Asymptotic Analysis of a Computational Method for Time- and Frequency-Dependent Radiative Transfer

We consider a time-dependent, energy-dependent, nonlinear radiative transfer problem in which opacities are large [O(ε−1)] and interior sources are small [O(ε)]. An asymptotic analysis of this problem as ε→0 leads to the equilibrium diffusion equation in the interior of the system, along with bounda...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Journal of Computational Physics 1998-10, Vol.146 (1), p.366-403
Hauptverfasser: Adams, Marvin L., Nowak, Paul F.
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:We consider a time-dependent, energy-dependent, nonlinear radiative transfer problem in which opacities are large [O(ε−1)] and interior sources are small [O(ε)]. An asymptotic analysis of this problem as ε→0 leads to the equilibrium diffusion equation in the interior of the system, along with boundary conditions and initial conditions for this equation. We apply the same asymptotic analysis to a discrete version of the problem, in which the frequency variable is discretized by the multigroup method, the direction variable by the discrete-ordinates method, the time variable by the fully implicit method, and the spatial variable by a subcell-balance method. We find that as ε→0 the discrete solution satisfies a robust discretized version of the correct equilibrium diffusion equation, with boundary conditions and initial conditions that are remarkably accurate. The analysis thus predicts that if a spatial grid is chosen that resolves interior temperature gradients, then the numerical method obtains an accurate solution in the interior of the system, even though the optical thickness of the spatial cells tends to ∞ and boundary layers in the transport solution are not resolved. We go a step further to analyze problems that are optically thin at some photon frequencies but thick at others, and show that once again the discrete solution is remarkably accurate. We present numerical results that verify these and other predictions of the analyses.
ISSN:0021-9991
1090-2716
DOI:10.1006/jcph.1998.6063