Prediction of the mechanical properties of PE/PP blends using artificial neural networks

Polymers have been widely used in industrial applications due to their good thermal and electrical insulation properties, low density and high resistance to chemicals, but they are mechanically weaker and exhibit lower strength and stiffness than metals. Polymer blends, however, offer enhanced mechanical properties. Accurate estimation of mechanical behavior through experimental testing is essential in blends characterizations and structural design. Since the process of experimentally investigating a blend's properties can be costly and time consuming, this paper explores the potential use of ANNs in the field of polymer characterization. It addresses the use of ANNs in predicting the tensile curves and mechanical properties of pure polyethylene PE, pure propylene PP and their blends (PE/PP). Blends of different proportions have been considered. The experimentally acquired data is used to train and test the neural network's performance. The key system inputs for the modeler are blend ratio and percent strain, and the system output is the stress. The ANN predicted outputs were compared and verified against the available experimental date. The study indicates that a multilayered ANN can simulate the effect of the polymer blending ratio on the mechanical behavior and properties to a high degree of accuracy. It also demonstrates that ANN approach is an effective analytical tool that can be adopted to reduce cost and time, and will broaden the range of potential applications of polymers in diverse fields.