Owen Peckham

Despite the lightweighting benefts that hollow structures afford, current Generative Design (GD) tools are not capable of synthesising them by default. This paper proposes an approach to generate hollow structures using an off-the-shelf GD tool and an innovative shelling method. The approach is used to create solid and hollow variants of a load bearing component. These are modelled using Finite Element Analysis (FEA) then Additively Manufactured (AM) and characterised via destructive load testing. FEA results show that the shelled structures are up to 2.5 x stiffer than their solid counterparts however destructive testing revealed small stiffness losses attributable to the AM process. Despite the physical testing results the method offers the potential to apply GD tools to industries where hollow tubes are accepted practice, enabling part consolidation capabilities to be leveraged.

This review explores the intersection of Artificial Intelligence (AI) and Generative Design (GD) in engineering within the mechanical, industrial, civil, and architectural domains. Driven by advances in AI and computational resources, this intersection has grown rapidly, yielding over 14,000 publications since 2016. To map the research landscape, this review employed semantic search and Natural Language Processing, parsing 14,355 publications to ultimately select the 88 most relevant studies through clustering and topic modelling. These studies were categorised according to AI and GD techniques, application domains, benefits, and limitations, providing insights into research trends and practical implications. The results reveal a significant growth in the integration of advanced generative AI methods, notably Generative Adversarial Networks for direct design generation, alongside the continued use of genetic algorithms and surrogate models (e.g., Convolutional Neural Networks and Multilayer Perceptrons) to manage computational complexity. Structural and aerodynamic applications were the most common, with benefits including improvements in computational efficiency and design diversity. However, barriers remain, including data generation costs, model accuracy, and interpretability. Research opportunities include the development of generalisable foundation surrogate models, the integration of emerging generative methods such as diffusion models and large language models, and the explicit consideration of manufacturability constraints within generative processes.

This paper demonstrates the unpredictability of outcomes that result from compounding variabilities when using generative design (GD) coupled with additive manufacturing (AM). AM technologies offer the greatest design freedom and hence are most able to leverage the full capability of generative design (GD) tools and thus maximize potential improvements, such as weight, waste and cost reduction, strength, and part consolidation. Implicit in all studies reported in the literature is the fundamental assumption that the use of GD, irrespective of user experience or approach followed, yields high-performing and/or comparable design outputs. This work demonstrates the contrary and shows that achieving high performance with GD tools requires careful consideration of study setup and initial conditions. It is further shown that, when coupled with the inherent variability of AM parts, the potential variation in the performance of the design output can be significant, with poorer designs achieving only a fraction of that of higher-performing designs. This investigation shows how AM by Material Extrusion (MEX), which is used to manufacture components with polylactic acid (PLA), varies through different design pathways, bridging MEX and GD. Through a practical study across nine independently generated designs, the breadth of performance—due to initial GD conditions and MEX part strength unpredictability—is shown to reach 592%. This result suggest that current GD tools, including their underlying workflows and algorithms, are not sufficiently understood for users to be able to generate consistent solutions for an input case. Further, the study purports that training and consideration on GD setup are necessary to apply GD toolsets to achieve high-performing designs, particularly when applied in the context of MEX.

Generative design is now a core capability of today's CAD systems enabling broader exploration of the design space. This paper presents a desktop review of 15 tools with the aim of characterising where the tools are used in the design process and to what extent they can be considered creative.The results showed that, despite their increasing capabilities, GD tools are unable to exhibit all phases of creativity and cannot be applied in all stages of the design process. This highlights the criticality of the human in design providing opportunities for GD to augment designer creativity.