Georgios Mavrogenis, "Potential applications of topology optimization in mechanical components", Master Thesis, School of Production Engineering and Management, Technical University of Crete, Chania, Greece, 2025
https://doi.org/10.26233/heallink.tuc.104271
This thesis is focused on the application of topology optimization in the design of mechanical components. Most of the scientific literature is limited on using Additive Manufacturing for producing topology optimized parts, due to their complexity.Although the Additive Manufacturing Methods have experienced significant growth and the cost of production has been extensively lowered, the traditional manufacturing methods as die-casting, progressive die forming and others, continue to constitute the main methods of producing parts in large scale productions.In this thesis the topology optimization method will be applied in parts, taking into consideration the manufacturability of them with traditional methods of manufacturing. Two main software will be used in this thesis, Solidworks for designing the parts and conducting the topology optimization and Ansys for the parametric optimization and the construction of the metamodel, utilizing Ansys OptiSlang.The workflow of the study starts with the topology optimization, from which the regions of the part that mass can be removed are identified. Then the part is imported into Ansys, on which the removal regions are modelled in a parametric way, so that a parametric optimization can be conducted and the best performing parameter set can be identified. After that the range of the parameters is defined and the model is imported into Ansys OptiSlang. The full fidelity model is simulated for a number of parameters set, a process that is necessary so that the metamodel can be build. After the successful build of the metamodel, the parametric optimization is conducted using the metamodel and evaluating the performance of thousands of parameter sets. The use of the metamodel allows for the completion of the parametric optimization by evaluating the performance of thousands of parameters sets in a short time period with very minimal computational needs. Finally, the performance of the best design extracted from the simulation of the metamodel is verified by comparing it with the high-fidelity model.