This study concerns the prediction of the hardness of reinforced high-density
polyethylene waste (HDPEw) composites. The locally sourced palm
inter-fruitlet membrane served as the reinforcing (filler) material while
the Yoghurt Can wastes constituted the polymer matrix. The palm
inter-fruitlet membrane, used for the study were pulverized and sieved to
fine particle sizes. All filler particles passed through a mesh of 250 ?m.
Consequently, the filler sample was characterized using DTA, TGA, and FTIR
techniques whereas SEM was used to study the morphology of the produced
composite. Different weight-percentage compositions of the filler were used
to produce the examined samples with the following formulations: 100 %
LDPEw, 6wt.%, 12wt.%, 18wt.%, and 24wt.% filler composites using the
compression moulding method. On the other hand, hardness, flexural, tensile,
and impact strengths were conducted to understand the mechanical behaviour
of the produced composites. Multiple regression and artificial neural
networks were used to predict the experimental hardness values in
consideration of other independent variables like composite formulations,
tensile, flexural, and impact strengths. The result of the TGA analysis
showed the weight loss and degradation of the organic constituents in the
filler while the DTA study revealed a variety of thermal occurrences and
transitions indicating dehydration, phase change, and filler disintegration.
The maximum hardness value of 76.67 HV was recorded for the composite with
24 wt.% filler while the composite formulation with 12 wt.% filler had the
highest flexural and impact strengths of 41.87 MPa and 0.4979 J/mm2
respectively. The composite composition with 18 wt.% filler gave the highest
tensile strength of 39.04 MPa. The unequal distribution of the filler within
the HDPEw matrix was revealed by the SEM micrographs. The more uniformly
dispersed composites with 12 and 18 wt.% fillers were seen to have improved
mechanical properties whereas the reverse was the case for the 24 wt.%
filler composite formulation which was found to exhibit directional
reinforcement zones. The mean squared error assessment of the predicted
hardness values indicated that predictions by multiple regression were more
accurate than those that were obtained by ANN. This outcome could be caused
by the relative linearity of the examined variables.