Faced with the proliferation of plastic waste (LDPE), the overexploitation of Sanaga sand, and the under-utilization of coconut husks in Cameroon, this study developed and characterized an innovative ternary composite. The aim was to establish an artisanal manufacturing process and optimize the proportions of Low-Density Polyethylene (LDPE) waste, Sanaga sand, and coconut husk powder for sustainable construction. The process, inspired by ecological paving, tested nine formulations (LDPE 20-40%, sand/coconut 0-10%). Samples were characterized for their apparent density, water absorption rate, and mechanical properties (bending). Results showed a decrease in density with LDPE and coir powder. Water absorption remained very low (max. 0.640%), guaranteeing excellent durability in damp conditions. Although maximum flexural strength was lower than for thermo-compressed composites (probably due to the absence of pressure compaction in the artisanal method), the incorporation of coir conferred a certain ductility. Some formulations (F0/40, F5/20) are potentially usable for T4/T5 or T5 type pavers, and indoor flooring (P2). Future optimization will include pressure compaction and improvement of the LDPE/coconut interface to enhance performance. This work contributes to sustainable waste management and the promotion of environmentally-friendly building materials.

The aim of this work is to take into account the effect of the stress gradient as a beneficial effect in fatigue, since its presence within a material systematically leads to an increase in the endurance limit in terms of local maximum stress. We have therefore proposed that the stress gradient be taken into account in the fatigue criterion using a statistical approach based on the variance method. This method assumes that the fracture plane is the one on which the variance of the equivalent stress is maximum. This method defines three types of equivalent stress on a plane: normal stress, shear stress and combined normal and shear stress. The results obtained using the simple variance method and the stress gradient method were compared for a cyclic biaxial loading condition. We found that in terms of prediction of the fracture plane, there is not much difference between the two formulations (the simple variance method and the one taking into account the stress gradient), but the maximum values of the variance for the variance method taking into account the stress gradient vary greatly compared to the case of the simple variance method.