Compression molded SMC(Sheet Molding Compound) parts, consisting of thermosetting unsaturated polyester resin, a particulate filler, and glass fiber reinforcement, have become increasingly important in the plastics industries due to their combination of outstanding specific stiffness and strength, moldability, and relative low costs. An important aspect influencing the mechanical and surface properties of molded SMC products is the distribution of reinforcing glass fibers. In this study, a method of quantitatively measuring the fiber volume distribution through digital image processing is proposed, and numerical prediction of fiber volume fractions is carried out with an in-house rigid thermo-viscoplastic FEM program. The validity of the image processing method is verified through comparisons with burning test measurements, and the image processing results are used for the analysis of molding parameters which influence the distribution of fiber volume fractions. The results show that a more uniform distribution of fibers can be obtained with smaller compression ratios and faster mold closing speeds, while mold temperatures do not largely affect the distribution. The prediction of fiber volume fraction distribution is carried out using a revised version of Hojo et al.'s fiber content ratio prediction model for long-fiber reinforced thermoplastics. This model consists of two coefficients that characterize the flow of fibers in the SMC charge flow. A numerical search scheme was implemented for determination of appropriate values for the two coefficients involved with the model based on minimizing the average error between simulation and experiment results. The influence of each coefficient on predictions of fiber volume fractions was numerically determined, and by the use of the chosen coefficient values it was found that simulation results agreed well with experimental observations. Also, the numerically determined coefficient values are compared to calculated separation coefficient and flow index values of Hojo et al.'s original model which had assumed the material as a power-law fluid. The comparison results display similar orders for the values, confirming the successful implementation of Hojo et al.'s model in the rigid thermo-viscoplastic FEM simulations, and demonstrating the ability of the current program to predict the distribution of fiber volume fractions. It is construed that the current study for determination and prediction of fiber volume fractions in compression molded SMC parts can aid in the design of successful SMC products.
