Gas transport model for chemical vapor infiltration
- PDF / 801,522 Bytes
- 7 Pages / 576 x 792 pts Page_size
- 84 Downloads / 251 Views
A node-bond percolation model is presented for the gas permeability and pore surface area of the coarse porosity in woven fiber structures during densification by chemical vapor infiltration (CVI). Model parameters include the number of nodes per unit volume and their spatial distribution, and the node and bond radii and their variability. These parameters relate directly to structural features of the weave. Some uncertainty exists in the proper partition of the porosity between "node" and "bond" and between intra-tow and inter-tow, although the total is constrained by the known fiber loading in the structure. Applied to cloth layup preforms the model gives good agreement with the limited number of available measurements.
I. INTRODUCTION For forced flow chemical vapor infiltration (CVI) the gas permeability of the preform and the partially dense composite is a critical factor in the success of the process. The gas permeability of the material controls the pattern of reactant flow through the preform and sets the limit for the ultimate achievable density. Previous efforts to estimate gas transport properties and their evolution during CVI densification have been limited. It is important to note that estimation of the initial properties of the preform is not sufficient for this application. The transport properties of the material change as density increases and a suitable model must track this change. In particular, the limiting behavior near full density is very important as this determines to a great extent the final microstructure of the composite. Early work by Naslain and co-workers1-2 modeled the preform as a collection of cylindrical, monosized pores and estimated the effective diffusion coefficient for the material as the pores filled during CVI. Convective flow of gas was not considered although this simple microstructure would allow a relatively straightforward extension of the model to include gas permeability. This simple microstructure, however, is not a good representation of real woven preforms and cannot provide quantitative estimates of transport properties. For randomly oriented, short fiber preforms Starr3 proposed a structure model based on an orthogonal array of cylinders and used this model to estimate surface area and gas permeability during CVI. This microstructure model is more realistic in that it includes a distribution of pore sizes. Developed for chopped fiber preforms the model does not represent anisotropic woven materials. Yu and Sotirchos4 proposed a generalized pore model for gas-solid reactions based on an interconnected network of cylindrical pores with a distribution of 2360 http://journals.cambridge.org
J. Mater. Res., Vol. 10, No. 9, Sep 1995 Downloaded: 13 Mar 2015
diameters. Percolation theory is used to predict creation of trapped porosity as density increases and the smaller pores are filled. This feature is an important advance since residual, inaccessible porosity is always present in "fully" infiltrated CVI processed composites. This approach has been tested using Mon
Data Loading...
