The Geometry of Reaction Trajectories and Attracting Manifolds in Composition Space

Zhuyin Ren,Stephen B. Pope
DOI: https://doi.org/10.1080/13647830500448297
IF: 1.644
2006-01-01
Combustion Theory and Modelling
Abstract:In numerical simulations of combustion processes, the use of dimension reduction to simplify the description of the chemical system has the advantage of reducing the computational cost, but it is important also to retain accuracy and adequate detail. Most existing dimension reduction methods assume the existence of low-dimensional attracting manifolds in the full composition space and try to approximate or directly identify the low-dimensional attracting manifolds. However, questions remain about the geometry of the reaction trajectories in the full composition space, the existence of the low-dimensional attracting manifolds in low-temperature regions, and the minimum dimension of the attracting manifold required for describing a particular chemical system. This paper tries to address some of these issues by studying the reaction trajectories starting from a wide range of different initial compositions for both H2/air and CH 4/air mixtures. Along each trajectory, we study the tangent bundle of the trajectory, the eigenvalues of the Jacobian matrices, and the singular values of the sensitivity matrices (i.e. sensitivity with respect to initial composition). It is shown that the dimension of the affine space containing a trajectory (or of the tangent bundle along a trajectory) is much smaller than the dimension of the full composition space. Even at low temperatures, the Jacobian matrices still have a significant number of large (in magnitude) negative eigenvalues, which implies the existence of fast time scales and low-dimensional attracting manifolds (even at low temperatures). The geometrical significance of sensitivity matrices is explored. Based on the sensitivity matrices, a new method is proposed to determine the minimum dimension of the attracting manifold required for describing a chemical system with prescribed accuracy, and to identify the ‘principal subspace’ which is an approximation to the tangent space of the attracting manifold. Keywords: Dimension reductionAttracting manifoldReduced chemistry Acknowledgments This work is supported by Department of Energy, grant number DE-FG02-90ER14128.
What problem does this paper attempt to address?