ˆ cp P…c i
S…c p †dcp decomposing the joint-pdf as P…Y
0 0 u7cu s  p †: Note that such a relation may be
P c …Y1p ; ¼; YNp ucp †P…c
…118† used when in premixed ¯ames the progress variable, based
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 223

Table 3
Tools for turbulent combustion modeling

Description Tools Modeling issues

Geometrical G-®eld with G ˆ Gp at the ¯ame Propagation speed ST


~ 1 u´7
2G=2t ~
~ G~ ˆ …r u =r †ST u7Gu
V_ i Turbulence
Flame surface S v_ i ˆ Combustion S
Algebraic closure or transport equation Total stretch ˆ curvature 1 strain rate
Small scales Scalar dissipation rate, r x~ ˆ rDu7Yu 2 Fast chemistry v~_ < x~
Algebraic closure or transport equation Provide the rate of micromixing
Statistical Probability
p
density function
~ F ; Z p ; x; t†
P…Y Micromixing
(1) Presumed 1 ¯amelets (1) Strategy to generate the pdf
(2) Transport equation ~ ˆ¼
(2) Solve 2P=2t
Conditional
 mean, CMC,
 R p presumed
p
PDF p
Micromixing and conditional source
YF uZ…x; t† ˆ Z p ˆ 10 YF P c …YF uZ p ; x; t†dYF
 
Fundamental links: S…x; t† ˆ u7cuuc ˆ c p P…c  p ; x; t† and r x~ ˆ krDu7cul s u7cu

on reactant mass fractions, and the temperature are indepen- global turbulent ¯ame speed ST. From experimental data
dent (compressibility effects, heat losses, non-unity Lewis [109,110] or theoretical analysis (Renormalization group
numbers, etc.). theory [111]), the following expression has been proposed:
The fundamentals of turbulent combustion modeling !n
clearly rely on pdf, ¯ame surface density, G-®eld and scalar ST u0
ˆ11a …123†
dissipation rate concepts. The previous relations may be SL SL
used to carefully compare the proposed closure schemes
where a and n are two model constants of the order of unity.
and Table 3 summarizes the tools and their related modeling
u' is the turbulent velocity (i.e the RMS velocity).
issues. Various expressions for mean reaction rates are
Unfortunately, the turbulent ¯ame speed ST is not a fully
displayed in Table 4. Major differences between the various
well de®ned quantity [112]. Experimental data exhibit a
approaches only appear when closing the unknown quanti-
large scatter because they depend on various parameters
ties, but at the light of these relations, many closures are
(chemistry characteristics, turbulence scales, ¯ow geometry,
essentially equivalent.
etc.). While this global approach is not particularly well suited
to close Favre averaged transport equations, it may be of
7. Reynolds-averaged models for turbulent premixed interest in the context of Large Eddy Simulation [79,113].
combustion
7.2. Eddy-Break-Up model
7.1. Turbulent ¯ame speed
Devised in Ref. [114], this model is based on a phenom-
Turbulent premixed ¯ames may be described in terms of a enological analysis of turbulent combustion assuming high

Table 4
 p † or surface
Exact expressions for the averaged reaction rate as a function of the mean scalar dissipation, x~ ; probability density function, P…c
density, S (c p) of iso-surface c ˆ c p : Connections with the G-equation are readily obtained with Eq. (117). kQls is a surface average (Eq. (91)),
kQls corresponds to generalized surface average and is de®ned in Eq. (96)

Tools Averaged reaction rate

R1
v_
v_
0 S…c p †dcp
u7cu s u7cu s
Scalar dissipation r R1 p p
x
~ ˆ r
 x~
0 krDu7cul s S…c †dc krDu7cul s

R1
Probability density function 0
 p †dcp
v_ …cp †P…c
R1
v_
v_
Flame surface density 0 S…c p †dcp ˆ u7cu
u7cu s u7cu s
224 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

Reynolds …Re @ 1† and DamkoÈhler …Da @ 1† numbers. The chemical features [115]. Eddy-Break-Up modeling tends to
reaction zone is viewed as a collection of fresh and burnt overestimate the reaction rate, especially in highly strained
gases pockets. Following the Kolmogorov cascade, turbu- regions, where the ratio 1 /k is large (¯ame-holder wakes,
lence leads to a break down of fresh gases structures. walls, etc.).
Accordingly, the mean reaction rate is mainly controlled
by the turbulent mixing time t t. When oxidizer is in excess,
the mean reaction rate is expressed as: 7.3. Bray±Moss±Libby model
q
Yf002 7.3.1. Introduction
v_ F ˆ 2CEBU r F
…124† Known under the initials of its authors, Bray, Moss and
tt
Libby, or, from the involved physical hypothesis, BiModal
where Y 00F denotes the fuel mass fraction ¯uctuations and Limit, this model, ®rst proposed in 1977 [116], has been the
CEBU is a model constant of the order of unity. The turbulent subject of a large amount of work leading to many improve-
mixing time, t t is estimated from the turbulence kinetic ments (see papers by Bray, Moss and Libby, and then by
energy k and its dissipation rate 1 according to: Bray, Champion and Libby). Combining a statistical
approach using probability density functions and a physical
k
tt ˆ analysis, this model has evidenced some special features of
1
turbulent premixed combustion (counter-gradient turbulent
as an approximation of the characteristic time of the integral transport, ¯ame turbulence generation, etc.). The presenta-
length scales of the turbulent ¯ow ®eld. tion is mainly limited here to basic concepts of the Bray±
The reaction rate may be recast in terms of progress vari- Moss±Libby (BML) formulation.
able, c, as: A one-step, irreversible chemical reaction between two
q reacting species, fresh gases (R) and combustion products
cf002
(P) is considered:
v_ ˆ 2CEBU r …125†
tt
R!P
Mass fraction ¯uctuations Yf 002 (or progress variable ¯uctua-
F
f002
tions c ) must be modeled and may be estimated from a Classical assumptions are made to simplify the formula-
balance equation (see Eq. (77)). Assuming an in®nitely thin tion: perfect gases, incompressible ¯ows, constant chemical
¯ame front (i.e. c ˆ 0 or c ˆ 1), cf 002
is easily estimated properties, unity Lewis numbers, etc. A progress variable, c,
because c2 ˆ c : of the chemical reaction is introduced where c ˆ 0 in fresh
gases and c ˆ 1 in fully burnt gases, as described in Section
r cf
002 ~ 2 ˆ r …ce2 2 c~2 † ˆ r c…1
ˆ r…c 2 c† ~ 2 c†
~ …126† 3.1.
The basic idea of the BML formulation is to presume the
The square root has been introduced for dimensional reasons
probability density function of the progress variable c at a
in Eqs. (124) and (125) but, unfortunately, Eqs. (125) and
given location …x; t† as a sum of fresh, fully burnt and burn-
(126) lead to inconsistencies because the c~ derivative of v_ ;
ing gases contributions (Fig. 15):
d v_ =d c;
~ is in®nite both when c~ ˆ 0 and when c~ ˆ 1 (Borghi,
1999, private communication). Then, a corrected version of  p ; x; t† ˆ a…x; t†d…cp † 1 b…x; t†d…1 2 cp †
P…c
the Eddy-Break-Up model, without the square root, is used |‚‚‚{z‚‚‚} |‚‚‚‚‚{z‚‚‚‚‚}
fresh gases burnt gases
for practical simulations:
1 1 g…x; t†f …cp ; x; t† …129†
v_ ˆ CEBU r c…1 ~ 2 c†~ …127† |‚‚‚‚‚{z‚‚‚‚‚}
k burning gases

or, in terms of fuel mass fraction:


! where a , b and g respectively denote the probability to
1 ~F
Y ~F
Y have, at location …x; t†; fresh gases, burnt gases and reacting
v_ F ˆ 2CEBU r 12 0 …128†
k YF0 YF mixture. d…cp † and d…1 2 cp † are respectively the Dirac delta
functions corresponding to fresh gases …c ˆ 0† and fully
where YF0 is the initial fuel mass fraction in the reactant burnt ones …c ˆ 1†:
stream, assuming excess of oxidizer. Normalization of the probability density function:
The EBU model was found attractive because the reaction
rate is simply written as a function of known quantities Z1
without any additional transport equation and is available  p ; x; t†dcp ˆ 1
P…c …130†
0
in most commercial CFD codes. The modeled reaction rate
does not depend on chemical characteristics and assumes a leads to the following relations:
homogeneous and isotropic turbulence. Some adjustments
of the model constant CEBU have been proposed to mimic a1b1gˆ1 …131†
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 225

Fig. 15. Probability density function in premixed turbulent


Fig. 16. Intermittency between fresh and fully burnt gases at a given
combustion. location in the reaction zone. This signal corresponds to a bimodal
(c ˆ 0 and c ˆ 1) probability density function.
Z1
f …cp ; x; t†dcp ˆ 1 …132† function of the Favre average progress variable c~ :
0
Z1
with f …0† ˆ f …1† ˆ 0:  p †dcp ˆ rb b
r c~ ˆ r c~ ˆ rcp P…c …138†
The balance equation for the progress variable c may be 0
written: where r b is the burnt gases density.
2rc r c~ r c~
1 7´…ruc† ˆ 7´…rD7c† 1 v_ …133† bˆ and a ˆ 1 2 …139†
2t rb rb
This equation is averaged and the mean reaction rate, at the
The mean density r may be written:
location …x; t† is: !
Z1 rec rec
Z1  p p
 p ; x; t†dcp r ˆ rP…c †dc ˆ aru 1 brb ˆ 1 2 r 1 r
v_ …x; t† ˆ v_ …c p †P…c …134† 0 rb u rb b
0
…140†
leading to Eq. (129):
Z1 One may also introduce the reaction heat release factor t ,
v_ …x; t† ˆ g…x; t† v_ …cp †f …cp ; x; t†dcp …135† de®ned as:
0
ru T
All studies devoted to this line of models are based on such a tˆ 21ˆ b 21 …141†
rb Tu
formulation. The objective is now to determine the unknown
functions a , b , g and the probability density function f. leading to:
Using the DamkoÈhler number, Da, comparing turbulence ru ˆ …1 1 t†rb ˆ r …1 1 tc†
~ …142†
and chemical time scales, and, the turbulent Reynolds
number Re, we will focus only on the case where corresponding to the perfect gases state law, assuming a
Re @ Da @ 1. In this situation, the combustion is controlled constant pressure P. Then, the probabilities a and b
by the turbulent transport and the reaction zone may be become:
assumed to be in®nitely thin. Accordingly, g ! 1 (i.e. 1 2 c~ …1 1 t†c~
a @ g and b @ g ). aˆ ; bˆ …143†
1 1 tc~ 1 1 tc~

The probability density function P…c† is determined and
7.3.2. BML model analysis depends only on the mean progress variable c~ (and on the
This model is developed under the assumption that heat release factor, t , which is ®xed for a given chemical
Re @ Da @ 1, corresponding to g ! 1. At a given location reaction). The BML model involves a presumed pdf (Eq.
in the ¯ow, an intermittency between fresh gases …c ˆ 0† (135)), however, the mean reaction rate cannot be calculated
and fully burnt ones …c ˆ 1† is observed and the probability from the pdf since g was neglected in Eq. (129)).
density function of the progress variable c reduces to: Starting from the conservative and non-conservative
 p ; x; t† ˆ a…x; t†d…cp † 1 b…x; t†d…1 2 cp † 1 O…1=Da†
P…c forms of the progress variable c balance equations:
…136† 2rc
1 7´…ruc† ˆ 7´…rD7c† 1 v_
which is `quasi-bimodal' for large DamkoÈhler numbers. 2t
Then at the point …x; t† inside the reaction zone, c looks as 2c
a telegraphic signal as displayed in Fig. 16. This signal also r 1 ru´7c ˆ 7´…rD7c† 1 v_
2t
satis®es:
and multiplying by c and adding these equations:
c2 ˆ c 1 O…1=Da† and c…1 2 c† ˆ O…1=Da† …137†
2rc 2
Under this assumption, a and b are easily determined as a 1 7´…ruc 2 † ˆ 7´…rD7c 2 † 2 2rD7c´7c 1 2cv_ …144†
2t
226 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

Subtracting the balance equation for c 2 (Eq. (144)) to the linking mean reaction rate and scalar dissipation may also be
balance equation for the progress variable c (Eq. (133)) recovered from the general relations between modeling
leads to a balance equation for c…1 2 c† : tools (Eq. (121)):
2  
‰rc…1 2 c†Š 1 7´‰ruc…1 2 c†Š 1 v_
2t v_ ˆ r x~ …153†
krDu7cul s u7cu s
ˆ 7´…rD7‰c…1 2 c†Š† 1 2rD7c´7c 2 2cv_ 1 v_ …145†
To link mean reaction rate and scalar dissipation rate using
Under the assumption of the BML model, the progress vari- the ¯amelet analysis, Eq. (153) requires estimates for v_ =u7cu
able c is equal to zero or to unity. Accordingly, c…1 2 c† ˆ 0 and rDu7cu; averaging of these quantities along iso-surface
and the balance Eq. (145) reduces to: c ˆ c p and integration over all the possible c p values.
2rD7c´7c ˆ 2cv_ 2 v_ …146† Assuming that in®nitely thin ¯ame elements may be viewed
as one-dimensional, steady state, premixed laminar ¯ame
leading to, after averaging: propagating at a given laminar ¯ame speed SL, the balance
2rD7c´7c ˆ …2c m 2 1†v_ …147† equation for the progress variable c in this 1D ¯ame is:

where a progress variable cm, de®ned as: r 0 SL 7´c ˆ 7´…rD7c† 1 v_ …154†


Z1
cv_ f …c†dc where r 0 denotes the density in the fresh gases. Integrating
cm ˆ Z01 …148† Eq. (154) across all the ¯ame front from fresh gases to burnt
v_ f …c†dc gases …21 # x # 11† and up to a given location x0, corre-
0 sponding to a progress variable c0 …21 # x # x 0 †; leads to:
has been introduced and characterizes the chemical reaction. Z1 1 Z1 v_
The mean reaction rate v_ becomes: r0 SL ˆ v_ dx ˆ dc …155†
21 0 u7cu
rx
v_ ˆ 2 …149†
2cm 2 1 Zx0
…rD7c† x0 ˆ r0 SL c0 2 v_ dx
where rx is the scalar dissipation rate of the progress 21
variable c. Z1  Zc0 v_
v_
2c 2c ˆ dc c 0 2 dc …156†
rx ˆ rx
e ˆ rD7c´7c ˆ rD …150† 0 u7cu 0 u7cu
2x i 2xi
As the ¯ame front is supposed to be thin and planar, all c p
The mean reaction rate v_ is then related to the dissipation
iso-surfaces have the same surface and all quantities are
rate x~ ; describing the turbulent mixing, and to cm, character-
constant along these surfaces, moreover the ¯ame is
izing the chemical reaction. A transport equation for the
supposed locally one-dimensional and u7cu ˆ 7c: Then:
scalar dissipation rate may be written and solved [117], or
one may postulate a linear relaxation of the ¯uctuations Z1 Z1
generated by micromixing (Section 9.2.2), leading to: kv_ =u7cul s dcp v_ =u7cudc
0
Z1 ˆ Z10
rc 002 krDu7cul s dcp rDu7cudc
rx ˆ …151† 0 0
tt
Z1 1
where a turbulent time scale, t t is introduced. Assuming an v_ dx
intermittency between fresh and burnt gases (c ˆ 0 or ˆ Z1 1 21
Z1 Z1 Z c 0
c ˆ 1), cf002 is given by Eq. (126). Then:
v_ dx c dc 2 …v_ =u7cu†dc dc 0
21 0 0 0
2 r c…1
~ 2 c†
~
v_ ˆ …152† Z1 1
2cm 2 1 tt
2 v_ dx
21 2
The Eddy-Break-Up model expression (Eq. (127)) is ˆ Z1 1 Z1 1 ˆ …157†
2cm 2 1
recovered (Section 7.2). The BML model may then be 2 cv_ dx 2 v_ dx
21 21
viewed as a theoretical derivation, where the assumptions
made are clearly stated, of the Eddy-Break-Up (EBU) where cm is de®ned as:
model, initially based on a phenomenological approach.
Z1 1
cv_ dx
7.3.3. Recovering mean reaction rate from tools relations
cm ˆ Z2111 …158†
The analysis developed in Section 7.3.2 is the usual
v_ dx
derivation of the BML model. However, expression (149) 21
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 227

Fig. 17. The Reynolds average c of the progress variable cis plotted as a function of the Favre average c~ for various values of the heat release
factor t , assuming a bimodal distribution (c ˆ 0 or c ˆ 1) of c (Eq. (165)).

Then, from Eq. (153): of the progress variable c, which may be recast in terms
rx of ¯ame surface density, exhibiting simple links between
v_ ˆ 2 …159† modeling tools.
2cm 2 1
recovering Eq. (149). This second analysis leads to the 7.3.4. Reynolds and Favre averaging
following comments: Assuming a bimodal distribution of the progress variable
 and Favre …c†
c, Reynolds …c† ~ averages are easily related.
² The de®nition of cm is slightly different in Eqs. (149) and From:
(159). In the ®rst one, the integration is performed over Z1
all possible values of the progress variable c at a given 
c ˆ cP…c†dc ˆb …163†
location …x; t†: In the second analysis, the integration is 0

performed along a normal direction to the ¯ame front, Z1


assumed to be a laminar one-dimensional premixed rc ˆ r c~ ˆ 
rcP…c†dc ˆ rb b ˆ rb c …164†
¯ame. The expressions (148) and (158) are identical 0

when: together with expressions (142) and (143), one easily


  obtains:
1 dc 21
f …c† ˆ …160† …1 1 t†c~
dl dx f c ˆ …165†
1 1 tc~
where …dc=dx†f describes the inner structure of the ¯ame-
let. corresponding to a model for the density/progress variable
² This second derivation clearly exhibits how a ¯amelet correlations (Eq. (15)):
model works (the turbulent ¯ame is viewed as a collec- tc…1
~ 2 c† ~
tion of thin laminar ¯ame elements). The cm parameter is r 0 c 0 ˆ 2r …166†
1 1 tc~
related to the inner structure of the ¯ame front and to the
Reynolds …c† and Favre …c† ~ averages of the instantaneous
properties of the chemical reaction.
progress variable c are compared on Fig. 17 for various
² Making use of the ¯ame surface density S as de®ned by
values of the heat release factor t . The discrepancy between
Eq. (116), the expression for the mean burning rate
becomes: the two quantities strongly increases with t . r 0 c 0 is plotted
as a function of c~ on Fig. 18.
v_ ˆ r 0 SL S …161†
where 7.3.5. Conditional averagingÐcounter-gradient turbulent
transport
2 1 2 1 ~ 2 c†
c…1 ~ The analysis of intermittency between fresh gases …c ˆ 0†
Sˆ r x~ < r
2c m 2 1 r0 SL 2cm 2 1 r0 SL tt and fully burnt gases …c ˆ 1† leads to the introduction of
…162† conditional averaging. The Favre average Q~ of any quantity
Q may be expressed as a function of the conditional
averages for fresh gases …Q u † and fully burnt products …Q b † :
² The BML model has been developed starting from a
statistical analysis (two peaks pdf), the mean reaction Z1
r Q~ ˆ rQ ˆ rQP…c†dc  ˆ aru Q u 1 brb Q b …167†
rate was written in terms of the scalar dissipation rate 0
228 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

Fig. 18. Correlation 2r 0 c 0 =r as a function of the Favre averaged progress variable c~ for various values of the heat release factor t , assuming a
bimodal distribution (c ˆ 0 or c ˆ 1) of c (Eq. (166)).

using Eqs. (142) and (143): using the same formalism:


Q~ ˆ …1 2 c† u
~ Q 1 c~Q b …168† ug
00 u 00 ˆ …1 2 c†u
i j
u
~ 0i u 0j 1 cu
b
~ 0i u 0j 1 c…1 ~ ubi 2 uui †…ubj 2 uuj †
~ 2 c†…
|‚‚‚‚‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚‚‚‚‚}
where Q u and Q b are de®ned as: intermittency

Z1 1 …172†
Q u ˆ QP c …Quc ˆ 0†dQ
21 where one may note a weighted mean between the Reynolds
u b
stresses in the fresh u 0i u 0j and in the burnt gases u 0i u 0j
Z1 1
Q b ˆ QP c …Quc ˆ 1†dQ representative of turbulent motions. The additional term
21 represents the intermittency between fresh and burnt gases.
P c …Quc† is the pdf of Q for the given value c of the progress
variable (conditional pdf, see Section 6.4.1). 7.3.6. Extensions to partially premixed combustion
The components u~k of the mean velocity vector u~ may be Some attempts have been made to extend BML modeling
written as a linear combination of their conditional fresh and concepts to partially premixed combustion, i.e. when reac-
burnt gases averages: tants are not perfectly premixed before burning [118]. Two
dif®culties are then encountered. First, in the de®nition of
~ u ui 1 c~u bi
u~i ˆ …1 2 c† …169† the progress variable c (Eq. (35)), unburnt and burnt gas
temperatures and fuel mass fractions Tu, Tb, YFu and YFb are
Then:
no longer constant and depend on local equivalence ratio. A
r ug
00 c 00 ˆ r …u
f ~ ˆ r …c~ubi 2 u~i c†
~i c† ~ ˆ r c…1 ~ ubi 2 uui †
~ 2 c†… balance equation may be still derived (see, for example, Ref.
i ic 2 u
[119]) but it incorporates additional terms and is not obvious
…170†
to close, especially when rich and lean zones coexist in the
which is the scalar turbulent ¯ux, generally modeled using a same ¯ow ®eld. Mixing should also be taken into account
gradient assumption: through a mixture fraction Z. The challenge is then to model
   p ; Z p ; x; t†: Lahjaily
the joint probability density function P…c
00 c 00 ˆ 2 mt
r ug
2c~
…171†
i et al. [118] write:
Sc 2xi
 p ; Z p ; x; t†
P…c
The two expressions (170) and (171) may describe opposite
¯uxes: consider a left-traveling one dimensional turbulent ˆ a…x; t†d…cp †P u …Z p ; x; t† 1 b…x; t†d…1 2 cp †P b …Z p ; x; t†
¯ame, because of thermal expansion the conditional velo-
city in the burnt gases, ubi ; is expected to be larger than the 1 g…x; t†F…cp ; Z p ; x; t†H…Z p 2 Zmin †
conditional velocity in the fresh gases, uui : According to Eq. |‚‚‚‚‚‚‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚‚‚‚‚‚‚}
burning zones
(170), the turbulent ¯ux, ug 00 00
i c is expected to be positive. On
the other hand, as the mean progress variable gradient is also 1 gm …x; t†Fm …cp ; Z p ; x; t†‰1 2 H…Z p 2 Zmin †Š (173)
|‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚‚}
positive, Eq. (171) leads to a negative value of ug 00 00
i c : This mixing without burning
situation, called `counter-gradient turbulent transport', is a
key point of the BML analysis and will be further discussed where P u …Z p ; x; t† and P b …Z p ; x; t† represent the mixture frac-
in Section 8. tion distributions in fresh and burnt gases respectively. F
The Reynolds stresses ug 00 u 00 may also be decomposed
i j and Fm are the distributions within the reaction zones,
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 229

Their derivation and their histories differ, but they are all based
on similar concepts, described in Section 6.3.2. These models
assume that the chemical reaction occurs in thin layers
separating fresh gases from fully burnt ones (high Dam-
koÈhler number limit). The reaction zone may then be viewed
as a collection of laminar ¯ame elements called ¯amelets.
The ¯ame surface density is here introduced at the light of
experimental data from Refs. [120,121]. The experimental
Fig. 19. Experimental burner. A propane/air premixed ¯ow is
burner is displayed on Fig. 19. A turbulent premixed
injected in a rectangular burner through a grid. The turbulent
propane/air ¯ame is stabilized behind a small cylinder
¯ame is stabilized behind a small cylinder (blockage ratio of 6%)
[120,121]. (blockage ratio of 6%). Flow rates are about 35±100 g/s,
corresponding to inlet velocities between 10 and 30 m/s
assuming to be ¯amelets, and within the non-reactive (turbulence levels from 5 to 10%). Equivalence ratios f
mixing zone between fresh and burnt gases. Combustion are in the range 0.7±1.1. Velocity (laser Doppler veloci-
is supposed to occur when Z p $ Zmin (H is the Heaviside metry), CH and C2 radical emission (reaction rate estima-
function verifying H…z , 0† ˆ 0 and H…z $ 0† ˆ 1). Under tion) and high-speed laser tomography (¯ame front
BML assumptions, the two last contributions are neglected characteristics) measurements have been performed and
(g p 1 and gm p 1) and dilution effects are incorporated are described in Refs. [120±123].
through P u …Z p ; x; t† and P b …Z p ; x; t†: In Fig. 20 (half burner), ¯ame surface density pro®les are
plotted as a function of the transverse coordinate for various
downstream locations and for two equivalence ratios.
7.4. Models based on the ¯ame surface area estimation

7.4.1. Introduction 7.4.2. Algebraic expressions for the ¯ame surface density S
Several ¯ame surface density models are now discussed. Assuming intermittency between fresh and burnt gases

Fig. 20. Transverse ¯ame surface density (S ) pro®les (m 21) plotted as a function of the transverse location for various downstream locations
(mm downstream the rod). (a) f ˆ 0:78; (b) f ˆ 0:9: Flow rate: 35 g/s [121].
230 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

~ A submodel is required to describe the wrink-


function of c:
ling length scale Ly, generally assumed to be proportional to
the integral length scale lt:
 n
S
Ly ˆ Cl lt L0 …178†
u
where Cl and n are two constants of the order of unity [75].
According to Gouldin [125,126], the ¯amelet orientation
factor s y is directly linked to the vectors normal to the
instantaneous ¯ame front, n, and to the mean ¯ame brush,
np (see Section 6.3.3):
* +
1 1
Fig. 21. De®nition of the ¯ame transit time tt is the ¯ame crossing ˆ …179†
frequency BML model. sy unp ´nu s

Despite of the same normal vectors n and np involved, the


(Section 7.3), Bray, Moss, Libby and their coworkers have ¯amelet orientation factor s y (Eq. (179)) and the ¯ame
proposed to describe the mean reaction rate v_ as the product surface wrinkling factor J (Eq. (97)) are different quanti-
of a ¯ame crossing frequency n and a local reaction rate per ties. These relations, combined with Eqs. (100) and (177),
¯ame crossing, wF: may relate the wrinkling length scale Ly to the thickness d B
v_ ˆ w F n …174† of the mean ¯ame brush, under the assumption of a in®nitely
thin ¯ame front:
Because c(t) may be viewed as a telegraphic signal (Fig. 16),
this crossing frequency is derived from a statistical analysis  2 c†
c…1   2 c†
c…1 
Sˆg ˆ Ju7cu
 < aJ …180†
of the telegraph equation, leading to: s y Ly dB
 2 c†
2c…1  where a is a model constant of the order of unity. Then:
nˆ …175†
T^  * +
g 1
where T^ is time scale for the ¯uctuations of c. Ly < n ´hni d …181†
a unp ´nu s p s B
This analysis is attractive because the crossing frequency
n may be easily obtained in experiments, for example from Assuming that a , g and s y are constant, d B/J measures the
time-resolved local temperature measurements (thermo- wrinkling length scale Ly of the ¯ame front.
couple). On the other hand, the ¯ame surface density S The agreement between this BML model and the exper-
and the local reaction rate per ¯ame crossing (wF) are not imental data is very good, as displayed on Fig. 22 where the
obvious to estimate. Eq. (175) is generally closed estimating ratio S=‰c…1  corresponding to g=s y Ly and assumed to
 2 c†Š;
T^ from a characteristic turbulent time t t. The reaction rate be constant, is displayed. Nevertheless, a submodel is
per crossing ¯ame, wF (Eq. (174)), is usually modeled as: required for the wrinkling length scale which increases
with the downstream location in our experiment.
r0 SL
wF ˆ …176† Estimating the local reaction rate per unit ¯ame area V_ c
dt =tt from the laminar ¯ame speed SL (V_ c ˆ r0 SL where r 0 is the
where r 0 is the unburnt gases density, SL and d l are respec- fresh gases density) the mean reaction rate becomes, when
tively the speed and the thickness of the laminar ¯ame. The nˆ1:
¯ame transit time tt measures the averaged time spend by a g u0 g u 0 …1 1 t†c…1
~ 2 c†
~
point in the ¯ow to cross a ¯ame front and corresponds to the v_ ˆ r0  ˆ r0
 2 c†
c…1
s y Cl lt s y Cl lt ~2
…1 1 tc†
mean transition time between c ˆ 0 and c ˆ 1 levels of the
progress variable, as shown in Fig. 21 (in practice, the c- …182†
signal is not exactly bimodal). As tt ˆ lt =u 0 is a turbulence characteristic time scale, the
This model has been latter rewritten in term of ¯ame mean reaction rate v_ is proportional to the intermittency
surface density, leading to the algebraic expression [124] between fresh and burnt gases, determined from c…1 ~ 2 c† ~
 2 c†
c…1  g 11t or c…1  and inversely proportional to t t. The physical
 2 c†;
Sˆg ˆ ~ 2 c†
c…1 ~ …177† analysis leading to the Eddy-Break-Up model is recovered
s y Ly ~2
s y Ly …1 1 tc†
and a similar expression for the reaction rate is found. The
where g is a constant of order unity. s y is a ¯amelet orienta- previous expression is slightly different than the one
tion factor measuring the mean angle of the instantaneous proposed in Section 7.3.3. The discrepancies are easily
¯ame front with the c surface and assumed to be an universal explained by the crude model used for the scalar dissipation
model constant …s y < 0:5†: Ly is a ¯ame front wrinkling of the progress variable to derive Eq. (162).
length scale and Eq. (165) has been used to replace c as a In Eq. (182), u 0 =lt is generally modeled from the
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 231

Fig. 22. S=…c…1  transverse pro®les in mm 21 corresponding to g=…s y Ly † in the Bray, Champion, Libby model (Eq. (177)), plotted as the
 2 c††
function of the mean progress variable c for several locations downstream from the rod: (a) f ˆ 0:78; (b) f ˆ 0:9 [121].

turbulence k± 1 model as 1=k: The ITNFS ef®ciency function where


G k [127] may be introduced in the turbulent time scale to  
lt
account for the reduced ability of small turbulent structures s ˆ log10 and
dl
to wrinkle the ¯ame front. ITNFS is one of the closure
! " !1=3 #! …185†
schemes for the ¯ame surface density balance equation u0 2 1 u0
(see Section 7.4.3 and Table 5). The mean reaction rate is s1 ˆ 1 2 exp 2
SL 3 2 SL
then [118,128,129]:
! When the length scale ratio lt/d t tends towards zero, G K also
lt u 0 1 decreases, reducing the effective ¯ame strain rate, as
v_ ˆ ar 0 G k ;  2 c†
c…1  …183†
dt SL k displayed on Fig. 23. G K only slightly depends on the vel-
ocity ratio u 0 =S L : The ef®ciency function does not reach a
where a is a model constant and the ef®ciency function G K constant level when lt/d l increases since increasing lt/d l,
depends on the length scale (lt/d l) and the velocity …u 0 =S l † keeping u 0 =SL constant, corresponds to an implicit increase
ratio comparing the turbulence and the laminar ¯ame char- of the turbulent Reynolds number Re.
acteristics. The ef®ciency function has been ®tted from DNS Because the ef®ciency function G K tends to counter-
data [127,130]: balance the known trend of Eddy-Break-Up modeling to
overestimate the mean reaction rate in highly strained
1 regions, this simple approach improves the accuracy of the
log10 …G k † ˆ 2 exp‰2…s 1 0:4†Š numerical predictions.
s 1 0:4
The BML model proposes a simple algebraic expression
! ! to estimate the ¯ame surface density s and the correspond-
u0
1 1 2 exp‰2…s 1 0:4†Š† s 1 s 2 0:11 ing reaction rate, but Bray, Moss, Champion and Libby
SL
have mainly focused their attention on a careful description
…184† of the turbulent ¯uxes using the balance equations for the
232 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

Table 5
Comparison of source (Si) and consumption (D) terms in the ¯ame surface density balance equation in different turbulent premixed combustion
models (details in text). k and 1 denote, respectively, the turbulent kinetic energy and its dissipation. Re is the turbulent Reynolds number. a 0,
b 0, g , l , CA, a, d, C, E and K are model constants, G k is the ef®ciency function in the ITNFS model [127] and depends on the length scale (lt/d l)
and the velocity (u 0 /SL) ratios comparing the turbulence and the laminar ¯ame characteristics. In the Choi model (CH) [137], u 0 denotes the rms
turbulent velocity and ltc is an arbitrary length scale introduced for dimensional consistency and combined to a 0 as a single arbitrary constant

Model S 1 ˆ km S S 2 ˆ kt S S3 D

 1=2
2uk 1 2 1 e2aR 2
CPB [134] A ik S a 0 CA S a 0 SL S
2xi n 
3…1 2 c†
1
…1 2 c†

SS L k
p
2uk 1 S 1C k 2
CFM1 Aik S a0 S b0 L S
2xi k 1 2 c
p
2uk 1 S 1C k 2
CFM2-a Aik S a0GK S b0 L S
2xi k 1 2 c
p
2uk 1 S 1C k 2
CFM2-b [130] Aik S a0GK S b0 L S
2xi k  2 c†
c…1 
p
u 0i u 0k 2uk p 1 F 1 0 0 2c SL Re
MB [117] E S a 0 Re S uc b0   S2
k 2xi k S L k i 2xi S 2g
 2 c†
c…1  1 1 d pL
k
1 S S
CD [136] a0 l S for k t # a0 K L b0 L S 2
k dL 1 2 c
 
1 1=2 SL
CH1 a0 S b0 S2
15n c…1
 2 c†
u0 SL
CH2 [137] a0 S b0 S2
ltc  2 c†
c…1 

Reynolds stresses r ug00 u 00 and the scalar ¯uxes r ug


i j
00 c 00 to take
i the consumption of ¯ame area. The modeled balance equa-
into account the possible occurrence of counter-gradient tion is rewritten as:
transport and ¯ame turbulence generation [131,132].  
2S n
The ¯ame surface density may also be derived from frac- 1 7´…U~ S† ˆ 7´ t 7S 1 S 1 1 S2 1 S3 2 D …187†
tal theories, leading to [133]
2t sS
  In this expression, the turbulent ¯ux of ¯ame surface
1 Louter D22
Sˆ …186† density is expressed using a classical gradient assumption,
L outer Linner n t is the turbulent viscosity and s S a ¯ame surface turbulent
where Linner and Louter are respectively the inner and outer Schmidt number. Five main closures are summarized:
cut-off length scales (the ¯ame surface is assumed to be
² The CPB model [134] is derived from the exact transport
fractal between these two scales). D is the fractal dimension
equation for the ¯ame surface density. The strain rate due
of the ¯ame surface. The cut-off scales are generally esti-
to thepturbulent
 ¯uctuations is estimated from the time
mated from the turbulence Kolmogorov lk and the integral lt
scale n=1 of the Kolmogorov structures. The turbulent
length scales.
strain rate is probably overestimated. Despite the fact that
the Kolmogorov structures contain the highest energy,
7.4.3. Flame surface density balance equation closures their lifetime is too short (because of viscous effects) to
The previously described balance equations for S (Eqs. actually affect the ¯ame front [65].
(89) and (90)) are unclosed and require modeling. In Table ² The coherent ¯ame model (CFM), developed by Candel
5, various closures found in the literature are compared and his coworkers following the initial work of Marble of
where S1 is the strain rate acting on the surface and induced Broadwell [84]. Three versions are presented in Table 5.
by the mean ¯ow ®eld. S2 is the strain rate due to the turbu- In the initial version (CFM1), the strain rate due to the
lent motions (Eq. (94)) and the third source term, S3, occurs turbulent ¯uctuations is estimated from the character-
only in the derivation proposed in Ref. [117]. D describes istic time of the integral length scale (k/1 ). In the two
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 233

Fig. 23. ITNFS ef®ciency function G K (Eq. (184)) as a function of the length scale ratio lt/d l for several values of the velocity ratio u 0 /SL (0.1;
1.0; 10; 11).

succeeding formulations (CFM2), the expression of the laminar ¯ame studies because, due to curvature and
turbulent strain rate acting on the ¯ame front is improved unsteady effects, a ¯ame is able to sustain higher strain
from direct numerical simulations and multi-fractal rates than expected.
analysis (ITNFS model, Eq. (184) [127]). The destruction ² The CH model [137] has been devised for spark-ignited
term differs in these two last formulations (CFM2a and engines to recover experimental data obtained in a closed
CFM2b). vessel [138]. The consumption term D is similar to the
² The MB model [117] is based on an exact equation for the one proposed in CFM2-b model whereas two formula-
scalar dissipation rate x : tions of the strain rate induced by turbulent motions are
proposed. The ®rst expression (CH1) corresponds to the
2c 00 2c 00 closure in CPB model, based on the Kolmogorov turbu-
r x~ ˆ rD …188† lent time scale. In CH2, the strain rate is only propor-
2x i 2xi
tional to the turbulence intensity u 0 and an arbitrary
assuming a constant density r . The transport equation is length scale, ltc, is incorporated in the model constant.
rewritten as a ¯ame surface density transport equation
under the ¯amelet assumption (see Section 6.5, Eq. Despite these comments, all these closures exhibit strong
(116)). This approach leads to a different expression for similarities. For example, the consumption term D is always
the source term S1 and an additional source term S3 is proportional to S 2. A comparison between these models to
found. In a ®rst analysis, this term S3, which does not predict turbulent ¯ame speed ST may be found in Ref. [130].
depend on the available ¯ame surface density S; might In the case of a statistically one-dimensional turbulent ¯ame
be viewed as an ignition term [135] involving a gradient propagating in a frozen turbulence, a KPP (Kolmogorov±
of the mean fuel mass fraction or of the mean temperature Petroski±Piskunov) analysis was used to analytically deter-
(i.e. fresh gases are ignited by heat transfers). However, mine the turbulent ¯ame speed ST as a function of the model
this analysis does not hold because the ¯ame surface parameters.It was found that only the CFM-2 formulation is
density balance equation is derived assuming an able to predict the so-called bending phenomenon, where
established ¯ame surface. In fact, as shown below, S3 the turbulent ¯ame speed decreases before the occurrence of
corresponds to an anisotropic contribution of the tur- ¯ame extinction when the turbulence level increases, as
bulent strain rate aT (see Eq. (205)). S3 seems to be negli- experimentally evidenced [110].
gible in practical simulations, at least to describe the A recent work, [135] has compared CPB, CFM1, MB and
¯ame front propagation in a homogeneous and isotropic CD models to predict a turbulent premixed jet ¯ame. The
turbulent ¯ow ®eld. CD predicts extremely high temperatures whereas CFM1,
² The CD model [136] is similar to the ®rst version of the MB and CPB provide reasonable predictions of mean vel-
coherent ¯ame model (CFM1). An additional term is ocities and temperatures. A slight overestimate (respectively
proposed to take into account ¯ame extinction under underestimate) of temperature is pointed out for CPB
excessively high strain rates. Such a term was tested in (respectively MB) and is probably due to the expression
the coherent ¯ame model but the choice of the critical for the strain rate. The MB closure [117] is found to be
strain rate is somewhat arbitrary. Moreover, this critical more sensitive to the inlet turbulent quantities than CFM1
strain rate cannot be deduced from planar strained but CFM2 models, incorporating an ef®ciency function
234 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

including the inability of small vortices to wrinkle the ¯ame This relation, con®rmed by direct numerical simulations
surface, have not been tested. Flame surface density models [140], shows that turbulent ¯uxes of c~ and S are closely
are extended to non-isenthalpic ¯ows (premixed reactants related. A counter-gradient turbulent transport will be
injected in a co-¯ow of air [139]). A pioneering work has observed simultaneously for these two scalar ®elds.
been ®rst conducted [118] extending the algebraic ¯ame
surface density BML model to non-isenthalpic ¯ows, 7.4.4.2. Strain rate induced by the mean ¯ow ®eld, AT. The
using similar concepts. A presumed pdf for a mixture frac- only unclosed quantities in the strain rate due to the mean
tion Z, determined from mean and rms values of Z, is ¯ow ®eld, AT, are the orientation factors knnls. Following
incorporated to account for dilution phenomena. [134], the vector normal to the ¯ame front, n may be split
into a mean component, M, and a ¯uctuation, m:
7.4.4. Analysis of the ¯ame surface density balance equation n ˆ M 1 m with knl s ˆ M and kmls ˆ 0 …195†
Following the description of the exact balance equation
for the ¯ame surface density S (Section 6.3.2) and the Then, the orientation factors become:
summary of the most popular closures (Section 7.4.3), the knnls ˆ MM 1 kmmls …196†
aim of this section is to carefully analyze this balance
equation. This analysis may be based on direct numerical Using the de®nition nu7cu ˆ 27c and assuming an
simulations [17] or on experimental data [120,121]. in®nitely thin ¯ame front leads after averaging to [134,141]:
Starting from the `propagative' form (Eq. (90)): knl s S ˆ MS ˆ 27c …197†
2S where c and c~ are related using the BML relation r c~ ˆ r b c
1 7´…u~ S† 1 7´…ku 00 l s S†
2t (Eq. (138)). Then, only the ¯uctuation cross products kmmls
remain unclosed. Several closure schemes have been
 S 1 k 7´u 00 2 nn : 7u 00 l s S
ˆ …7´u~ 2 knnl s : 7u†
|‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚} |‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚} proposed:
AT aT

² In Ref. [134] one assumes an isotropic distribution of the


2 7´‰kwnl s SŠ 1 kw7´nl s S …189†
|‚‚‚{z‚‚‚} |‚‚‚{z‚‚‚} ¯uctuating components of the normal vector n:
propagation curvature
dij
each unclosed term may be investigated as follows: kmi mj ls ˆ …1 2 Mk Mk † …198†
3

7.4.4.1. Turbulent transport. The turbulent ¯ux of ¯ame


² In Ref. [117], a relation between orientation factors and
surface density is generally expressed using gradient
Reynolds stresses is indirectly proposed:
transport:
n ug
00 u 00
ku 00 l s S ˆ 2 t 7S …190† kni nj ls ˆ
i j
…199†
sS 2k
Bidaux and Bray (1994, unpublished work, see for example where k is the turbulent kinetic energy.
[75,140]) have shown from a simple BML-type approach ² From experimental data [120], a much more complicated
that the turbulent ¯uxes of the mean progress variable …c† ~ model is proposed that will not be described here because
and of the ¯ame surface density (S ) are closely related. its practical implementation is probably not so easy.
Assuming that the conditional velocity on the ¯ame These authors have also shown that the isotropic
surface, kul ~ s is a linear function of the conditional fresh assumption made in Ref. [134] is clearly wrong for the
…u u † and burnt gases velocities: turbulent ¯ame stabilized downstream a small rod. On the
other hand, a slight modi®cation of Ref. [117] leads to
~ s ˆ u u 1 cp …u b 2 uu †
kul …191† very good results [121]:
p X 2
where c denotes the c-level chosen to de®ne the ¯ame front. uf00k
The BML relation (Eq. (169)): k±i ug
00 u 00
i j
kni ni ls ˆ ; kni nj±i ls ˆ …200†
4k 2k
~ u u 1 c~u b
u~ ˆ …1 2 c† …192†
leads to: These orientation factors have very important effects and
may lead to surprising results. In the ¯ame-holder stabilized
ku 00 ls ˆ kuls 2 u~ ˆ …cp 2 c†…
~ u b 2 u u † …193† turbulent ¯ame investigated in Ref. [121], the main velocity
gradient is the transverse gradient of the mean axial velocity,
From Eq. (170) the ¯ame surface density turbulent ¯ux
corresponding to the mixing layer shear stress. But, because of
becomes:
the low value of the corresponding orientation factor, its
f00 l S ˆ …c p 2 c†
~ g contribution to the strain rate AT induced by the mean ¯ow
ku s u 00 c 00 S …194†
~ 2 c†
c…1 ~ ®eld is not that important. The strain rate AT is dominated by
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 235

Fig. 24. (a) Transverse pro®les of the mean velocity gradients and the corresponding strain rate AT. (b) Components of the source term ATS due
to the strain rate induced by the mean ¯ow ®eld. Data are plotted as a function of the transverse coordinate y for x ˆ 80 mm and f ˆ 0:9 [121].

the axial gradient of the mean axial velocity, induced by the In most models, this term is generally assumed to be
thermal expansion due to the combustion heat release, as proportional to the inverse of a turbulent time scale, either
illustrated in Fig. 24. Notice also that the strain rate AT the Kolmogorov time scale (CPB model) or the integral time
cannot be reduced to the simpli®ed expression ®rstly scale 1 /k (CFM, MB and CD models). This turbulent time
proposed [84]: may corrected with the ef®ciency function, G k, of the ITNFS
closure (Eq. (184)). Nevertheless, aT is always modeled
~
AT ˆ u7Uu …201† being isotropic despite the orientation factors nn occurring
Recent works [142] have shown that u7Uu ~ may probably be in expression (202).
viewed as a model for the total strain rate AT 1 aT and not only Using the previous splitting of the normal vector to the
for the strain rate AT due to the mean ¯ow. This ®nding ¯ame front, n, combined with the geometrical relation
explains why, in previous versions of the coherent ¯ame (197), leads to:
model where the orientation factors kninjls were not * + * +
2u 00 1 2u 00i 2c 2c
incorporated, the mean strain rate AT was not included a T S ˆ d ij i S 2
because the simple expression (201) clearly overestimates AT. 2xj s S 2xj s 2xi 2xj
* + * +
2u 00i 2c 2u 00i 2c
7.4.4.3. Strain rate aT due to turbulent motions. The source 1 mj 1 mi
2xj s 2xi 2xj 2xj
term due to the strain rate aT is:
* +
* + 2u 00
2u 00i 2u 00 2 mi mj i S …203†
a T S ˆ k7´u 00 2 nn : 7u 00 l s S ˆ d ij 2 ni nj i S 2xj s
2xj 2xj s
…202† This interesting relation decomposes the source term due to
236 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

This stands as the main dif®culty of the propagative


approach for the ¯ame surface density balance equation
because various effects are incorporated in w (strain rate,
curvature effects, etc.).
Assuming a constant displacement speed w equal to the
laminar ¯ame speed SL, the normal propagation term
becomes:
 ˆ 2S L 7 2 c …206†
7´‰kwnl s SŠ ˆ S L 7´‰knl s SŠ ˆ S L 7´‰27cŠ
where the geometrical relation (197) is used.
This term is generally neglected in models (see Table 5).
Fig. 25. Flame front curvature analysis. Experimental data from Refs. [120,121] in a V-shape turbu-
lent ¯ame show that it is not always negligible and is of the
the strain rate induced by turbulent motions, aTS , into two same order as the curvature term.
parts: a contribution depending only on turbulent character- The propagation/curvature term becomes:
istics that may be assumed to be isotropic (®rst and last
terms) and an anisotropic contribution, where the mean kw7´nl s S ˆ S L k7´nl s S …207†
¯ame surface orientation occurs through the gradient of where the only unknown is the mean ¯ame front curvature
the mean progress variable (second, third and fourth k7´nls The curvature is positive when the ¯ame front is
terms). A priori, this last part is not proportional to the convex towards the fresh gases, which is, a priori, the case
available ¯ame surface density S . when the mean progress variable c is close to zero (see Fig.
When velocity and normal vector ¯uctuations are 25). On the other hand, curvatures are probably negative
supposed uncorrelated (this assumption is well veri®ed in (¯ame convex towards the burnt gases) when c < 1: As-
direct numerical simulation data), the previous expression suming that the mean curvature is of the order of the inverse
may be reduced using: of the wrinkling length scale Ly, we have:
* + * 00 +
2u 00i 2u i 1 1
mj ˆ kmj ls ˆ0 …204† lim k7´nl s ˆ and lim k7´nl s ˆ 2
2xj s 2xj s 
c!0 Ly 
c!1 Ly
from the mj de®nition. Then: Then, a simple linear model may be proposed:
* + * + * +
2u 00i 2u 00i 1 2u 00i 2c 2c cp 2 c
aT S ˆ d ij S 2 m i mj S2 k7´nl s < …208†
2xj s 2xj s S 2xj s 2xi 2xj Ly
|‚‚‚‚‚‚‚‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚‚‚‚‚‚‚‚} |‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚}
isotropic contribution anisotropic contribution
Replacing Ly by its value as a function of the ¯ame surface
…205† density S from BML modeling (Eq. (177)) leads to the
This simple analysis explains the third source term, S3 found following model for the curvature term in the S -equation:
in Ref. [117] from the derivation of a balance equation for cp 2 c 2
the scalar dissipation rate x~ of the progress variable c. This kw7´nl s S < bS L S …209†
 2 c†
c…1 
term S3 corresponds to an anisotropic contribution depend-
ing on 7c and on the strain rate due to turbulent motions. where b is a model constant.
Assuming a gradient type closure for the turbulent transport This term is positive in the fresh gas side of the turbulent
 c and appears as
of the fresh gases, S3 is proportional to 7c´7 ¯ame brush and becomes negative in the burnt gas side. This
a model for the anisotropic contribution of the turbulent trend is in agreement with the ®ndings of direct numerical
strain rate. Nonetheless, Mantel and Borghi [117] have simulations [17] and experimental measurements [120,121],
shown that this term is negligible in numerical simulations as displayed in Fig. 26. The expression differs from classical
of a ¯ame propagating in an homogeneous and isotropic closure where kw7´nls is always negative (term D in
turbulent ¯ow ®eld. This ®nding is probably questionable Table 5).
in other con®gurations and further investigations are A linear relaxation is retained to model the mean curva-
required. ture k7´nls (Eq. (208)). This type of closure is common in
turbulent reacting ¯ows and possesses similarities with the
7.4.4.4. Propagation and curvature terms. These two terms relaxation model used for the scalar dissipation rate (Section
are analyzed together because they derive from the laminar 9.2) or for the pdf balance equation (Section 9.8.2). The
¯ame propagation and are related to the ¯ame front links between these closures are further discussed in Section
displacement speed w. The modeling of these terms 9.8.5.
requires the description of the displacement speed w that The previous analysis was based on both a theoretical
may have values far from the laminar ¯ame speed SL [78]. analysis of the exact S -balance equation and experimental
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 237

combustion has started but is not yet fully established. This


®nding displays one of the dif®culties of ¯ame surface
density models. In their present formulations, these models
are unable to describe the ¯ame stabilization because the S -
equation (Eq. (189)) is derived assuming that the ¯ame does
exist. All source terms in this equation are proportional to S
or to S 2 and the equation cannot generate ¯ame surface
when there is no initial ¯ame surface. In addition, an initia-
tion effect must also be incorporated to account for ignition
time delay in the local reaction rate per unit surface, VÇ i (Eq.
(85)), to recover the observed results.
Fig. 26. Transverse pro®les of propagation (27´[knlsS ]), curvature
(k7´nlsS ) and combined (k7´nlsS 2 7´[knlsS ]) terms (m 22) in the
7.4.6. A related approach: G-equation
¯ame surface density balance equation plotted as a function of the
 The consumption term modeled as 2S 2/
mean progress variable c:
As described in Section 6.3.1, the premixed turbulent
 and the new proposed closure, 7 2 c 1 b…cp 2 c†
(1 2 c)  S 2 =c…1
 2 c†
 ¯ame may be described using a level set approach. Most
where b ˆ 0:4 and cp ˆ 0:5 are also displayed. f ˆ 0:90; x ˆ of the modeling issues discussed above are then recast in
80 mm [121]. terms of G-equation and modeling for the turbulent burning
velocity ST [89].
data [120,121]. An improved version of the closed balance
equation may be proposed: 8. Turbulent transport in premixed combustion
2S 8.1. Introduction
1 7´…u~ S† 1 7´…ku 00 l s S†
2t
! Turbulent ¯uxes of the progress variable c, r ug
00 c 00 ; are
u 0 lt 1 i
~ S 1 Gk
ˆ …7´u~ 2 knnl s : 7u† ; S 1 S L 7 2 c generally modeled using a gradient transport hypothesis as
SL dl k
for inert scalars (Section 7):
 
S2 00 c 00 ˆ 2 mt
r ug
2c~
1 bSL …cp 2 c†
 (210) i …211†
 2 c†
c…1  Sct 2xi
where mt is the turbulent viscosity given by the turbulence
where G k is the ITNFS ef®ciency function [127]. Orientation model and Sct is a turbulent Schmidt number. Theoretical
factors, kninjls may be described using a formulation [143] and experimental studies [19,20] have evidenced in
previously discussed. The turbulent ¯ux ku 00 ls S is described some turbulent premixed ¯ames counter-gradient turbulent
using a classical gradient expression.
transport where the turbulent ¯uxes r ug 00 00
i c and the mean
progress variable c~ gradient, 2c=2x ~ i ; have the same sign in
7.4.5. Flame stabilization modeling some regions and cannot be approximated with Eq. (211).
The ¯ame surface density ®elds obtained from laser This phenomenon is known as counter-gradient turbulent
tomography and the mean reaction rate estimated from transport or counter-gradient turbulent diffusion.
CH radical emission are compared in Fig. 27. According This surprising ®nding was explained by the work of
to Eq. (85), reaction rate and ¯ame surface density are Bray, Moss and Libby discussed in Section 7.3. In their
roughly proportional, excepted close to the stabilization formulation, the turbulent ¯uxes of the progress variable c
rod, the ¯ame surface density S is high, whereas the mean are directly connected to the conditional mean velocities in
reaction rate remains low. In this zone, fresh and burnt gases fresh …uui † and burnt gases …ubi † (Eq. (170) Section 7.3.5):
are separated by an interface (high surface densities) where  
r ug
00 00
i c ˆ r c…1 ~ ubi 2 uui
~ 2 c† …212†

Even though conditional velocities are not obvious quanti-


ties, this expression is useful to explain counter-gradient
turbulent transport: Because of the thermal expansion due
to combustion heat release, the burnt gas conditional vel-
ocity, ubi ; is likely to be greater than the fresh gas conditional
Fig. 27. Flame surface density (half top) and mean reaction rate, velocity, uui : Then, in opposition with the modeled expres-
estimated form CH radical emission (half bottom) ®elds are sion (Eq. (211)), the turbulent ¯uxes of the progress variable
compared for f ˆ 0:90: Flame surface density data, extracted c have the same sign as the mean gradient …2c=2x ~ i †: Counter-
from two different data sets, are not available from 30 to 70 mm gradient transport also increases when the heat release factor
downstream the rod [121]. t , de®ned by Eq. (141), increases [144].
238 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

[140]. The reduced costs of 2D simulations allow the inves-


tigation of a large range of ¯ame and turbulence parameters.
In these simulations, a planar laminar premixed ¯ame is
®rstly superimposed on a homogeneous isotropic turbulent
¯ow ®eld (Fig. 28). The ¯ame front is progressively
wrinkled by turbulent motions, and the turbulence decays
in time. After a time of the order of the eddy-turnover time
of the largest turbulence structures, the ¯ame may be
assumed to be in equilibrium with the turbulent ¯ow ®eld
and relevant modeling information is extracted from DNS.
Fig. 28. Numerical con®guration. A plane laminar premixed ¯ame As the numerical con®guration is statistically one-
is initially superimposed to an homogeneous and isotropic turbu- dimensional in the propagating direction, quantities such
lence.
as mean progress variable or mean turbulent ¯uxes may
be extracted from averaging in the perpendicular direction.
The various versions of this BML model propose an alge-
braic closure for the mean reaction rate of the progress 8.2.2. Results
variable c, but focus attention on the scalar turbulent trans- Turbulent ¯uxes ru 00 c 00 extracted from the direct numer-
port (closure schemes for the turbulent ¯uxes r ug
00 c 00 balance
i ical simulations conducted by Trouve [145] (denoted CTR)
equations), whereas other models may lead to more sophis- and Rutland [146] are displayed in Fig. 29. The ®rst data-
ticated reaction rate formulations, retaining a simple gradi- base clearly exhibits gradient turbulent transport, whereas
ent closure (Eq. (211)) for these ¯uxes. the second one corresponds to a counter-gradient situation.
The occurrence of counter-gradient turbulent transport The main discrepancy between the two databases lies in the
have been analyzed using direct numerical simulation initial turbulence level which is higher in the CTR simula-
(DNS) [140]. The results demonstrate the power of DNS tion …u 00 =SL ˆ 10† than in the Rutland database …u 00 =SL ˆ 1†:
to help in the modeling of turbulent combustion. This Mean and conditional average velocities across the turbu-
section is devoted to these results, obtained from a direct lent ¯ame brush are displayed in Fig. 30 for the two DNS
solution without any closure models, of the instantaneous simulations. These results lead to the following comments.
balance equations.
² As expected, in the Rutland database, the burnt gas con-
8.2. Direct numerical simulation analysis of turbulent ditional velocity, ub ; is higher than the fresh gas condi-
transport tional velocity, uu ; leading to a counter-gradient turbulent
transport in agreement with expression (212).
8.2.1. Introduction ² On the other hand, in the CTR database, the fresh gas
Gradient and counter-gradient turbulent scalar transport conditional velocity is higher than the burnt gas con-
was observed in DNS of premixed ¯ame/three-dimensional ditional velocity. This result is ®rst surprising: ub is
turbulence interactions [145,146]. Then, simulations of two- expected to be larger than uu because of thermal expan-
dimensional ¯ame/turbulence interactions were reported sion due to the heat release. However, this observation is

Fig. 29. Turbulent ¯ux ru 00 c 00 displayed as a function of the mean progress variable c~ (bold lines) in the tri-dimensional direct numerical
simulations from CTR (Ð) and Rutland (- - -). Positive (respectively negative) values of ru 00 c 00 denote a counter-gradient (respectively
gradient) turbulent transport. The corresponding turbulent ¯uxes estimated using the BML expression (Eq. (212)) are displayed for comparison
using thin lines. Velocities are made dimensionless using the laminar ¯ame speed SL [140].
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 239

Fig. 30. Conditional velocities across the turbulent ¯ame brush displayed as a function of the mean progress variable c~ in the Rutland (top) and
the CTR (bottom) databases. Favre averaged velocity, u~ (Ð Ð ), fresh gases conditional velocity, uu (´´´), burnt gases conditional velocity, ub (±)
and ¯ame front conditional velocity, kuls (- - -). Velocities are made dimensionless using the laminar ¯ame speed SL [140].

in agreement with expression (212) and a gradient turbu- 8.3. Physical analysis
lent transport. Note that Eq. (212) overestimates the
turbulent ¯ux ug 00 00
c (see Fig. 29) because, due to the A simple physical analysis is now reported to derive a
¯ame thickness that must be resolved in the numerical criterion predicting the occurrence of counter-gradient
simulation, the progress variable c is not fully bimodal turbulent diffusion [140].
(c ˆ 0 or c ˆ 1). In both cases, the Bray±Moss±Libby Following Bidaux and Bray (1994) (unpublished work
model remains able to predict the turbulent ¯ux type already presented in the BML model context Section
(gradient or counter-gradient). 7.4.4), turbulent ¯uxes of the progress variable c, ug 00 c 00 ;
i
are directly connected to the surface-averaged ¯uctuating
The two-dimensional simulations conducted in Ref. velocity, ku 00i ls (Eq. (194)). Thus, a model for ug
00 00
i c may be
[140], referenced as CRCT in Fig. 31 have been used to deduced from a model for ku 00i ls involving the conditional
analyze the occurrence of counter-gradient turbulent fresh and burnt gases mean velocities:
transport. The gradient turbulent transport is clearly
 
enhanced by an increase in the turbulence level u 0 /SL and ku 00i ls ˆ kui ls 2 u~i ˆ …cp 2 c†
~ ubi 2 uui …213†
decreasing values of the heat release factor t . The
`increase', in terms of velocity ratio u 0 /SL, of the line
delimiting gradient and counter-gradient turbulent transport In the following, the ¯ow ®eld is assumed to be statistically
when lt/d l decreases is due to the reduced ability of small one-dimensional and only the turbulent transport in the
scale turbulence motions to wrinkle the ¯ame front. This propagating direction, ug 00 00
c will be described. Our analysis
phenomenon has already been discussed [65] and included is based on the two limiting cases pictured in Fig. 32:
in the ITNFS model [127]. Three- and two-dimensional Low turbulence level: The ¯ame front remains smooth
direct numerical simulations lead to very similar results and the velocity jump between fresh and burnt gases, ub 2
for this problem. u u ; is determined primarily by thermal expansion and its
240 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

Fig. 31. Premixed turbulent combustion diagram. The DNS ¯ame-¯ow conditions are plotted as a function of the velocity ratio, u 0 /SL, and
length-scale ratio, lt/d L. The Classical Klimov±Williams criterion and the criterion from Ref. [65] are given to show the domain of validity of
¯amelet combustion. Also plotted are the DNS conditions of the Rutland (CR, t ˆ 2:3) and CTR …t ˆ 3† simulations. As the turbulence is
decaying in the CTR simulation, CTR conditions are displayed as an almost vertical line. The symbols W …t ˆ 3† and A …t ˆ 6† correspond to
the CRCT DNS. In 2D DNS, the turbulence decay is smaller and is not represented. Filled (open) symbols denote gradient (counter-gradient)
turbulent diffusion. The transition criterion, NB ; tsL =2au 0 ˆ 1 (Eq. (220)), separating CGD (below) from GD (above) is plotted for …t ˆ 3†
and …t ˆ 6† [140].

value is close to the one obtained in a plane and laminar ² At the leading edge of the turbulent ¯ame (near c~ ˆ 0),
¯ame …ub 2 uu < tSL †: Eq. (213) becomes: the ¯ame front is convected towards the fresh gas with a
mean velocity estimated by 2u 0 (see Fig. 32). Then,
ku 00 ls ˆ …cp 2 c†
~ tSL …214†
ub 2 uu < 2u 0 …215†
High turbulence level: Due to strong viscous dissipation
of turbulent eddies in the hot burnt gas, the ¯ame front where u 0 denotes the rms velocity in the fresh gases.
motions are assumed to be dominated by the turbulence ² At the trailing edge of the ¯ame brush …c~ < 1†; the ¯ame
properties taken upstream of the ¯ame. front is convected by turbulent motions towards the burnt

Fig. 32. Two limiting cases: counter-gradient transport promoted by thermal expansion (left); gradient transport due to turbulent motions
(right).
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 241

Fig. 33. DNS estimated ef®ciency function a (Eq. (21)) as a function of the length scale ratio lt/d l comparing the integral turbulent length scale
and the laminar ¯ame thickness [140].

gases with a mean speed estimated by 1u 0 (see Fig. 32): and expression (219) may be used to derive a criterion deli-
neating gradient and counter-gradient regimes. The Bray
u b 2 uu < 1u 0 …216†
number:
leading to the simple linear model: tSL
NB ˆ …220†
ku 00 ls ˆ 22…cp 2 c†
~ au 0 …217† 2au 0
where a is an ef®ciency function, similar to the ITNFS is greater (respectively lower) than unity when a counter-
model [127], accounting for the weak ability of small gradient (respectively gradient) turbulent transport is
scale turbulent motions to wrinkle and convect the expected. This criterion is well veri®ed by direct numerical
¯ame front. a is expected to be of order unity for large simulation results, as shown in Fig. 31. The ef®ciency func-
turbulent length scales and vanishes when turbulent tion a has been estimated from DNS (Fig. 33). Recent
eddies are too small to affect the ¯ame front. The factor experimental results [147] have con®rmed these ®ndings.
2 has been introduced assuming cp < 0:5:
8.3.1. Comments
00 ² Following Fig. 31 and criterion (220), in practical appli-
Then, modeling ku ls as a sum of these two contributions
cations turbulent transport may be counter-gradient, or
leads to:
close to the transition between gradient and counter-
ku 00 ls ˆ …cp 2 c†…
~ tSL 2 2au 0 † …218† gradient regimes. In many cases, the heat release factor
t is about 5±7 leading to a transition between gradient
and the turbulent ¯ux becomes:
and counter-gradient situations when u 0 =SL is of the order
ug
00 c 00 ˆ c…1 ~ tSL 2 2au 0 †
~ 2 c†… …219† of 3. Nevertheless, the possible occurrence of counter-
gradient transport is generally neglected in modeling.
This simple model is well veri®ed in direct numerical simu- ² The mean progress variable gradient may be estimated
lations [140] and has also been recovered when applying a as:
second order modeling (i.e. balance equations for turbulent
scalar ¯uxes) to stagnating ¯ames in the limit of small turbu- 2c~ ~ 2 c†
c…1 ~
< …221†
lence intensities [131]. The turbulent ¯ux may be viewed as 2x db
the sum of two contributions acting in opposite directions, where a length scale d b characterizing the ¯ame brush is
one induced by turbulent motions and the other by thermal introduced. Then, the gradient type contribution in Eq.
expansion. Then, the turbulent transport is analyzed as (219) corresponds to a Prandt±Kolmogorov turbulence
follows: for a suf®ciently high turbulence level, the ¯ame modeling:
is unable to impose its own dynamics to the ¯ow ®eld and
p 2c~
the turbulent transport is of the gradient type, as for any ~ db au 0 < 2adb k
~ 2 c†
2c…1 …222†
inert scalar. On the contrary, when the turbulence level 2x
p
remains low, the thermal expansion due to heat release where k is the turbulent kinetic energy and u 0 ˆ k.
dominates and the ¯ame is able to impose its own dynamics ² Recent works [148,149] have reported regimes, corre-
leading to a counter-gradient turbulent transport. Counter- sponding to low turbulence levels, where combustion
gradient turbulent diffusion occurs when ug 00 c 00 is positive
instabilities occur and wrinkle the ¯ame front, acting in
242 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

8.4. External pressure gradient effects


The previous analysis is completed by investigating the
effects of externally imposed pressure gradients on turbulent
premixed ¯ames. Counter-gradient turbulent transport was
®rst explained [19,151] by differential buoyancy effects of
pressure gradients on pockets of heavy and cold fresh gases
and on pockets of light and hot burnt gases. In many
combustion systems, ¯ames are ducted and submitted to
strong pressure gradients due to thermal expansion and lead-
ing to ¯ow accelerations. In Ref. [20], it is experimentally
shown that counter-gradient transports are enhanced in
ducted ¯ames.
In Refs. [148,152], the same type of DNS as [140] was
conducted, introducing an externally imposed pressure
gradient (in fact, due to technical reasons, a constant
acceleration). The main conclusions, displayed in Figs. 34
and 35, are:

² A favorable pressure gradient, i.e. a pressure decrease


from unburnt to burnt gases, is found to decrease the
¯ame wrinkling (see Fig. 34), the ¯ame brush thickness,
and the turbulent ¯ame speed ST (Fig. 35). It also
promotes counter-gradient turbulent transport.
² On the other hand, adverse pressure gradients tend to
increase the ¯ame brush thickness and turbulent ¯ame
speed (Fig. 35), and enhance classical gradient turbulent
transport. As proposed in Ref. [153], the turbulent ¯ame
speed is modi®ed by a buoyancy term linearly dependent
on both the imposed pressure gradient and the integral
Fig. 34. Superimposed instantaneous temperature and vorticity length scale lt.
®elds at the same time. (a) No imposed pressure gradient±gradient ² A corrected Bray criterion (Eq. (220)) has been proposed
turbulent transport; (b) favorable imposed pressure gradient± to account for the pressure gradient effects. We do not
counter-gradient turbulent transport. A planar laminar ¯ame give more details because this modi®ed criterion needs to
separating fresh gases (left) from burnt gases is initially super- be validated and improved (length scale effects are not so
imposed to an homogeneous and isotropic turbulent ¯ow ®eld clear and various analyses are possible, see [152]).
…u 00 =SL ˆ 5† [152].
All these results suggest that counter-gradient turbulent
an opposite direction than counter-gradient diffusion transport should be expected in most ducted turbulent ¯ames
reducing the ¯ame front wrinkling. These instabilities and some experimental observations are now reported.
should also be included in combustion models [150].
² Expression (219) has been derived to analyze the occur- 8.5. Counter-gradient transportÐexperimental results
rence of counter-gradient transport, but is not suited for The V-shape turbulent ¯ame was described in Section
numerical simulations. For instance, such a simpli®ed 7.4.1. The thermal expansion modi®es the ¯ame dynamics
formulation is unable to predict a gradient transport in as is clearly apparent on high speed tomography ®lms
the vicinity of the leading edge …c~ < 0† always observed, (scheme in Fig. 36, see also Ref. [121]): in the ®rst part
even in counter-gradient situations. of the chamber (region 1), because of the rod wake, the

Table 6
Geometrical analysis of the scalar turbulent transport in a V-shaped premixed ¯ame (see Fig. 36). G and CG denote, respectively, a gradient and
a counter-gradient turbulent diffusion

2c~ 2c~
Zone ub 2 uu ug
00 c 00 vb 2 vu (when y . 0) (when y . 0) vg
00 c 00
2x 2y

1 ,0 .0 G .0 ,0 G
2 .0 .0 CG .0 ,0 G
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 243

Fig. 35. Reduced turbulent ¯ame speed ST/SL plotted as a function of the reduced time t=tf ˆ SL t=dL ; where tf ˆ dL =SL is a ¯ame time, for
different values of the externally imposed pressure gradient. (a) initial turbulent level u 00 =SL ˆ 5; no pressure gradient (Ð) and increasing
favorable pressure gradient (´´´ and - - -); (b) initial turbulent level u 00 =SL ˆ 2; no pressure gradient (Ð) and increasing adverse pressure
gradient (´´´ and - - -). Markers correspond to the Libby theory [152,153].

coherent structures embedding the ¯ame front turn clock- phenomenon may be recast in terms of turbulent trans-
wise (counter-clockwise) in the upper (lower) ¯ame sheet, port using a simple geometrical analysis based on the
as in classical Von KaÂrmaÂn vortex streets, except for Bray±Moss±Libby relation (212) and summarized in
their symmetry due to pressure waves. When the center- Table 6. Accordingly, the transverse turbulent ¯ux
line velocity increases, because of thermal expansion in vg00 00
c is always of gradient type, but the change in rota-
burnt gases (region 2), the upper (lower) coherent struc- tion of the coherent structures corresponds to a transi-
tures start to turn counter-clockwise (clockwise). This tion between gradient and counter-gradient transport for
the downstream turbulent ¯ux ug 00 00
c : The turbulent ¯uxes
are of gradient type just behind the rod, as expected to
ensure the stabilization of the ¯ame, and becomes of
counter-gradient type further downstream.

8.6. To include counter-gradient turbulent transport in


modeling
Except for the work of Bray, Moss, Libby and their
co-workers [19,128,129,132,151], very few works have
been devoted to the effective modeling of counter-
gradient transport. The approach of Bray et al. is mainly
Fig. 36. A rough scheme of the coherent structures dynamics based on second order modeling, using balance equations
observed using laser tomography [121]. for the turbulent ¯uxes ug00 c 00 : These equations are easily
i
244 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

derived from momentum and progress variable balance ² The mean pressure gradient term (VI) is easily known
equations: under a BML assumption (Section 7.3). Making use of
Eq. (165) yields:
2r ug
00 c 00
i 2r u~j ug 00 00
ic
1 ~ 2 c†
c…1 ~
2t
|‚‚{z‚‚} 2x j c 00 ˆ c 2 c~ ˆ t …224†
|‚‚‚{z‚‚‚} 1 1 tc~
…I† …II†

2r uf00 00 00
j ui c 00 00 2c~ ~i
00 00 2u 2p ² The ¯uctuating pressure term (VII) is more dif®cult to
ˆ2 2 r ug
i uj 2 rugjc 2 c 00 understand and to model. In Ref. [151], this term is
2xj 2xj 2xj 2xi
|‚‚‚{z‚‚‚} |‚‚‚{z‚‚‚} |‚‚‚{z‚‚‚} |‚{z‚} neglected, however this assumption is not supported by
…III† …IV† …V† …VI†
DNS results [152]. The mean pressure term (VI) and the
¯uctuating pressure term (VII) probably need to be modeled
2p 0 2Jk 00 2tik
2 c 00 2 u 00i c 1 r ug00 v
i _ (223) together as c 00 2p=2xi : In Ref. [128], the pressure gradient
2xi 2xk 2xk |‚{z‚} term closure is carefully discussed. Recently, a model based
|‚{z‚} |‚‚{z‚‚} |‚{z‚} …X†
…VII† …VIII† …IX† on a partitioning of each pressure ¯uctuation covariance
into contributions from reactants, products and thin ¯ame-
where the RHS terms correspond, respectively, to turbulent lets was proposed [132]. The comparison of this new
transport of r ug
00 c 00 (III), c-gradient
i ~ effects (IV), mean vel- model with DNS results are encouraging and con®rm the
ocity gradient effects (V), the action of mean (VI) and ¯uc- importance of the intermittency between the conditional
tuating (VII) pressure gradients, r ug 00 00
i c turbulent dissipation
mean pressure gradients in reactants and products.
(VIII and IX) and reaction rate (X). This balance equation is, ² Turbulent dissipation terms (VIII and IX) are generally
of course, unclosed and each term may be extracted from expressed together using small scale dissipation rate
direct numerical simulations [132,140,152]. assumptions [151].
The discussion of the closure schemes for this equation is
beyond the scope of the present review and the reader may
Second order closures for turbulent ¯uxes and Reynolds
®nd relevant information in Refs. [128,132,151]. Some
stresses require nine additional balance equations in 3D simu-
comments may be made:
lations (three for progress variable ¯uxes, ug 00 c 00 ; and six for
i
² The ug00 c 00 turbulent transport (III) is generally modeled
i
g00 00
Reynolds stresses, u u ). Because of very high computa-
i j
using a classical gradient expression (gradient turbulent tional costs, model closures and implementation dif®culties,
transport at the third order). very few simulations have been conducted using the second
² Mean progress variable gradient terms (IV) needs order formulation. Recently [129], very promising results
Reynolds stress modeling for ug 00 00
i u j : Then, new balance were obtained, particularly in predicting experimental results
equations are derived, and closed, for these quantities [20].
(second order turbulence model).
² The mean velocity gradient term (V) is closed because 8.7. Towards a conditional turbulence modeling?
turbulent ¯uxes ug 00 c 00 are provided by their balance
j The BML formulation (Section 7.3.5) directly distin-
equations. guishes between the properties of fresh and burnt gases,

Fig. 37. Axial velocity u histograms for four transverse locations 80 mm downstream the rod in the V-shape turbulent ¯ame (Fig. 19): burnt
gases (y ˆ 0 mm; -´-); reaction zone (y ˆ 5 mm; ´ ´´ and y ˆ 12 mm; Ð ); unburnt gases (y ˆ 18 mm; Ð) [121].
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 245

Table 7
Modeling strategies for non-premixed turbulent ¯ames. Z is the mixture fraction, x ˆ Du7Zu 2 its scalar dissipation rate, P~ denotes pdfs, VÇ , an
integrated amount of heat release, S , the density of ¯ame surface and Y iC ; a conditional moment. PaSR stands for partially stirred reactor,
turbulent micromixing was expressed with a linear relaxation (IEM-LMSE) D7 2 Yi < …Y~ i 2 Yi †=tt , where tt < …k=1† is an eddy break-up mixing
time; other formulation exist [96] (from Ref. [55])

Major assumptions Flame structure in mixture Turbulent combustion model


fraction space

In®nitely fast chemistry Species are function of Z: Presumed Pdf:


Yi ˆ Yi …Z p † ~ p † from Z~ and Zf
P…Z 002
R
Chemical equilibrium [35] Y i ˆ Z Yi …Z p †P…Z
~ p †dZ p

1D strained reaction zones Species are functions of Z and x : Flamelet


R Rmodeling:
Yi ˆ Yi …Z p ; xp …t†† Y~ i ˆ Z x Yi …Z p ; xp †P…Z
~ p ; xp †dZ p dxp [165]
Diffusion balances chemistry Steady [63] or unsteady [177]
as in a 1D counter-¯ow ¯ame
Single-step or complex Total heatRrelease is a function of x : Coherent ¯ame model:
chemistry V_ …x st † ˆ 1 1
21 v_ T …j; x st †dj Solve for the density of ¯ame surface S
Constant …Le i ˆ 1† or j coordinate across the ¯amelet [88] v_ T ˆ V_ …x †S [84]
complex transport properties
…Lei ± 1†

Diffusion is captured via Solve for conditional


 moments: Conditional moment closure:
micromixing modelling Y iC …Z p † ˆ Yi uZ ˆ Z p presumed  p
R pdf P…Z †p [102,103]
Y~ i ˆ 10 Y Ci …Z p †P…Z
~ †dZ p

Simple transport properties Solve for representative PaSR: Presumed pdf 1 PaSR modelling:
…Lei ˆ 1† dYi =dZ ˆ …Y~ i 2 Yi 1 tt v_ PaSR †=…Z~ 2 Z† MIL/PEUL [68,191]
i P R
to get v_ i …Z; tt †PaSR v_ i ˆ k Z …v_ PaSR
i
 p ; tpt †dZ p dtt
…Z p ; tpt ††k P…Z

Single-step or complex Solve for Monte-Carlo particles: Pdf methods:


R R
chemistry dYik =dt ˆ …Y~ i 2 Yik †=tt 1 v_ ki Y~ i ˆ Y1 ¼ Yn Yip P…Y
~ 1p ; ¼; Ynp †dY1p ¼dYnp
~ 1p ; ¼; Ynp †
to get P…Y [93]
Chemical source is closed
(only for pdf methods)

this was achieved using conditional averaging (Eq. (168)). which is easy from c and Q balance equations [141]. These
This approach is very attractive because turbulence charac- equations remain to be closed and a few attempts have been
teristics may differ in fresh and in burnt gases as shown in conducted in this direction [91,148,154], but no clear
Fig. 37, where velocity histograms obtained using laser conclusions can be found in the literature.
Doppler velocimetry in the V-shape turbulent ¯ame All these works devoted to turbulent transport suggest
(Section 7.4.1) are displayed. that turbulent combustion modeling might probably be
The velocity distribution is almost Gaussian in the greatly improved by advancing the description of turbulent
unburnt and the burnt gases, but becomes clearly bimodal transport itself. This point motivates large eddy simulation
in the reaction zone denoting an intermittency between fresh (LES) for turbulent combustion modeling [155].
and burnt gases, according to Eq. (172). This bimodal distri-
bution does not correspond to the assumptions made in the
derivation of most turbulence models, such as k± 1 . 9. Reynolds averaged models for non-premixed
A conditional approach to determine the conditional turbulent combustion
averages of a quantity Q in fresh …Q u † and burnt …Q b †
gases is probably a promising way leading to a straight- 9.1. Introduction
forward description of turbulent transport (Eqs. (170) and
(172)). The objective is then to derive balance equations for Much work has been devoted to the numerical modeling
quantities such as: of non-premixed combustion systems, mainly assuming a
chemistry much faster than mixing and molecular diffusion.
rcQ ˆ r c~Q b ; r…1 2 c†Q ˆ r …1 2 c†
~ Q u This `mixed is burnt' regime is easily described from the
246 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

turbulent mixing of conserved scalars (Section 3.2) and gradient transport closure for the turbulent ¯uxes (Section
mixing problems have been the subject of many discussions 6), one may write:
[156±158].
There exist strong motivations for improving non-premixed 2Z~
r ~ Z~ ˆ rn
1 r u´7  t 7 2 Z~ …225†
and partially premixed turbulent combustion modeling: 2t

² The development of new combustion technologies for 2Zf


002
aircraft engines, and more generally for gas turbines r ~ Zg
1 r u´7  t 7 2 Zf
00 2 ˆ rn 002 1 2rn ~ 2 2 2r x~
 t u7Zu
2t |‚‚‚{z‚‚‚} |{z}
operating in the non-premixed regime, implies the accu- Production Dissipation
rate determination of the position in the ¯ow where …226†
combustion starts and the control of pollutants emission.
Crucial points which cannot be addressed using the in®- The ®rst term on the RHS is the turbulent transport, the
nitely fast chemistry hypothesis. second in Eq. (226) is the production of ¯uctuations by
² Many practical systems include liquid injection of the the mean gradients, the last is the scalar dissipation rate x~
fuel, followed by non-premixed and partially premixed remaining unclosed. r x~ ˆ rDu7Zu 2 was discussed for
combustion. premixed turbulent ¯ames, in the EBU, BML and S models
² Even in burners operating in the premixed regime, the (Section 7).
premixing of the reactants is not always complete at the
molecular level and some partial premixing may be 9.2.2. Balance equation and simple relaxation model for x~
observed. Sometimes, partial premixing is even desirable A transport equation may be derived for x~ from the Z
to limit pollutant emissions (strati®ed charge engines). balance equation (Eq. (41)):
!
As for premixed combustion (Sections 7 and 8), the 2r x~ 2r u~ j x~ 2 00 2Z 00 2Z 00
1 1 rDu j
modeling of turbulent diffusion ¯ames relies on simplifying 2t 2xj 2xj 2xi 2xi
assumptions for both chemistry and transport. Depending on ! !
the simpli®cations made for these mechanisms, various 2 2Z 00 2Z~ 2Z~ 2Z 00 2u 00j
approaches for laminar ¯ames and models for turbulent ˆ 22 rDu 00j 22 rD
2xj 2xi 2xi 2xi 2xj 2xi
¯ames are obtained (see Table 7). Hypotheses formulated |‚‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚‚} |‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚}
…I† …II†
to construct models for non-premixed turbulent ¯ames may
be organized into three major groups: ! !
2Z~ 2Z 00 2u j 00
2u~j 2Z 00 2Z 00
22 rD 22 rD
² Assumption of in®nitely fast chemistry (mixed is burnt). 2xj 2xi 2xi 2xi 2xi 2xj
|‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚} |‚‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚‚}
² Finite rate chemistry assuming a local diffusive±reactive …III† …IV†
budget similar to the one observed in laminar ¯ames
(¯amelet assumption). 2u 00j 2Z 00 2Z 00
2 2 rD
² Finite rate chemistry with treatment of molecular and 2xi 2xi 2xj
heat transport separated from chemical reaction (CMC, |‚‚‚‚‚‚{z‚‚‚‚‚‚}
…V†
pdf method; Section 6). Diffusion is then addressed using
! !
turbulent micromixing modeling, while chemical sources 2 2Z 00 2 2Z 00
can be dealt with in an exact and closed form (only for 2 2 D2 r
2xj 2xi 2xj 2xi
pdf). |‚‚‚‚‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚‚‚‚‚}
…VI†
The proposed closures are all based on a particular !
description of turbulent mixing, hence the basic concepts 2 2Z 2r 2 2Z
2 D rD (227)
useful to capture fuel/air turbulent mixing are ®rst r 2xi 2xi 2xj 2xj
|‚‚‚‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚‚‚‚}
presented. …VII†

9.2. Fuel/air mixing modeling The term (I) on the RHS describes curvature effects of the
mean mixture fraction ®eld, and as (II) and (III), contains
9.2.1. Introduction
correlations between the Z ®eld and the velocity ®eld.
The mean value of the mixture fraction Z~ gives an indica-
Compared to other terms, these correlations vanish for suf®-
tion of the local mean fuel/air mixing in turbulent ¯ows. In
ciently large Reynolds numbers. (IV) corresponds to the
addition, the structure and the properties of the ¯ame depend
correlations between mean ¯ow motion and ¯uctuations.
on Zf002 ; measuring the degree of mixing between reactants.
(V) is the strain rate of the scalar Z ®eld, already discussed
A simple description of turbulent mixing is thus obtained for ¯ame surface density modeling in premixed ¯ames
from the two ®elds: Z~ and Zf 002 : Introducing the classical
(Section 7.4.4). (VI) is the dissipation rate of the scalar
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 247

dissipation rate and (VII) is negligible when density gradi- 9.3. Models assuming in®nitely fast chemistry
ents are kept small.
To derive a linear relaxation closure for the scalar dissi- 9.3.1. Eddy dissipation model
pation rate, a homogeneous Z steady ®eld in equilibrium The eddy dissipation model (EDC) is a direct extension to
(production ˆ dissipation) is ®rst considered. Then, non-premixed ¯ame of the eddy break up (EBU) closure,
Eq. (227) reduces to a balance between the straining rate initially devoted to turbulent premixed combustion [160]
of Z and the dissipation rate of the scalar dissipation rate: (Section 7.2). The fuel burning rate is calculated according
to:
! ! !
2u 00j 2Z 00 2Z 00 2 2Z 00 2 2Z 00 1 ~O
Y Y~P
22rD ˆ 2D2 r rv_ F ˆ ar min Y~ F ; ;b …231†
2xi 2xi 2xj 2xj 2xi 2xj 2xi k s …1 1 s†
…228†
where a and b are adjustable parameters of the closure. In
the ¯uctuating velocity gradient is assumed to be of the Eq. (231), the reaction rate is limited by a de®cient species.
order of the inverse of the small scales characteristic time, To account for the existence of burnt gases bringing the
i.e. …2u 00j =2xi † , …1=tk † , …1=n†1=2 : The remaining part of the energy to ignite the fresh reactants, this species may be
strain rate term is proportional to x~ ; then, the reaction products. A priori, this model does not respect
the response of diffusion combustion in mixture fraction
2u 00j 2Z 00 2Z 00 r x~ r x~ space and may generate mean fuel mass fraction values
22rD , , Rl lower than Y~ IFCM : Actually, for large a , Eq. (231) is dif®cult
2xi 2xi 2xj tk …k=1† F
to justify.
where Rl is a Reynolds number based on the Taylor
microscale. A quadratic behavior in x~ ; with a direct depen- 9.3.2. Presumed pdf: in®nitely fast chemistry model
One of the ®rst descriptions of non-premixed combustion
dence on …Zf 002 †21 is anticipated for the dissipation rate of the
was given in Ref. [35] which assumed an in®nitely fast
scalar dissipation rate:
single step chemical reaction (Section 3.2). Considering
! ! the piecewise relations (Fig. 5, Table 1):
2 2Z 00 2 2Z 00 r x~ 2
2rD2 , Rl
2xj 2xi 2xj 2xi YF ˆ YFIFCM …Z†; YO ˆ YOIFCM …Z†; T ˆ T IFCM …Z† …232†
Zf 002

relating the fuel and oxidizer mass fractions, and the


The equilibrium condition (Eq. (228)) leads to: temperature to the mixture fraction in the case of in®nitely
fast chemistry, mean quantities may be directly obtained
Zf
002
with a b -pdf (Section 6.4.2) and Eq. (230):
x~ ˆ …229†
…k=1† Z1
Y~ IFCM
F
~ p ; x; t†dZ p
ˆ YFIFCM …Z p †P…Z …233†
recovering the widely used linear relaxation model. 0
Eq. (229) is a very simpli®ed description of micromixing.
Z1
The scalar dissipation rate depends on the detail of the char- Y~ IFCM ˆ ~ p ; x; t†dZ p
YOIFCM …Z p †P…Z …234†
O
acteristics of the turbulence (velocity and scalar energetic 0
spectrum), the linear relaxation is thus a ®rst approximation
Z1
that may be improved, for instance, by solving a modeled
T~ IFCM ˆ ~ p ; x; t†dZ p
T IFCM …Z p †P…Z …235†
balanced equation for x~ : Such a closed equation may be 0
found in Ref. [159].
The mean ¯ame structure is then known from an in®nitely
The knowledge of Z~ and Zf 002 from their closed transport
~ p ; x; t†; fast chemistry model (IFCM) without solving balance equa-
equation may be used to presume the pdf of Z, P…Z
tions for mean thermochemical quantities. The inputs of this
with a b -shape (see Section 6.4.2). Because the internal
structure of diffusion `mixed is burnt' closure are Z~ and Zf 002 determining the b -
 ¯ames is easily obtained from condi-
pdf chosen to capture the detail of fuel/air mixing. IFCM is
tional statistics as Yi uZ ˆ Z p (Section 3.2), many models
therefore a two-equations model for non-premixed turbulent
(¯amelet, CMC) incorporate a b -pdf to calculate means
combustion. It is very popular and is usually coupled with a
quantities:
low Mach number solution of the turbulent ¯ow [161], so
z‚‚‚}|‚‚‚{
Mixing that the thermodynamics of the ¯ow is fully known from Eq.
Z1
Y~ i …x; t† ˆ ~ p ; x; t† dZ p
p
…Yi uZ ˆ Z † P…Z …230† (235). IFCM is an interesting ®rst guess to provide the
0 |‚‚‚{z‚‚‚} global ¯ame structure for the maximum of heat that can
Flame structure
be released, a valuable result for certain aspects of design.
The simplest of these models invokes the in®nitely fast Note that the piecewise relations are exact when the chemi-
chemistry assumption. cal time is in®nitely small. Therefore, for very large
248 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

Fig. 38. Scatter plots of major species mole fractions and temperature as functions of mixture fraction: (a) Data from three-dimensional direct
numerical simulations of Montgomery et al. [164]; (b) Raman scattering measurements of Barlow et al. [163] in H2/argon±air ¯ame. Solid lines
indicate chemical equilibrium.

DamkoÈhler numbers (Section 3.2), IFCM is an accurate [163,164]). In these measurements and calculations, the
description of a turbulent diffusion ¯ame. Unfortunately response of the ¯ame in mixture fraction space lies in the
such a zero order model does not exist for premixed turbu- vicinity of the curves given by YiIFCM (Fig. 38).
lent combustion. To handle multi-step chemistry, in®nitely Models have been proposed for such ¯ames [63,165]. For
fast chemistry may be replaced by a chemical equilibrium a given state of mixing in the turbulent ¯ow, thus given
condition [162]. values of Z and x , ¯amelet models are derived assuming
This model lacks any prediction capacities when ignition, that the local balance between diffusion and reaction is
quenching or even small ®nite rate chemistry effects exist. similar to the one found in a prototype laminar ¯ame for
the same values of Z and x . Flamelet models are therefore
9.4. Flamelet modeling constructed from an asymptotic view of diffusive-reaction
layers as given by Fig. 6. The two control parameters of
9.4.1. Introduction planar and steady laminar strained ¯ames are used: the
Experiments in jets ¯ames and direct numerical simula- mixture fraction Z and its scalar dissipation rate x (Section
tions suggest that there exist situations in burners where the 3.2). In a turbulent ¯ow, these two quantities ¯uctuate in
chemistry is fast, but not in®nitely fast (see for example space and time, but when the joint pdf P…Z ~ p ; xp ; x; t† is
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 249

known, the mean properties of the ¯ame may be calculated other ¯ow time scale of the problem. The hypotheses involved
as: in SLFM can be disputed, leading to various improvements:
Z Z
Y~ i ˆ ~ p ; xp ; x; t†dxp dZ p
YiSLFM …Z p ; xp †P…Z …236† ² Eq. (237) has been obtained neglecting diffusion in the
Zp xp
direction tangential to the iso-Zst surface, arguing that
YiSLFM …Z p ; xp † is the local ¯ame structure in mixture fraction when the mixing element is suf®ciently thin and features
space and P…Z ~ p ; xp ; x; t† captures the statistics of fuel/air weak curvature, the gradients measured along the stoi-
mixing. SLFM stands for steady laminar ¯amelet model. chiometric surface are much smaller than those in the
This model may be viewed as a direct improvement of the perpendicular direction (Section 3.2). Thus, when using
in®nitely fast chemistry model (IFCM), since it uses the Eq. (237) to describe a prototype ¯ame of turbulent
same formalism, but with an additional parameter: the scalar ¯amelets, one supposes that the mixing ®eld may be
dissipation rate x , thereby including ®nite rate chemistry reduced to a steady one-dimensional structure. In conse-
effects. For a given chemistry, and therefore a given chemi- quence, the validity of Eq. (237) in a turbulent ¯ow also
cal time t c, when the DamkoÈhler number Dap ˆ …tc xst †21 is depends on the properties of micromixing, and up to now,
large, IFCM is recovered. An increase in x is followed by it is not obvious how to draw conclusions since one
®nite rate chemistry effects, or even quenching when x would need to measure the scalar dissipation rate in
becomes too large (Fig. 6), then, YiSLFM …Z p ; xp † features turbulent ¯ames. Some experimental results are available
mixing without reaction (Fig. 5). [169], but more works are required to conclude on the
The inputs of SLFM are similar to those of IFCM: Z; ~ Zf 002
dimensionality of scalar micromixing in ¯ames.
to which x~ is added. Two issues emerge: ² There are other issues related to the multi-dimensional
character of diffusion ¯ames. Straining cannot be
1. YiSLFM …Z p ; xp † must be determined and tabulated under uniformly distributed along the ¯ame sheet, leading to
particular hypothesis, choosing a given laminar ¯ame ¯amelet interactions when the distribution of x is non-
prototype. uniform on the iso-Zst. This transverse loss or gain of heat
~ p ; xp ; x; t† must be presumed using the mean values
2. P…Z modi®es the structure of the ¯amelet in the normal direc-
~ Zf
available to quantify fuel/air mixing (i.e. Z; 002 and x~ ). tion to the stoichiometric surface [55].
² Reference states at in®nity used to tabulate the ¯amelets
9.4.2. Flame structure in composition space, YiSLFM …Z p ; xp † may have to account for partial premixing [170,171].
YiSLFM …Z p ; xp † may be tabulated from solutions of coun- Consider the simple case of a jet ¯ame, where close to
ter-¯ow diffusion ¯ames (Fig. 4 [63]), a ¯ame con®guration the nozzle inlet, pure fuel and pure air react to form
widely studied experimentally [166,167]. Assuming within products. Moving downstream, turbulent diffusion
the turbulent ¯ow thin quasi-one-dimensional structures mixes these products with air, on the air side, and with
convected and stretched by the ¯uid motions, and neglecting fuel, on the fuel side. Further downstream, the reactants
higher order terms, the equations for the species and feeding the reaction zone are not likely to be either pure
temperature become (Section 3.2): fuel or pure air. This situation is strongly enhanced in
  2 ! ¯ows where recirculation zones are found to stabilize
2Yi x 2 Yi
ˆ v_ i 1 combustion. In Ref. [171], this shortcoming of SLFM is
2t Lei 2Z 2
overcome introducing transient ¯amelets, for which
!
2T XN
hn v_ n 22 T reference states at in®nity vary according to the value
ˆ2 1x ~
of a progress variable c:
2t nˆ1
Cp 2Z 2
² Flamelet libraries can be calculated in physical space or
where x ˆ …l=rCp †u7Zu 2 : in mixture fraction space. It is usually observed that
These equations were used to discuss laminar diffusion the decay of OH towards equilibrium is predicted by
¯ames (see Eqs. (47) and (50) and Section 3.2). Omitting the the ¯amelet solution in mixture fraction space using the
time derivative (steady ¯amelet), for a given value of x scalar dissipation rate as a control parameter sensitive to
corresponding to local micromixing conditions, one has to species boundary conditions, a trend that is not fully
solve for: reproduced by the ¯amelet solution in physical space
where the input parameter is the strain rate [171].
x 22 Yi
v_ ˆ 2 …237† ² Unsteadiness is also an important aspect. Time-depen-
Lei 2Z 2 dent ¯amelets have been used to include unsteady effects
The solution of this equation for given concentrations and [172±174]. As shown in Ref. [175], when a high value of
temperatures boundary conditions, and various x provides a the scalar dissipation rate is imposed to the ¯ame for a
Flamelet Library Yi(Z,x ). A variety of techniques are avail- suf®ciently short period of time, extinction may not be
able to build these libraries [168]. completed, with a limiting frequency at which the ¯ame
In SLFM, the characteristic time required to balance diffu- almost behaves like a steady state ¯ame. Unsteady ¯ame-
sion and reaction is assumed to be much smaller than any lets were used [176] to simulate extinction and re-ignition
250 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

Fig. 39. Three successive times of two-dimensional unsteady quenching of a planar diffusion ¯ame pinched by a pair of vortices, con®guration
studied in Ref. [53]. Left, bold: iso-reaction rate, dotted: vorcity, zoom: region of the ¯ow where the centerline response of the ¯ame versus
inverse mixture fraction dissipation rate is plotted. Right, solid line: 1D ¯amelet library, diamond: top left 2D ¯ame, square: middle left 2D
¯ame, triangle: bottom left 2D ¯ame.

in a turbulent jet ¯ame, history effects were included the reaction zone is fully dominated by unsteady effects.
using a Lagrangian time measured along the stoichio- ² Another related point of interest in ¯amelet theory is the
metric line. This work was pursued [177] introducing response of the turbulent ¯ame when quenching zones
as an additional control parameter, the diffusion time develop. In other words, one may discuss the assumption
needed to exchange mass and energy over a distance that the occurrence of local extinction at some points
DZ in mixture fraction space. does not prevent the use of a ¯amelet model for the
This is more generally related to questions arising remaining part of the turbulent ¯ame. Again using
concerning the determination of quenching limits along DNS, constant density ¯ames near extinction were
with the accuracy of quenching predictions using ¯amelet studied [182]. A critical DamkoÈhler number at which
theory a problem which can be addressed numerically, extinction occurs is determined as a function of a ¯ame
for instance by post-processing DNS databases to study thickness parameter, de®ned as the ratio between the rms
the reactive/diffusive layer in terms of ¯amelets mixture fraction and the reaction zone width in mixture
[70,74,178±181]. Numerical simulations of ¯ame/vortex fraction space. As expected for these ¯ames featuring
interaction, used to construct a combustion diagram strong unsteadiness effects, the value of the critical
([70], Fig. 10), show that when strong unsteadiness extinction DamkoÈhler number was different from the
and/or curvature effects appear, the laminar ¯amelet one predicted by laminar ¯ame theory. This observation
assumption does not always predict quenching (Fig. was explained by statistical variability. Consequently, an
10). Two critical DamkoÈhler numbers, DaLFA and Daext, extinction may be observed with DamkoÈhler numbers
are easily derived from the simulations to ®nd the limit larger or smaller than Dapq : In this last study, the lower
conditions where unsteady (DaLFA) and extinction (Daext) value of Daext differs from that of steady ¯amelets and
effects become important. These numbers are then exceeds the value Dapq given by ¯amelet theory.
compared with Dapq ; the quenching DamkoÈhler derived Other DNS results have shown that the scalar dissipation
from asymptotic analysis [30]. It is found that rate controlling the growth of the ¯ame hole is lower than
…DaLFA =Dapq † < 2 while …Daext =Dapq † < 0:4: Therefore, the one that should be applied to ®rst quench the ¯ame
Eq. (237) is an interesting approximation for Da p larger [38]. In Fig. 39, a DNS database is further analyzed [53]
than Dapq ; twice the asymptotic quenching value in these in which a diffusion ¯ame is pinched by a pair of vortices.
simulations. Below this value, even in the simple con®g- The value of x when the ¯ame extinguishes is in perfect
uration of ¯ame/vortex interaction, the time evolution of agreement with x q measured in a laminar ¯amelet library
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 251

sheet-like properties, they provide interesting results


(Fig. 40) and further re®nements of ¯amelet models are
under development [184]. It is also worthwhile to note
that with SLFM, the turbulent micromixing of chemical
species does not need any particular treatment, since diffu-
sion is directly included and coupled with chemistry in Eq.
(237).

9.4.3. Mixing modeling in SLFM


In steady laminar ¯amelet models (SLFM), the fuel/air
~ p ; xp ; x; t†:
turbulent mixing is captured via the joint pdf P…Z
Most of the ¯amelet models explicitly suppose that the
mixture fraction and its dissipation rate are two uncorrelated
quantities:
Fig. 40. Top: Favre-averages mean values of temperature and OH
P…Z ~ p ; x; t†P…
~ p ; xp ; x; t† < P…Z ~ xp ; x; t†
mole fraction along the centerline of a jet ¯ame. Bottom: radial
pro®les of mean temperature and OH mole fraction. Comparison ~ p ; x; t† (see Section
A b -function is assumed to presume P…Z
between experiments and ¯amelet modeling [177]. ~ xp ; x; t†
6.4.2) and a log-normal distribution is used for P…
[63]:
(Fig. 39, top). But once a hole exists in the ¯ame, the  
scalar dissipation rate xqEd at the quenching location is ~ xp ; x; t† ˆ p 1
P… exp 2
1
…ln x p
2 g …x; t††2

smaller than x q and varies depending on the structure of 2pxp s…x; t† 2s 2 …x; t†
the edge ¯ame. The ¯uxes of heat along the stoichio-
metric surface are responsible for this departure from The two parameters s and g are provided by the ®rst and
one-dimensional ¯amelet behavior. When the diffusion second moments of the scalar dissipation rate:
!
¯ame and its extremity have reaction zones of the same s2
thickness, for x . x q the response of the reaction rate x~ ˆ exp g 1 and xf002 ˆ x~ 2 …exp…s 2 † 2 1
2
follows the laminar ¯amelet one (Fig. 39, middle). Latter
after quenching, xqEd decreases and a partially premixed In Ref. [63], s 2 ˆ 2 is proposed.
front develops at the edge of the burning zone, then the Another alternative is to estimate a mean scalar dissipa-
reaction rate reaches values above the laminar ¯amelet tion rate under stoichiometric condition [89]. Starting from
behavior (Fig. 39, bottom). These values are representa- the solution of a steady strained planar counter-¯ow ¯ame:
tive of the existence of burning in a premixed regime at r Zj
1 p a r…j p † p
the extremity of the diffusion ¯ame. Therefore, the Z…h† ˆ erfc…h= 2†; where h…j† ˆ dj
quenching scalar dissipation rate x q is the relevant quan- 2 D 0 r0
tity to describe the quenching of a burning ¯amelet, but a is the strain rate, j the coordinate normal to the stoichio-
diffusion ¯ame quenching leads to edge ¯ame combus- metric plane and the assumption r 2 D < cst ˆ r20 D0 was
tion (partially premixed ¯amelets). used. The corresponding scalar dissipation rate is given by:
² When partially premixed ¯amelets are expected in the a
combustion system, for instance at the base of a lifted x…Z† ˆ exp…22‰erfc21 …2Z†Š2 †
2p
turbulent jet-¯ame, it may be interesting in the modeling
to use premixed ¯amelets instead of diffusion ¯amelets. ˆ x0 exp…22‰erfc21 …2Z†Š2 † ˆ x0 F…Z† …238†
Bradley and coworkers have proposed ¯amelet closures 21
along these lines where chemical sources are tabulated erfc denotes the reciprocal of the complementary error
using one-dimensional premixed ¯ames [183]. The function and x 0 is the maximum value of the scalar dissipa-
chemical sources are parametrized in terms of the tion rate. Mean and conditional scalar dissipation rate are
mixture fraction Z, that determines the equivalence then related by:
ratio of the ¯amelet whose progress of reaction is given ~ Zf
002 †
x~ st ˆ x~ F…Z; …239†
by a progress variable c. The mean burning rates are then
obtained by averaging the source terms with a presumed where F is given by:
form for the joint-pdf of Z and c.
~ Zf
002 † ˆ F…Zst †
F…Z; Z1 …240†
Within the family of closures for non-premixed turbulent ~ p †dZ p
F…Z p †P…Z
0
¯ames, models based on Eq. (237) represent great progress
compared to the in®nitely fast chemistry hypothesis. When and to retain in the ¯amelet library the pro®les
the chemistry is fast enough with mixing elements featuring ~ p † (Eq. (236)).
YiSLFM …Z p ; x~ st † for averaging with P…Z
252 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

Here again, the stumbling block is the estimation of the balance equation for the ¯ame surface density S may be
mean scalar dissipation rate x~ : The linear relaxation model derived (Section 6.3.2).
may be written (Eq. (229)): In ¯ame surface density models, under a ¯amelet assump-
tion, it is assumed that the interface between fuel and
Zf
002
oxidizer in non-premixed burners behaves, for any values
x~ ˆ Cx
…k=1† of Z, as S…Z p † < u7Zu ˆ S (Section 6.3.2). The mean burn-
ing rate v_ i is then expressed as the product of S by V_ i the
where Cx is a constant depending on the topological and
local reaction rate per unit of ¯ame area. S accounts for
spectral properties of the mixture fraction ®eld. More
¯ame turbulence interaction and V_ i for the chemistry.
recently, it was proposed to choose [33]:
According to Eq. (121)
…DZF †2  
x~ ˆ Cx v_ i
…k=1† v_ i ˆ u7Zu ˆ V_ i S …242†
u7Zu s
Cx < 2; DZF is the thickness of the reaction zone of these
laminar ¯ame measured in mixture fraction space. The reac- The local burning rate V_ i is estimated from a one-dimen-
tion zone is centered at the location Z ˆ Zst : As the stoichio- sional laminar ¯ame as:
metric surface is close to the air side for hydrocarbon ¯ame  
v_ i Z1 v_ Z1 1
…Zst p 1†; one may write: V_ i ˆ < i
dZ ˆ v_ i dj …243†
u7Zu s 0 u7Zu 21
DZF < 2Zst
which provides an interesting approximation of this reaction where j is the coordinate along the normal to the ¯ame
thickness. This last expression was successful to model front.
turbulent ¯ame lift-off [185]. In diffusion ¯amelets, where v_ i depends on the mixture
fraction Z and the scalar dissipation x , expressions (96), (91)
and the relation:
9.4.4. Conclusion
The laminar ¯amelet
 model
 presumes the conditional   Z
mean of species rYi uZ ˆ Z p averaging over the response v_ i uZ p ˆ v_ i …Z p ; xp †P c …xp uZ p †dxp …244†
x
of laminar prototype ¯ames:
  Z lead to:
rYi uZ ˆ Z p ˆ r ~ xp †dxp
YiSLFM …Z p ; xp †P…
1 Z1 Z
xp
V_ i ˆ v_ i …Z p ; xp †P c …xp uZ p †P…Z
 p †dxp dZ p …245†
In our generic classi®cation of turbulent combustion model- u7Zu 0 x
ing (Section 6) and Eq. (230), SLFM may be viewed as a
presumed pdf technique involving conditional mean values Then, using Eq. (243):
determined from laminar ¯ame solutions. Another alterna-
tive is to build models where quantities are estimated intro- Z1 1 Z
V_ i < v_ i …xp ; j†P c …xp uZ p †dxp dj
ducing a direct treatment of micromixing and small scale 21 x
diffusion (see Sections 9.6±9.8).
Z Z1 1 
<  xp †dxp
v_ i …xp ; j†dj P… …246†
9.5. Flame surface density modeling, coherent ¯ame model x 21
|‚‚‚‚‚‚‚{z‚‚‚‚‚‚‚}
Flame surface density concepts were ®rstly introduced for V_ i …xp †

the coherent ¯ame model (CFM) [84]. The balance equation


for the ¯ame surface density S was based on phenomeno- The reaction rate per unit of ¯ame area should, therefore, be
logical considerations starting from balance equation for a estimated from the integrated reaction rate of a ¯amelet
material surface where combustion effects have been in- submitted to a scalar dissipation rate x with the distribution
 xp †: In practice, V_ i is estimated as V_ i ˆ V_ i …x~ †; where x~

tuitively added [84,186]. Recent works [88,187,188] have
provided an exact balance equation, identifying the ¯ame is the averaged scalar dissipation rate. The coherent ¯ame
surface to the stoichiometric iso-surface Z ˆ Z st : The ¯ame model, then, corresponds to a simpli®ed version of the
surface density is then de®ned as: steady laminar ¯amelet model. The integrated burning rate
  V_ i may also be expressed as a function of the species mol-
S ˆ u7Zud…Z 2 Z st † ˆ u7ZuuZ ˆ Z st P…Z st † …241† ecular ¯uxes, see Eq. (53) in Section 3.2.
The introduction of effects of unsteadiness was realized in
 st † denotes the probability of ®nding Z ˆ Zst :
where P…Z Ref. [186], following the analysis proposed in Ref. [172] in
Starting from this de®nition and the balance equation for the context of SLFM, a ¯ame time response is then intro-
the mixture fraction Z (Eq. (41)), an exact but unclosed, duced when calculating V_ i:
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 253

tion occurs only when ignition is faster than turbulent


fuel/air mixing, the probability to ®nd ignition at a
particular location is:
Z1 1
a…Z; x; t† ˆ  tt ; x; t†dtt

tig …Z;x;t†

where P… tt ; x; t† is the pdf of the micromixing times


having a mean value equal to the integral length scale
time (k/1 ). Since t ig depends on position in mixture
fraction space and mean temperature, a is also a func-
Fig. 41. Sketch of MIL in mixture fraction space.
tion of Z and of the local ¯ow conditions through posi-
tion and time. Once P…  tt ; x; t† and P…Z
~ p ; x; t† are known,
9.6. MIL model the mean burning rate of fuel is estimated as:

MIL (`ModeÁle Intermittent Lagrangien' or Lagrangian Z1 Z1 1


p p ~ p  p
Intermittent Model) was proposed by Borghi in Ref. [68]. v~_ F …x; t† ˆ v_ MIL p
F …Z ; tt †P…Z ; x; t†P…tt ; x; t†dtt dZ
p
0 tig …Z p ;x;t†
As in the steady laminar ¯amelet model (SLFM) and the …247†
conditional moment closure (CMC), MIL incorporates the
¯ame structure in Z space. Arguing that mixing occurs with- A partially stirred reactor (PaSR), combining mixing
out reaction for large values of the scalar dissipation rate, and chemical effects, is introduced to calculate
corresponding to fast mixing, the ¯ame structure is v_ MIL p p
F …Z ; tt † [68], leading to:
constructed dynamically from two possibilities of diffusion
combustion: Mixing without reaction (before ignition) and dYFp …Y~ p 2 YF † 1 tpt v_ MIL
in®nitely fast chemistry (after ignition) (Fig. 41). The tran- ˆ F F
…248†
dZ …Z~ 2 Z p †
sition between mixing without combustion and a `mixed is
burnt' regime is controlled by the position of an ignition This technique is close to the one chosen in Ref. [190]
time delay, t ig, within the distribution of fuel/air turbulent for modeling NOx production in turbulent jet ¯ames. Eq.
mixing times tk , tt , 11, where tk . …n=1†1=2 is the (248) describes a local ¯ame element, where molecular
Kolmogorov time. diffusion, D7 2YF, is modeled as a linear relaxation term
The ignition delay t ig is associated to a scalar dissipation …Y~ F 2 YF †=tt : Solving Eq. (248) with boundary condi-
rate xig …tig < x21
ig †: The main objective of MIL is to account tions similar to the ones of ¯amelet libraries provides
for unsteadiness in the coupling between small scale diffu- v_ MIL p p
F …Z ; tt †:
sion and chemistry, by means of the spectral distribution of A generic shape for P… tt ; x; t† was proposed [189,191]:
micromixing times. For a given value of tig < x21 ig ; mixing
times smaller than t ig (t t , t ig or x . x ig) prevent ignition …q21†    
 tt ; x; t† ˆ …q 2 1† t …q 2 1† tt q
and contribute to pure mixing. Whereas for x , x ig, corre- P… exp 2
tqk q tk
sponding to t t . t ig, mixing is slower than chemistry and
combustion develops. q…x; t† is determined so that Y~ MIL calculated along the MIL
F
The inputs of MIL are: trajectory in mixture fraction space (Fig. 41), weighted by
the b -pdf, reproduces the mean Y~ F :
1. A table of ignition delay times t ig depending on mixture Fig. 42 shows an example of numerical simulation of a
fraction, fuel mass fraction and mean temperature, tig ˆ lean premixed prevaporized (LPP) combustion chamber
~ Y~ F ; Z†: This look-up table is easily derived from
tig …T; using MIL. The LPP system is divided into three parts:
partially stirred reactor (PaSR) calculations, where T~ ®rst, liquid fuel is injected and vaporized (prevaporized
and Y~ F are the mean conditions in the PaSR. Detailed tube), then is mixed with air in the premixing tube before
chemistry can be used to construct such a t ig table entering the main combustion chamber. Combustion is
[189]. expected to reduce pollutant emission in a lean premixed
2. The pdf of turbulent mixing times P…  tt ; x; t† (equiva- regime. The premixing tube operates for a large range of
lent to the pdf of the inverse of x ). inlet temperatures (from 300 to 800 K) and a large range of
3. The pdf of the mixture fraction P…Z ~ p ; x; t†: pressures (from 1 to 40 bar). For the highest temperature and
4. The mean concentration of fuel Y F : ~ pressure conditions, some undesirable effects, like auto-
ignition or ¯ashback, may be observed in the premixed
The output is the mean burning rate v~_ F to calculate tube. These ®ndings are reproduced with turbulent combus-
new Y~ F values, accounting for auto-ignition and transi- tion closures like MIL describing strongly coupled micro-
ent burning in turbulent ¯ames. Assuming that combus- mixing and ignition phenomena [191].
254 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

Fig. 42. Snapshot of temperature in a TURBOMECA-ONERA LPP engine computed with MIL [191]. 1: Vaporized tube, 2: premixed tube,
3: ¯ame tube. Combustion starts in the premixed tube.

9.7. Conditional moment closure of species concentrations and temperature for a given
level of mixture fraction may be included in the condi-
In®nitely fast chemistry (IFCM), steady laminar ¯amelets tional burning rate by using second order CMC. This
(SLFM), and MIL suggest that non-premixed turbulent leads to a variety of modeling re®nements where more
¯ames may be conveniently studied using conditional aver- transport equations for conditional quantities are studied,
aging in the mixture
 fraction
 space. Instead of getting Qi ˆ modeled and solved  [105]. 
rYi uZ ˆ Z p = ruZ ˆ Z p from laminar ¯amelets, the CMC ² The estimation of rxuZ ˆ Z p from a b -pdf is feasible
approach proposes to solve a balance equation for Qi for simple shapes of P…Z ~ p †; but as shown from the pdf
[102,103,105], which is written, neglecting two terms  
balance equation (Eq. (110)), the detail of rxuZ ˆ Z p is
(Eq. (114)):
  2Q   2Q linked to the detail of the shape of the pdf, that may not be
ruZ ˆ Z p i
ˆ 2 rui uZ ˆ Z p i correctly captured with a presumed pdf.
2t 2xi ² Within a homogeneous ®eld, the CMC equation reduces
|‚‚‚‚‚‚{z‚‚‚‚‚‚}
…I† to:
  22 Q     2Q   22 Q  
i i
1 rxuZ ˆ Z p i
1 v_ i uZ ˆ Z p ruZ ˆ Z p ˆ rxuZ ˆ Z p p2
1 v_ i uZ ˆ Z p
p2
2Z ‚} |‚‚‚‚{z‚‚‚‚} 2t 2Z
|‚‚‚‚‚‚‚{z‚‚‚‚‚‚
…II† …III† …251†
…249† This last formulation may be compared with the ¯amelet
In addition to turbulent transport included in (I), micro- equation (Eq. (237)) replacing Qi by Yi [107]. These two
mixing (II) and the chemical source (III) are unclosed. In turbulent combustion models have similarities in term of
the simplest version of CMC, the ¯uctuations of chemical their general formalism, but their underlying physical
source in mixture fraction space are neglected leading to: assumptions strongly differ. CMC incorporates diffusion
      as turbulent micromixing and decouples it from chemis-
v_ i uZ ˆ Z p ˆ v_ i Y1 uZ ˆ Z p ; ¼; YN uZ ˆ Z p …250† try. On the other hand, the ¯amelet assumption deals with
a coupled representation of diffusion and reaction as in a
As for any combustion models, the scalar dissipation rate is laminar ¯ame for a given value of x , and therefore, a
unknown
 (II).
 With CMC, the conditional average given thickness of the laminar diffusive zone.
rxuZ ˆ Z p needs closure. This conditional mean is an
important ingredient of the pdf transport equation and Once the number M of conditional values required to
micromixing models are available (Section 9.8.1). Simi- capture the conditional means of the N chemical species is
larly, when the pdf is known via its presumed b -shape determined, CMC modeling requires the solution of N £ M
balance equations. CMC was also extended to reacting
 value of x may be approxi-
(Section 6.4.2), the conditional
mated from P…Z~ p † to obtain rxuZ ˆ Z p [192]. particles [193].
CMC has been the subject of many works and re®nements
9.8. Pdf modeling
during the past years [105]. The following points have been
addressed: In premixed turbulent combustion modeling, BML and
related closures assume a bimodal probability density func-
² The approximation of the conditional chemical source tion (pdf) of the progress variable. In non-premixed ¯ames,
using Eq. (250) is much less restrictive than to write v~_ i < using in®nitely fast chemistry (IFCM), steady laminar
v_ i …Y~ 1 ; ¼; Y~ N † but implies a weak level of ¯uctuations for ¯amelet (SLFM), or conditional moment closure (CMC),
a given value of Z. When calculating the chemical source, the pdf of the mixture fraction was assigned a b -shape.
the conditional  i uZ p † is approximated by a peak
 pdf Pc …Y  The objective of PDF modeling is to relax all hypotheses
located at Yi uZ ˆ Z p : Accordingly, this method is concerning the shape of pdfs. Once a methodology has been
expected to provide interesting results for conditions developed to calculate pdfs, it is possible to construct turbu-
not too far from in®nitely fast chemistry. The ¯uctuations lent combustion closures in which all the values taken by
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 255

species is resolved (Eq. (113)). Recent developments


have also included time scale information into the pdf
[196,197].

There are two major issues with pdf modeling:

² Models must be proposed for capturing micromixing and


pressure ¯uctuation and viscous effects (the latter only
when the velocity ®eld is incorporated into the pdf).
² A new numerical treatment is required and leads to the
decomposition of the pdf into a large number of com-
putational particles required to succeed in the computa-
tional implementation of pdf methods [93,198,199].
The chemical composition and velocities of these par-
ticles evolve according to simple differential equations,
resulting from a discrete Monte Carlo simulation of the
continuous pdf equation.

The main advantage of the pdf method lies in the


possibility of treating complex chemical sources directly
(Section 6).

9.8.1. Turbulent micromixing


From the discussion of fundamentals properties of ¯ames
and of turbulent combustion models, the scalar dissipation
rate appears as a key quantity in both premixed and non-
premixed turbulent systems. Using pdfs and CMC, one
Fig. 43. Two-dimensional pdf of mixture fraction and reaction needs to express the conditional mean value of the scalar
progress variable for a round jet non-premixed ¯ame [104]. dissipation rate (or the conditional mean value of diffusion).
This term represents turbulent micromixing [95,96].
species and temperature in mixture fraction space may be Using turbulence modeling, such as the k± 1 closure, a
reached (Figs. 5 and 43). As in ¯amelet modeling and CMC, mean mechanical time tt < …k=1† and a microscale time
a detailed description of chemistry may be incorporated in tk < …n=1†1=2 may be estimated. A global picture of mixing
the modeling. Differential diffusion is easily introduced in should therefore distinguish between macro and micro-
SLFM when tabulating counter-¯ow ¯ames, these prefer- mixing [200]:
ential diffusion effects may be more dif®cult to introduce The macromixing is organized from a cascade mechanism
in CMC, in the context of pdfs, some works have been made between large and small scales. The study of this cascade in
in this direction [194,195]. a fully developed turbulence shows that the mixing
Two levels of complexity may be chosen: frequency fZ is a function of the characteristic length lZ
[201]:
² One may solve the balance equation for the joint pdf of  1=2 !2=3
the thermochemical variables, species, temperature, etc. 1 hk
fZ …lZ † ,
(Eq. (111) in Section 6): n lz
2 ~ p The speed at which Z is mixed increases when the charac-
r P…Z ; x; t†
2t teristic length lZ decreases, suggesting that the cascade
mechanism is limited by the mixing frequency fZ(lZ) of the
XN
2 h  
~ p ; x; t†
i
ˆ 2r p 2u´7Y i uY…x; t† ˆ Y p 1 v_ pi P…Z largest scales. When the Schmidt number of Z is greater than
2Y i
iˆ1
unity (n . Di), the impact of the molecular dissipation is
" ! # less than the effect of viscous dissipation, then the smallest
XN X N
2 2 2Yi 2Yj p ~ p
2r D Y…x; t† ˆ Y P…Z ; x; t† scales of Z are smaller than the Kolmogorov scales. For
iˆ1 jˆ1
2Yip 2Yjp 2xk 2xk
those small scales (lZ , h k) the mixing frequency is of the
…252† order of a constant fZ …lZ † ˆ …1=n†1=2 : This cascade process
from large to small scale does not change the energy in the
² Velocity ®eld statistics may also be included into the pdf, system and therefore does not affect Zf 002 :

then the balance equation for the joint pdf of velocity and The molecular diffusion acting at smaller scales makes
256 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

Fig. 44. Left: Scatter plot of D7 2Y. Right: …D7 2 YuY…x; t† ˆ Y p †: These two terms were measured in DNS of an inert mixing layer [234].

the gradients smoother and then induces the micromixing, at Z ˆ 0 and Z ˆ 1), the behavior at the edges of the sample
phenomenon required to put the reactants in contact with the space is not of prime importance to capture the major prop-
reaction zone. When n ˆ D; this process starts at the erties of micromixing, and a ®rst approximation is given by:
Kolmogorov scale, when n ± D; the Batchelor scale ! !
becomes the relevant limiting length:   Z 2 Z p Z 2 Z p
2
D7 ZuZ ˆ Z < p < …253†
tt k=1
lB , …nD2 =1†1=4 ˆ Sc21=2 hk
All these approximations are given for non-reactive ¯ows to This simple model cannot capture the time evolution of the
exactly quantify the effect of combustion is an arduous task. pdf in homogeneous mixing problems and has many short-
This dif®culty is also encountered in ¯amelet or ¯ame comings, for instance it relaxes the ¯uctuations without
surface modeling, but appears in a different form. modifying the shape of the pdf [94±96,204].
In pdf modeling,
 the micromixing
 closure must estimate
the term D7 2 Yi uY ˆ Y p describing the molecular diffu-
9.8.3. GIEM model
sion of each chemical species, or mimic its impact on the An improvement of the IEM model was recently
pdf. Various techniques have been proposed [95]. discussed [205], a b -pdf is used to reproduce the correct
behavior of conditional diffusion. The mixture fraction is
9.8.2. Linear relaxation model, IEM/LMSE chosen as a shadow ®eld to estimate micromixing, this
In b -pdf modeling (Section 9.2.2), the scalar dissipation closure is also called a `Beta-Mapping-Closure'. In a homo-
rate of the mixture fraction was expressed as x~ ˆ …Zf 002
=t †:
t geneous ®eld, the time evolution of the mixture fraction pdf
The extension of this linear relaxation to conditional diffu- is:
sion provides the linear mean square estimation (LMSE) or
interaction by exchange with the mean (IEM) model. First 
~ p ; x; t†
2P…Z 2 h ~ p ; x; t†
i
proposed in the pdf context by Dopazo and O'Brien [202], ˆ 2 p D7 2 ZuZ ˆ Z p P…Z
2t 2Z
IEM was also developed and used in chemical engineering
applications by Villermaux [203].  
Fig. 44 shows the conditional mean of the diffusive since D7 2 ZuZ ˆ 0 ˆ 0; one may write:
budget of the mixture fraction Z measured in a DNS of a !
  1 2 ZZ ~ 1
p
mixing layer. A quasi-linear response is observed close to D7 2 ZuZ ˆ Z p ˆ 2 P…Z ; x; t†dZ 1
the mean, with a slope that can be approximated using the ~ p ; x; t† 2t
P…Z 0
inverse of tt < …k=1†: Since the gradient of Z must relax …254†
to zero for Z ˆ 0 and Z ˆ 1; this quasi-linear
 response

is combined with the conditions D7 2
ZuZ ˆ 0 ˆ When P…Z~ p ; x; t† is given by a b -function parameterized
 
D7 2 ZuZ ˆ 1 ˆ 0: For a `Gaussian-type pdf' (no peaks with Z~ and Zf002 ; this equation determines …D7 2 ZuZ ˆ Z p †.
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 257

The LMSE/IEM model leads to: review), the simplest one consists of writing:
!
  1 Z 1 1 Z 11
D7 2 ZuZ ˆ Z p < …Z~ 2 Z p † …255† T…Z 1 ; Z 11 ! Z p † ˆ d Z p 2 …257†
tt 2

Making a direct analogy GIEM proposes: meaning that the two volumes interact to generate a volume
  with Z p ˆ …Z 1 1 Z 11 †=2: Eq. (257) was ®rst proposed by
D7 2 ZuZ ˆ Z p < C…Z p ; t†…BZ …t† 2 Z p † …256† Curl [206] to simulate the production of droplets of a given
diameter by coalescence of two droplets of various sizes and
where C(Z p, t) is a pseudo-mixing frequency depending both re-dispersion of the two droplets.
on the position in mixture fraction space and time. It allows These closures are clearly linked to a purely stochastic
for reproducing the correct behavior at the edges Z p ˆ 0 and view of micromixing and are well suited to Monte Carlo
Z p ˆ 1 and its time dependence gives the correct behavior simulations [93]. Their applications to ¯ames can be disputed
during the relaxation of the pdf. BZ(t) denotes the point easily since mixing described via stochastic processes does not
where the diffusive ¯ux is zero, relaxing the constraint always account for the presence of reaction zones. However,
BZ …t† ˆ Z~ of IEM, C(Z p, t) and BZ(t) are determined from much work has been done to improve these models which
Eqs. (254) and (256). have then been quite successful in the calculation of jet
The same treatment is applied to any reactive species Yi, ¯ames [97]. Fig. 45 shows pdfs of the temperature calcu-
using the same pseudo-mixing frequency (an assumption lated from Eq. (110) [207] using the coalescence-dispersion
made in most micromixing models): model [206] and the k±1 closure for the velocity ®eld [208].
  For the same calculations, Fig. 46 illustrates the impact of
D7 2 Yi uYi ˆ Yip ; Z ˆ Z p ˆ C…Z p ; t†…BYi …t† 2 Yip † micromixing modeling on temperature and CO concentra-
tions predictions, the reduced chemical scheme of Ref.
The parameter BYi …t† is then estimated from the fact that [209] has been utilized with the LMSE (IEM) and Curl's
micromixing by itself should not change the mean: coalescence re-dispersion formulation.
! Micromixing modeling is obviously one of the greatest
2Yi C…Z; t†Yi challenges of turbulent combustion modeling, and strong
ˆ …2D7 2 Yi † < 0 ! BYi …t† ˆ
2t C…Z; t† efforts have been made in this direction. Many types of mixing
models were proposed, a full review is beyond the scope of this
This attractive model conserves the simplicity of linear paper. Detailed discussion of mixing models in the context
relaxation modeling, but overcomes some of its drawbacks. of pdfs may be found in the following list of references
As Zf002 is an input of the closure, the mean scalar dissipation [93,95±97,104,194,196,205,206,208,210±236].
rate x~ should be previously modeled. This approach has
similarities with the estimation of the conditional values 9.8.5. Interlinks PDF/¯ame surface modeling
of x in CMC using b -pdf [192] and is not a full closure The unclosed molecular diffusion term in the pdf balance
of micromixing per se. Eq. (109) may be recast as:
!
1
…7´…rD7Z††uZ ˆ Z p
9.8.4. Stochastic micromixing closures r
Micromixing modeling may also be developed by choos- !
ing an ad hoc stochastic process mimicking the relaxation of 1 2 p
ˆ2 …rDu7Zu†u7ZuuZ ˆ Z
pdfs due to small scale diffusion. These closures are usually ru7Zu 2n
presented in the context of non-continuous interactions of  
volumes of ¯uid. When only diffusion is acting, the generic 2 D7´nu7ZuuZ ˆ Z p
form of these models is written [94]:
   
1 2
~ p ; t†
2P…Z 2b ZZ ˆ2 …rDu7Zu† 1kD7´nl s u7ZuuZ ˆ Z p
ˆ P…Z ~ 11 ; t†
~ 1 ; t†P…Z ru7Zu 2n s
2t tZ  
! ˆ 2‰kvn ls 1 kD7´nl s Š u7ZuuZ ˆ Z p
~ p†
 T…Z 1 ; Z 11 ! Z p †dZ 1 dZ 11 2 P…Z …258†
The molecular diffusion is decomposed into a normal contri-
b is a parameter of the model and T(Z 1, Z 11 ! Z p) is the bution, that may be expressed in terms of a normal displace-
probability that a volume of concentration Z 1 interacts with ment speed wn and a curvature term D7´n; according to
a volume at Z 11 to evolve into Z p. The dif®cult point lies in Eq. (38). The diffusion coef®cient D is assumed constant
the choice of the transition probability T. Multiple choices and kD7´nl s ˆ Dk7´nl s : The ¯ame surface averaged curva-
are possible and have been made (see Ref. [95] for a ture may be modeled from geometrical considerations
258 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

Fig. 45. Comparison between measurements (left) and numerical predictions (right) of unweighted temperature pdf at different locations in the
Masri et al. [207] `L' jet ¯ame (location x=D ˆ 20). Pdf calculations by Jones and Kakhi [208] using the Curl mixing model [206].

[88,121] as in Section 7.4.4 (Eq. (208)): It was shown in (Section 9.8.2) that for a non-reacting
scalar, the IEM model provides an interesting estimation
Z 2 Z 1 of the micromixing term around Z ˆ Z p when the turbulent
k7´nl s < 2 …259† mixing time is correctly calculated [234] but becomes de®-
Ly
cient when Z p tends towards 0 or 1, where u7Zu is expected to
where Ly is a wrinkling length scale of the ¯ame surface relax to zero. This shortcoming is attenuated in the ¯ame
and Z 1 is the Z level where mean curvatures are equal surface density context where the surface averaged curva-
to zero. Mean curvatures are supposed positive (i.e. ture k7´nl s is multiplied by the ¯ame surface density (Eq.
convex toward Z ˆ 0 region) and negative for Z ˆ 1. (90)), ensuring zero values at either side of the ¯ame brush.
The comparison of Eqs. (258) and (259) shows that The turbulent mixing time t t and the¯ame wrinkling  length
scale, L , are then related as L < D u7ZuuZ ˆ Z p t ; where
the geometrical assumption used to model the mean  y  y t
curvature corresponds to an LMSE±IEM model of the u7ZuuZ ˆ Z p characterizes laminar ¯ame elements.
molecular diffusion, when neglecting the normal displace- Expressing the mixing time from linear relaxation hypo-
ment speed kwn ls (Section 9.8.2): 2 thesis, tt < Zf 002 =x~ and using Eq. (116) under a ¯amelet

! assumption yields Ly < Zf 002 =u7Zu: Wrinkling length scales

1 Z 2 Z p depend on scalar ¯uctuations given by Eq. (226), eviden-


…7´…rD7Z††uZ ˆ Z p ˆ …260† cing, once again, the need for modeling scalar dissipation
r tt
rates.
2
In non-premixed turbulent ¯ames, the mean normal displace-
ment speed kwn ls is expected to be negligible against curvature
9.8.6. Joint velocity/concentrations pdf modeling
effects [88] because such ¯ames are unable to propagate. This is When the joint pdf of velocity and species enters the
not the case in premixed combustion, where the IEM±LMSE model problem, one needs to model additional viscous and
requires a correction to account for kwn ls that is of the order of the pressure terms related to velocity ¯uctuations (see Eq.
laminar ¯ame speed [237]. (113)). In turbulent non-reactive ¯ows, experiments have
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 259

Fig. 46. Radial pro®les of CO mass fraction and temperature at different locations in Masri et al. [207] `B' jet ¯ame, pdf calculations [208]. Two
mixing models have been used: ®lled circles: measurements; solid line: coalescence±dispersion micromixing Curl model [206]; dashed line:
LMSE micromixing model (IEM). Reduced chemical scheme of Jones and Lindstedt [209].

shown that the velocity ¯uctuations have pdfs that may be viscous unclosed terms of Eq. (113) may be written:
approximated with Gaussian functions. Closures for the " !
velocity pdf may then be obtained using a Langevin type 1 2p 0 p p
uu ˆ u ; Z ˆ Z
model, a simple stochastic model reproducing this velocity r 2xi
#
distribution, as done in Ref. [238].  
 p ; Z p ; x; t†
2 n7 2 u 0i uu ˆ up ; Z ˆ Z p P…u
Basic properties of stochastic methods [239] show that an

equation with a diffusive term for a pdf P: 2  p p
 p ; Z p ; x; t† 1 C0 1
ˆ Gij …upj 2 uj †P…u ‰P…u ; Z ; x; t†Š
2upi
 p†
2P…Z 2 2 …261†
 p †† 1 1 2 …B…Z p ; t†P…Z
ˆ 2 p …A…Z p ; t†P…Z  p ††
2t 2Z 2 22 Z p The physical properties of this closure are included in the tensor
Gij. The simplest model considers a linear relaxation for the
velocity ®eld, then Gij ˆ …1=2 1 3C0 =4†…1=k†dij ; where C0 is a
is well reproduced by a stochastic process, Z(t), similar to constant. More sophisticated methods have been proposed
the Langevin process: [240], where a parallel is drawn between Reynolds stress
closures (modeled equations for u 0i u 0j ) and the tensor Gij [241].
p As already indicated, a numerical solution for the pdf
dZ…t† ˆ A…Z…t†; t†dt 1 B…Z…t†; t†dW…t†
implies the development of accurate Monte Carlo solutions
[93]. The pdf ®eld is decomposed into a set of stochastic
where dW…t† is a white noise (Wiener process). Using this particles whose evolution simulates the turbulent ¯ame.
equivalence, the Langevin model [238] for the pressure and When a Lagrangian context is adopted [242], each particle
260 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

n is characterized by its position xpn …t†; its velocity upn …t† and from well established turbulent models (mostly k, 1 ), simply
concentration(s) Ynp …t†: The Langevin model given in Eq. re-written in terms of Favre averaged quantities. So far,
(261) corresponds to the system for the discrete particles: combustion effects on the ¯ow (¯ame induced turbulence
8
>
>
dxin generation, higher viscous dissipation, etc.) are not inten-
>
> ˆ uin
>
> dt tionally included in the calculations.
>
>   !  
< duin 1 2p 22 ui dWi There are many open questions and challenges left. One
ˆ2 1n 1 Gij …ujn 2 uj † 1 …C0 1†1=2
>
> dt r 2xi 2xj 2xj dt of them is the understanding of the coupling between spray
>
>
>
>
> dY p ~
…Y 2 Yi † p and combustion including detailed chemistry. DNS with a
>
: i ˆ i 1 v_ i …Yip †
dt …k=1† two-way coupling between a dilute spray and a carrier phase
were recently performed to progress in this direction [243]
…262†
and closures exist [243±245], but more work is needed to
The last dif®culty is the determination of the mean pressure
reach a level where liquid atomization, vaporization
 i †: Methods have been proposed to calculate it
®eld …2p=2x
together with ¯ames are properly described.
from Monte Carlo simulations [101]. In most of Monte
RANS combustion models will remain useful for the next
Carlo simulations, the Lagrangian simulation is coupled
few years. However, large eddy simulations (LES) stands as
with Eulerian calculations of means values providing
very promising technique for turbulent combustion:
the mean pressure ®eld and the dissipation rate of the
velocity ¯uctuations. However, full pdf methods includ- ² Combustion ¯ow ®elds generally exhibit large scale
ing a model for the turbulent frequency following a motions [26].
¯uid particle have also been developed with success [197]. ² LES appear as a promising tool to capture combustion
Fully detailed reviews on pdf methods may be found in instabilities.
Refs. [93±95]. ² Flames are mainly driven by mixing. LES is a good
candidate to capture unsteady turbulent mixing.
² LES may directly provide part of the description of turbu-
10. Conclusion
lence/combustion interactions because zones of fresh and
Modeling turbulent combustion is a challenging task. In burnt gases, having different turbulence characteristics, are
turbulent ¯ames, various dif®culties (strong heat release, instantaneously identi®ed at the level of the resolved grid.
complex chemistry, large range of time and length scales,
LES is at a very early stage for combustion applications and
etc.) are added to the complexity of constant-density turbu-
only few works have been done in this direction, mainly
lent ¯ows.
devoted to feasibility tests (two-dimensional simulations,
A review of the most classical Reynolds (or Favre) aver-
constant density ¯ows, etc.). Nonetheless, the results of these
aged Navier±Stokes (RANS) models has been proposed.
preliminary tests suggest that LES will rapidly become a
Three main ingredients must be modeled:
complementary way to carefully simulate and understand
² Reynolds stresses: ug ~i u~j : turbulent combustion systems. So far LES modeling is essen-
i uj 2 u
² Turbulent transport of species mass fractions: ug ~i Y~ k : tially based on the same modeling strategies as the ones
i Yk 2 u
² Mean reaction rate of species: v_ k : described here for RANS turbulent combustion modeling.

Most works have focus on re®ned descriptions of the


References
mean reaction rate v_ k : To provide this reaction rate, tools
have been proposed, which are based on a pure statistical or
[1] LinÄaÂn A, Williams FA. Fundamental aspects of combustion.
more oriented geometrical view of turbulent ¯ames. All Oxford: Oxford University Press, 1993.
these various models are related via a scalar dissipation [2] Clavin P. Premixed combustion and gasdynamics. Annu Rev
rate, a key quantity representing micromixing of the reac- Fluid Mech 1994;26:321±52.
tants occurring at the small scales where chemical reactions [3] Buckmaster J, Ludford G. Theory of laminar ¯ames.
develop. The links and the similarities between combustion Cambridge: Cambridge University Press, 1993.
models have been demonstrated here. The precise knowl- [4] Clavin P. Dynamics of combustion fronts in premixed gases:
edge of these relations may be useful to understand the exact from ¯ames to detonations. Proc Comb Inst 2000;28:569±85.
implications of the underlying physical hypothesis. Major [5] Roberts WL, Driscoll JF. A laminar vortex interacting with a
models are then directly derived from the proposed expres- premixed ¯ame: measured formation of pockets of reactants.
Combust Flame 1991;87:245±56.
sions, which illuminate unexpected links between well-
[6] Mueller CJ, Driscoll JF, Reuss DL, Drake MC, Rosalik ME.
known closure schemes. Vorticity generation and attenuation as vortices convect
Few studies have focussed on species turbulent transport, through a premixed ¯ame. Combust Flame 1998;112:342±58.
generally modeled using a gradient expression, whereas [7] Smooke MD, Mitchel RE, Keyes DE. Numerical solution of
counter-gradient turbulent transport is known to appear in two-dimensional axisymmetric laminar diffusion ¯ames.
some situations. Reynolds stresses are generally described Combust Sci Technol 1989;67:85±122.
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 261

[8] Givi P. Model free simulations of turbulent reactive ¯ows. [29] Peters N. The turbulent burning velocity for large-scale and
Prog Energy Combust Sci 1989;15:1±107. small-scale turbulence. J Fluid Mech 1999;384:107±32.
[9] Poinsot T. Using direct numerical simulations to understand [30] LinÄaÂn A. The asymptotic structure of counter¯ow diffusion
premixed turbulent combustion. Twenty-Sixth Symposium ¯ames for large activation energies. Acta Astronaut
(International) on Combustion. Pittsburgh: The Combustion 1974;1007(1).
Institute, 1996 p. 219±32. [31] LinÄaÂn A, Orlandi P, Verzicco R, Higuera FJ. Effects of non-
[10] Baritaud T, Poinsot T, Baum M. Direct numerical simulation unity Lewis numbers in diffusion ¯ames. Studying turbulence
for turbulent reacting ¯ows. Paris: Editions Technip, 1996. using numerical databasesÐV. In: Moin P, Reynolds WC,
[11] Poinsot T, Trouve A, Candel S. Applications of direct 1994. Stanford, CA: CTR, Stanford University. p. 5±18.
numerical simulations of premixed turbulent combustion. [32] Barlow RS, Fiechtner GJ, Carter CD, Chen J-Y. Experiments
Prog Energy Combust Sci 1996;21:531±76. on the scalar structure of turbulent CO/H2/N2 jet ¯ames.
[12] Vervisch L, Poinsot T. Direct numerical simulation of non- Combust Flame 2000;120(4):549±69.
premixed turbulent ¯ame. Annu Rev Fluid Mech [33] Bray KNC, Peters N. Laminar ¯amelets in turbulent ¯ames.
1998;30:655±92. In: Libby PA, Williams FA, editors. Turbulent reacting
[13] BarreÁre M, Prud'homme R. Equations fondamentales de ¯ows. London: Academic Press, 1994. p. 63±113.
l'aeÂrothermochimie. Paris: Masson, 1973. [34] Cuenot B, Poinsot TJ. Asymptotic and numerical study of
[14] Williams FE. Combustion theory. 2nd ed. Reading, MA: diffusion ¯ames with variable Lewis number and ®nite rate
Addison-Wesley, 1985. chemistry. Combust Flame 1996;104(1):111±37.
[15] Kuo KK. Principles of combustion. New York: Wiley, 1986. [35] Burke SP, Schumann TEW. Diffusion ¯ames. Ind Engng
[16] Poinsot T, Veynante D. Theoretical and numerical combus- Chem 1928;20:998±1004.
tion. Philadelphia, PA: Edwards, 2001. [36] Kim JS, Williams FA. Extinction of diffusion ¯ames with
[17] Trouve A, Poinsot T. The evolution equation for the ¯ame nonunity Lewis numbers. J Engng Math 1997;31:101±18.
surface density. J Fluid Mech 1994;278:1±31. [37] Clavin P, LinÄaÂn A. Theory of gaseous combustion. An intro-
[18] Favre A. Statistical equations of turbulent gases. In: SIAM, ductive course. In: Verlarde MG, editor. Non equilibrium
editor. Problems of hydrodynamics and continuum cooperative phenomena in physics and related ®elds. New
mechanics. Philadelphia: SIAM, 1969. p. 231±66. York: Plenum Press, 1983.
[19] Bray KNC, Libby PA, Masuya G, Moss JB. Turbulence [38] Favier V, Vervisch L. Edge ¯ames and partially premixed
production in premixed turbulent ¯ames. Combust Sci Tech- combustion in diffusion ¯ame quenching. Combust Flame
nol 1981;25:127±40. 2001;125(1/2):788±803.
[20] Shepherd IG, Moss JB, Bray KNC. Turbulent transport in [39] LinÄaÂn A. Ignition and ¯ame spread in laminar mixing layer.
con®ned premixed ¯ame. 19th Symposium (International) on In: Buckmaster J, Jackson TL, Kumar A, editors. Combus-
Combustion. Pittsburgh: The Combustion Institute, 1982. tion in high speed ¯ows. Dordrecht: Kluwer Academic
p. 423±31. Publishers, 1994. p. 461.
[21] Piomelli U, Chasnov JR. Large eddy simulations: theory and [40] Phillips H. Flame in a buoyant methane layer. Tenth Sym-
applications. In: HallbaÈck H, Henningson DS, Johansson posium (International) on Combustion. Pittsburgh: The
AV, Alfredsson PH, editors. Turbulence and transition Combustion Institute, 1965.
modelling, Dordrecht: Kluwer Academic Publishers, 1996. [41] Kioni PN, Rogg B, Bray KNC, LinÄaÂn A. Flame spread in
p. 269±336. laminar mixing layers: the triple ¯ame. Combust Flame
[22] Ferziger J. Large eddy simulation: an introduction and 1993;95:276.
perspective. In: MeÂtais O, Ferziger J, editors. New tools in [42] Plessing T, Terhoeven P, Peters N, Mansour MS. An exper-
turbulence modelling. Berlin: Les Editions de PhysiqueÐ imental and numerical study on a laminar triple ¯ame.
Springer, 1997. p. 29±47. Combust Flame 1998;115(3):335±53.
[23] Lesieur M. Recent approaches in large-eddy simulations of [43] Kioni PN, Bray KNC, Greenhalgh DA, Rogg B. Experimen-
turbulence. In: MeÂtais O, Ferziger J, editors. New tools in tal and numerical studies of a triple ¯ame. Combust Flame
turbulence modelling. Berlin: Les Editions de PhysiqueÐ 1998;116:192±206.
Springer, 1997. p. 1±28. [44] Dold JW. Flame propagation in a nonuniform mixture:
[24] Lesieur M, MeÂtais O. New trends in large-eddy simulations analysis of a slowly varying triple ¯ame. Combust Flame
of turbulence. Annu Rev Fluid Mech 1996;28:45±82. 1989;76:71±88.
[25] Veynante D, Poinsot T. Reynolds averaged and large [45] Hartley LJ, Dold JW. Flame propagation in a nonuniform
eddy simulation modeling for turbulent combustion. In: mixture: analysis of a propagating triple-¯ame. Combust
MeÂtais O, Ferziger J, editors. New Tools in Turbulence Sci Technol 1991;80:23±46.
Modelling. Berlin: Les Editions de PhysiqueÐSpringer, 1997. [46] Buckmaster J. Edge-¯ames and their stability. Combust Sci
p. 105±40. Technol 1996;115:41±68.
[26] Coats CM. Coherent structures in combustion. Prog Energy [47] Ghosal S, Vervisch L. Theoretical and numerical investiga-
Combust Sci 1996;22:427±509. tion of a symmetrical triple ¯ame using a parabolic ¯ame tip
[27] Menon S, Jou WH. Large eddy simulations of combustion approximation. J Fluid Mech 2000;415:227±60.
instability in an axisymmetric ramjet combustor. Combust [48] Bourlioux A, Cuenot B, Poinsot T. Asymptotic and nu-
Sci Technol 1991;75(1±3):53±72. merical study of the stabilization of diffusion ¯ames by hot
[28] Ghosal S, Moin P. The basic equations for the large eddy gas. Combust Flame 2000;120(1/2).
simulation of turbulent ¯ows in complex geometry. J Comp [49] Ruetsch GR, Vervisch L, LinÄaÂn A. Effects of heat release on
Phys 1995;118:24±37. triple ¯ame. Phys Fluids 1995;6(7):1447±54.
262 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

[50] Domingo P, Vervisch L. Triple ¯ames and partially premixed combustion. Twenty-Fifth Symposium (International) on
combustion in autoignition of nonpremixed mixtures. Twenty- Combustion. Pittsburgh: The Combustion Institute, 1994.
Sixth Symposium (International) on Combustion. Pittsburgh: [71] Cook A, Riley JJ. Direct numerical simulation of a turbulent
The Combustion Institute, 1996. p. 233±40. reactive plume on a parallel computer. J Comp Phys
[51] Plessing T, Terhoven P, Peters N, Mansour MS. An experi- 1996;129(2):263±83.
mental and numerical study on a laminar triple ¯ame. [72] Libby PA, Williams FA. Turbulent combustion: fundamental
Combust Flame 1998;115(3):335±53. aspects and a review. In: Libby PA, Williams FA, editors.
[52] Echekki T, Chen JH. Structure and propagation of methanol± Turbulent reacting ¯ows. London: Academic Press, 1994.
air triple ¯ames. Combust Flame 1998;114(1/2):231±45. p. 2±61.
[53] Favier V, Vervisch L. Effects of unsteadiness in edge-¯ames [73] Cuenot B, Egolfopoulos FN, Poinsot T. An unsteady laminar
and liftoff in non-premixed turbulent combustion. Twenty- ¯amelet model for nonpremixed combustion. Combust
Seventh Symposium (International) on Combustion. Pitts- Theor Model 2000;4(1):77.
burgh: The Combustion Institute, 1998. p. 1239±45. [74] Mahalingam S, Chen JH, Vervisch L. Finite-rate chemistry
[54] Ghosal S, Vervisch L. Stability diagram for lift-off and blow- and transient effects in direct numerical simulations of turbu-
out of a round jet laminar diffusion ¯ame. Combust Flame lent nonpremixed ¯ames. Combust Flame 1995;102(3):285±
2001;124(4)646±55. 97.
[55] Vervisch L. Using numerics to help understanding of nonpre- [75] Bray KNC. The challenge of turbulent combustion. Twenty-
mixed turbulent ¯ames. Proc Combust Inst 2000;28:11±24. Sixth Symposium (International) on Combustion. Pittsburgh:
[56] Borghi R. ReÂactions chimiques en milieu turbulent. PhD The Combustion Institute, 1996.
Thesis. Universite Pierre et Marie Curie, Paris 6, 1978. [76] Williams FA. Turbulent combustion. In: Buckmaster JD,
[57] Villasenor R, Pitz RW, Chen JY. Interaction between chemi- editor. The mathematics of combustion. Philadelphia:
cal reaction and turbulence in supersonic non-premixed SIAM, 1985. p. 116±51.
H2 ±air combustion. In: 29th Aerospace Science Meeting, [77] Kerstein AR, Ashurst W, Williams FA. Field equation for
Reno (Nevada), AIAA paper 91-0375, 1991. interface propagation in an unsteady homogeneous ¯ow
[58] Nieuwstadt FTM, Meeder JP. Large-eddy simulation of air ®eld. Phys Rev A 1988;37(7):2728±31.
pollution dispersion: a review. In: Metais O, Ferziger J, [78] Poinsot T, Echekki T, Mungal MG. A study of the laminar
editors. New tools in turbulence modelling. Berlin: Les ¯ame tip and implications for premixed turbulent combus-
Editions de PhysiqueÐSpringer, 1997. p. 264±80. tion. Combust Sci Technol 1991;81(1±3):45.
[59] Meeder JP, Nieuwstadt FTM. Subgrid-scale segregation of [79] Smiljanovski V, Moser V, Klein R. A capturing±tracking
chemically reactive species in a neutral boundary layer. In: hybrid scheme for de¯agration discontinuities. Combust
Chollet JP, Voke PR, Kleiser L, editors. Direct and large Theor Model 1997;1(2):183±215.
eddy simulation II. Dordrecht: Kluwer Academic Publishers, [80] Bourlioux A, Moser V, Klein R. Large eddy simulations of
1997. p. 301±10. turbulent premixed ¯ames using a capturing/tracking hybrid
[60] Bray KNC. Turbulent ¯ows with premixed reactants in approach. Sixth International Conference on Numerical
turbulent reacting ¯ows. In: Libby PA, Williams FA, editors. Combustion, New Orleans, Louisiana 1996.
Topics in applied physics, vol. 44. New York: Springer, [81] Kim W-W, Menon S, Mongia HC. Large-eddy simulation of
1980. p. 115. a gas turbine combustor ¯ow. Combust Sci Technol
[61] Borghi R. On the structure and morphology of turbulent 1999;143:25±62.
premixed ¯ames. Rec Adv Aerosp Sci 1985:117±38. [82] Piana J, Veynante D, Candel S, Poinsot T. Direct numerical
[62] Borghi R, Destriau M. Combustion and ¯ames, chemical and simulation analysis of the G-equation in premixed combus-
physical principles. Paris: Editions Technip, 1998. tion. In: Chollet JP, Voke PR, Kleiser L, editors. Direct and
[63] Peters N. Laminar ¯amelet concepts in turbulent combustion. large eddy simulation II. Dordrecht: Kluwer Academic
Twenty-First Symposium (International) on Combustion. Publishers, 1997. p. 321±30.
Pittsburgh: The Combustion Institute, 1986. p. 1231±50. [83] Fichot F, Lacas F, Veynante D, Candel S. One-dimensional
[64] Tennekes H, Lumley JL. A ®rst course in turbulence. propagation of a premixed turbulent ¯ame with a balance
Cambridge: MIT Press, 1972. equation for the ¯ame surface density. Combust Sci Technol
[65] Poinsot T, Veynante D, Candel S. Quenching processes and 1993:90.
premixed turbulent combustion diagrams. J Fluid Mech [84] Marble FE, Broadwell JE. The coherent ¯ame model of non-
1991;228:561±606. premixed turbulent combustion. Project Squid TRW-9-PU,
[66] Roberts WL, Driscoll JF, Drake MC, Goss LP. Images of the Project Squid Headquarters, Chaffee Hall, Purdue Univer-
quenching of a ¯ame by a vortex: to quantify regimes of sity, 1977.
turbulent combustion. Combust Flame 1993;94:58±69. [85] Candel S, Poinsot T. Flame stretch and the balance equation
[67] Bilger RW. The structure of turbulent non premixed ¯ames. for the ¯ame area. Combust Sci Technol 1990;70:1±15.
Twenty-Second Symposium (International) on Combustion. [86] Pope SB. The evolution of surfaces in turbulence. Int J
Pittsburgh: The Combustion Institute, 1988. Engng Sci 1988;26(5):445±69.
[68] Borghi R. Turbulent combustion modelling. Prog Energy [87] Vervisch L, Bidaux E, Bray KNC, Kollmann W. Surface
Combust Sci 1988;14:245±92. density function in premixed turbulent combustion model-
[69] Y.Y. Lee, Nonpremixed reacting ¯ows near extinction. PhD ing, similarities between probability density function and
Thesis. Cornell University, 1994. ¯ame surface approach. Phys Fluids A 1995;7(10):2496±
[70] Cuenot B, Poinsot TJ. Effects of curvature and unsteadiness 503.
in diffusion ¯ames, implications for turbulent diffusion [88] Van Kalmthout E, Veynante D. Direct numerical simulations
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 263

analysis of ¯ame surface density models for non premixed sium (International) on Combustion. Pittsburgh: The
turbulent combustion. Phys Fluids A 1998;10(9):2347±68. Combustion Institute, 1984. p. 505±12.
[89] Peters N. Turbulent combustion. Cambridge: Cambridge [110] Abdel-Gayed RG, Bradley D. Combustion regimes and the
University Press, 2000. straining of turbulent premixed ¯ames. Combust Flame
[90] Weller HG, Marooney CJ, Gosman AD. A new spectral 1989;76:213±8.
method for calculation of the time-varying area of a laminar [111] Yakhot V, Orszag CG, Thangam S, Gatski TB, Speziale
¯ame in homogeneous turbulence. 23rd Symposium (Inter- CG. Development of turbulence models for shear ¯ows by
national) on Combustion. Pittsburgh: The Combustion a double expansion technique. Phys Fluids 1992;4(7):
Institute, 1990. p. 629±36. 1510.
[91] Weller HG. The development of a new ¯ame area combus- [112] Gouldin FC. Combustion intensity and burning rate integral
tion model using conditional averaging. Thermo-Fluids of premixed ¯ames. Twenty-Sixth Symposium (Interna-
Section Report TF/9307. Imperial College of Science Tech- tional) on Combustion. Pittsburgh: The Combustion Insti-
nology and Medicine, 1993. tute, 1996. p. 381±8.
[92] O'Brien EE. The probability density function (pdf) approach [113] Im HG, Lund TS, Ferziger JH. Large eddy simulation of
to reacting turbulent ¯ows. In: Williams FA, Libby PA, turbulent front propagation with dynamic subgrid models.
editors. Turbulent reacting ¯ows. London: Academic Press, Phys Fluids A 1997;9(12):3826±33.
1980. p. 185. [114] Spalding DB. Mixing and chemical reaction in steady
[93] Pope SB. Pdf method for turbulent reacting ¯ows. Prog con®ned turbulent ¯ames. 13th Symposium (International)
Energy Combust Sci 1985;11:119±95. on Combustion. Pittsburgh: The Combustion Institute,
[94] Kollmann W. The pdf approach to turbulent ¯ow. Theor 1971. p. 649±57.
Comp Fluid Dynam 1990;1:285±349. [115] Said R, Borghi R. A simulation with a cellular automaton for
[95] Dopazo C. Recent developments in pdf methods. In: Libby turbulent combustion modelling, Twenty-Second Sym-
PA, Williams FA, editors. Turbulent reacting ¯ows. London: posium (International) on Combustion. Pittsburgh: The
Academic Press, 1994. p. 375±474. Combustion Institute, 1988 p. 569±77.
[96] Dopazo C, Valino L, Fuego F. Statistical description of the [116] Moss JB, Bray KNC. A uni®ed statistical model of the
turbulent mixing of scalar ®elds. Int J Mod Phys B premixed turbulent ¯ame. Acta Astronaut 1977;4:291±319.
1997;11(25). [117] Mantel T, Borghi R. A new model of premixed wrinkled
[97] Chen JY, Kollmann W. Pdf modeling of chemical nonequi- ¯ame propagation based on a scalar dissipation equation.
librium effects in turbulent non-premixed hydrocarbon Combust Flame 1994;96(4):443.
¯ames. Twenty-Second Symposium (International) on [118] Lahjaily H, Champion M, Karmed D, Bruel P. Introduction
Combustion. Pittsburgh: The Combustion Institute, 1988. of dilution in the BML model: application to a stagnating
[98] Abramowitz M, Stegun IA. Handbook of mathematical func- turbulent ¯ame. Combust Sci Technol 1998;135:153±73.
tions. NY, USA: Dover, 1970. [119] Poinsot T, Veynante D, Trouve A, Ruetsch G. Turbulent
[99] Bradley D, Lawes M, Scott MJ. Combust Flame ¯ame propagation in partially premixed ¯ames. Proceedings
1994;99:581±90. of the 1996 Summer Program. Center for Turbulence
[100] Lundgren TS. Distribution function in the statistical theory of Research, 1996. p. 111±36.
turbulence. Phys Fluids 1967;10(5):969±75. [120] Veynante D, Duclos JM, Piana J. Experimental analysis of
[101] Pope SB. Computations of turbulent combustion: progress ¯amelet models for premixed turbulent combustion, Twenty-
and challenges. Twenty-Third Symposium (International) Fifth Symposium (International) on Combustion. Pittsburgh:
on Combustion. Pittsburgh: The Combustion Institute, 1990. The Combustion Institute, 1994.
[102] Klimenko AY. Fluid Dynam 1990;25:327. [121] Veynante D, Piana J, Duclos JM, Martel C. Experimental
[103] Bilger RW. Conditional moment closure for turbulent react- analysis of ¯ame surface density model for premixed turbu-
ing ¯ow. Phys Fluids 1993;5(2):327±34. lent combustion. Twenty-Sixth Symposium (International)
[104] Chen J-Y, Kollmann W. Comparison of prediction and on Combustion. Pittsburgh: The Combustion Institute,
measurement in non-premixed turbulent ¯ames. In: Libby 1996. p. 413±20.
PA, Williams FA, editors. Turbulent reacting ¯ows. London: [122] Duclos JM. Etude theÂorique et expeÂrimentale d'une ¯amme
Academic Press, 1994. p. 213±308. turbulente stabiliseÂe par un barreau cylindrique. PhD Thesis.
[105] Klimenko AY, Bilger RW. Prog Energy Combust Sci Ecole Centrale Paris, 1997.
1999;25(6):595±687. [123] Martel C. Etude expeÂrimentale de la combustion turbulente
[106] Smith NSA. Conditional moment closure of mixing and reac- preÂmeÂlangeÂe. Analyse de modeÁles. PhD Thesis. Ecole
tion in turbulent nonpremixed combustion. In: Moin P, Centrale Paris, 1998.
Reynolds WC, editors. Annual research briefs. Stanford, [124] Bray KNC, Champion M, Libby PA. The interaction
CA: CTR, Stanford University, 1996. p. 85±99. between turbulence and chemistry in premixed turbulent
[107] Swaminathan N, Bilger RW. Assessment of combustion ¯ames. In: Borghi R, Murphy SN, editors. Turbulent reacting
submodels for turbulent nonpremixed hydrocarbon ¯ames. ¯ows, Lecture Notes in Engineering, vol. 40. Berlin:
Combust Flame 1999;116(4):519±45. Springer, 1989. p. 541±63.
[108] Borghi R. Turbulent premixed combustion: further discus- [125] Gouldin FC. Review of closure models for the ¯amelet
sions on the scales of ¯uctuations. Combust Flame regime of premixed turbulent combustion. Joint Meeting of
1990;80:304±12. the British and German Sections of the Combustion Institute,
[109] Abdel-Gayed RG, Bradley D, Hamid MN, Lawes M. Lewis Queens' College, Cambridge, UK, 1993. p. 52±5.
number effects on turbulent burning velocity. 20th Sympo- [126] Gouldin FC, Miles PC. Chemical closure and burning rates in
264 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

premixed turbulent ¯ames. Combust Flame 1995;100(1/ tion in premixed turbulent combustion. In: Center for Turbu-
2):201. lence Research, editor. Proceedings of the Summer Program,
[127] Meneveau C, Poinsot T. Stretching and quenching of ¯ame- Stanford, Stanford: Center for Turbulence Research, 1994.
lets in premixed turbulent combustion. Combust Flame p. 95±124.
1991;86:311±32. [146] Rutland C, Cant RS. Turbulent transport in premixed ¯ames.
[128] Bailly P. Contribution l'eÂtude de l'interaction turbulenceÐ In: Moin P, Reynolds WC, editors. Studying turbulence
combustion dans les ¯ammes turbulentes de preÂmeÂlange aÁ using numerical databasesÐV. Stanford, CA: CTR, Stan-
l'aide de modeÁles du second ordre. PhD Thesis. Universite de ford University, 1994. p. 75±94.
Poitiers, 1996. [147] Kalt PAM. Experimenal investigation of turbulent scalar ¯ux
[129] Bailly P, GarreÂton D, Simonin O, Bruel P, Champion M, in premixed ¯ames. PhD Thesis. University of Sydney, 1999.
Deshaies B, Duplantier S, Sanquer S. Experimental and [148] Boughanem H. Evaluation des termes de transport et de
numerical study of a premixed ¯ame stabilized by a rectan- dissipation de surface de ¯amme par simulation numeÂrique
gular section cylinder, Twenty-Sixth Symposium (Interna- directe de la combustion turbulente. PhD Thesis. University
tional) on Combustion. Pittsburgh: The Combustion of Rouen, France, 1998.
Institute, 1996. p. 923±30. [149] Boughanem H, Trouve A. The occurence of ¯ame instabil-
[130] Duclos JM, Veynante D, Poinsot T. A comparison of ¯amelet ities in turbulent premixed combustion. 27th Symposium
models for premixed turbulent combustion. Combust Flame International on Combustion 1998.
1993;95(1/2):101±18. [150] Paul RN, Bray KNC. Study of premixed turbulent combus-
[131] Bray KNC, Champion M, Libby PA. Premixed ¯ames in tion including Landau±Darrieus instability effects. Twenty-
stagnating turbulence part iv: a new theory for the Reynolds Sixth Symposium (International) on Combustion. Pittsburgh:
stresses and Reynolds ¯uxes applied to impinging ¯ows. The Combustion Institute, 1996. p. 259±66.
Combust Flame 2000;120(1/2):1±18. [151] Bray KNC, Moss JB, Libby PA. Turbulent transport in
[132] Domingo P, Bray KNC. Laminar ¯amelet expressions for premixed turbulent ¯ames. In: Zierep J, Oertel H, editors.
pressure ¯uctuation terms in second moment models of Convective transport and instability phenomena. Germany:
premixed turbulent combustion. Combust Flame University of Karlsuhe, 1982.
2000;121(4):555±74. [152] Veynante D, Poinsot T. Effects of pressure gradients on
[133] Gouldin FC, Bray KNC, Chen JY. Chemical closure model turbulent premixed ¯ames. J Fluid Mech 1997;353:83±114.
for fractal ¯amelets. Combust Flame 1989;77:241±59. [153] Libby PA. Theoretical analysis of the effect of gravity on
[134] Cant RS, Pope SB, Bray KNC. Modelling of ¯amelet surface premixed turbulent ¯ames. Combust Sci Technol
to volume ratio in turbulent premixed combustion. 23rd 1989;68:15±33.
Symposium (International) on Combustion, Orleans. Pitts- [154] Chen JY, Lumley JL, Gouldin FC. Modeling of wrinkled
burgh: The Combustion Institute, 1990. p. 809±15. laminar ¯ames with intermittency and conditional statistics.
[135] Prasad ROS, Gore JP. An evaluation of ¯ame surface density 21st Symposium (International) on Combustion. Pittsburgh:
models for turbulent premixed jet ¯ames. Combust Flame The Combustion Institute, 1986. p. 1483±91.
1999;116(1±2):1±14. [155] Boger M, Veynante D. Large eddy simulations of a turbulent
[136] Cheng WK, Diringer JA. Numerical modelling of SI engine premixed V-shape ¯ame. Eighth European Turbulence
combustion with a ¯ame sheet model. SAE paper 910268. Conference, Barcelona 2000.
International Congress and Exposition, Detroit, 1991. [156] Koochesfahani MM, Dimotakis PE. Mixing and chemical
[137] Choi CR, Huh KY. Development of a coherent ¯amelet reactions in turbulent liquid mixing layer. J Fluid Mech
model for a spark ignited turbulent premixed ¯ame in a 1986;170:83±112.
closed vessel. Combust Flame 1998;114(3/4):336±48. [157] McMurty PA, Riley JJ, Metcalfe RW. Effects of heat release
[138] Checkel MD, Thomas A. Turbulent combustion of premixed on the large-scale structure in turbulent mixing layer. J Fluid
¯ames in closed vessels. Combust Flame 1994;96(4):351± Mech 1989;199:297±332.
70. [158] Hermanson JC, Dimotakis PE. Effects of heat release in
[139] Prasad ROS, Paul RN, Sivathanu YR, Gore JP. An evaluation turbulent, reacting shear layer. J Fluid Mech 1989;199:
of combined ¯ame surface density and mixture fraction 333±75.
models for nonisenthalpic premixed turbulent ¯ames. [159] Jones WP. Turbulence modelling and numerical solution
Combust Flame 1999;117(3):514±28. methods for variable density and combusting ¯ows. In:
[140] Veynante D, Trouve A, Bray KNC, Mantel T. Gradient and Libby PA, Williams FA, editors. Turbulent reacting ¯ows,
counter-gradient scalar transport in turbulent premixed London: Academic Press, 1994. p. 309±68.
¯ames. J Fluid Mech 1997;332:263±93. [160] Magnussen BF, Hjertager BH. On the mathematical model-
[141] Dopazo C. On conditioned averages for intermittent turbu- ing of turbulent combustion with special emphasis on soot
lent ¯ows. J Fluid Mech 1977;81:433±8. formation and combustion. Sixteenth Symposium (Interna-
[142] Trouve A. The production of premixed ¯ame surface area in tional) on Combustion. Pittsburgh: The Combustion Insti-
turbulent shear ¯ow. Combust Flame 1994;99:687±96. tute, 1976. p. 719±29.
[143] Libby PA, Bray KNC. Countergradient diffusion in premixed [161] Ferziger JH, Peric M. Computational methods for ¯uid
turbulent ¯ames. AIAA J 1981;19:205±13. dynamics. Berlin: Springer, 1996.
[144] Masuya G, Libby P. Non gradient theory for oblique turbulent [162] Coupland J, Priddin CH. Modelling of ¯ow and combustion
¯ame with premixed reactants. AIAA J 1981;19:205±13. in a production gas turbine combustor. In: Durst F, Launder
[145] Trouve A, Veynante D, Bray KNC, Mantel T. The coupling BE, Lumley JL, Schmidt FW, Whitelaw JH, editors. Turbu-
between ¯ame surface dynamics and species mass conserva- lent shear ¯ows 5th. Berlin: Springer, 1987. p. 310±23.
D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266 265

[163] Barlow RS, Dibble RW, Chen JY, Lucht RP. Combust Flame [182] Lee YY, Pope SB. Nonpremixed turbulent reacting ¯ow near
1990(82). extinction. Combust Flame 1995;101:501±28.
[164] Montgomery CJ, Kosaly G, Riley JJ. Direct numerical simu- [183] Bradley D, Gaskell PH, Gu XJ. The mathematical modeling
lation of turbulent H2 ±O2 combustion using reduced chem- of liftoff and blowoff of turbulent non-premixed methane jet
istry. In: AIAA Paper 93-0248. 31st Aerospace Sciences ¯ames at high strain rate. Proc Combust Inst 1998;27:1199±
Meeting and Exhibit, Reno, NV, 1993. p. 1±10. 206.
[165] Peters N. Prog Energy Combust Sci page 1984:319. [184] Cuenot B, Egolfopoulos FN, Poinsot T. An unsteady laminar
[166] Sun CJ, Sung CJ, Wang H, Law CK. On the structure of ¯amelet model for non-premixed combustion. Combust
nonsooting counter¯ow ethylene and acetylene diffusion Theor Mod 2000;4:77±97.
¯ames. Combust Flame 1996;107(4):321±35. [185] MuÈller CM, Breitbach H, Peters N. Partially premixed turbu-
[167] Sung CJ, Liu JB, Law CK. Structural response of counter- lent ¯ame propagation in jets ¯ames. Twenty-Fifth Sympo-
¯ow diffusion ¯ames to strain rate variations. Combust sium (International) on Combustion. Pittsburgh: The
Flame 1995;102:481±92. Combustion Institute, 1994. p. 1099±106.
[168] Rogg B. The Cambridge universal laminar ¯amelet computer [186] Fichot F, Delhaye B, Veynante D, Candel S. Strain rate
code. In: Peters N, Rogg B, editors. Reduced kinetic mechan- modelling for a ¯ame surface density equation with applica-
isms for applications in combustion systems, Appendix C. tion to non-premixed turbulent combustion, 25th Symposium
Berlin: Springer, 1993. (International) on Combustion. Pittsburgh: The Combustion
[169] Nandula SP, Brown TM, Pitz RW. Measurements of scalar Institute, 1994.
dissipation in the reaction zones of turbulent nonpremixed [187] van Kalmthout E, Veynante D, Candel S. Direct numerical
H2 ±air ¯ames. Combust Flame 1994;99:775±83. simulation analysis of ¯ame surface density equation in non-
[170] Bish ES, Dahm WJA. A strained dissipation and reaction premixed turbulent combustion, Twenty-Sixth Symposium
layer formulation for turbulent diffusion ¯ames. Western (International) on Combustion. Pittsburgh: The Combustion
States Section Fall Meeting. Pittsburgh: The Combustion Institute, 1996.
Institute, 1993. [188] Van Kalmthout E, Veynante D. Analysis of ¯ame surface
[171] Ferreira JC. Flamelet modelling of stabilization in turbulent density concepts in non-premixed turbulent combustion
non-premixed combustion. PhD Thesis. Swiss Federal Insti- using direct numerical simulation. Eleventh Symposium on
tute of Technology, ETH, Zurich, 1996. Turbulent Shear Flows, Grenoble, France 1997.
[172] Haworth DC, Drake MC, Pope SB, Blint RJ. The importance [189] Obounou M, Gonzalez M, Borghi R. A lagrangian model for
of time-dependent ¯ame structures in stretched laminar predicting turbulent diffusion ¯ames with chemical kinetic
¯amelet models for turbulent jet diffusion ¯ames. Twenty- effects. Twenty-Fifth Symposium (International) on Combus-
Second Symposium (International) on Combustion. Pitts- tion. Pittsburgh: The Combustion Institute, 1994. p. 1107±13.
burgh: The Combustion Institute, 1988. p. 589±97. [190] Broadwell JE, Lutz A. A turbulent jet chemical reaction model:
[173] Pitsch H, Wan YP, Peters N. SAE Paper, 952357, 1995. NOx production in jet ¯ames. Combust Flame 1998:114.
[174] Zhang Y, Rogg B, Bray KNC. 2-d Simulation of turbulent [191] Ravet F, Vervisch L. Modeling non-premixed turbulent
autoignition with transient laminar ¯amelet source term combustion in aeronautical engines using pdf-generator.
closure. Combust Sci Technol 1995;105:211±27. 36th Aerospace Sciences Meeting and Exhibit AIAA paper
[175] Darabiha N. Transient behaviour of laminar counter¯ow 98-1027, Reno, NV 1998.
hydrogen±air diffusion ¯ames with complex chemistry. [192] Girimaji SS. Phys Fluids 1992;4:2529.
Combust Sci Technol 1992;86:163±81. [193] Smith NSA, Ruetsch GR, Oefelein J, Ferziger JH. Simulation
[176] Mauss F, Keller D, Peters N. A lagrangian simulation of ¯ame- and modeling of reactive particles in turbulent nonpremixed
let extinction and re-ignition in turbulent jet diffusion ¯ames. combustion. In: Moin P, Reynolds WC, editors. Studying
Twenty-Third Symposium (International) on Combustion. turbulence using numerical databasesÐVII. Stanford, CA:
Pittsburgh: The Combustion Institute, 1990. p. 693±8. CTR, Stanford University, 1998. p. 39±60.
[177] Pitsh H, Chen M, Peters N. Unsteady ¯amelet modeling of [194] Kerstein AR. A linear-eddy model for turbulent scalar trans-
turbulent hydrogen±air diffusion ¯ames. Twenty-Seventh port and mixing. Combust Sci Technol 1988;60:391±421.
Symposium (International) on Combustion. Pittsburgh: The [195] Fox RO. The lagrangian spectral relaxation model for differ-
Combustion Institute, 1998. p. 1057±64. ential diffusion in homogeneous turbulence. Phys Fluids
[178] Mell WE, Nilsen VN, KosaÂly G, Riley JJ. Direct numerical 1999;11(6):1550±71.
simulation investigation of the conditional moment closure [196] Fox RO. The spectral relaxation model of the scalar dissipa-
model for nonpremixed turbulent reacting ¯ows. Combust tion rate in homogeneous turbulence. Phys Fluids
Sci Technol 1991;91:179±86. 1995;7(5):1082±94.
[179] Delhaye B. Etude des ¯ammes de diffusion turbulentes. [197] Van Slooten Jayesh PR, Pope SB. Advances in pdf modeling
Simulations directes et modeÂlisation. PhD Thesis. Ecole for inhomogeneous turbulent ¯ows. Phys Fluids 1998;10(1):
Centrale Paris, 1994. 246±65.
[180] Delhaye B, Veynante D, Candel S. Simulation and modeling [198] Spielman LA, Levenspiel O. A Monte Carlo treatment for
of reactive shear layers. Theoret Comput Fluid Dynam reacting and coalescing dispersed phase systems. Chem
1994;6:67±87. Engng Sci 1965;20:247±54.
[181] Chang CHH, Dahm WJA, Trygvason G. Lagrangian model [199] Valino L. A ®eld Monte Carlo formulation for calculating the
simulations of molecular mixing, including ®nite rate chemi- probability density function of a single scalar in turbulent
cal reactions, in a temporally developing shear layer. Phys ¯ow. Flow Turb Combust 1998;60(2):157±72.
Fluid 1991;3(5):1300±11. [200] Fox RO. Computational methods for turbulent reacting ¯ows
266 D. Veynante, L. Vervisch / Progress in Energy and Combustion Science 28 (2002) 193±266

in the chemical process industry. Revue de L'Institut Fran- [224] Gao F, O'Brien EE. Joint probability density function of a
cËais du PeÂtrole 1996;51(2). scalar and its gradient in isotropic turbulence. Phys Fluids
[201] Corrsin S. The isotropic turbulent mixer. Part II. Arbitrary 1991;3(6).
Schmidt number. AIChE J 1964;10:870±7. [225] Gao F, O'Brien EE. A mapping closure for multispecies
[202] Dopazo C. Non-isothermal turbulent reactive ¯ows: stochas- ®ckian diffusion. Phys Fluids 1991;3(3).
tic approaches. PhD Thesis. University of New York, Stony [226] Kerstein AR. Linear-eddy modelling of turbulent transport.
Brook, 1973. Part 6. Microstructure of diffusive scalar mixing ®elds. J
[203] Villermaux J. Micromixing phenomena in stirred reactors. Fluid Mech 1991;231:361±94.
Encycl Fluid Mech 1986:707. [227] Chen J-Y, Kollmann W. Pdf modeling and analysis of ther-
[204] Eswaran V, Pope SB. Direct numerical simulation of the mal NO formation in turbulent nonpremixed hydrogen±air
turbulent mixing of a passive scalar. Phys Fluids jet ¯ames. Combust Flame 1992;88:397±412.
1988;31(3):506±20. [228] Jiang T-L, Givi P, Gao F. Binary and trinary scalar mixing by
[205] Tsai K, Fox RO. The bmc/giem model for micromixing in Fickian diffusionÐsome mapping closure results. Phys
non-premixed turbulent reacting ¯ows. Ind Engng Chem Res Fluids A 1992;4(5):1028±35.
1998:6. [229] Kerstein AR. Linear-eddy modelling of turbulent transport.
[206] Curl RI. Dispersed phase mixing. Theory and effects in Part 7. Finite rate chemistry and multi-stream mixing. J Fluid
simple reactors. AIChE J 1963;175(9). Mech 1992;240:289±313.
[207] Masri AR, Bilger RW, Dibble RW. Turbulent non-premixed [230] Fox RO. Improved Fokker±Planck model for the joint scalar,
¯ames of methane near extinction. Combust Flame scalar gradient pdf. Phys Fluids 1994;1(6):334±48.
1988;73:261±8. [231] Masri AR, Subramaniam S, Pope SB. A mixing model to
[208] Jones WP, Kakhi M. Pdf modeling of ®nite rate chemistry improve the pdf simulation of turbulent diffusion ¯ames.
effects in turbulent nonpremixed jet ¯ames. Combust Flame Twenty-Sixth Symposium (International) on Combustion.
1998;115(1/2):210±29. Pittsburgh: The Combustion Institute, 1996.
[209] Jones WP, Lindstedt RP. Global reaction schemes for hydro- [232] Fox RO. On velocity-conditioned scalar mixing in homo-
carbon combustion. Combust Flame 1988;73:233±49. geneous turbulence. Phys Fluids 1996;8(10):2678±91.
[210] Dopazo C. Relaxation of initial probability density functions [233] Fox RO. On the relationship between lagrangian micromix-
in the turbulent convection of scalar ®elds. Phys Fluids ing models and computational ¯uid dynamics. Chem Engng
1979;22(20). Process 1998;37:521±35.
[211] Janicka J, Kolbe W, Kollmann W. Closure of the transport [234] ReÂveillon J, Vervisch L. Subgrid mixing modeling: a
equation for the probability density function of turbulent dynamic approach. AIAA J 1998;36(3):336±41.
scalar ®elds. J Non-Equilib Thermodyn 1979;4(47). [235] Subramaniam S, Pope SB. A mixing model for turbulent
[212] Pope SB. An improved turbulent mixing model. Combust Sci reactive ¯ows based on Euclidean minimum spanning
Technol 1982;28:131±5. trees. Combust Flame 1998;115(4):487±514.
[213] Pope SB. Consistent modeling of scalars in turbulent ¯ows. [236] Subramaniam S, Pope SB. Comparison of mixing model
Phys Fluids 1982;26:404±8. performance for nonpremixed turbulent reacting ¯ows.
[214] Leonard A, Hill J. Direct numerical simulation of turbulent Combust Flame 1998;117(4):732±54.
¯ows with chemical reaction. J Scient Comput [237] Anand MS, Pope SB. Calculations of premixed turbulent
1988;3(1):25±43. ¯ames by pdf methods. Combust Flame 1987:67.
[215] Kraichnan RH. Closures for probability distributions. Bull [238] Haworth DC, Pope SB. A generalized Langevin model for
Am Phys Soc 1989;34:2298. turbulent ¯ows. Phys Fluids 1986;29:208±28.
[216] Dutta A, Tarbell JM. Closure models for turbulent reacting [239] Gardiner CW. Handbook of stochastic methods. Berlin:
¯ows. AIChE J 1989;35(12). Springer, 1997.
[217] Gao F. An analytical solution for the scalar probability [240] Pope SB. On the relationship between stochastic Lagrangian
density function in homogeneous turbulence. Phys Fluids models of turbulence and second-moment closures. Phys
1991;3(4):511±3. Fluids 1994;6:973±85.
[218] Valino L, Dopazo C. A binomial sampling model for scalar [241] Wouters HA, Peeters TWJ, Roekaerts D. On the existence of
turbulent mixing. Phys Fluids A 1990;2(7):1204±12. a stochastic Lagrangian model representation for second-
[219] Norris AT, Pope SB. Turbulent mixing model based on moment closures. Phys Fluids 1996;8(7):1702±4.
ordered pairing. Combust Flame 1991;83:27±42. [242] Pope SB. Lagrangian pdf methods for turbulent ¯ows. Annu
[220] Valino L, Dopazo C. A binomial langevin model for turbu- Rev Fluid Mech 1994;26:23.
lent mixing. Phys Fluids A 1991;3(12):3034±7. [243] ReÂveillon J, Vervisch L. Spray vaporization in non-premixed
[221] Valino L, Ros J, Dopazo C. Monte Carlo implementation and turbulent ¯ames: a single droplet model. Combust Flame
analytic solution of an inert-scalar turbulent-mixing test 2001;121(1/2):75±90.
problem using mapping closure. Phys Fluids A [244] Hollmann C, Gutheil E. Modeling of turbulent spray diffu-
1991;3(9):2191±8. sion ¯ames including detailed chemistry. Twenty-Sixth
[222] Pope SB. Mapping closure for turbulent mixing and reaction. Symposium (International) on Combustion. Pittsburgh: The
Theoret Comput Fluid Dynam 1991;2:255±70. Combustion Institute, 1996. p. 1731±8.
[223] Gao F. Mapping closure and non-gaussianity of the scalar [245] Calimez X. Simulation aÁ petite eÂchelle par une meÂthode VOF
probability density functions in isotropic turbulence. Phys d'eÂcoulement diphasiques reÂactifs. PhD Thesis. Ecole
Fluids A 1991;3(10):2438±44. Centrale Paris, 1998.

You might also like