Function-oriented in-line quality assurance of hybrid sheet molding compound

The use of fiber-reinforced polymers (FRPs) increases steadily on global scale. The combination of discontinuous SMC (DiCo-SMC) and continuous SMC (Co-SMC) in a new, hybrid material class (CoDiCo-SMC) promises reasonable manufacturing costs as well as high local stiffness and strength. Traditionally, harsh requests for defect-free FRP components, leading to high quality requirements, are a cost driver in this industry. Further, occurring manufacturing deviations jeopardize the fulfillment of the functionality of manufactured components. To overcome these limitations, an approach for a component-specific, function-oriented in-line quality assurance is proposed. For this type of quality assurance, in-line measurement results are integrated into functional models, such as finite element (FE) models. Surrogate models of these functional models accelerate the evaluation to adhere to the cycle time within production. In the present work, component-specific, function-oriented in-line quality assurance was exemplarily implemented for the new CoDiCo-SMC material class. Three different measurement technologies were used to quantify three relevant manufacturing deviations (glass fiber fraction, Co-SMC patch pose, delamination). Terahertz spectroscopy was used for the first time to in-line measure local variations of the glass fiber fraction in DiCo-SMC. Pulsed phase thermography was used to quantify delamination and an industrial camera to measure Co-SMC patch pose. For each measurement technology, the measurement uncertainty was quantified according to the “International Guide to the Expression of Uncertainty in Measurement (GUM)”. The measurement results were further processed in a parameterized FE model and aggregated to a functional prediction. Using the measurement results and respective FE-modeled functions, surrogate models were trained through input-output relations. Within this work, the predicted component function is also understood as measurement. Hence, the measurement uncertainties of both FE models and surrogate models were determined. The presented approach was validated using two exemplary test specimens. The results show that in particular the measurements of local glass fiber fractions and the Co-SMC patch pose allow conclusions to be drawn on the component-specific stiffness. The determined measurement uncertainties do currently not allow for an industrial application. Making use of component-specific functional information in production allows to reduce common safety factors during the design phase of FRP components.