Document Type : Special Issue: Mechanics of Advanced Fiber Reinforced Composite Structures
Authors
1
Department of Mechanical Engineering, Vishwakarma Institute of Information Technology, Pune 411048, India
2
Department of Mechanical Engineering, Cummins College of Engineering for Women, Pune 411052, India
3
Department of Computer Engineering, Vishwakarma Institute of Information Technology, Pune 411048, India
Abstract
Wood/PLA biocomposite filament is a 3D printing material that blends Polylactic Acid (PLA), a biopolymer, with wood powder acting as reinforcement. This combination results in a sustainable 3D printing filament that has grown in popularity in recent years due to its eco-friendliness and the natural appearance of 3D-printed parts. To assess the suitability of wood/PLA biocomposite for various additive manufacturing applications, it is essential to determine its mechanical properties. This study employs fused deposition modelling (FDM) as the additive manufacturing process and focuses on assessing the mechanical properties (tensile, flexural, and impact) of 3D-printed biocomposite. The Taguchi L27 design of the experiments is utilized, and the key process parameters under consideration are infill pattern, layer thickness, raster angle, nozzle temperature, and infill density. A layer thickness of 0.3 mm and an infill density of 100% yielded the highest tensile strength of 42.46 MPa, flexural strength of 83.43 MPa, and impact strength of 44.76 J/m. The dataset has been carefully prepared to facilitate machine learning for both training and testing, and it contains the experimental results and associated process parameters. Four distinct machine learning algorithms have been selected for predictive modeling: Linear Regression, Support Vector Machine (SVM), eXtreme Gradient Boosting (XGBoost), and Adaptive Boosting (AdaBoost). Given the intricate nature of the dataset and the presence of nonlinear relationships between parameters, XGBoost and AdaBoost exhibited exceptional performance. Notably, the XGBoost model delivered the most accurate predictions. The results were assessed using the coefficient of determination (R2), and the achieved values for all observed mechanical properties were found to be greater than 0.99. The results signify the remarkable predictive capabilities of the machine learning model. This study provides valuable insights into using machine learning to predict the mechanical properties of 3D-printed wood/PLA composites, supporting progress in sustainable materials engineering and additive manufacturing.
Keywords
Main Subjects
)