Experimental and artificial neural network ANN investigation of bending fatigue behavior of glass fiber/polyester composite shafts

Abstract

This study includes an experimental investigation for rotating bending fatigue in five types of composite material shaft with comparison to a standard type made of pure polyester material. In addition, an artificial neural network ANN prediction is compared with experimental part. The five types vary according to the type of glass fiber fabric (thin and thick fiber fabric) and fiber volume fraction (10.2 and 14.6% for thin fibers, and 14.6, 29.3 and 43.9% for thick fibers). Tensile tests are performed to identify the range of applied fatigue stress for every type. Specimens weight is measured in order to calculate the ratio of strength to weight. The rotating bending fatigue tests are performed to record the fatigue behavior for each type. Results show that the type with (29.3%) volume fraction of thick fiber has the maximum tensile strength to weight ratio with (290%) improvement compared with the standard type. The best fatigue behavior is for type with (14.6%) volume fraction of thin glass fibers. This type has the maximum value of fatigue stress at life limit of (106 cycles) with (362%) improvement compared with the standard type. Then, the types with (29.3%) and (43.9%) volume fractions of thick glass fibers make (356%) and (331%) improvement of fatigue stress, respectively. The ANN gives satisfactory results and it is a good prediction tool for fatigue life of composite materials.