During the production of fiber-reinforced plastics using resin transfer molding (RTM), various characteristic defects and flaws can occur, such as fiber displacement and fiber waviness. Particularly in high-pressure RTM (HP-RTM), fiber misalignments are generated during infiltration by local peaks in the flow rate, leading to a significant reduction in the mechanical properties. To minimize or avoid this effect, the manufacturing process must be well controlled. Simulative approaches allow for a basic design of the mold filling process; however, due to the high number of influencing variables, the real behavior cannot be exactly reproduced. The focus of this work is on flow front monitoring in an HP-RTM mold using phased array ultrasonic testing. By using an established non-destructive testing instrument, the effort required for integration into the manufacturing process can be significantly reduced. For this purpose, investigations were carried out during the production of test specimens composed of glass fiber-reinforced polyurethane resin. Specifically, a phased array ultrasonic probe was used to record individual line scans over the form filling time. Taking into account the specifications of the probe used in these experiments, an area of 48.45 mm was inspected with a spatial resolution of 0.85 mm derived from the pitch. Due to the aperture that had to be applied to improve the signal-to-noise ratio, an averaging of the measured values similar to a moving average over a window of 6.8 mm had to be considered. By varying the orientation of the phased array probe and therefore the orientation of the line scans, it is possible to determine the local flow velocities of the matrix system during mold filling. Furthermore, process simulation studies with locally varying fiber volume contents were carried out. Despite the locally limited measuring range of the monitoring method presented, conclusions about the global flow behavior in a large mold can be drawn by comparing the experimentally determined results with the process simulation studies. The agreement between the measurement and simulation was thus improved by around 70%.
Read full abstract