Isothermal Chemical Vapor Infiltration Modeling by Random Walks in CMT 3D Images at Two Scales

Abstract

International audienceThe production cycle of high-quality Ceramic-Matrix Composites (CMCs) often involves interphase or matrix deposition by Chemical Vapor Infiltration (CVI). This costly step has motivated many modeling approaches, in order to provide guidelines for process control and optimization. In this context, numerical tools for direct modeling of isothermal, isobaric CVI in complex 3D images of the composite architecture, acquired e.g. by X-ray Computerized Microtomography (CMT) have been developed. To address inter- and intra-bundle length scales inherent to a composite with a woven textile reinforcement, a numerical strategy has been set up, based on two numerical tools. They solve diffusion-reaction equations and handle simultaneously the progressive evolution of the porous structure. They involve distinct random walk methods and image handling routines. The small-scale program uses Pearson random walks simulating rarefied gas transport; the fluid/solid interface is explicitly represented as a set of triangles through a Simplified Marching Cube approach. Direct simulation of CVI in intra-bundle pores is possible with such a tool. Effective laws for the evolution of porosity, surface and transport properties as infiltration proceeds are inferred from these simulations by averaging and are considered as inputs for the next modelling step. The large-scale solver uses Brownian motion simulation; the porous medium is considered as a continuum with locally heterogeneous and anisotropic diffusivities, and the deposition reaction is handled through a survival probability computation. Simulation of the infiltration of a whole composite material part is possible with this program. Validation of these tools on test cases, as well as some examples on actual materials, are shown and discussed