Spatial Surface Mass Flux on Confined Fibers: DSMC Validation of a Kinetic-Theory Model for Forced-Flow CVI Deposition

Abstract

Chemical vapor infiltration (CVI) is widely used to manufacture ceramic matrix composites, but deposition non-uniformity and premature pore closure still limit processing time and cost. CVI operates at low pressures and in microscale pores, where non-equilibrium gas transport develops, making common continuum assumptions unreliable. Direct simulation Monte Carlo (DSMC) can resolve pore-scale rarefied transport and gas–surface interactions; however, it is too expensive for furnace-scale modeling. Practical CVI modeling relies on kinetic-theory-based macroscopic closures, which are often poorly described and rarely tested. Here, DSMC is used to test kinetic-theory surface-flux models at varying rarefaction and flow conditions. Local surface mass flux and deposition vary strongly with velocity but show no explicit dependence on Kn, in contrast to the channel mass flow rate exhibiting strong rarefaction dependence. The drifting-Maxwellian model accurately predicts face-resolved surface flux and deposition: upstream flux increases while downstream flux decreases with flow speed. At CVI-relevant low Mach numbers, the surface-averaged deposition remains close to the no-flow prediction due to a cancellation over a symmetric surface, explaining why total densification can appear consistent with diffusion-only models in the CVI literature. Finally, MTS–H₂ simulations confirm that deposition is predicted by the kinetic-theory flux multiplied by an adsorption probability for heterogeneous reaction.