A self-contained composition space solution method for strained and curved premixed flamelets

Abstract

Solving flames in scalar or composition space usually requires the modeling of scalar dissipation rates, or scalar gradients, which appear in the Jacobian of the transformation from the physical-space coordinates to the scalar space coordinates attached to the flame surface. Recently, Scholtissek et al. (2019) have discussed a self-contained solution for freely propagating premixed flames, in which the gradient distribution of the coordinate is also solved in scalar space. This approach is here extended to include curvature, strain and effects of unsteadiness. The resulting method is applied to a constant pressure and homogeneous ignition case, a stagnation flame, a tubular flame and a spherical expanding flame, capturing accurately effects of unsteadiness, strain and curvature. The results are systematically validated against physical-space solutions and experiments. In addition to the possibility of exploring these various canonical problems with a single set of equations, it is shown that another major outcome of this new formulation lies in the possibility of studying steady flames subjected to negative strain.

Derivation and analysis of two-dimensional composition space equations for multi-regime combustion using orthogonal coordinates

“…This already indicates a suitable mapping between time and reaction progress variable, which has already been utilized for unsteady flamelet modeling [35] and modeling of homogeneous reacting mixtures [21,29] in previous works. This aspect is revisited and discussed further below.…”

“…(23) the transient terms compensate each other and vanish. It has been shown in previous works [28,29] that composition space models can accurately recover characteristics of transient combustion processes (e.g. auto-ignition, spherical expanding flames) based on this approximation.…”

“…This has several reasons: first, 2D composition space models rely on closure strategies for the scalar dissipation rates (or gradients) of the conditioning variables and the so-called cross-terms which additionally appear in the equations. Only recently it has been shown that a closure for the progress variable gradient (or its scalar dissipation rate) requires an additional equation to be solved [28,29], since no general model exists for this quantity. Second, while the classical flamelet equations are simple to solve and implement, 2D composition space models are significantly more complex.…”

Machine learning for detailed chemistry reduction in DNS of a syngas turbulent oxy-flame with side-wall effects

“…Numerous successful strategies have been discussed in the literature to reduce detailed chemical kinetics. They are based mainly on fuel lumping, time-scale and principal component analysis or direct numerical treatment of the chemical responses from graph analysis or genetic algorithms, sometimes associated with tabulation methods [3][4][5][6][7][8][9][10][11][12][13][14][15][16][17].…”

Chemistry reduction using machine learning trained from non-premixed micro-mixing modeling: Application to DNS of a syngas turbulent oxy-flame with side-wall effects

“…Thereby, more or less automated reduction techniques of these reference detailed chemical mechanisms have been discussed in the literature. These reduction approaches may combine fuel lumping, time-scale and principal component analysis or a direct numerical treatment of the chemical signals from graph analysis or genetic algorithms, approaches which may also be associated with systematic tabulation techniques [7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23] (not exhaustive list).…”