New bio-based honeycomb panel with a core in Arundo donax rings and flax/epoxy composite skins: Experimental tests and numerical model validation

Honeycomb sandwich panels are widely used in multiple building sector applications like non-structural lightweight panels for partitions, floating floors, insulation panels, or furniture. These panels typically consist of three layers: two thin rigid external layers and a thick low-density internal layer with a honeycomb geometric shape. To reduce environmental footprint, eco-friendly materials are encouraged. In this study, we propose an innovative sandwich panel with a core made of Arundo donax rings and skins made of flax fibre-reinforced epoxy composite. Arundo donax is a highly invasive reed in southern Europe. Its rapid growth makes it environmentally advantageous as it can absorb a significant amount of CO ₂ which is then sequestered in the panel, enhancing the panel’s sustainability compared to existing alternatives. The honeycomb panels are produced through thermocompression and mechanically tested using pull-off tests and three-point bending. A key novelty of this research is the detailed investigation of the lateral surface roughness of the rings, which has never been studied before. The study explores how different roughness levels affect the adhesion between the core and the skins, the adhesion between the individual Arundo donax rings within the core, as well as the mechanical bending properties of the panel. The results show that polishing treatment significantly impacts all the flexural mechanical properties, especially when performed on lateral surfaces of the rings. In particular, the polishing treatment has increased the flexural strength by 24.25 %, shifting from 19.06 MPa to 25.16 MPa. Numerical simulations are conducted employing Finite Elements with cohesive surface contacts between skins and Arundo donax rings and within the rings. By adjusting the parameters of the cohesive interface, the model can account for different roughness levels and accurately replicate the experimental results. Once calibrated, the model can serve as a design tool.