International Communications in Heat and Mass Transfer 110 (2020) 104400

Contents lists available at ScienceDirect

International Communications in Heat and Mass Transfer

journal homepage: www.elsevier.com/locate/ichmt

An extended weighted-sum-of-gray-gases model to account for all CO2 ‐ H2O T

molar fraction ratios in thermal radiation
Hadi Bordbara, , Guilherme C. Fragab, Simo Hostikkaa
⁎

a
Department of Civil Engineering, Aalto University, Finland
b
Mechanical Engineering Department, Universidade Federal do Rio Grande do Sul, Brazil

ABSTRACT

All published weighted sum of gray gases models (WSGGM) were either developed for a limited number of molar fraction ratios, MR, or include MR as a variable in
their formulations. Either way, they are not able to adequately support moisture-rich regions of combustion environments, such as the outer regions of unwanted
fires, fires during water-based suppression, and some air-injection regions of furnaces. In this article, we provide an extension to a previously published WSGGM by
coupling it to a new accurate set of WSGGM coefficients for pure carbon dioxide and water vapor. The coupling follows a linear interpolation methodology, which is
justified by a detailed analysis of the line-by-line (LBL) absorption spectra and by plotting the total emissivity of CO2 ‐ H2O mixtures with large and small values of
MR for various temperatures and path lengths. The proposed model is discussed and validated in four benchmarks, using a solution obtained by LBL integration as
reference. The results indicate that, while the previously available WSGGMs either have excessive computational costs or yield inaccurate results in the regions of
large MR, the new model can be safely used for all gas compositions in a computationally efficient manner.

1. Introduction these cases, these WSGGMs can not be safely applied. Another approach
for dealing with this limitation is to adopt the double integration
As a member of a more generic category of global spectral models, methodology, in which WSGGM coefficients for the individual species
the weighted-sum-of-gray-gases model (WSGGM) aims at obtaining are combined using probabilistic arguments (see, for example, [11,12]).
total radiative quantities such as the radiative heat source and heat flux However, while this in principle allows for considering the full range of
by solving the spectrally-integrated radiative transfer equation (RTE) MRs, it entails a significant increment in the computational cost as the
for a small number of gray gases [1]. Due to its simplicity and lower radiative transfer equation must be solved for each gray gas of H2O and
computational cost compared to other global models (such as the CO2. Andre et al. [13] and Consalvi et al. [14] presented a detailed
spectral line-based WSGGM [2] and the full-spectrum k-distribution assessment of computational performance of the double integration
method [3]), the WSGGM has been more widely used for radiative WSGGM [12] for high temperature [13] and high pressure [14] com-
transfer calculations involving CO2 ‐ H2O mixtures in large-scale com- bustion systems in comparison with some other global models such as
bustion systems. Most WSGGMs (e.g., [4–6]) were obtained for a few different versions of full spectrum correlated-k method (FSCK).
specific molar fraction ratios MR = Yw/Yc (where Yw and Yc are the In this regard, the present study proposes an extension to the pre-
molar fractions of water vapor and carbon dioxide, respectively), which viously published WSGGM of Bordbar et al. [9], originally developed
restricts their range of application. More recent WSGGMs [7–10] have for 0.01 ≤ MR ≤ 4.0, to cover all values of MR. The accuracy of the
attempted to overcome this limitation by including a dependence on WSGGM of [9] has been proven by several comparative studies [15,16].
MR in their formulations, but the difficulty of fitting emissivity data To the best of the authors' knowledge, this is the first attempt to provide
over all possible molar ratio values—i.e., from MR → 0 for pure CO2 to a WSGGM that is applicable for the entire range of MR, beside the
MR → ∞ for pure H2O—makes even these models limited to a certain computationally demanding double integration models.
range of MR. A summary of the applicability range of the most widely
used varying MR WSGGMs is provided in Table 1. 2. Methodology
Although the domain-average value of MR in combustion systems
usually lies within the limits of the models in Table 1, it is possible that 2.1. WSGGM coefficients for pure species
the molar fraction ratio locally exceeds the limits, such as in moisture-
rich regions of flames and in the outer regions of unwanted fires. In Following the same methodology used in [9], two new sets of

⁎
Corresponding author.
E-mail address: [email protected] (H. Bordbar).

https://doi.org/10.1016/j.icheatmasstransfer.2019.104400

0735-1933/ © 2019 Elsevier Ltd. All rights reserved.

H. Bordbar, et al. International Communications in Heat and Mass Transfer 110 (2020) 104400

Table 1
Range of molar fraction ratio supported by recent WSGGMs that account for varying MR.
Johansson et al. [7] Kangwanpongpan et al. [8] Bordbar et al. [9] Guo et al. [10]

0.125 ≤ MR ≤ 2.0 0.125 ≤ MR ≤ 4.0 0.01 ≤ MR ≤ 4.0 0.05 ≤ MR ≤ 2.0

WSGGM coefficients are generated for pure CO2 and pure H2O to sup- 2.2. Coupling methodology
port cases where only one of these species exist. First, line-by-line (LBL)
absorption spectra at atmospheric pressure are obtained for each spe- For conditions where 0 < MR < 0.01 or MR > 4.0 (i.e., outside
cies from the HITEMP 2010 database [17] as described in [18], the applicability range of the WSGGM of Bordbar et al. [9]), the cor-
adopting a spectral resolution of 0.02 cm−1 within a wavenumber range relations in Tables 2 and 3 are coupled to those of [9] by linearly in-
150 cm−1 ≤ η ≤ 15000 cm−1. A resolution of 0.02 cm−1 is reported as terpolating the values of κp, i and ai. For 0 < MR < 0.01, the upper
the optimal spectral resolution for obtaining the LBL absorption spectra and lower bounds of this interpolation are the results of the WSGGM of
of combustion gases, and using finer spectral resolution has found to [9] for MR = 0.01 and the coefficients of Table 2 for MR = 0, re-
have a negligible influence on the results of LBL spectral data while spectively. For MR > 4.0, the lower bound is again the results of [9]
significantly increasing the computing time [19]. Considering 43 evenly for MR = 4.0, while the upper bound is taken as the coefficients of
distributed temperature points between 300 and 2400 and by using the Table 3 for a very large molar fraction ratio represent pure H2O. Our
LBL spectra, a total emissivity ε database is constructed, where ε is numerical tests showed that using a value of MR = 108 to represent this
given as [3]. upper bound (H2O) consistently provides very good accuracy in all of
our validation tests.
0
Ib [1 exp( p pt YL)] d This interpolation approach is based on the assumption that the
= ,
Ib d absorption spectra of CO2 ‐ H2O mixtures with MR < 0.01 or
0 (1)
MR > 4.0 are linearly scalable with MR; thus, for instance, for large
in which Ibη is the Planck function, κpη is the pressure-based spectral MR values (> 4.0) one could obtain the absorption spectrum of a
absorption coefficient, pt is the total pressure, Y is the molar fraction, mixture by a linear interpolation between the spectra of MR = 4.0 and
and L is the path length. The emissivity database is generated by LBL of pure H2O. This is physically justified because, in these ranges of MR,
integration for 20 path lengths between 0.01 m and 60 m, with pt = 1 the mixture mainly consists of only one species (H2O for large MR and
atm and Y = 1. Then, the following equation, which expresses the total CO2 for small MR), so the effect of mixing of the absorption spectra of
emissivity for the WSGGM, is fitted to the database to obtain the the individual gases is marginal. Nonetheless, this, like any other si-
weighting factors ai and pressure-based absorption coefficients κp, i of milar assumption used in development of global models—such as the
each gray gas i, correlation or scaling assumption used in FSCK or FSSK methods [3]—is
not completely true and its accuracy needs to be checked. For that
Ng purpose, Fig. 1 compares the real LBL absorption spectra of a CO2 ‐ H2O
= ai [1 exp( p, i pt YL) ]. mixture with MR = 10 and the spectra recreated by linearly inter-
i=1 (2) polating between MR = 4.0 and pure H2O. Two spectral regions con-
taining strong H2O or CO2 absorption lines are shown in this figure, and
Similarly to [9], the number of gray gases Ng of the model is taken as
for both of them, as well as for the three different medium temperatures
4 and the weighting factor of the gray gases is given as a polynomial
considered, the retrieved spectra of the mixture exhibit a good agree-
function of the temperature T,
ment with the LBL data.
5 j 1 The linear interpolation assumption is further assessed by ex-
T amining the total emissivity of CO2 ‐ H2O mixtures with MR > 4.0.
ai = bi, j ,
j =1
Tref (3) This is done in Fig. 2, where ε predicted by the extended WSGGM is
compared to results obtained by LBL integration. For small and inter-
with bi, j the polynomial constants of the model and Tref = 1200 K. As is mediate path lengths, the extended WSGGM gives a very accurate ap-
common in the WSGGM, the weighting coefficient for the transparent proximation of the total emissivity for the entire temperature range for
windows (i = 0) is determined from energy conservation considera- which the model was developed. Moreover, compared to the WSGGMs
tions as a0 = 1 − ∑i=1Ng ai. Tables 2 and 3 report the obtained coeffi- presented by Smith et al. [4] and Yin et al. [5], that are also applicable
cients for pure CO2 and pure H2O, respectively. to MR > 4 and are briefly described in Section 3, the present model
It should be noted that WSGGM coefficients for these individual performs significantly better for all the temperatures, path lengths and
species already exist in the literature—for example, in [4,5,12]. How- molar fraction ratios. Both the present extended WSGGM and the
ever, they have been developed with different conditions than those of double integration method perform equally good in most of the tem-
[9] (e.g., different number of gray gases or distinct ranges of tem- perature and path length range especially in small to intermediate path
perature applicability), thus making the coupling to already existing length. However, as expected the double integration WSGGM [12]
correlations not sensible. provides slightly more accurate total emissivities for large path lengths,
but at the price of higher computational time as will be explained more

Table 2
WSGGM coefficients for pure CO2 at 1 atm.
i κp, i (m−1 atm−1) bi, 1 bi, 2 bi, 3 bi, 4 bi, 5

−2 −1 −1
1 3.272772 × 10 8.495135 × 10 −1.496812 1.361406 −5.551699 × 10 8.076589 × 10−2
2 4.229655 × 10−1 −1.103102 × 10−1 9.363958 × 10−1 −1.250799 6.527827 × 10−1 −1.206959 × 10−1
3 4.905367 1.731716 × 10−1 −5.174223 × 10−1 8.256840 × 10−1 −4.998864 × 10−1 1.008743 × 10−1
4 1.085440 × 102 3.995426 × 10−2 1.423006 × 10−1 −1.649481 × 10−1 5.140768 × 10−2 −3.497246 × 10−3

2
H. Bordbar, et al. International Communications in Heat and Mass Transfer 110 (2020) 104400

Table 3
WSGGM coefficients for pure H2O at 1 atm.
i κp, i (m−1 atm−1) bi, 1 bi, 2 bi, 3 bi, 4 bi, 5

−2 −1 −1
1 8.047859 × 10 6.670204 × 10 −1.228413 1.428908 −6.267906 × 10 9.628539 × 10−2
2 9.557208 × 10−1 2.343433 × 10−1 −3.192256 × 10−1 8.867348 × 10−1 −5.927787 × 10−1 1.185824 × 10−1
3 8.005283 −1.793041 × 10−1 1.683454 −2.136989 1.020422 −1.723960 × 10−1
4 7.613186 × 101 3.455969 × 10−1 −7.510442 × 10−1 6.313180 × 10−1 −2.416500 × 10−1 3.530972 × 10−2

in the results of 1D test cases in the next section. Similar results are Yw = 10−4. Fig. 3 compares the extended WSGGM proposed in this
observed for mixtures with MR < 0.01 and are omitted here for paper to the variable-MR correlations presented in Table 1 in terms of
brevity. the predicted radiative heat source, Sr, and radiative heat flux, qr. The
figure also reports the results obtained by three additional formula-
tions: the one by Yin [5], in which fixed-MR coefficients are provided
3. Results and discussion for a number of different gas compositions alongside an algorithm to
define which of them to be used; a linear interpolation (carried out
The extended WSGGM proposed in this paper is tested for four test following a similar procedure as described in Section 2.2) between the
cases, all consisting of a one-dimensional medium slab bounded by two more outdated set of coefficients of Smith et al. [4], which were gen-
parallel black walls at 400 K, separated by a distance X = 1 m for test erated for MR = 1.0, MR = 2.0, MR → 0 and MR → ∞; and a double
cases 1–3 and X = 10 m for the test case 4. All the cases are subjected to integration methodology similar to the one outlined in [11], but em-
the same non-homogeneous, symmetrical temperature profile, given as ploying the pure CO2and H2O coefficients in Tables 2 and 3.
Tx = 400 K + 1400 Ksin2 (2 x ) , where x = x / X is the dimensionless For the first case, all WSGGMs yield physically sensible results, even
distance from the left wall. The medium composition of each case is though the medium has a molar ratio well below the recommended MR
defined as to consider the conditions MR < 0.01 (case 1), MR > 4.0 ranges of the correlations in Table 1. A normalized error of each model
(case 2 and 4), and 0 ≤ MR ≤ ∞ (case 3). can be defined as δϕ = |ϕWSGGM − ϕLBL|/ max (|ϕLBL|), where ϕWSGGM
The RTE is solved with the discrete ordinates method, following the and ϕLBL are the values of the radiative quantity ϕ (Sr or qr) obtained
methodology and discretization parameters outlined in [6]. The re- with the WSGGM and the LBL method, respectively, and max(|ϕLBL|) is
ference solution, to which all WSGGMs are compared, is provided via the maximum absolute value of ϕLBL in the domain. Table 4 reports the
the LBL integration method as described in [20]. maximum and domain-average δSr and δqr for all WSGGMs, where it
The first test case represents a scenario in which MR → 0, where the can be seen that for test case 1 the most accurate radiative transfer
medium is a homogeneous CO2 ‐ H2O ‐ N2 mixture with Yc = 0.5 and

Fig. 1. The spectral absorption of a mixture with MR = 10: comparison of the real LBL profiles with those retrieved by linear interpolation between MR = 4.0 and
pure H2O.

3
H. Bordbar, et al. International Communications in Heat and Mass Transfer 110 (2020) 104400

Fig. 2. Comparison of the total emissivity predicted by the extended WSGGM proposed in this paper and by existing models that comport MR > 4 with ε generated
by LBL integration.

solutions are achieved with the model of [7], with the extended case, all WSGGMs in Table 1 lead to nonphysical results, so they are
WSGGM developed in this paper and with the double integration omitted. Similarly, the WSGGMs of Smith et al. [4] and Yin [5], al-
model. though capable of capturing well the overall spatial variation of Sr and
Fig. 4 shows the results of the second test case, which consists of a qr, fail at particular locations, where the radiative source experiences
homogeneous CO2 ‐ H2O ‐ N2 mixture with Yc = 10−4 and Yw = 0.5, abrupt peaks or valleys and, correspondingly, the radiative heat flux
illustrating the condition of MR → ∞. Note that results of the WSGGMs profile has quick changes in its slope. Such locations correspond to
in Table 1 are not included in Fig. 4 because those models yield phy- regions where the formulations of [4,5] transition from one set of fixed-
sically unrealistic profiles of Sr and qr (i.e., the values of these quantities MR WSGGM coefficients to another, resulting in a mismatch in the
are several order of magnitudes larger than the LBL results). Con- values calculated for κp, i and ai in one grid point to the next and
versely, the models based on fixed-MR correlations [4,5] are able to yielding large local errors (especially for Sr), as depicted in Table 4. The
provide fairly accurate results; however, as indicated by the errors re- same does not happen for the double integration and the extended
ported in Table 4, they are much less accurate than the extended WSGGMs, and no spurious oscillations of Sr and qr can be seen in the
WSGGM. Finally, the results of the double integration approach are results of theses models.
almost as accurate as the current extended model but with much higher As the WSGGMs are widely used to model thermal radiation in large
computational cost, as it will be discussed later. scale combustion systems where the medium are usually optically thick,
In the third test case, the medium is a non-homogeneous CO2 ‐ H2O the performance of the present extended WSGGM is assessed in test case
mixture with Yw (x ) = 10 4 + (1 10 4)sin2 ( x ) and 4 with X = 10 m representing an optically thick condition. All the other
Yc (x ) = 1 Yw (x ) , representing a scenario where both MR → 0 and settings of the fourth test case are the same as the second test case and
MR → ∞ occur in the domain. Fig. 5 and Table 4 present the results and therefore MR > 4.0. The radiative heat flux and heat source of test
the associated errors for this case, respectively. As in the second test case 4 are shown in Fig. 6. As seen, the accuracy of the present extended

Fig. 3. Results of the first test case (MR → 0).

4
H. Bordbar, et al. International Communications in Heat and Mass Transfer 110 (2020) 104400

Table 4
Domain-average and maximum (in parenthesis) normalized error of and average computational time required for the different WSGGMs.
Case 1 Case 2 Case 3 Case 4 t a
( × 10 3s)
δSr(%) δqr(%) δSr(%) δqr(%) δSr(%) δqr(%) δSr(%) δSr(%)

Extended WSGGM 5.79 7.20 5.52 4.60 6.57 8.16 4.76 10.4 5.9
(9.14) (21.3) (7.54) (8.33) (11.1) (23.8) (6.58) (21.8)
Smith et al. [4] 8.53 16.3 8.20 6.21 12.9 25.3 7.38 25.4 3.2
(16.3) (34.0) (12.0) (14.1) (31.9) (51.3) (27.8) (30.6)
Yin [5] 17.0 24.9 11.5 9.24 12.6 7.47 7.26 26.8 3.8
(39.5) (38.3) (21.5) (17.5) (43.1) (19.0) (26.5) (32.3)
Johannson et al. [7] 5.79 5.21 – – – – – – 2.6
(9.65) (8.80)
Kangwanpongpan et al. [8] 16.1 30.8 – – – – – – 3.2
(31.5) (43.2)
Bordbar et al. [9] 6.81 5.54 – – – – – – 5.1
(9.56) (9.45)
Guo et al. [10] 60.4 90.1 – – – – – – 2.7
(154) (138)
Double integration 5.77 7.28 4.50 5.24 2.74 3.02 6.68 7.26 16
(9.18) (21.5) (6.34) (8.92) (6.55) (5.81) (10.6) (19.5)

a
For the LBL integration method, t = 66 s .

Fig. 4. Results of the second test case (MR → ∞).

Fig. 5. Results of the third test case (0 ≤ MR ≤ ∞).

5
H. Bordbar, et al. International Communications in Heat and Mass Transfer 110 (2020) 104400

Fig. 6. Results of the fourth test case (MR → ∞, X = 10 m).

WSGGM is quite good and in a similar level of double integration The extended model was then tested for four non-isothermal cases
method. This figure and the errors reported in Table 4 for this test case and compared to a reference solution provided by the LBL integration
prove that the present extended WSGGM perform equally good in op- method, as well as to other WSGGMs developed for a wide range of MR.
tically thin and thick regimes. For a case where MR → 0, most models performed well, while none of
Finally, the computational cost of different models should be the varying-MR WSGGMs available in literature were capable of pro-
checked and compared. Table 4 reports the average computational time viding physically realistic results in a scenario with MR → ∞. Two
t required by each WSGGM for simulating the three test cases, from other WSGGMs based on fixed-MR correlations were also tested and
which one may conclude that all but the double integration model have obtained fairly accurate solutions for the latter case, but were out-
similar processing costs; differences in t between these models are at- performed by the extended WSGGM proposed in this paper. Finally, for
tributed to the different number of gray gases used in the different a more challenging case with a varying molar fraction ratio, in which
models and the amount of operations necessary for determining κp, i and both MR → 0 and MR → ∞ occur in the domain, only the extended
ai. Conversely, the average computational time associated to the double WSGGM and a model based on the double integration of the pure CO2
integration model is 3 to 6 times that of the other WSGGMs, due to the and H2O coefficients provided physically realistic and reasonably ac-
need of accounting for a much greater number of gray gases—5 gases curate results. While the double integration model led to similar (as in
and transparent windows for both CO2 and H2O, totaling Ng = 24 cases 1, 2 and 4) or more accurate (as in case 3) results than the present
(besides a global transparent window for the mixture), compared to extended WSGGM, its computational cost was significantly higher,
Ng = 4 for the extended WSGGM, for instance. However, despite this which limits its usage in large scale industrial cases. Therefore, the new
substantially longer t , the gained accuracy of double integration model extended WSGGM provides a computationally efficient solution with
compared to the extended WSGGM for the third case is quite small, sufficient accuracy for most engineering applications that require the
while for the first, second and fourth cases both models perform simi- calculation of radiative transfer, and can be particularly useful for the
larly. modeling of combustion systems with heterogeneous conditions of gas
concentrations, including locally moisture rich regions.
4. Conclusions
Acknowledgments
With recent improvements of computational power, the role of
numerical modeling in analysis and design of the energy conversion Authors HB and SM greatly acknowledge the support of the
systems, which include complex physical phenomena such as gas par- Academy of Finland under grant no. 314487. Authors GCF thanks CNPq
ticle flow[23,24], combustion and thermal radiation [21,22], got more for a doctorate scholarship.
and more highlighted. Thermal radiation is the dominent heat transfer
mechanism in many combustion systems [21]. In this regard, devel- References
opment of simple and efficient methodlogies for complex physical
phenomena such as spectral thermal radiation is on great importance [1] M.F. Modest, J. Heat Transf. 113 (1991) 650–656, https://doi.org/10.1115/1.
[22]. In this paper, an extension to a published WSGGM has been 2910614.
[2] V.P. Solovjov, B.W. Webb, F. Andre, Handbook of Thermal Science and
proposed, making it able to cover all possible molar fraction ratios for Engineering, Springer International Publishing, 2017, pp. 1–74, https://doi.org/10.
gas mixture of carbon dioxide and water vapor. The extension is based 1007/978-3-319-32003-8_59-1.
on a linear interpolation between the existing formulation and newly [3] M.F. Modest, Radiative Heat Transfer, Academic Press, 2013.
[4] T.F. Smith, Z.F. Shen, J.N. Friedman, J. Heat Transf. 104 (1982) 602–608, https://
developed WSGGM coefficients for pure CO2 and pure H2O. The va- doi.org/10.1115/1.3245174.
lidity of this interpolation approach has been justified by a detailed [5] C. Yin, L.C.R. Johansen, L.A. Rosendahl, S.K. Kær, Energy Fuels 24 (2010)
analysis of the absorption spectra of gas mixtures with large (> 4.0) or 6275–6282, https://doi.org/10.1021/ef101211p.
[6] L.J. Dorigon, G. Duciak, R. Brittes, F. Cassol, M. Galarça, F.H. França, Int. J. Heat
small (< 0.01) molar fraction ratios, which showed that the spectra of Mass Transf. 64 (2013) 863–873, https://doi.org/10.1016/j.ijheatmasstransfer.
the mixture is approximately linearly scalable with MR. Furthermore, a 2013.05.010.
good agreement between the total emissivity predicted by the extended [7] R. Johansson, B. Leckner, K. Andersson, F. Johnsson, Combust. Flame 158 (2011)
893–901, https://doi.org/10.1016/j.combustflame.2011.02.001.
WSGGM developed in this paper and results obtained by LBL integra- [8] T. Kangwanpongpan, F.H. França, R.C. da Silva, P.S. Schneider, H.J. Krautz, Int. J.
tion for MR > 4.0 and MR < 0.01 was also verified.

6
H. Bordbar, et al. International Communications in Heat and Mass Transfer 110 (2020) 104400

Heat Mass Transf. 55 (2012) 7419–7433, https://doi.org/10.1016/j. 2139–2150, https://doi.org/10.1016/j.jqsrt.2010.05.001.

ijheatmasstransfer.2012.07.032. [18] H. Bordbar, T. Hyppänen, Int. J. Heat Mass Transf. 127 (2018) 870–884, https://
[9] M.H. Bordbar, G. Wecel, T. Hyppänen, Combust. Flame 161 (2014) 2435–2445, doi.org/10.1016/j.ijheatmasstransfer.2018.06.093.
https://doi.org/10.1016/j.combustflame.2014.03.013. [19] H. Chu, F. Liu, H. Zhou, Int. J. Heat Mass Transf. 54 (2011) 4736–4745, https://doi.
[10] J. Guo, X. Li, X. Huang, Z. Liu, C. Zheng, Int. J. Heat Mass Transf. 90 (2015) org/10.1016/j.ijheatmasstransfer.2011.06.002.
218–226, https://doi.org/10.1016/j.ijheatmasstransfer.2015.06.052. [20] F.R. Coelho, F.H.R. França, Int. J. Heat Mass Transf. 127 (2018) 105–114, https://
[11] M.K. Denison, B.W. Webb, J. Heat Transf. 117 (1995) 788, https://doi.org/10. doi.org/10.1016/j.ijheatmasstransfer.2018.07.075.
1115/1.2822652. [21] M.H. Bordbar, T. Hyppänen, Multiscale Numerical Simulation of Radiation Heat
[12] F. Cassol, R. Brittes, F.H. França, O.A. Ezekoye, Int. J. Heat Mass Transf. 79 (2014) Transfer in Participating Media, Heat Trans. Eng. 34 (2013) 54–69, https://doi.org/
796–806, https://doi.org/10.1016/j.ijheatmasstransfer.2014.08.032. 10.1080/01457632.2013.695210.
[13] F. Andre, F.R. Coehlo, J.-L. Consalvi, F.H.R. Franca, M. Galtier, F. Nmira, [22] H. Bordbar, A. Maximov, T. Hyppänen, Improved banded method for spectral
V.P. Solovjov, B.W. Webb, 9th International Symposium on Radiative Transfer, thermal radiation in participating media with spectrally dependent wall emittance,
RAD-19, (2019), pp. 407–414, https://doi.org/10.1615/rad-19.500. Appl. Energy 235 (2019) 1090–1105, https://doi.org/10.1016/j.apenergy.2018.11.
[14] J.-L. Consalvi, F. Andre, F.R. Coehlo, F.H.R. Franca, M. Galtier, F. Nmira, 033.
V.P. Solovjov, B.W. Webb, 9th International Symposium on Radiative Transfer, [23] H. Bordbar, P. Zamankhan, Dynamical states of bubbling in vertically vibrated
RAD-19, (2019), pp. 423–430, https://doi.org/10.1615/RAD-19.520. granular materials. Part I: Collective processes, Commun. Nonlinear Sci. Numer.
[15] H. Chu, Y. Feng, W. Cao, F. Ren, M. Gu, Acta Phys. Sin. 66 (2017) 94207, https:// Simul. 12 (3) (2007) 254–272, https://doi.org/10.1016/j.cnsns.2005.04.001.
doi.org/10.7498/aps.66.094207. [24] M.H. Bordbar, P. Zamankhan, Dynamical states of bubbling in vertical vibrated
[16] V. Kez, F. Liu, J. Consalvi, J. Ströhle, B. Epple, J. Quant. Spectrosc. Radiat. Transf. granular materials. Part II: Theoretical analysis and simulations, Commun.
172 (2016) 121–133, https://doi.org/10.1016/j.jqsrt.2015.11.002. Nonlinear Sci. Numer. Simul. 12 (3) (2007) 273–299, https://doi.org/10.1016/j.
[17] L. Rothman, I. Gordon, R. Barber, H. Dothe, R. Gamache, A. Goldman, V. Perevalov, cnsns.2005.03.008.
S. Tashkun, J. Tennyson, J. Quant. Spectrosc. Radiat. Transf. 111 (2010)