| Author | S. Adjerid, J. E. Flaherty, M. S. Shephard, Y. J. Wang, W. Hillig, J. Hudson, N. Patibandla |
|---|---|
| Title | Adaptive Numerical Techniques for Reactive Vapor Infiltration |
| Year | 1994 |
| Journal | J Ceramic Engineering and Science Proceedings |
| Volume | 15 |
| Pages | 924-931 |
| Issue | 5 |
| Abstract | Patibandla, et al. [1,2] describes a reactive vapor infiltration (RVI) process for manufacturing fiber-reinforced ceramic composites where silicon carbide (SiC) or alumina (Al2)3) fibers are mixed with molybdenum (Mo) powder and pressed at room temperature to form a porous preform. The preform is exposed to a silicon tetra-chloride (SiCl4) and hydrogen (H2) flow where molecular-surface reactions liberate Si which, when absorbed into the preform, reacts with Mo to form a molybdenum di-silicide (MoSi2) matrix. As a first step in modeling the RVI process, we present a mathematical model of the diffusion of Si into a compressed-powder Mo pellet to form the MoSi2 matrix. The production of an intermediate (Mo5Si3) silicide layer, the growth of the MoSi2 layer, and the volume expansion of the pellet are predicted. The model, consisting of a nonlinear ordinary and partial differential system, is solved using a state-of-the-art adaptive software system [3] that includes capabilities for automatic quadtree-structured mesh generation, mesh refinement/coarsening (h-refinement), method order variation (p-refinement), and mesh motion (r-refinement). Computational solutions of one- and two-dimensional problems indicate that the adaptive software is a robust and effective tool for addressing composite-processing problems. When compared with experimental observations, the mathematical model predicts a parabolic growth rate of the silicide layer and the volume expansion of the pellet to a high degree of accuracy. Anticipated applications of the adaptive software and enhancements to the mathematical model are described in a final section. |
| PDF File |  Download |