MULTI-OPTIMIZATION OF EMPIRICAL MODELS FOR THE MATERIAL EXTRUSION PROCESS
Schuravi N. Mallian1*, Boppana V. Chowdary2*
1,2Faculty of Engineering, The University of the West Indies, Trinidad
1Email: schuravi_z@hotmail.com *(Corresponding author)
2Email: Boppana.Chowdary@sta.uwi.edu
Abstract:
Advances in materials and manufacturing technology and increased competitiveness has led to companies needing to manufacture products more efficiently and rapidly to meet growing market demands. The Additive Manufacturing (AM) process is ideally suited to the fabrication of complex geometries usually impossible with traditional methods furthermore it is capable of fabricating entire assemblies in step without the need for tooling or human involvement.
Due to the flexibility and advantages over conventional methods AM has garnered significant attention from the manufacturing sector in meeting market demands. Of the array of available AM processes, Material Extrusion (ME) utilizes a heated thermoplastic filament to construct parts or assemblies via a layer by layer deposition method. This process is not without its own flaws, suffering from accuracy, build time, strength etc., due to the conflicting nature of the process parameters of ME. Therefore, it is critical to understand the shortfalls of ME and classify the factors that directly influence the performance of a part.
This paper focuses on the enhancement of the performance measures of the part in terms of build time, material consumption and max torsional stress. This is accomplished by understanding the influence of the process parameters such as raster width, raster angle, part orientation and layer thickness on the performance measures via statistically valid models and optimization methods. This was accomplished using a Box-Behnken design for the experimental design followed by the multi-objective optimization of the empirical models from which the optimum process settings was determined.
This study has shown that complex a non-linear relationship exists between the process parameters and performance measures. Results show that the Artificial Neural Network had a better fit when compared to the Response Surface Methodology model and it can be a promising alternative for the prediction and optimization of the ME process.
Keywords: Additive Manufacturing, Material Extrusion, Multi-objective optimization
https://doi.org/10.47412/WIZL8999
