Lecture 12.

Modeling of Turbulent Combustion

X.S. Bai Modeling of TC

Content

• direct numerical simulation (DNS)

• Statistical approach (RANS)

– Modeling of turbulent non-premixed flames
– Modeling of turbulent premixed flames

• Large eddy simulation

X.S. Bai Modeling of TC

 Direct Numerical Simulation: DNS

• Solve the entire set of governing ∂ρ

equations + ∇ ⋅ ( ρv ) = 0,
∂t
– Down to the smallest flow
scales ∂ρYi ∂ρu jYi ∂ ⎛ ∂Yi ⎞
+ = ⎜⎜ i
ρD ⎟⎟ + ωi , i = 1,… N
– Down to the fine reaction ∂t ∂x j ∂x j ⎝ ∂x j ⎠
zones
∂ρ v
+ ∇ ⋅ ρ vv = ∇ ⋅ ( pI + τ ) , (i = 1, 2 ,3)
∂t
Dh Dp ⎛ N
⎞
ρ − = −∇ ⋅ ⎜ q + ρ ∑ YV i i i ⎟ + Qr + τ : ∇v
h
1 Dt Dt ⎝ i =1 ⎠
p=ρ Ru T ,
Wmix
XiX j
( ) ∇p ρ N
N
∇X i = ∑ V j − Vi + (Yi − X i ) + ∑ Yi Y j ( f i − f j )
N
⎛ p⎞ N Dij p p j =1
h ≡ ∑ Yi ⎜⎜ u si + uci + ⎟⎟ = ∑ Yi hi j =1

i =1 ⎝ ρ ⎠ i =1
T

f ,i (Tref )
p
hi = u si + uci + = ∫ pi
c dT + h 0

ρ Tref

X.S. Bai Modeling of TC

Principles of DNS

• Governing equations (N+5, N+4)

– Continuity equation, 1
– Momentum equations, 3
– Species transport equations, N (number of species)
– Enthalpy tranpsort equation, 1
– Equation of state, 1
– Calorific equation of state, 1
– Transport coefficients, N+2
• Independent variables to be simulated (2N+9)
– Density, pressure, temperature, 3
– Velocity components, 3
– Species mass fractions, N
– Enthalpy, 1
– Transport coefficients, N+2

X.S. Bai Modeling of TC

Principles of DNS

• Fully resolving all flow

scales
– Kolmogrov scales:
length, time,
velocity
– All flame scales:
reaction zones

X.S. Bai Modeling of TC

Principles of DNS

• Fully resolving all flow

scales
– Kolmogrov scales:
length, time,
velocity
– All flame scales:
reaction zones

X.S. Bai Modeling of TC

Cost of DNS to resolve one large eddy

l0
∝ Rel 0 3 / 4 ;
η
v0
∝ Rel 01/ 4 ;
vη
τ0
∝ Rel 01/ 2 ;
τη
Assuming the smallest grid is η and smallest time step is τ η
l0 τ0
Computational cost for 1-D ∼ ∼ Rel 0 5 / 4
η τη
⎛l 0 ⎞⎟ τ 0
2

Computational cost for 1-D ∼ ⎜⎜ ⎟⎟ ∼ Rel 0 2

⎜⎝ η ⎠⎟ τ
η

⎛l 0 ⎞⎟ τ 0
3

Computational cost for 1-D ∼ ⎜⎜ ⎟⎟ ∼ Rel 011/ 4

⎜⎝ η ⎠⎟ τ η

X.S. Bai Modeling of TC

Cost of DNS to resolve one large eddy

Total number of spatial mesh points x time steps needed for resolving one large eddy
scales of flames with different spatial dimensions and Reynolds numbers

Re 0
1-D 2-D 3-D

1 1 1 1
10 17.8 100 562
100 316 10,000 316227
1000 5623 1000,000 177,827,900
10,000 100,000 100,000,000 100,000,000,000

• DNS with detailed chemistry for an SI engine takes 30 years

• DNS is used for 2D
• DNS is used for low Reynolds number flames

X.S. Bai Modeling of TC

DNS of hydrogen flame, Mizobuchi et al, 29th symp

T=1000K

N.F.I iso-surfaces

H2 Jet flame: 9 species, 17 reactions, 30Dx30D, 22.8 million grids

N.F.I.: normalized flame index – square of concentration gradient

X.S. Bai Modeling of TC

Statistical methods (SM):
Ensemble Averages and Modeling

(Reynolds averaged Navier-Stokes

equations: RANS)

X.S. Bai Modeling of TC

Principles of ensemble averages

• Turbulent
flame is a
random
process

• Only the
statistical
mean field
is solved

X.S. Bai Modeling of TC

Ensemble average
u

M
1
Reynolds decomposition: u = u + u ′, u =
M
∑u
m =1
m

ρu
Favre decomposition: u = u + u ′′, u =
ρ

X.S. Bai Modeling of TC

 Cost of Statistical Methods to resolve one large eddy

Total number of spatial mesh points x time steps needed for resolving one large eddy
scales of flames with different spatial dimensions and Reynolds numbers

Re 0
1-D 2-D 3-D SM

1 1 1 1 1
10 17.8 100 562 1
100 316 10,000 316227 1
1000 5623 1000,000 177,827,900 1
10,000 100,000 100,000,000 100,000,000,000 1

X.S. Bai Modeling of TC

Governing equations for the mean flame

~
∂ ρ ∂ ρuj
Mass: + =0
∂t ∂ xj
~~
Momentum: ∂ρu~i ∂ρ uiu j ∂p ∂ ρui′′u′′j
+ =− −
∂t ∂x j ∂xi ∂x j

~ ~
Species: ∂ ρ Yi ∂ ρ u~ jYi ∂ ⎛ '' '' ⎞
+ =− ⎜ ρu jYi ⎟ + ω i
∂t ∂ xj ∂ xj ⎝ ⎠

Energy equation: similar as above

X.S. Bai Modeling of TC

Modeling issues

∂ρYi ∂ρYi u j ∂ ⎛ ∂Yi ⎞ ∂ ⎛ ′′ ⎞+ω

+ = ρ
⎜ i D ⎟ + ⎜ − ρ Y u ′′
j⎟
∂x j ⎜⎝ ∂x j ⎟⎠ ∂x j ⎝
i i
∂t ∂x j ⎠

Turbulent transport flux Turbulent reaction

rate

Turbulence models Combustion

e.g. K-epsilon model models

X.S. Bai Modeling of TC

Modeling of Turbulent Non-premixed flames

• Flame sheet model

• Flamelet models
• Eddy dissipation concept model
• Conditional moment closure models
• Probability density function models

X.S. Bai Modeling of TC

 T Direct photo CH

Turbuelnt
Combustion
of a fuel jet

X.S. Bai Modeling of TC

 Presumed PDF Burke-Schumann
model

X.S. Bai Modeling of TC

Burke-Schumann flame sheet model

• In 1970s Bilger advocated - in diffusion flames there is such as

‘magic’ variable called mixture fraction (Z). All the species mass
fractions, temperature, density etc, are uniquely related to Z …
• Burke-Schumann were the first one found this magic relationship

⎧ Z − Z st ⎧0.233(1 − Z / Z st ) Z ≤ Z st
⎪ Z ≥ Z st YO 2 = ⎨
YF = ⎨ 1 − Z st
⎩ 0 Z > Z st
⎪⎩ 0 Z < Z st
⎧ 1− Z
⎧ 1− Z ⎪⎪1 − Z (Tst − TFu ) + TFu Z ≥ Z st
⎪ Z ≥ Z st
YP = ⎨1 − Z st T =⎨ st
⎪⎩ Z / Z st Z
Z < Z st ⎪ (Tst − TOu ) + TOu Z < Z st
⎪⎩ Z st

X.S. Bai Modeling of TC

Ensemble average of flame sheet in turbulent flows

Z Z Zm Z Measurement
at a flow field point

ΔZ

t n p
N M M 1

∑ ∑ ∑
n( Z m )
∫ p(Z )ZdZ
1 1
Z= Z (ti ) ≡ n( Z m ) Z m ≡ Z m ΔZ ≡
N N NΔZ
i =1 m =1 m =1 0

N M M 1

∑ ∑ ∑
n( Z m )
∫ p(Z )Y (Z )dZ
1 1
Y= Y ( Z (ti )) ≡ n( Z m )Y ( Z m ) ≡ Y (Z m ) ΔZ ≡
N N NΔZ
i =1 m =1 m =1 0

Probability density function: pdf

X.S. Bai Modeling of TC

How to obtain PDF? Presumed PDF approach

Z a (1 − Z )b
Presumed PDF: p( Z ) = 1

∫
0
Z a (1 − Z )b d Z

∫
1
Z = Z p ( Z )dZ ,
0
g = Z' = Z − Z
2
( ) = ∫0 (Z − Z ) p(Z )dZ
2 1 2

a, b ⇔ Z , g Two equations, two unknowns

~
Mixture ∂ ρ g ∂ ρu j g ∂ ⎛ ρu′′ Z '2 ⎞ + P − ρ χ
fraction + =− ⎜ j ⎟
variance: ∂t ∂ xj ∂ xj ⎝ ⎠

X.S. Bai Modeling of TC

 Numerical implementation (flame sheet model)

~
∂ ρ ∂ ρuj
Mass: + =0
∂t ∂ xj
~~
Momentum: ∂ρu~i ∂ρ uiu j ∂p ∂ ρui′′u′′j
+ =− −
∂t ∂x j ∂xi ∂x j

Mixture ~ ∂ ρ u~ Z~
fraction:
∂ ρZ
∂t
+
∂ xj
j
=−
∂
∂ xj
(
ρu j Z ′′ , ) g − equation

1
Flame sheet
relation: ∫
T = T ( Z ) p ( Z )dZ ,....
0

X.S. Bai Modeling of TC

 Presumed PDF flamelet model

X.S. Bai Modeling of TC

 Influence of finite rate chemistry on flamelet structure

• Chemical CH4/air diffusion flame, p=1 bar, Tu=300 K

kinetics does not
affect the flame
shape and flame
height very much
!!!
• Chemical
reaction does
affect the
species and
temperature H
distribution a lot
!!!

δZ

X.S. Bai Modeling of TC

Influence of the finite rate chemistry
on maximum species mole fraction and T

• CH4/air diffusion flame, p=1 bar, Tu=300 K

X.S. Bai Modeling of TC

The flamelet library

• The flamelet equation can be derived using Crocco transformation

1 d 2Yi
χ 2
= wi
2 dZ
• Flamelet library

Yi = fi ( Z , χ ), T = fT ( Z , χ ), ρ = f ρ ( Z , χ )

– How to get ?

Solve the above flamelet equation using detailed chemical

kinetic mechanisms!!

X.S. Bai Modeling of TC

Numerical implementation

1∞
• Ensemble average
∫∫
~ 1
Yi = ℘( Z , χ ) f ρ ( Z , χ ) f ( Z , χ )dχdZ
i
ρ
0 0
1∞
ρ=
∫∫℘(Z , χ ) f ρ (Z , χ )dχdZ
0 0

• Presumed PDF
– How to get ? ℘( Z , χ )

Similar to flame sheet model. But here there are four unknown
parameters. One needs 4 transport equations.

X.S. Bai Modeling of TC

 Numerical implementation

Transport
Continuity + equations for
momentum k-epsilon the mean and
equations variance of
mixture
fraction, and
scalar
Ensemble averages dissipation
1∞ rate
ρ=
∫∫℘(Z , χ ) f ρ (Z , χ )dχdZ
0 0

~
Yi ....
X.S. Bai Modeling of TC
 Other modeling approaches

X.S. Bai Modeling of TC

Direct modeling of mean reaction rates: Eddy dissipation concept model

~ ~
Species: ∂ ρ Yi ∂ ρ u~ jYi ∂ ⎛ '' '' ⎞
+ =− ⎜ ρu jYi ⎟ + ω i
∂t ∂ xj ∂ xj ⎝ ⎠

Mixing and reaction zone

1 ⎛ Y ⎞
ωi = CEDC min⎜⎜ Y F, O ⎟⎟
t0 ⎝ γ ⎠

Fuel u0 air ‘Mixed is burned’

model
l0

X.S. Bai Modeling of TC

 Modeling of turbulent premixed flames

X.S. Bai Modeling of TC

photo CH2O CH

Vo=0.45 m/s, phi=1.17; Vin=120m/s, phi=1.0

X.S. Bai Modeling of TC
 Modeling of turbulent premixed flames

• Desirable Models
– taking into account the basic features of turbulent
premixed flames
• wrinkling
• stretch
• local extinction, re-ignition
with reasonably detailed chemistry
• local flame structure
• ...
– Computationally inexpensive
– Valid for wide parameter range

X.S. Bai Modeling of TC

 Modeling of turbulent premixed flames
a unified model does not exist

• Examples of models • Resolved issues

• k-ε model – Mean flame position
• global chemistry + – Mean major species
EDC/EBU ... • CO2, O2, UHC, …
• detailed chemistry + – Mean temperature
G-equation +
presumed PDF +
flamelet library • Unresolved issues
• BML ... – intermediate species
• Flame surface density • CO
models • NOx
• soot
– flame dynamics

X.S. Bai Modeling of TC

Direct modeling of mean reaction rates: flame surface density model

~ ~
Species: ∂ ρ Yi ∂ ρ u~ jYi ∂ ⎛ '' '' ⎞
+ =− ⎜ ρu jYi ⎟ + ω i
∂t ∂ xj ∂ xj ⎝ ⎠

V ρu AL S LYF ,u
sL ωF
Σ V
unburned burned ⎛ AL ⎞
= ρu S LYF ,u ⎜ ⎟
⎝V ⎠
l0
= ρu S LYF ,u Σ
mean reaction zone

X.S. Bai Modeling of TC

 Flamelet library approach

• Mean flame brush

– ensemble of laminar flamelets

• global structure
– Wrinkling and fluctuating laminar
flamelets

• local structure
– stretched local laminar flamelet

unburned
burned

X.S. Bai Modeling of TC

 Stretched laminar flamelet library

X.S. Bai Modeling of TC

 Influence of flame stretch on Laminar flames

• 1-D geometry
• Counterflow fresh-to-
burned configuration
• Counterflow fresh-to-
fresh twin-flame
configuration

• Detailed chemical kinetic

mechanisms (up to C3)
• Peters’ group (Lecture
notes in physics m15)

• Numerical code
• Chemkin
• Cantera

X.S. Bai Modeling of TC

 Level-set Based Flamelet Library Approach

Counterflow DNM
with detailed chemistry Level-set G formulation

Structures of laminar flamelet Statistics of flamelets

(quenching & species distributions) (fluctuations and wrinkling)

Ensemble average based on presumed PDF

Mean Turbulent Flame

X.S. Bai Modeling of TC

 Mean Flame Position – Level-set G-equation

~ ~
~ ~ ∂G ∂G ∂xi
G ( xi , t ) = G0 = 0 ⇒ + =0 (1)
∂t ∂xi ∂t
~
∂G
∂xi
ni = − ~ ~ ( 2)
∂G ∂G
∂x j ∂x j
dxi ~
= ui + ni sT (3)
dt
~ ~ ~
∂G ~ ∂G ∂G
Insert (3) in (1) ⇒ + ui = − sT ni
∂t ∂xi ∂xi
~ ~ ~ ~
∂G ~ ∂G ∂G ∂G
Use (2) ⇒ + ui = sT
∂t ∂xi ∂x j ∂x j

X.S. Bai Modeling of TC

 A test case: Bluff-body stabilized premixed flames

VR-1 LDA data: u/SL = 10 - 14; l/δL = 40 - 200

Thin reaction & flamelet regime (Peters) !

X.S. Bai Modeling of TC

 Previous RANS: CO Simulation

EDC

X.S. Bai Modeling of TC

RANS with new FLA: profiles at x=150 mm
(1)no stretch & wrinkling; (2)with stretch, no wrinkling; (3) with stretch & wrinkling
Nilsson & Bai 29th symp

X.S. Bai Modeling of TC

RANS with new FLA: profiles at x=350 mm
(1)no stretch & wrinkling; (2)with stretch, no wrinkling; (3) with stretch & wrinkling

Nilsson & Bai 29th symp

X.S. Bai Modeling of TC

Large eddy simulations: LES
• Filter away the small scales

• Retain the large eddies (larger than Taylor micro scales)

• Large scale unsteady motion is resolved

• Eddies smaller than the filter size need to be modeled

• Flame thickness is typically thinner than the LES grid size

• Models are needed to account for the unresolved scales

• Models are similar to the RANS models

• Computational cheaper than DNS, but more expensive than

RANS

X.S. Bai Modeling of TC

 Large Eddy Simulation of bluff-body flame

Flame surface G=0

Streamwise vorticity 500 1/s

Flame fluctuations, large scale wrinkling are captured !

X.S. Bai Modeling of TC

LES of HCCI engine

X.S. Bai Modeling of TC