A multi-objective optimization framework for natural frequency of 3D printed structures

• Main Author: Jerin Wesley R
• Other Authors: Moon Seung Ki
• Format: Thesis-Master by Research
• Language: English
• Published: Nanyang Technological University 2021
• Subjects: Engineering::Mechanical engineering
• Online Access: https://hdl.handle.net/10356/153345
• Institution: Nanyang Technological University

Additive manufacturing (AM) is gaining momentum from being considered as a rapid prototyping tool to final part fabrications. AM is also quickly gaining popularity as it has enabled designers to design complicated and organic shapes which cannot be manufactured using traditional manufacturing methods. Common uses of AM largely include the fabrication of lattice structures to reduce weight of components. However, the reduction in mass can potentially make components susceptible to vibrations. This is especially true for applications within the automotive industry. When external forces excite the component to its first natural frequency, a phenomenon known as resonance occurs. Reaching resonance can be detrimental as it can cause damage to parts. Traditional methods to overcome this issue is to increase the mass or by applying damping mechanisms. In this research, the objective is to develop a framework for selecting suitable lattices based on the results from design of experiments (DOE) and multiobjective optimization. In the proposed framework, a genetic algorithm (GA) is applied to solve a multiobjective optimization problem. This proposed framework is validated by using a case study to obtain high first natural frequency and minimum mass. An exploratory study on the implementation of 2 different lattice types (Surface lattice & Volume lattice) is performed to understand their impact on the first natural frequency using Polyamide 11 material. The lattices are printed with the selective laser sintering (SLS) method. Free-free vibration simulations are performed using ABAQUS and experiments are performed to compare with the simulation results. In the case study, 3 lattice design parameters are changed and both mass and frequency responses are monitored using DOE. Results from DOE are used in the GA to identify a set of Pareto-optimal points that match the criteria of high frequency and low mass. The Pareto-optimal points are then ranked using the desirability function by assigning weights. The results show the solid part has the highest first natural frequency and weight while the latticed part has a much lower weight (35% reduction) and lower first natural frequency (20% reduction). The difference in simulation and experimental results increase as the design parameters get bigger with a maximum of 17%. The main contributions of this research include the proposed methodology through which lattices can be selected and applied onto parts using simulations and an evolutionary algorithm to identify optimal designs.