Phd in INDUSTRIAL AND MANAGEMENT ENGINEERING

Thesis title: Generative Design: perspectives of applicability in the field of mechanical engineering

Generative Design is a computational design methodology that has gathered the interest of engineering designers dealing with the increasing complexity of design problems focused on engineering performance in an environment characterized by intense competition, where cost reduction and lead time minimization are critical objectives. Generative Design has its roots in the interest towards biological systems regarding evolution, cooperation, and organic design. Indeed, Evolutionary Algorithms, Swarm Intelligence, and Topological Optimization constitute the elements able to generate novel and unbiased solutions, along with the latest trend in engineering design: the Deep Generative Models. Providing a concise definition, a Generative Design system constitutes a framework able to generate spatial shapes from problem and/or process requirements, providing automation and autonomy in the design process. This work adopts a four-phase Generative Design Method leveraging Reverse Engineering, Parametric Modeling, Genetic Algorithms, and Generative Design founded on density-based Topological Optimization to generate mechanical components within the fields of Cultural Heritage and automotive, highlighting the modularity and the generalization capabilities. In particular, case studies concerning the design synthesis of supporting structures for ancient artifacts are considered, as well as multi-material layout optimization in automotive components. The presented workflow aims to propose a structured way to leverage these technologies moving forward with respect to conventional engineering design processes. The procedure starts from the clarification of the design task, the acquisition of the design requirements in terms of artifact features and characteristics, exhibition and manufacturing requirements. Later, leveraging Reverse Engineering for the acquisition of the digital model, Parametric Modeling and Finite Element Analysis ensure a simulation-driven approach. A Genetic Algorithm is in charge of defining Pareto-optimal configurations. At this point, the Generative process takes place involving Topological Optimization on an open design space, accounting also for Design for Additive Manufacturing methodologies. Shifting towards executive design, a multi-material optimization process could take place. Finally, the surfaces of the model could be reconstructed and the final Finite Element Analysis validation could be performed. Since mechanical components are generated accounting for Additive Manufacturing, they can be directly produced. The present work is structured as follows: Chapter 1 provides a common definition and framework for Generative Design systems, reporting also a brief historical perspective on the engineering design process, and then reports the state of the art of Generative Design methodologies. Chapter 2 presents the proposed Generative Design methodology, structured in four phases. Chapters 3 to 6 report the case studies on which the methodology has been tested. In particular, Chapter 3 provides a setup and preliminary test of Phases One and Two on a simplified artifact, while Chapter 4 provides a setup and preliminary test of Phase Three on a simplified automotive seat cushion and frame. Chapters 5 and 6 apply the proposed methodology to two retrievals from the Museo Civico dell’Agrocimino of Soriano nel Cimino in Italy: a metallic hoe and an Aryballos. Finally, the conclusions are outlined in Chapter 7.