Journal of Composite Materials

The effect of gravity on free surface shape during the filling of a disk-shaped cavity is experimentally studied. A disk-shaped mold cavity is constructed to measure spreading of the fluid front due to gravity. Fluid front sensors are mounted on the top and bottom mold walls at three radial locations. The time the fluid front reaches these sensor locations is recorded. This data is used to calculate the radial distance between the top and bottom of the fluid front at each sensor location. The important parameters that govern the fluid front dynamics are identified as the Reynolds, Bond, and Capillary numbers. Spreading is found to be mainly dependent on the Bond number, such that a large change in spreading is observed due to a small change in Bond number. Spreading increases with decreasing Capillary number and remains nearly constant over the range of Reynolds numbers studied.

Conformation of fabrics to complex molds during composite processing induces significant fabric deformation and local shear, which in turn alter the processability of these preforms from their unsheared, flat configuration. The present work (Parts I and II) establishes that although composite mold processing can often be described generally by a percolation flow assumption (e.g., Darcy or Poiseuille flow), changes in microarchitecture of fabric in shear result in markedly different flow fronts. We reiterate our earlier finding that use of a transformed Darcy law (i.e., mathematical transformation of tensor of undeformed permeability, to the sheared configuration) does not accurately predict permeability for sheared fabrics. In essence, the effect of the change in microarchitecture of the fabric is not captured by mathematical transformation of the tensorial permeability. Wealso point out the deficiencies of semi-empirical approaches in determining sheared fabric permeability. We then develop a 3D fabric model, which is used to quantify the effects of nesting and changes in gap architecture with shear angle. We show that nesting produces gaps in molds in commonly-used permeability experiments which easily exceed single-layer fabric thicknesses when more than a few layers are used, but that this condition is easily detectable in an experiment (i.e., Poiseuille flow between topmost layer and mold top is easily detected). We also show that shear angle (in our case,. = 0°, 15°, and 30°) produces little difference in nesting, though it significantly alters fabric microstructure and the sizes and shapes of intralayer gaps. In Part II of this paper, we use this fabric model to predict fabric permeability. Our work suggests that departure from the more traditional approach of generation of a large suite of data from permeation experiments to determine manufacturability of preforms, in favor of computational simulation of fabric geometries, is well-justified.

Use of the creeping flow assumption provides a computationally efficient and accurate means of predicting flow fronts in reinforcement media in many technologically important polymer processes. In the case of fabrics, this creeping flow is shown to proceed primarily through the gaps in fabrics, with capillarity playing little if any role for commonly used materials. This finding has important implications for selection of strategy in modeling manufacturing processes of such materials. Namely, in the presence of fabric deformation which induces local shear, changes in the gap architecture greatly affect the flow patterns, and are not well predicted by tensor transformation of Darcy-type permeabilities. Asimple, classic flow model is adapted to the case of fabrics penetrated by low-pressure viscous liquids after careful analysis of fabric architecture. An applicability range of this creeping flow model, the variable gap model, is developed. The present paper gives the model assumptions, and confirmation of its agreement with more sophisticated calculations. A demonstration of the approach for an unbalanced fabric (Knytex 24 5×4 unbalanced plain-woven glass fabric) shows excellent prediction of the bounds on flow behavior, and supports our earlier experimental findings on flow front orientation. This approach also shows clear superiority to semi-empirical, geometry-based models, since no fitting parameters at all are used in the modeling: only the constituent materials’ geometry and properties are needed. This methodology is better able to predict trends in flow fronts, both qualitatively and quantitatively, than semi-empirical fitting. Extension of this work to realistic production processes is planned. 1Author to whom correspondence should be addressed. 1285 Journal ofCOMPOSITE MATERIALS, Vol. 35, No. 14/2001