Professional Documents
Culture Documents
Stanciu PDF
Stanciu PDF
Stanciu PDF
Abstract
The objective of this paper is to investigate the sources of volumetric irreversibilities in
both laminar and turbulent diffusion flames. The theoretical background of analysis
relies on the local exergy transport equation, which allows the microscopic formulation
of the well-known Gouy-Stodola theorem. For laminar reacting flows, the volumetric
entropy generation rate expression includes the viscous, thermal, diffusion and chemical
components. Their expressions show that the corresponding irreversibilities are
uncoupled if the combustion process occurs at constant pressure. The numerical
simulation of a methane-air combustion process shows that the thermal, chemical and
diffusive irreversibilities represent, in order of enumeration, the predominant
irreversibilities in the laminar diffusion reacting flows. In the case of turbulent diffusion
flames, the viscous, thermal, diffusion and chemical mean components have to be
expressed in accordance with the combustion model. Two combustion models are used:
the multi-species approach based on the eddy-break formulation of mean reaction rate,
and the assumed probability density function for a conserved scalar that relies on the
flame sheet model. For a diffusion methane-air jet flame, the distribution of mean
irreversibility components is presented. Taking into account the technical importance of
diffusion flames, the analysis could serve to improve the combustion geometry and the
flow condition.
Key words: Reacting flows, entropy generation, diffusion flames, combustion methane-
air
#
Author to whom all correspondence should be addressed.
∂τ ∂x α
( )
N
i=1,N (2) h * = ∑ Yi h i(o) + h i(s) + 12 u α u α =
i =1
[ ] ⎛ ⎞ (9)
∂
∂τ
( )
ρu β +
∂
∂x α
(V)
ρu α u β + Pδ αβ − τ βα =0
N
⎜ (o )
T
⎟
= ∑ Yi ⎜ h i + ∫ c p,i dT ⎟ + 12 u α u α
i =1 ⎜ ⎟
β=1,3 (3) ⎝ T0 ⎠
(o)
where h i represents the energy of formation of
( ) ∂ ⎡ N
∂
∂τ
ρe * +
∂x α ⎢⎣
∑
⎢ρu α h * + ρ i ϑ (αi ) h i + i-specie and RM is the universal gas constant.
Although the ideal gases mixture hypothesis
i =1
(V)
]
+ q& α − τ αβ u β = 0 (4a)
restraints the generality of the analysis, its
important role in the technical combustion
processes is well - known.
Of course the total energy e* and the total
In this formulation, the mathematical model
enthalpy h* appearing in the energy Eq. (4a)
of reacting flows includes all the chemical
[
= ρu α h * − h (o) − T0 s − s (o ) ( )] (20) where gi is the specific free enthalpy of the i-
component. In this case, the derivatives
The exergy diffusion flux appears because of the (∂µi/∂xα)T can be easily calculated and the
mass, heat and work interactions. Its components expression of the volumetric entropy generation
can be expressed as: rate (24) becomes (Stanciu et al., 2000):
Fα(d ) (ex ) =
N
∑ ρiϑα(i) (ex h )i + (Ex Q )α + (Ex τ )α =
Ω)
S& (gen ( ) + (S& ) + (S& ) + (S& )
Ω)
= S& (gen
V
( Ω)
gen Q
( Ω)
gen D
( Ω)
gen CH (26)
i =1
−λ
i =1
∂T ⎛ T0 ⎞
⎜1 − ⎟ − ταβ u β
(S& )
(Ω)
gen V =
τ βα ∂u β
T ∂x α
>0 (27)
∂x α ⎝ T ⎠
represents the viscous part of the volumetric
Finally, the exergy source contains both the entropy generation rate. It models the flow
volumetric production and the volumetric irreversibilities due to the shear stress tensor. The
dissipation rates. The volumetric rate of chemical following two terms:
to physical exergy conversion:
{ [
Π (ex ) = − ∆G (o) + R MT ln K p (T0 ) ω > 0 ]} (22) (S& ) + (S& )
(Ω)
gen Q
(Ω)
gen D =
λ
2
∂T ∂T
∂x α ∂x α
+
T
N ⎡ R ∂Yi ∂P R i ∂Yi ∂Yi ⎤
is proportional to the chemical exergy of the
reactants, ζch=-{∆G(o)+RMTln[Kp(T0)]}, which is
+ ∑ ρD im ⎢ pi + ⎥>0
∂x α ∂x α Yi ∂x α ∂x α ⎦
i =1 ⎣
released at a rate equal to the chemical reaction
rate, ω. But more important for our goal is the model the irreversibilities due to the heat and
volumetric rate of exergy dissipation, which has mass transfer interactions. Although the diffusion
the expression: phenomena due to the pressure and temperature
gradients were neglected in Fick’s law (5), these
∆(ex) = T0S& (gen
Ω)
>0 (23) irreversibilities are still coupled through the
product (∂Yi/∂xα . ∂P/∂xα) which has an
showing that it is proportional to the volumetric unknown sign. But neglecting the diffusion due
entropy generation rate. Of course the relation to the pressure gradients, as in Eq. (5), means
(23) represents the local formulation of the that:
Gouy-Stodola theorem.
dP P << dYi Yi
2.3 The entropy generation rate exp-
ression The above hypothesis has a wide validity
Eq. (19) shows that at each point of the because many engineering applications of
flow, the exergy variation depends strongly on combustion processes happen at P≈const. In this
4 Int.J. Applied Thermodynamics, Vol.4 (No.1)
case, the thermal and diffusion irreversibilities The flame chemistry is modeled by a single
uncouple so that: step chemical reaction like (R1) and the reacting
flow is governed by Eq. (1-4b). The numerical
(S& )
(Ω)
gen Q =
∂T ∂T
λ
T x α ∂x α
∂ 2
>0 (28) calculation was performed with the computer
code FLUENT 5.1.
Distribution of the viscous part of
(S& )
N
R i ∂Yi ∂Yi
(Ω)
gen D = ∑ ρD im
Yi ∂x α ∂x α
>0 (29) volumetric entropy generation rate has very
small values so it wasn’t plotted. Of course the
i =1
highest dissipation rates take place near burner
Finally, the last term in the expression of the entrance region because of the inlet velocity
volumetric entropy generation rate: difference between the two coaxial jets. A
viscous dissipation of the same magnitude also
(S& )
(Ω)
gen CH =
Aω ω N
T
=
T i =1
(ν ′i − ν ′i′ )µ M,i > 0
∑ (30) occurs around the adiabatic wall of the burner
where the non-slip boundary condition generates
high velocity gradients.
model the chemical irreversibilities due to
Figure 2a presents the distribution of the
reaction (R1). It can be seen that this term
thermal component of volumetric entropy
depends on the chemical reaction rate, and the
generation rate. By far this kind of dissipation is
molar chemical potential µM,i of each species. more important than the viscous one. The
2.4 Numerical simulation of laminar extremely high temperature gradients in the
diffusion flame irreversibilities flame front proximity generate the maximal
thermal dissipation in this area. Another place of
As an application of the above relations, let
significant thermal dissipation is located around
us consider the coaxial jet diffusion flame of
the fuel entrance because of the heat transfer
methane-air laminar combustion. Burner
from the flame. As expected, the maximal values
geometry and boundary conditions are presented
of thermal irreversibilities decrease along the
in Figure 1. The fuel, having the inlet velocity of
front of the flame due to the cutting down of
0.5 m/s, enters into the burner through a circular
temperature gradients. The diffusion component
inlet with diameter d=2 mm and the surrounding
variation of the volumetric entropy generation
air meets the methane jet with an inlet velocity of
rate is presented in Figure 2b. As in previous
0.05 m/s. The two coaxial jets are bounded by an
cases the most important diffusion dissipations
adiabatic outer wall.
are located around the flame front, but their
values are smaller than in the previous case.
Figure 2c shows the distribution of chemical
volumetric entropy generation rate. The highest
values of this irreversibility component are found
in the flame front where the most important part
of the chemical transformations happens. It can
also be seen that in the sites where they take
place, the chemical irreversibilities remain the
most significant among the flow irreversibilities.
The distribution of the total volumetric
entropy generation rate is presented in Figure 3.
It is easy to point out that around the flame front,
this distribution is identical with that of the
chemical volumetric rate, which rules in this
area. Along the normal direction at the front of
the flame, the methane’s concentration
continuously decreases and the reaction rate
slows down. As a consequence, the chemical
dissipation becomes smaller and smaller so that,
in these sites, the shape of the volumetric entropy
generation rate is formed by the thermal
component of the irreversibilities. It is surprising
that the diffusion part of volumetric dissipations,
that are greater than in the case of premixed
Figure 1. Burner geometry and boundary flame (Stanciu et al., 2000), doesn’t seriously
conditions for laminar flame calculation affect the shape of irreversibility distribution.
(S& )
gen V
(S& )
gen Q (S& )
gen D (S& )
gen CH S& gen
Entropy
flux
Error
For the sake of clarity, the smallest values of 3.1.1 The multi-species model
volumetric entropy generation rate were removed This model uses the average of full reacting
from Figures 2 and 3. flow equations. At this moment it is widely
Reacting flow irreversibility model, as well accepted that Favre decomposition of all
as the precision of calculation can be verified instantaneous properties, except density and
with the aid of the entropy transport Eq. (16). pressure, provides the most convenient way for
Taking into account that vertical boundaries are obtaining the averaged continuity (32), species
adiabatic and the outlet boundary is far from the (33), momentum (34) and enthalpy (35)
flame front, the integral of this equation reduces equations which become:
to:
∂ρ ∂( ρ~ uα )
r v (Ω) + =0 (32)
∫∫ ρsu ⋅ ndΣ = ∫∫∫ S& gen dΩ (31) ∂τ ∂x α
∂ΩC ΩC
gradient through the turbulent diffusion the local parameters σ , µ , α 0 and α1 are
coefficient DT, and the Reynolds heat flux vector determined functions of mean mixture fraction
q (αR ) is connected to the mean enthalpy gradient and mixture fraction variance (Elghobashi,
1977), (Lockwood, 1977) for which the
using the turbulent heat transfer coefficient λT . corresponding averaged transport equations are
These turbulent quantities are computed with the added.
aid of turbulent Schmidt and Prandtl numbers.
For standard and realizable K-εK models, these 3.2 The mathematical model of turbulent
turbulent quantities are constant, (i.e. ScT =0.7 reacting flow irreversibilities
and PrT=0.85), while for RNG K-εK model they The first step in modeling the flow
vary following an algebraic relation. irreversibilities is to average the instantaneous
In the case of diffusion flames, the entropy Eq. (16) which leads to:
characteristic chemical kinetics time is much ~
∂ (ρ ~s ) ∂ ⎡ ~ ~ N
∂Yi ~
smaller than the turbulent mixing time so that the
mixing always controls the combustion rate. The ∂τ
+ ⎢ ρ u α s + ρu ′α′ s ′′ −
∂x α ⎣⎢
∑ ρ D im
∂x α
si −
i =1
eddy break-up models are mixing controlled
combustion techniques that determine the mean N
∂Yi′′ ⎤
reaction rate ω~ as a function of mean mass − ∑ ρD im
∂x α ⎥
& (Ω)
s ′i′ + q& T ⎥ = Sgen (42)
~ i =1 ⎦
fraction field, Yi and the characteristic time of
turbulence, K/ε. For this work we selected the It is not very easy to model the correlations
well-known eddy-break formulation of appearing on the left hand side of the above
Magnussen and Hjertager (Magnusen and equation. But as in the laminar case, with some
Hjertager, 1976). simplifications imposed by the boundary
conditions, it can be used for checking the
& (Ω ) ⎛ & ( Ω ) ⎞
Sgen = ⎜ Sgen ⎟ + ⎛⎜ Sgen
& (Ω) ⎞
⎟ + ⎛⎜ Sgen
& (Ω ) ⎞
⎟ + ⎛⎜ Sgen
& (Ω ) ⎞
⎟ (44) is the dissipation rate of fluctuating i-component
⎝ ⎠V ⎝ ⎠Q ⎝ ⎠D ⎝ ⎠CH
mass fraction variance K (ψi) = 12 ρYi′′ 2 ρ . As in
In order to find the averaged expressions of the previous case, the first terms of the right hand
volumetric entropy generation rate components side of Eqs. (46) and (47) model the mean
(27)-(30), the instantaneous temperature, as well motion field irreversibilities, while the second
as the instantaneous mass fraction, are terms take into account the mean dissipations of
decomposed into their mean and fluctuating fluctuating field. For instance, the mean chemical
~
parts. So using the hypothesis T ′′ T << 1 , and source term is modeled as:
dropping the serial decomposition of
−1 ~ −1 ~ −1
T = T (1 + T ′′ / T ) at the first term, the (S& )
(Ω)
gen CH
~ N
ω
≅~ ∑
T i =1
(ν ′i − ν ′i′ )µ~ M,i (48)
following expression for volumetric viscous part
of entropy generation rate is obtained: where the mean chemical potential is simply
⎛⎜ S& (Ω) ⎞⎟ ≅ ⎛⎜ S& (Ω) ⎞⎟ + ⎛⎜ S (Ω) ⎞⎟
&
≅
~
computed as µ
~ ~ ~ ~
(
M ,i = h M ,i − Ts M ,i T, p, X i )
⎝ gen ⎠ V ⎝ gen ⎠ VM ⎝ gen ⎠ VT
(45) Correlating the irreversibility model and the
~τ ( V ) ~
βα ∂u β ρε mathematical formulation of diffusive reacting
≅ ~ + ~K flow, it can be seen that only the mean motion
T ∂x α T
irreversibilities, denoted by subscript M, and the
The first term of the above expression models the turbulent viscous one, identified by subscript VT
viscous irreversibilities of mean motion field due can be computed. For both thermal turbulent and
to the gradients of averaged velocity. It is the diffusion turbulent irreversibilities, the
homologue of the term modeling the laminar mathematical formulation of reacting flow does
viscous irreversibilities because it is generated by not give any information, because its procedures
( ) ⎛ dT ⎞ f (ξ )
1 2 where the subscript st denotes the stoichiometric
& (Ω) ρ~
Sgen
Q
= χ
2 ∫
ρc p ⎜⎜ ⎟⎟
⎝ dξ ⎠ T 2 (ξ )
dξ (51) conditions.
0 Unfortunately, the expressions (51), (52)
and (58) include both the mean motion and the
(S& ) ρ ~ ⎡ N Ri
1 2⎤
⎛ dYi ⎞
⎥ f (ξ )dξ
∫∑ turbulent parts of volumetric irreversibility
(Ω)
gen D = χ ⎢ ⎜⎜ ⎟⎟ (52)
2 ⎢ Y
0 ⎣ i =1 i ⎝ dξ ⎠ ⎥
⎦
components because this model doesn’t have the
Figure 5a. Viscous mean motion and viscous turbulent volumetric irreversibilities
Figure 5c. Diffusion mean motion and diffusion turbulent volumetric irreversibilities
RNG 8.1 x10-6 0.0011 0.456 28.499 0.0283 2.896 41.967 73.847 69.397 6.4
K-εK
Realiz.
K-εK 8.9x10-6 0.0019 0.508 31.752 0.0304 3.0486 39.824 75.16 70.792 6.1
τβα Stress tensor Launder, B. E., Morse, A., Rodi, W., Spalding,
D.B., 1972, “The prediction of free shear flows –
τ Time
A comparison of six turbulence models”,
ω Reaction rate Technical report, NASA SP – 311.
ΩC Computational domain
Libby, P., Williams, F, A., 1993, Turbulent
Subscripts Reacting Flows, Academic Press, New York.
.M Mean Lockwood, F., C., 1977, “The Modeling of
.T Turbulent part Turbulent Premixed and Diffusion Combustion
in the Computation of Engineering Flows”,
CH Chemical part
Combustion & Flame, 29, pp 111-122.
D. Diffusion
F Fuel Magnusen, B. F., Hjertager, B.,H., 1976, “On the
M Molar quantities Mathematical Model of Turbulent Combustion,
O Oxydant with Special Emphasis on Soot Formation and
Q. Thermal Combustion”, 16th Symp. On Combustion (The
Combustion Institute).
T Turbulent
V Viscous Peeters, T.W.J., 1995, “Numerical modeling of
in inert turbulent natural-gas diffusion flames”, Ph.D.
thesis, TU Delft.
Superscripts
Sanders, J.P.H., 1994, “Scalar transport and
ϕ Reynolds mean part of ϕ
flamelet modeling in turbulent jet diffusion
~
ϕ Favre mean part of ϕ flames”, Ph.D. thesis, Technische Universiteit
ϕ ′′ Favre fluctuating part of ϕ Eindhoven.
Ω Volumetric Sciubba, E., 1994, “Entropy Generation Rates as
V Viscous part a True Measure of Thermal and Viscous Losses
R Reynolds part in Thermochemical Components adn Systems”,
Workshop on Heat Transfer and
References Thermodynamics, Bucharest.
Abou-Ellail, M. M., Salem, H., 1990, “A skewed Stanciu, D., 1999, " The Study of Viscous and
PDF combustion model for jet diffusion flames”, Thermal Irreversibilities in Turbomachineries",
Journal of Heat Transfer, 112, pp. 1002-1007. Ph.D. Thesis, Polytechnic Univ. of Bucharest.
Bejan, A., 1983, Entropy generation through Stanciu, D., Isvoranu, D., Marinescu, M., 2000,
fluid flow and heat, John Wiley & Sons, New “Entropy generation rates in flames”,
York. Proceedings of ECOS 2000, pp. 333-342
Int.J. Applied Thermodynamics, Vol.4 (No.1) 17
Stanciu, D., Isvoranu, D., Marinescu, M, 2000*, Development and Validation”, Computer Fluids
“Second Law Analysis of the Turbulent Flat 24 (3).
Plate Boundary Layer”, Int. Journal of Applied Vilcu, R., 1988, Irreversible Thermodynamics,
Thermodynamics, Vol. 3, No. 3, pp 99-104. Ed. Tehnica, Bucharest.
Shih, T.,H., Liou, W. W., Shabbir, A., Zhu, J., Williams, F.A., 1985, Combustion Theory,
1995, “A New K-ε Eddy Viscosity Model for Benjamin / Cummings Publishing Co., Menlo
High Reynolds Numbers Turbulent Flows-Model Park, CA, second edition.