CLEAR project (Defects in materials: identification of critical sizes and shapes) develops an innovative methodology to identify critical defects in materials through their impacts on the structural behaviour. Combining numerical modelling, machine learning, optimisation and experiments, it is applied on promising steel-concrete composite (SC) structures, a high TRL material. Aligned with CONNECT-NM RL3, the methodology aims to enhance safety, reduce costs, accelerate the characterization of the performance and support the standardization of the materials, considering the impact of their defects.