This thesis explores the application of topology optimization for large-scale glass structures in architecture, addressing the limitations of traditional casting methods and exploring the potential of 3D printed glass. Previous research emphasized the importance of manufacturing methods, but highlighted the lack of transparency in topology optimized cast glass. This study uses 3D printing to overcome these challenges, focusing on the unique properties and manufacturing techniques of 3D printed glass, including the limitations of its brittle nature and different tensile and compressive strengths.

An extensive literature review provides a foundation for glass properties, manufacturing techniques and the principles of topology optimization. The research advances the use of SIMP methodology and adapts it to 3D printed glass constraints such as overhang, path continuity and nozzle size. Therefore, it incorporates an adapted layer-to-layer overhang filter and addresses the island eect and path control using advanced computing techniques.

The implementation of these methodologies is described in detail, with specific adjustments to the overhang angles and connection strategies for island structures. Testing within a predefined design domain evaluates the limitations and capabilities of the proposed solutions, culminating in the selection and 3D printing of a feasible design.

The results validate the approach and highlight the need for further research, especially regarding the anisotropy of glass layers.

This research is concluded with a discussion of the broader academic implications and future research opportunities, supported by additional insights and findings presented in the appendix.

This work demonstrates significant advances in structural glass architecture and paves the way for innovative applications of 3D printed glass.

This research addresses the implementation of learning algorithms and generative design in string-based topology exploration methods. It aims to generate diverse structural patterns for shells and surface structures that align architectural, engineering, and construction objectives. By integrating reinforcement learning (RL) and quad-mesh grammars, surface topology is explored through quantitative metrics, demonstrating the strength and generality of this approach. The research ultimately promotes creative exploration during the conceptual stages of structural design, emphasizing collaboration between form-designers and form-analyzers to harness emerging computational techniques.

The quad-mesh grammar was first formulated within a Markovian decision framework to integrate with open-source RL Python packages. States, actions, and rewards were defined with sufficient generality to avoid over fitting while evaluating the RL agent's ability to navigate between two specified string-action sequences and their associated mesh layouts. Initially, simple tasks involving four design steps were tested, followed by more generalized target terminal states with longer design sequences. The impact of different reward structures and varied model parameter setups on convergence and accumulated rewards was also analyzed.

The findings indicate that reward functions based solely on topological and grammatical characteristics did not fully guide the agent from an initial coarse mesh to a target state. However, extended design episodes demonstrated potential for improved RL outcomes. The DQN struggled with non-optimal policies due to negative rewards and sparse positive reinforcement, suggesting that customized model architectures or alternative RL algorithms could enhance performance. The exploration phases yielded suboptimal but diverse mesh configurations, highlighting the need for additional structural and geometric parameters, as well as more complex grammar operations to improve diversity while mitigating computational challenges. These insights underscore the importance of balancing feasibility, exploration, and optimization in computational design workflows.

Climate change and extreme heat are critical issues which have been faced all over the world. Consequently, designers strive to monitor and assess the performance of facades by developing environmental assessments at the early design stages, since design changes do not require many resources. Rapid urban expansion in many parts of the world is leading to increased exposure to extreme natural hazards, exacerbated by climate change. It is essential to come up with strategies for mitigating the vulnerability of the built environment. The concept of thermal resilience and adaptation to climate change have gained ground and international attention in the Architecture, Engineering and Construction (AEC) industry. Resilience is a multi-facet property which defines the vulnerability of the built environment. Although the qualitative assessment of resilience value, the quantification of urban resilience is not yet representative enough and there is a lack of calculating the resilience in the built environment. However, designers are called to develop building and planning proposals with taking into consideration the thermal resilience of buildings against extreme hazards. This thesis aims to fill the gap between the qualitative and quantitative evaluation of thermal resilience in buildings by considering the operational building performance and the thermal performance of the building envelope in case of extreme heat waves. Towards this direction, the most influential parameters of thermal resilience are identified by implementing a sensitivity analysis process, in the first part. Secondly, a quantification method is presented and the thermal resilience performance for buildings in Amsterdam is calculated. Last, this thesis attempts to develop a computational workflow in order to assist designers and engineers in defining the thermal resilience index from the early design stage. Defining a less computational cost and time-consuming workflow is also a goal. Due to time limitations, the multi-facet aspect of resilience and the difficulty of quantification of its indicators, this research focuses on the ex-ante evaluation of the building envelope by identifying its vulnerability to extreme heat waves.

This thesis continues the investigation in the direction of exploring the potential regarding the use of Topology Optimization techniques for the design of cast glass structures. Previous theses in TU Delft have underlined the large potential for the design of these megaliths, but at the same time they have also underlined the strong limitations that derive from the use of commercial software as the tool for it.

The limitations are directly related to the brittle nature of glass which results in significantly different behavior regarding its maximum tensile and compressive allowable limits. This renders fundamental to be able to evaluate both of these criteria during the optimization process. If this is not possible, as it was the case in the previous theses, a secondary (post-processing) phase should be integrated in the process in order to alleviate the peak stresses that may occur in the structure. This increases significantly the time and effort needed for the design and, therefore, it was underlined as an issue to be tackled in further exploration.

This thesis aspires to address this problem with the creation of a customized optimization tool that takes all the structural constraints into consideration and, additionally, integrates the criteria specifically related to the glass manufacturing process, such as the overall annealing time needed. The tool is created in Matlab with the use of Finite Element Method equations in order to develop the structural model. The results of the structural analysis were validated through comparison with results obtained through ANSYS.

The literature review covers a wide scope of topics. Firstly, the glass properties and the casting process are investigated in order to properly indicate the criteria and constraints that arise in every phase. The second part refers to topology optimization. A comparative review of the different algorithmic methodologies is realized and SIMP is selected as the most appropriate for the project. Additionally, the different categories of formulation for the optimization problem – stress, compliance and volume based – are discussed in order to select the appropriate objective and constraints. At the same time, a review of the previous theses is realized in order to indicate which method was used in every case and how the constraints were integrated in the process every time.

In the end two different algorithms are developed based on two different problem formulations; one with compliance objective and a second one with volume objective. The aim is to investigate if the volume objective optimization can be a robust alternative to the classical compliance approach leading to more lightweight structures which at the same time fulfill the criteria regarding their feasibility to be manufactured.

Firstly, the performance of the algorithm in relation to each objective and constraint individually is evaluated though application in a smaller scale benchmark problem. The results showed that all the setups work and, therefore, they can be used for the final design experiments. However, it also indicated that some constraints, such as stress, cannot be applied individually but they always have to be combined with another constraint that guides the optimization in order to lead in a reasonable result. Afterwards, a combination of objective and constraints for each of the two aforementioned formulations – compliance and volume - is tested and applied in the case study example which refers to a slab that serves as a small pedestrian bridge inside the British Museum.

The results validate the estimation that a volume-based problem formulation can offer a robust result which resembles the result obtained from the traditional compliance-based formulation. Moreover, the result in the volume-based case is clearer and sharper and for this reason it was selected in the end for implementation in the final design. The formulation is then used in combination with different glass types, boundary conditions and design domain in order to conclude to the final shape of the slab.