RADIATION STATISTICS IN HOMOGENEOUS ISOTROPIC

TURBULENCE
C.B. da Silva1, I. Malico2, P.J. Coelho1 and J.C.F. Pereira1
1
Mechanical Engineering Department, Technical University of Lisbon,
Av. Rovisco Pais, 1049-001 Lisboa, Portugal
2
Physics Department, University of Évora, R. Romão Ramalho, 59, 7000-671 Évora, Portugal

Abstract

An analysis of the interaction between turbulence and radiation in homogeneous and isotropic
turbulence has been carried out. A direct numerical simulation code was used to generate
instantaneous turbulent scalar fields, and the radiative transfer equation was solved to provide
statistical data of the radiation intensity and absorption coefficient, as well as correlations related to
radiative emission and absorption. In addition, the time averaged radiative transfer equation was
solved and the mean radiation intensity, mean absorption coefficient, and mean emission and
absorption terms were computed and compared with those derived from the statistical data. An
analysis of the number of samples required to achieve statistically meaningful results is presented.
The influence of the optical thickness of the medium, mean and variance of the temperature, and
variance of the mean molar fraction of the absorbing species were studied. The moments of the
radiation intensity, Planck and incident mean absorption coefficient and emission and absorption
correlations relevant to the turbulence – radiation interaction were calculated. It was found that in
all cases the correlation between fluctuations of the absorption coefficient and fluctuations of the
radiation intensity is small, which supports the optically thin flame approximation, and justifies the
good predictions achieved using the time averaged radiative transfer equation.

1 Introduction

The interaction between turbulence and radiation (TRI) is a relevant issue in turbulent reactive
flows, yielding a significant increase of the radiative heat fluxes in comparison with laminar flows
(Li and Modest, 2005, Coelho, 2007). However, our knowledge about such interaction is still
limited. Direct numerical simulation (DNS) provides fundamental and reliable insight on turbulent
flows, but it can only be applied to simple geometries and low Reynolds number flows, because of
the high computational requirements. Recently, DNS has been used to investigate TRI in simple
premixed and diffusion combustion systems (Wu et al., 2005, Deshmukh et al., 2005).

Silva et al., 2006, have reported statistical data of the radiation intensity field in a homogeneous
isotropic turbulent non-reactive flow using a pseudo-spectral code for the DNS. In the DNS
calculations, a transport equation for a passive scalar was solved. The instantaneous scalar field
computed from DNS was used as input data for the radiative transfer calculations, i.e., the
instantaneous fields of temperature and molar fraction of an absorption species were determined
from that scalar field, prescribing the mean value and the variance of the temperature and of the
molar fraction of the species. Then, these instantaneous scalar fields were used to solve the radiative
transfer equation (RTE) in a narrow band, along a large number of lines of sight, in order to collect
statistical data. A statistical narrow band model was used to calculate the radiative properties of the
medium. However, some features of the radiative calculations reported in that work were not fully
consistent with the requirements of the flow simulation, such as the definition of the boundary
conditions, and the dependence of the statistical data on the number of samples was not
investigated. These drawbacks are eliminated in the present work, which extends the analysis to the
full spectrum rather than just one band, and examines the influence of several parameters on the
radiation statistics and correlations. In addition, predictions obtained from the solution of the time
averaged form of the RTE are included.

2 Theory

The DNS calculations were carried out using a standard pseudo-spectral code in which the temporal
advancement is made with an explicit 3rd order Runge-Kutta scheme (Canuto, 1988). The physical
domain is a periodic cubic box of side 2π. A DNS simulation of statistically steady (forced)
homogeneous isotropic turbulence was carried out using a uniform mesh with 1923 grid nodes. The
instantaneous field of a passive scalar was computed and taken as input data for the radiative
transfer simulations. The analysis was performed using up to 40 instantaneous fields after all the
turbulence quantities are statistically stationary. Details of the simulation may be found in Silva and
Pereira, 2007.

The radiative transfer calculations were performed using also a cubic box. The size of the radiation
domain is different from the size of the flow domain, and it was defined in a way different from that
formerly used in Silva et al., 2006. Here, the length of the side was taken as the ratio of the optical
thickness of the medium, which was prescribed, to the Planck mean absorption coefficient in the
absence of turbulent fluctuations. The mean and the variance of temperature and of the molar
fraction of an absorbing species, taken as CO2, were also prescribed. Data from the flow domain
(DNS) were rescaled into the radiation domain as reported in Silva et al., 2006, providing the
instantaneous fields of temperature and molar fraction of CO2. These two scalars were assumed to
be fully correlated. This assumption is justified by the experimental data in many reactive flows,
which reveals a strong correlation between temperature and molar fraction of the species, and by the
Burke-Schumann theory for diffusion flames (Kuo, 1986). The radiative properties of the medium
were evaluated using the correlated k-distribution (CK) method (Goody et al., 1989).

Under the conditions of homogeneous and isotropic turbulence, the statistical data computed from a
time series of scalar data along a single optical path parallel to a coordinate axis is identical to the
statistical data calculated from all optical paths parallel to the coordinate axes at a given time, as
illustrated in figure 1. The statistical data reported below was obtained from the DNS data using all
the available optical paths parallel to the coordinate axes, which are statistically indistinguishable.
This means that 6×1922×Nt ≈ 2.2×105×Nt samples are used to obtain the results described below, Nt
being the number of instantaneous fields considered.

The integration of the RTE along a line of sight yields

s s s
()
$% 0 !" ()
0
() $% s*
( )
I i ,)* k (s ) = I i ,)* k (0 ) exp &' ( k i s * ds * # + ( k i s * I b,)* k s * exp &' ( k i s ** ds ** # ds *
!"
(1)

where ki is the absorption coefficient associated with the ith quadrature point, and Δνk is the kth
wavenumber interval length. In the DNS calculations, periodic boundary conditions have been
enforced at the boundaries of the computational domain, in order to describe a homogeneous
isotropic flow. Therefore, a similar boundary condition should be used in the radiative transfer
calculations if the DNS data are taken as an input. This periodic boundary condition requires that
the radiation intensity entering the calculation domain at s = 0 is equal to the leaving intensity, i.e.,
I i ,"! k (s ) = I i ,"! k (0 ). Accordingly, the radiation intensity entering the domain is calculated from
this condition. Such a condition is actually found in real radiative transfer problems in the limit of a
transparent medium or of a homogeneous and isothermal optically thick medium. The integrals in
equation (1) were numerically evaluated using Simpson’s rule, and the parameters of the CK
method were interpolated from the tabulated data from Soufiani and Taine, 1997, using cubic
splines in order to keep the order of accuracy of the radiative calculations consistent with the order
of accuracy of the DNS solver (see Silva et al., 2006, for details).

The statistical results obtained from the solution of the RTE along a sufficiently large number of
optical paths, using the instantaneous scalar data from DNS, were compared with the solution of the
time averaged form of the RTE. If the optically thin flame approximation (Coelho, 2007) and the
CK model are used, the integration along a line of sight yields (Coelho, 2004)

[ ( )
I = ## % i I i ,"$ (s ) = ## % i I i ,"$ (0 ) exp(! k i s )+ k i I b,"$ k k i (1 ! exp(! k i s ))
k k
] (2)
k i k i

where ωi is the quadrature weight and I i ,"! k is the time-averaged spectral radiation intensity
integrated over the kth band for the ith quadrature point. The mean values of the absorption
coefficient and emission term are evaluated from integration of the instantaneous values of these
quantities weighted by a pdf. The blackbody radiation intensity is only a function of temperature,
while the absorption coefficient depends on the temperature and on the molar fraction of CO2. Since
the mean value and the variance of these scalars is prescribed and they are fully correlated, the
probability density function (pdf) of ki and the joint pdf of ki and I b,"! k may be easily determined.

3 Results and Discussion

The standard radiative transfer calculations were carried out assuming that the mean temperature of
the medium is 1500 K, and that the medium is a mixture of CO2 and N2, the mean molar fraction of
CO2 being 0.10. The rms of temperature and of CO2 molar fraction were taken as 150 K and 0.01,
respectively. The optical thickness of the medium is equal to one. In all other calculations where the
influence of one variable was studied, only the variable under consideration was changed, while the
remaining ones take the standard values mentioned above.

Figure 2 shows the normalized values of the mean, root mean square (rms), skewness and flatness
of the radiation intensity leaving the computational domain as a function of the number of
instantaneous fields. The normalized values are defined as follows:

12 32 2
I I b (T ) &$ I ' 2 #! I b (T ) I ' 3 &$ I ' 2 #! I ' 4 &$ I ' 2 #! (3)
% " % " % "

The results plotted in figure 2 show that Nt = 1 is sufficient to accurately determine the normalized
mean and rms of the radiation intensity, but not enough to obtain statistically independent results for
the skewness and flatness. However, when Nt exceeds 20 to 25 the influence of Nt on the results
becomes marginal. The pdfs of the radiation intensity and of the Planck mean absorption
coefficient, κP, which are given in figure 3 for Nt = 10, 20, 30 and 40, confirm good convergence of
the results and show that the shape of the two pdfs is similar.

The discussion below may be more clearly understood if equation (2) is rewritten for the time
averaged RTE using total properties and total radiation intensity rather than the CK model. In that
case, equation (2) reads as

( )
I = I (0 ) exp(! k G s )+ kI b k G (1 ! exp(! k G s )) (4)
 Figure 1: Equivalence between temporal and Figure 2: Radiation intensity as a function of the
spatial statistics under homogeneous and number of instantaneous fields.
isotropic turbulence.

Figure 3: Probability density functions of the radiation intensity and Planck mean
absorption coefficient as a function of the number of instantaneous fields, Nt.

The Planck mean and the incident mean absorption coefficients, κP and κG, respectively, are defined
as follows:

" "
$P = ! $# I b# d# !0 I b# d# (5a)
0

" " " "

$G = ! $# I# d# !0 I# d# = ! $# G# d# !0 G# d# (5b)
0 0

The second equality in equation (5b) is only valid if the radiation intensity does not change with the
direction, which is valid on a statistical basis in the case of a homogeneous and isotropic medium,
i.e., different directions are statistically indistinguishable. The following definitions have also been
used
 "
$ Ib = $ P Ib = ! $# I b# d# (6a)
0

"
$ I = $G I = ! $# I# d# (6b)
0

Applying the periodic boundary condition, I = I (0 ) , to equation (4) yields I = kI b k G . This

solution may be rewritten as

( )
I I b (T )= " I b " P I b ! (" P " G )! (I b I b (T )) (7)

Figure 4 shows statistical data for the radiation intensity, normalized according to equation 3, as a
function of the optical thickness of the medium, mean temperature, rms of temperature, and rms of
CO2 molar fraction. Figure 4 also shows the normalized mean radiation intensity predicted from the
solution of the time averaged RTE using the CK model.

The change of the optical thickness of the medium was accomplished by modifying the size of the
domain, while keeping the Planck mean absorption coefficient of the medium, evaluated at the
mean temperature and mean CO2 molar fraction, unchanged. This is the reason why the mean
radiation intensity does not change with the optical thickness, as shown in figure 4(a). The
statistical data exhibit a small change of I , which may be attributed to statistical errors, but the
predictions based on the solution of equation 4 do not. The rms of the radiation intensity increases
marginally with the increase of the optical thickness of the medium. The skewness is different from
zero, i.e., the pdf is slightly asymmetric, and the flatness is just a little above 3, which means that
the pdf is not exactly a Gaussian, but it is close to a Gaussian. These moments of the radiation
intensity remain approximately constant when the optical thickness changes.

Figure 5 shows the influence of the same variables on several correlations relevant in the emission
and absorption terms of the RTE. Figure 5(a) shows that ! I b ! P I b is constant and equal to
approximately 0.94 when the optical thickness of the medium varies keeping ! P (T ) constant.
Predictions based on the assumption of Gaussian scalars distribution, which are also shown in
figure 5(a), closely match this value. The invariance with the optical thickness is expected, since
neither the absorption coefficient nor the temperature have been changed. If the absorption
coefficient and the blackbody radiation intensity are expanded as a sum of a mean value and a
fluctuation, the following relation holds

! I b ! P I b = 1 + ! " I b" ! P I b (8a)

and, similarly,

! I !G I = 1+ ! " I " !G I (8b)

The absorption coefficient of the medium and the blackbody radiation intensity are local quantities,
i.e., they depend only on the local temperature and absorption coefficient of the medium. Moreover,
the absorption coefficient generally increases with the decrease of the blackbody radiation intensity,
i.e., they are anti-correlated (Coelho, 2007). Therefore, " ! I b! <0, yielding ! I b ! P I b < 1 , as found
above. Moreover, ! I b! is also independent of the optical thickness of the medium, as expected.
 3.2 3.4
a) 3.3 b)
3.1 3.2
3.0 3.1
3.0
1.04
1.03 1.10
1.02
1.01 1.05
1.00

0.34 1.00

0.32
0.5
0.30
0.28 1/2 0.4
I I b (T ) I2 Ib T
0.26
Skewness Flatness 0.3
0.24
0.22 I I b (T ) Predictions
0.2
0.20
0.18 0.1
0.1 1 10 100 600 800 1000 1200 1400 1600 1800
Optical thickness Temperature [K]
3.5 3.120
c) 3.110 d)
3.100
3.0
3.090
3.080
2.5 3.070

1.05 1.04

1.00 1.03

0.95 1.02
0.310
0.3 0.308
0.306
0.2
0.304
0.1
0.196
0.0 0.194
0.192
-0.1 0.190
0 50 100 150 200 0.000 0.005 0.010 0.015 0.020 0.025
rms ( T ) [K] rms ( xCO )
2

Figure 4: Statistics of radiation intensity.

In contrast to the blackbody radiation intensity, the radiation intensity is not a local quantity, but
depends on the temperature and species concentrations along the optical path. As a consequence of
this, the correlation between the absorption coefficient and the radiation intensity is expected to be
small, which justifies the so-called optically thin flame approximation (Coelho, 2007). Therefore
! I ! G I is close to unity, i.e., " ! I ! is indeed small.

The influence of the temperature on the radiation statistics, correlations and absorption coefficients
is illustrated in figures 4(b), 5(b) and 6(a), respectively. The rms of temperature was also changed to
keep T !2 T 2 constant. Since the optical thickness of the medium is now maintained constant, and
the absorption coefficient depends on the temperature, the absorption coefficient changes, as well as
the size of the domain. Figure 6(a) shows that the mean absorption coefficients, ! P and ! G ,
decrease with the increase of the temperature, as pointed out above. Still, the rms of the Planck
mean absorption coefficient does not change significantly. Therefore, the second term on the right
 1.00
1.05
0.98
Ib / I
P b I
1.00
0.96 Ib / I
P b
(Predictions)

0.94 0.95

0.92 2 0
_I _
b/
_
P I b_2 _I _/ _ I _2
G
2

0.0
-0.20

-0.2
-0.40

-0.4
-0.60

-0.6 -0.80

-0.8 a) -1.00 b)

-1.0 -1.20
0 1 10 100 600 800 1000 1200 1400 1600 1800
Optical thickness Temperature [K]

1.00
1.00

0.95
0.95

0.90 0.90

0.8 0.20

0.4 0.00
-0.20
0.0
-0.40

-0.4 -0.60
-0.80
-0.8
c) -1.00 d)
-1.2 -1.20
0 50 100 150 200 0 0.005 0.01 0.015 0.02 0.025
rms ( T ) [K] rms ( x CO )
2

Figure 5: Correlations between the absorption coefficient and the radiation intensity.

3.0 1.00 1.00

P 0.99
0.98
G 0.98
2.5 rms ( P) 0.96 0.97
rms ( G) 0.96
2.0 0.94 0.95
0.20 0.94
1.5
0.16 0.30
0.12 0.25
1.0
0.20
0.08 0.15
0.5 0.10
0.04
a) b) 0.05 c)
0.0 0.00 0.00
600 800 1000 1200 1400 1600 1800 0 50 100 150 200 0 0.005 0.01 0.015 0.02 0.025
Temperature [K] rms ( T ) [K] rms ( xC O 2)

Figure 6: Statistics of the Planck mean and incident mean absorption coefficients.
of equation (8a) increases in modulus with the increase of the temperature, but it is negative, and
therefore ! I b ! P I b decreases, as shown in figure 5(b). A similar behaviour is found for
! I ! G I , which departs from 1 if the temperature increases, which implies an increase of the
absolute value of " ! I ! , although its normalized value is rather small.

Now, ! I b may be approximated by the following expression (see, e.g., Coelho, 2007, for details)

& T '2 (' T'#

( I b = ( P I b (T )$1 + 6 2 + 4 P ! (9)
$ T (P T !
% "
in which correlations of order higher than 2 were neglected. Since T ! 2 T 2 does not change, only
the last term on the right of equation (9) varies in the present case. From the previous analysis, we
conclude that ! ' T ' ! T increases in modulus when the temperature increases, but it is negative.
Therefore, ! I b ! I b (T ) decreases with the increase of the mean temperature. Consequently, it can
be concluded from equation (7) that I I b (T ) also decreases with the increase of the mean
temperature, because ! G " ! P and I b I b (T ) does not change. This is in agreement with the
findings from both the statistical analysis of the solution of the RTE based on DNS data, and from
the predictions based on the solution of the time averaged RTE along with the assumption of
Gaussian scalar distributions (see figure 4b). In addition, figure 4(b) shows that the skewness and
the flatness are close to those found for standard conditions, except at T = 1650 K, where both the
skewness and the flatness indicate that the pdf of the radiation intensity is not as close to a Gaussian
as in the other cases.

The influence of the rms of temperature is shown in figures 4(c), 5(c) and 6(b). In this case, the
mean temperature is kept equal to 1500 K, and so the intensity of temperature fluctuations is
changing. Figure 6(a) shows that the mean values of both the Planck mean and the incident mean
absorption coefficients are approximately constant, while the rms increases with the increase of the
rms of temperature. Hence, as in the previous case, the second term on the right of equation (8a)
increases in modulus with the increase of the rms of temperature, but it is negative, and therefore
! I b ! P I b decreases, as shown in figure 5(c). This means that the increase of the second term on
the right of equation (9) is overshadowed by the decrease of the third term. The absorption term,
! I ! G I , exhibits the same behaviour, ! " I " ! G I being again very small.

The increase of the temperature fluctuations yields an increase of both the mean and rms of
radiation intensity, as expected (Coelho, 2007). The increase is not large, because the turbulence
intensity is relatively low. It is about 3% if the turbulence intensity is 10%. In this case, the second
term on the right of equation (9) is equal to 0.06. However, the third term is negative, as explained
above, and thus the term into parenthesis is only marginally greater than unity. The predictions
based on the time averaged RTE reproduce the observed trend. The asymmetry of the radiation
intensity increases with the increase of the rms of temperature, and the flatness does not exhibit a
monotonous behaviour, but it is close to 3, as in the case of a Gaussian pdf.

Finally, figures 4(d), 5(d) and 6(c) show the influence of the rms of the CO2 molar fraction. Its
mean value was set to 0.1. Figure 6(c) shows that ! P decreases with the increase of the rms of CO2
concentration. This is due to the anti-correlation between the absorption coefficient and the
blackbody radiation intensity, on the one hand, and to the nonlinearity between the blackbody
radiation intensity and the temperature, on the other hand. The calculations based on assumed
Gaussian scalars distribution correctly predicted that evolution, but slightly overpredict the
statistical data. The rms of the Planck mean absorption coefficient also decreases with the increase
of the rms of CO2 concentration, as revealed by figure 6(c). Moreover, the ratio of the rms to the
mean value of the Planck mean absorption coefficient also decreases with the increase of the rms of
CO2 concentration. Accordingly, the analysis of equation (9) shows that ! I b ! P I b increases, as
confirmed by the results plotted in figure 5(d). Moreover, I b I b (T ) remains constant while
! P ! G decreases. Therefore, the first and the second terms on the right of equation (7) have
opposite contributions to I I b (T ). Figure 4(c) shows that I I b (T ) increases slightly with the
increase of the rms of the CO2 molar fraction, which means that the contribution of the first term in
equation (7) dominates. The normalized correlation ! I ! G I increases also slightly. However, in
all studied cases, the correlation between fluctuations of the absorption coefficient and fluctuations
of the radiation intensity is small, which supports the optically thin flame approximation, and
justifies the good predictions achieved using the time averaged radiative transfer equation.

4 Conclusions
The interaction between turbulence and radiation in homogeneous and isotropic turbulence was
studied. The influence of the optical thickness of the medium, mean and variance of the
temperature, and variance of the mean molar fraction of the absorbing species were investigated. It
was found that about 2×105 samples, corresponding to a single instantaneous field, provide accurate
results for the mean and variance of the radiation intensity leaving the domain, but more than 20
instantaneous fields are needed to obtain accurate values for the higher moments of the radiation
intensity, as well as a converged probability density function of the radiation intensity. If the optical
thickness of the medium is changed while the Planck mean absorption coefficient for the mean
properties is fixed, the moments of the radiation intensity and the emission and absorption
correlations remain approximately constant. The increase of the temperature of the medium,
keeping the same intensity of temperature fluctuations, causes a decrease of the mean values of the
normalized radiation intensity and Planck mean absorption coefficient, as well as the emission and
absorption normalized correlations. If the temperature if left constant while its rms increases, then
the mean normalized radiation intensity increases, the mean value of the Planck mean absorption
coefficient remains approximately constant, and the normalized emission and absorption
correlations decrease. The increase of the rms of the molar fraction of the absorbing species, while
keeping its mean value constant, yields an increase of the normalized mean radiation intensity, a
decrease of the mean value of the Planck mean absorption coefficient and an increase of the
normalized emission and absorption correlations. In all cases, the correlation between fluctuations
of the absorption coefficient and fluctuations of the radiation intensity is small, which supports the
optically thin flame approximation, and justifies the good predictions achieved using the time
averaged radiative transfer equation.

Acknowledgements
This work was developed within the framework of project POCI/EME/59879/2004, which is
financially supported by FCT-Fundação para a Ciência e a Tecnologia, programme POCI 2010
(29.82% of the funds from FEDER and 70.18% from OE).
References

Canuto, C., Hussaini, M.Y., Quarteroni, A., Zang, T.A., 1988, Spectral Methods in Fluid
Mechanics, Springer-Verlag, New York, pp. 201-212.

Coelho, P.J., 2004, Detailed Numerical Simulation of Radiative Transfer in a Non-Luminous

Turbulent Jet Diffusion Flame, Combustion and Flame, 136, 481-492.

Coelho, P.J., 2007, Numerical Simulation of the Interaction between Turbulence and Radiation in
Reactive Flows, Progress in Energy and Combustion Science, 33, 311-383.

Deshmukh, K.V., Haworth, D.C. and Modest, M.F., 2007, Direct Numerical Simulation of
Turbulence–Radiation Interactions in Homogeneous Nonpremixed Combustion Systems, Proc.
Combustion Institute, 31, 1641-1648.

Goody, R.M., West, R., Chen., L and Chrisp, D, 1989, The Correlated-k Method for Radiation
Calculations in Nonhomogeneous Atmospheres, J. Quant. Spectrosc. Radiat. Transfer, 42, 539-550.

Kuo, K.K., 1986, Principles of Combustion, John Wiley & Sons, New York.

Li, G. and Modest, M.F., 2005, Numerical Simulation of Turbulence–Radiation Interactions in

Turbulent Reacting Flows, in Modelling and Simulation of Turbulent Hest Transfer, Sundén B. and
Faghri, M. (eds.), Southampton, WIT Press, pp. 77-109.

Silva, C.B., Malico, I., Coelho, P.J. and Pereira, J.C.F., 2006, An Exploratory Investigation of
Radiation Statistics in Homogeneous Isotropic Turbulence, in Computational Thermal Radiation in
Participating Media II, Lemonnier, D., Selçuk, N. and Lybaert, P. (eds.), Lavoisier, pp. 215-224.

Silva, C.B. and Pereira, J.C.F., 2007, Analysis of the Gradient Diffusion Hypothesis in Large Eddy
Simulations Based on Transport Equations, Physics of Fluids, 19, 035106.

Soufiani, A, Taine, J., 1997, High Temperature Gas Radiative Property Parameters of Statistical
Narrow-Band Model for H2O, CO2 and CO, and Correlated-k Model for H2O and CO2, Int. J. Heat
and Mass Transfer, 40, 987-991.

Wu, Y., Haworth, D.C., Modest, M.F. and Cuenot, B., 2005, Direct Numerical Simulation of
Turbulence/Radiation Interaction in Premixed Combustion Systems, Proc. Combustion Institute, 30,
639-646.