École Nationale Supérieure de
Techniques Avancées

Abstract

Spacecraft thermal protection systems make extensive use of fibrous materials for insulation and reinforcement. Recent efforts have been made to improve microscale modeling of such materials to inform macroscale models. Here, we investigate the effective properties of computational models of typical fibrous materials using the PuMA (Porous Microstructure Analysis) software. We characterize the effective thermal conductivity as a function of material sample volume. We then propose and evaluate two Multilevel Monte Carlo methods to estimate mean effective properties at a reduced computational cost, with one method showing potential for significant cost reduction.