Professional Documents
Culture Documents
AN EXPLORATORY INVESTIGATION OF
RADIATION STATISTICS IN HOMOGONEOUS
ISOTROPIC TURBULENCE
Abstract
A fundamental study of radiation statistics in homogeneous isotropic turbulence is presented. A
pseudo-spectral code is used to simulate isotropic turbulence by means of DNS of the full Navier-
Stokes equations. The instantaneous scalar data is used to calculate the radiation intensity along a line
of sight using the statistical narrow band model. The mean, variance, skewness and flatness of
radiation intensity were obtained for conditions observed downstream of the flame tip of a piloted
turbulent jet flame, where the statistics of the flow field are close to the ones found in isotropic
turbulence. The joint probability density function between the temperature and the radiation intensity
is presented, as well as the spectra for the radiation intensity. The present one way coupling
philosophy used to connect isotropic turbulence data with radiation computations shows the correct
trends and allows one to study the detailed effects of the turbulent characteristics upon the structure of
the radiation intensity field.
106 Reλ=95.6
(-5/3)
4
10
E(k)/(εν5)1/4
102
0
10
10-2
10-4
10-6
0.5 1 1.5 2
kη
Fig. 1 - Energy spectrum for the present simulation.
In the present work, the radiative properties of the medium are evaluated using the SNB
model [14]. Moreover, a cold boundary is assumed, i.e. Iν(0) = 0. Therefore, Eq. (2) is
integrated over a narrow band, so that the mean radiation intensity over a narrow band of
width ∆ν is given by
1 s ∂ τ ∆ν (s ′ → s )
I ∆ν (s ) =
∆ν ∫∆ν Iν (s ) dν = ∫0 I b,∆ν (s ′)
∂ s′
ds ′ (3)
In the case of a homogeneous gas layer at total pressure p, the mean transmissivity is given
by [14]
γ δ
τ ∆ν (0 → s ) = exp − 2 1 + x s p s k − 1
(4)
δ γ
where k and 1/δ are parameters of the model, which were taken from the data of Soufiani and
Taine [15]. For a given absorbing species, these parameters are tabulated as a function of
temperature and spectral location. The parameter γ is a function of temperature, pressure and
species concentration, also given in [15]. In the case of a non-homogeneous medium, the
Curtis-Godson approximation [16] is employed.
The integrals in Eqs. (3) and (4) are numerically evaluated using Simpson’s rule, and the
parameters k and 1/δ are interpolated from the tabulated data using cubic splines. This keeps
the order of accuracy of the numerically evaluated radiation intensity consistent with the
order of accuracy of the DNS solver. Therefore, Eq. (3) is discretized as
N ( )
I b,∆ν (s n ) + 4 I b,∆ν s n+1 2 + I b,∆ν (s n +1 )
I ∆ν (s N +1 ) = ∑ 6
×
(5)
n =1
[τ ∆ν (s n+1 → s N +1 ) − τ ∆ν (sn → s N +1 )]
where N is the total number of grid points along the optical path in the DNS calculations. In
the DNS calculations, the boundaries of the computational domain are periodic. Therefore,
the instantaneous scalar data at the first point n =1, where the radiation intensity is
determined, are identical to those at n = N+1. The temperature and the species concentration
at points sn+1/2 are not available, and so they are interpolated from the DNS data using again
cubic splines.
4. Computational details
The computational domain in the DNS calculations is a cubic box of side 2π. The selection
of the size of the radiation domain, also taken as a cubic box, is not so straightforward. Here,
the size of the radiation domain was chosen in such a way that the ratio of the size of the box
in the DNS calculations to the integral length scale, L11, is equal to the ratio of the size of the
radiation domain to the integral length scale in the radiation calculations The implications of
this choice deserve further examination, which will be investigated in the future.
In order to rescale the data from isotropic turbulence simulations obtained in a cubic box
of side 2π into a radiation domain with a different size, kinematic similarity between the two
flows was assumed. Therefore, the instantaneous temperature field used in the radiation
r
computations, Trad( x ), was determined using the instantaneous temperature field from the
r
DNS data, TDNS( x ), through
r r r ′2 >
< Trad
Trad ( x ) =< Trad ( x ) > +TDNS ( x ) (6)
′2 >
< TDNS
r r
where <Trad( x )> is the mean temperature at point x , and < Trad ′ 2 > and < TDNS
′ 2 > are the
variance of the temperature fields from the radiation and from the isotropic turbulence
simulation, respectively. Similarly, for the field of the molar fraction of the absorbing species,
we use
It was assumed that the temperature and the absorbing species fields are fully correlated.
This is consistent with combustion models that relate the instantaneous thermochemical state
of the gaseous mixture to a single scalar, typically mixture fraction, e.g., the laminar flamelet
model. The experimental data reveals a very strong correlation between the temperature and
the molar fractions of H2O and CO2, which supports this assumption. Although radiation from
H2O could easily be included in the calculations, only radiation from CO2 is considered here.
This is mainly to reduce the computational requirements. Accordingly, the same scalar field
from the DNS calculations was used to prescribe both temperature and molar fraction of CO2,
r r
by setting TDNS( x ) = xs,DNS( x ).
The present radiative transfer calculations consider a purely idealized scenario, and no
extrapolation to practical applications is envisaged. Nevertheless, to ensure that the input data
considered in the radiative calculations are typical of those found in real flames, the mean and
rms of temperature and CO2 molar fractions were prescribed from the experimental data
available for a piloted CH4/air turbulent jet diffusion flame, the so-called flame D [17-19].
The mean and variance of these scalars, which are taken from a point in the flame, are
prescribed over the entire radiation domain. The integral length scale in the radiation
calculations is also taken from the experimental data [19].
The experimental data at x/D = 75, and for several radial locations, was used. This station
is located downstream of the flame tip, and it is the station at a largest distance from the
burner exit where experimental data is available. Hence, this is where the topology and
statistics of the flow are closer to the ones found in isotropic turbulence [20]. Radiative
calculations are performed for every radial location using the instantaneous scalar data
obtained from a unique DNS simulation by means of Eqs. (6) and (7). The mean and variance
of temperature and CO2 molar fraction used in these equations, which correspond to a single
point in flame D, are the same over all the radiation domain.
It is important to stress that the values predicted here cannot be compared with the
measurements and predictions reported in [19], not only because the radiation from H2O was
not taken into account in the present work, but also because the values in that reference were
obtained for the actual flame, in which the mean temperature and the species concentration
change along the optical path in the radial direction. On the contrary, the present values are
obtained assuming that the mean temperature and CO2 concentration taken from the
measurements at a fixed r/D remain unchanged in the radiation domain, and only the
instantaneous values change. This is consistent with the direct numerical simulation of
homogeneous isotropic turbulence.
Under the conditions of homogeneous and isotropic turbulence, the statistical data
computed from a temporal 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. The statistical data reported below was obtained from the
DNS data at a given time, using all the available optical paths parallel to the coordinate axes,
which are statistically indistinguishable. This means that 3×N2 ≈ 1.1×104 samples are used to
obtain the results described below. It is assumed that there is no radiation entering the
radiation domain, as stated above. The influence of this assumption on the computed
statistical data will be the subject of further research.
100
10-1
-2
10
10-3
10-4
10-5
-6
10
0 2 4 6 8 10
r/D
Fig. 2 - Mean, variance, skewness and flatness of radiation intensity as predicted by the
present model using the jet flame D data at x/D=75.
Figure 2 shows the mean, variance, skewness and flatness of the mean spectral radiation
intensity over the narrowband centred at 2.7 µm. The variance of the radiation intensity is
defined by <I’2>, where the brackets represent an average over all the samples from the
radiation domain computations. The skewness, S(I) = <I´3>/<I´2>3/2, represents the degree of
symmetry of the probability density function (PDF) of the radiation intensity (S = 0 for a
symmetric PDF), and the flatness factor, F(I) = <I´4>/<I´2>2, is a measure of the degree of
intermittence of a given variable (F = 3 for a Gaussian function).
Both the mean and variance of the mean spectral radiation intensity over a narrowband
attain their maxima at the centre of the jet, where the temperature and CO2 concentration are
highest, as shown in Fig. 2. The temperature and the CO2 molar fraction decrease along the
radial direction in the flame, and so do <I> and <I’2>. However, the turbulence intensity,
taken as ′ 2 > < Trad > , increases along the radial direction. This implies an increase of
< Trad
the influence of turbulence on radiation as r/D increases, as shown in Table 2. This trend is
also observed in [19], although the values cannot be compared, as explained above.
If there were no fluctuations of the absorption coefficient of the medium, then the
temperature self-correlation would be fully responsible for the influence of turbulence on
radiation. The temperature self-correlation is given by [22]
<T4 > < T ′2 > < T ′3 > < T ′4 >
= 1+ 6 +4 + (8)
< T >4 < T >2 < T >3 < T >4
The two first terms on the right side are generally dominant. If the PDF of temperature is
Gaussian, then the third term on the right side is zero and the last one is equal to 3 ×
(< T ′ 2 >/<T>2)2. Table 2 confirms the importance of the temperature self-correlation, and
demonstrates that the fluctuations of the absorption coefficient of the medium contribute to
enhance the influence of turbulence on radiation. The mean spectral radiation intensity over a
narrowband increases by 34% due to turbulence at r/D = 0, while it increases by 92% at r/D =
8.33. This supports the results from [19, 21], who concluded that the investigation of flame
radiation along chord-like paths is important to understand TRI, providing a more challenging
test to predictive models than diametric paths.
r/ D ′ 2 > < Trad >
< Trad < T 4 > < T >4 (a)
< T 4 > < T >4 (b)
< I ′2 > < I >
The skewness and flatness, on the other hand, seem to increase slightly with r/D. At the
centreline, S(I)=0.44 and F(I)=2.8, whereas at r/D=8.33 we have S(I)=1.50 and F(I)=5.2. It is
not clear why the asymmetry and intermittency of the radiation intensity increase with the
distance from the flame axis.
Figure 3 shows the contour plots of temperature and blackbody radiation intensity obtained
from the DNS computations in the first (y,z) plane of the domain. The contours of the mean
spectral radiation intensity over the narrowband in the same plane are also shown using the
flame data at x/D = 75 and r = 0. The correlation coefficient between T and I is equal to about
50%. The correlation coefficients at other r/D locations are similar. It should be noticed that
the radiation intensity at a grid node in the first (y,z) plane depends on the temperature and
CO2 molar fraction along the optical path in x direction, and not only on the local properties.
Therefore, it is not surprising that this correlation is not very high.
To have a global picture of this result, Figure 4 shows the joint PDF of T and I from all the
available samples. In agreement with the previous results, the joint PDF of these quantities
shows that the correlation between them is associated with extreme events of both variables.
The spectrum for the mean spectral radiation intensity over a narrowband at x/D=75 and
for several radial locations of flame D is shown in figure 5. The spectrum E(K) is defined by
r r E(K2D ) r ' r
< Iˆ( K ' ) Iˆ( K ) >= δ (K + K ) (9)
π K2D
r r
where Iˆ( K ) is the two dimensional Fourier transform of I ( x ) , given by
+∞ +∞ rr r r
r 1 r
Iˆ( K ) = ∫∫ I ( x ) exp(−ik .x )dKdx (10)
(2π ) 2 −∞ −∞
r r
and K = ( K1 , K 2 ) is the two-dimensional wave number vector of norm K = | K | . Thus, the
energy spectrum represents the spectral distribution of the energy associated to each wave
number. Figure 5 shows that the overall amount of radiative energy decreases with the radial
position, as expected, and in agreement with Fig. 2. It seems that the slope of the radiation
intensity spectrum, at the inertial range region, changes slightly with the distance from the jet
centreline. Moreover, the characteristic bump caused by the forcing at low wave numbers
seems to be attenuated near the jet centreline. These are only some of the issues we intend to
explore in future works.
2 -1
T (K) Ib (W/m sr cm ) I (W/m sr cm )
2 -1
Fig. 3 - Contours of temperature (left), blackbody radiation intensity (centre) and mean
spectral radiation intensity over a narrowband (right) in the first (y,z) plane of the domain.
10
0 r/D=0.0
r/D=2.7
0.2 10
-2
r/D=5.6
r/D=8.3
-4
10
I(W/m sr cm )
-1
-6
0.15 10
-8
E(K) 10
2
-10
10
0.1
-12
10
-14
10
0.05
-16
10
K
Fig. 4 - Joint PDF between the temperature Fig. 5 - Spectra for the mean spectral radiation
and the mean spectral radiation intensity over intensity over a narrowband for several radial
a narrowband obtained with data taken from locations obtained with the present model
three boundary planes of the turbulent box. using the flame D data at x/D=75.
6. Conclusion
Classical one point statistics, correlations and power spectra of radiation intensity were
computed from an idealized model combining direct numerical simulations of isotropic
turbulence, comprising temperature and concentration fields, coupled with radiative transfer
calculations. The statistics demonstrate that the turbulence has a strong influence on the
radiation for the studied cases, yielding an increase of radiation intensity due to turbulence
that ranges from 34% to 92%. The increase is largest away from the flame axis, where the
turbulence intensity is also highest. It is hoped that the numerical tools employed here, when
further exploited, will shed more light into the complex nature of the turbulence/radiation
interactions. In particular, the influence of the chemical composition of the medium, mean gas
temperature, Reynolds number, shape of the kinetic energy and temperature spectra, and
turbulence intensity will be investigated.
7. Acknowledgement
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
[1] KOUNALAKIS, M.E., GORE, J.P., FAETH, G.M. “Turbulence/radiation interactions in nonpremixed
hydrogen/air flames”, 22nd Symposium (Int.) on Combustion, pp. 1281-1290, 1988.
[2] KOUNALAKIS, M.E., SIVATHANU Y.R., FAETH, G.M. “Infrared radiation statistics of nonluminous
turbulent diffusion flames”, J. Heat Transfer, vol. 113, pp. 437-445, 1991.
[3] KRITZSTEIN, F., SOUFIANI, A. “Infrared gas radiation from a homogeneously turbulent medium”, Int.
J. Heat and Mass Transfer, Vol. 36, pp. 1749-1762, 1993.
[4] CHAN, S.H., PAN, X.C. “A general semicausal stochastic model for turbulence/ radiation interactions in
flames”, J. Heat Transfer, Vol. 119, pp. 509-516, 1997.
[5] VINCENT, A., MENEGUZZI, M. “The spatial structure and statistical properties of homogeneous
turbulence”, J. Fluid Mech., Vol. 225, pp. 1-20, 1991.
[6] JIMENEZ, J., WRAY, A.A., SAFFMAN, P.G., ROGALLO, R.S. “The structure of intense vorticity in
isotropic turbulence”, J. Fluid Mech., Vol. 255, pp. 65-90, 1993.
[7] GIRIMAJI, S.S., POPE, S.B., “Propagating surfaces in isotropic turbulence”, J. Fluid Mech., Vol. 234,
pp. 247-277, 1992.
[8] WU, Y., HAWORTH, D.C., MODEST, M.F., CUENOT, B. “Direct numerical simulation of turbulence/
radiation interaction in premixed combustion systems”, Proc. Combustion Institute, Vol. 30, pp. 639-646, 2005.
[9] DA SILVA, C.B., PEREIRA, J.C.F. “On the local equilibrium of the subgrid-scales: the velocity and
scalar fields”, Phys. Fluids, Vol. 17, 108103, 2005.
[10] CANUTO, C., HUSSAINI, M.Y., QUARTERONI, A., ZANG, T.A., “Spectral Methods in Fluid
Mechanics”, Springer-Verlag, New York, pp. 201-212, 1988.
[11] ALVELIUS, K. “Random forcing of three-dimensional homogeneous turbulence”, Phys. Fluids, Vol.11,
pp. 1880-1889, 1999.
[12] POPE, S. “Turbulent flows”, Cambridge University Press, 2000.
[13] MODEST, M.F. “Radiative Heat Transfer”, McGraw-Hill, New York, 2nd Edition, 2003.
[14] MALKMUS, W. “Random Lorentz band model with exponential tailed S-1 line-intensity distribution
function”, J. Optical Society of America, Vol. 57, pp. 323-329, 1967.
[15] SOUFIANI, A., TAINE, J. “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,
Vol. 40, pp. 987-991, 1997.
[16] YOUNG,S.J. “Nonisothermal band model theory”, J. Quant. Spectroscopy and Radiative Transfer, Vol.
18, pp. 1-28, 1977.
[17] BARLOW, R.S. FRANK, J.H., “Effects of turbulence on species mass fractions in methane/air jet
flames”, Proc. Combustion Institute, Vol. 27, pp. 1087-1095, 1998.
[18] International Workshop on Measurement and Computation of Turbulent Nonpremixed Flames,
http://www.ca.sandia.gov/TNF/DataArch/FlameD.html.
[19] ZHENG, Y., BARLOW, R.S., GORE, J.P., “Spectral radiation properties of partially premixed
turbulent flames”, J. Heat Transfer, Vol. 125, pp. 1065-1073, 2003.
[20] DA SILVA, C.B., MÉTAIS, O. “On the influence of the coherent structures upon interscale interactions
in turbulent plane jets”, J. Fluid Mechanics, Vol. 473, pp. 103-145, 2002.
[21] ZHENG, Y., SIVATHANU, Y.R., GORE, J.P., “Measurements and stochastic time and space series
simulations of spectral radiation in a turbulent non-premixed flame”, Proc. Combustion Institute, Vol. 29, pp.
1957-1963, 2002.
[22] COX, G., “On radiant heat transfer from turbulent flames”, Combustion Science and Technology, Vol.
17, pp. 75-78, 1977.