The multispectral gas radiation modeling: A new theoretical framework based on a multidimensional approach to k-distribution methods

- We introduce a new framework for gas radiation: the Multispectral approach.
- We build spectral intervals using concepts from Functional Data Analysis.
- Over those spectral intervals, absorption coefficients are scaled.
- The model outperforms usual Scaled-k models in highly nonuniform media.

A number of radiative heat transfer models dealing with uniform - homogeneous isothermal - or nonuniform gaseous media can be found in the literature. Usually, these models are considered to be distinct from each other by their developers, and hence the large number of different denominations. For a potential user, it makes difficult the choice of an optimal model for a given problem. The aim of the present work is to demonstrate that all k-distribution models, i.e. from the most accurate line-by-line up to the simplest correlated or Scaled-k ones, are based on a restricted set of concepts shared by all of them. By proposing the most general formulation of multidimensional k-distributions, we clearly define clustering techniques to build from reference line-by-line data approximate gas radiation models fully controllable in terms of accuracy. These techniques are presented in details and assessed in several test cases for demonstrating their relevance in radiative heat transfer.