In the present study, heat conductivity of an aircraft-grade BA9916 resin with high-toughness was characterized under the curing condition, so as to support curing modeling for this resin and its carbon fiber composites and avoid time- and labor-consuming experiments for manufacturing process design. Thermal-related properties, including density, curing kinetics, glass transition temperature, specific capacity and thermal diffusivity were measured to obtain thermal conductivity of the material. However, the BA9916 resin was toughened via addition of thermoplastic particles, resulting in much higher viscosity before completely cured than that of common epoxy resins. As a result, it was challenging to directly measure certain thermal properties of the neat resin. To settle this problem, the BA3202 unidirectional carbon fiber composite prepreg with the BA9916 resin was employed as a media to obtain corresponding properties of the resin through experiments and analytical calculation. Derived material properties of the resin were then input to the user-defined material subroutine UMAT to predict thermal response of the composite under various curing conditions, with the maximum error of 6.82% validated via experiments. Hence, the acquired characteristics can be utilized for numerical analysis of various composites composed of BA9916 resin, obviating the need for repeated physical experiments that are time- and resource-consuming.

The prepreg compression molding process has received increasing attention from industry due to its cost-effectiveness and ability to produce complex structural shapes, and the design of the initial blank geometry is critical for the efficient production of woven composite parts using the automated manufacturing process. To design the optimal blank geometry that meets the structure requirements, and minimize trimming and waste of the edge material after preforming in the prepreg compression molding process, a blank geometry design method was developed based on finite element analysis (FEA) of preforming and a modified non-orthogonal material model. Meanwhile, whether normal vectors of all shell elements of the preformed prepregs pointing to one side of the produced structure was analyzed to automatically detect appearance of wrinkles and overlaps. An optimal blank geometry can be designed by modifying edge elements of the prepreg model through iterations of the preforming simulation. By comparing with the experimental results, the blank span length, appearance and yarn angles predicted by the preforming model were validated, and capability of the modeling-based design method to optimize the prepreg blank geometry for minimum material waste during preforming was verified.