Multi-scale reliability-based design optimization of tow-steered composite laminates

3D printers and automated fibre placement machines have allowed the introduction of Variable Angle Tow (VAT) composites that, allowing the fibres to be relaxed along curvilinear patterns, offer greater tailoring capabilities than classic CFRP laminates. Nevertheless, the steering of brittle fibres is not flaw-exempt and, in fact, is greatly affected by the printer signature. In this work, we explore the use of multi-scale, high-order finite elements based on the Carrera Unified Formulation (CUF) to demonstrate the importance of manufacturing defects on the mechanical response and the reliability-based design of VAT composite laminates. In detail, the effect of fibre waviness, gaps, overlaps and micro-scale defects variation on the buckling and vibration response as well as on failure onset are investigated by using Monte Carlo analysis and stochastic random fields. Eventually, the use of metamodels and evolutionary algorithms are proposed and validated for uncertainty quantification and robust design. One of the main scope of the work is to show that high-order, multi-scale structural modelling based on component-wise kinematics can be helpful to broaden the design space that classical structural theories may have shrunk.