Characterization and modeling of race-tracking in liquid composite moldingprocesses

Liquid composite molding (LCM) processes require the impregnation of a poly meric resin through a porous preform, usually composed of glass, carbon, or Kevlar fibers. Numerical mold-filling simulations are currently being deve loped to predict the flow through LCM mold cavities and are a powerful desi gn tool. The accuracy of such simulations are very sensitive to the permeab ility components, input provided by the user to represent the resistance to flow provided by the porous preform. Air channels can be present within a mold cavity, either unintentionally formed, or intentionally placed to enha nce the mold filling process. Such channels provide paths of relatively low flow resistance, and can drastically alter flow front advancement and inje ction pressures, this effect being commonly referred to as 'race-tracking'. Race-tracking must be considered in numerical mold-filling simulations, on e common approach being the application of 'equivalent permeabilities' to f inite elements within a model that physically represent the air channels pr esent. A planar rectangular mold cavity is studied here, having a single ai r cavity of known dimensions running along one side of the mold. This geome try provides the simplest flow configuration, all Darcy velocity components being in-plane. In this paper, the equivalent permeability magnitudes are based upon steady-state, fully developed flow through a rectangular duct. D etailed flow visualization experiments have been completed, recording how-f ront advancement and injection-pressure histories, for two different prefor m types, and a variety of mold cavity thicknesses, preform volume fractions air-channel widths. Significant effort was made to provide exact compariso ns of both flow front advancement and injection pressure, between the exper iments and simulations based on the equivalent permeability approach. The c omparisons are very good, the magnitude of the deviations being close to wh at should be expected from natural permeability variations from preform to preform. Sources for experimental error have been examined, the major limit ation of these experiments being caused by the complex interaction of the a crylic mold, and stiffening bars used. The equivalent permeability approach has been shown to model flow front advancement and injection pressures ver y well within a Darcy's law based mold filling simulation, for the volume f ractions studied (0.50 to 0.16). This study has been limited to low volume fractions, and care should be taken when extending the equivalent permeabil ity approach to woven and stitched preform styles. An alternative numerical method was investigated, modeling flow in the preform by using Darcy's law , and one-dimensional Stokes flow in the air channel, with a variable sourc e term to model the transverse how into the preform. The results from the t wo simulations are in excellent agreement, demonstrating that the equivalen t permeability approach employing Darcy's law over the entire domain, model s the transverse flow into the preform from the air channel accurately. (C) 1999 Elsevier Science Ltd. All rights reserved.