Assessing daylight in architecture increasingly depends on Radiance-based simulations that, although highly accurate, are too computationally demanding for rapid iteration during early-stage design. As layouts shift, openings are repositioned, or geometries become more complex, the turnaround time of physically based simulation limits the designer’s ability to explore alternatives. Surrogate models offer a promising solution, yet most existing approaches rely on artificial neural networks (ANNs) trained on tightly controlled parametric datasets and evaluated only within their training distribution. As a result, their reliability for real, highly variable design geometries remains uncertain.

This thesis investigates graph neural networks (GNNs) as a new surrogate modelling paradigm for daylight prediction. Unlike vector-based ANNs, GNNs represent spatial relationships explicitly, allowing the model to “reason” about how light propagates through space rather than inferring patterns solely from coordinate inputs. To the best of the author’s knowledge, this is the first formulation of daylight factor (DF) prediction as a graph-learning task, representing rooms as heterogeneous graphs in which window and sensor nodes exchange geometric and photometric information such as distance, direction, and relative orientation.

A deliberately restricted training setup, limited to simple square rooms, is used to test whether GNNs can generalise beyond the conditions they are trained on. The models are then evaluated on a wide range of unseen geometries, including elongated, asymmetric, rotated, and self-occluding layouts, alongside ANN baselines from the daylight literature. The proposed model, WindowGraphNet, consistently preserves the spatial structure of DF distributions across these unseen cases, whereas coordinate-anchored ANNs often fail when room form or orientation deviates from their training domain. The GNN’s main limitation occurs in deeply recessed or fully occluded areas, where long light paths fall beyond
the model’s message-passing range, leading to systematic overestimation of low-illumination regions.

These findings demonstrate that generalisation in surrogate daylight modelling is governed primarily by the representation: embedding relational and geometric structure enables robustness that conventional ANNs cannot achieve through data alone. WindowGraphNet therefore establishes a new methodological foundation for surrogate daylight tools, with an extensible graph structure that can be expanded to incorporate surface nodes, external obstructions, or climate-based inputs. With further training on
diverse datasets, such graph-based surrogates have the potential to deliver fast, geometry-aware daylight feedback directly within architectural design environments, supporting more informed and iterative decision-making in early-stage design

This study investigates the development of sustainable offshore fish farms in Sisal, Yucatán. Local fishermen face seasonal restrictions on fishing due to environmental regulations. Given the socioeconomic dependence of the region on fishing, the community of Sisal has been experiencing increasing instability of livelihoods. Offshore fish farming has emerged as a potential solution to this challenge, offering an alternative income source outside the traditional fishing season. However, previous industrial attempts to introduce fish cages failed due to a lack of local engagement and inadequate design, leaving Sisal residents sceptical. To address these past issues, this research seeks to design affordable, durable, and locally accepted fish cages that meet the unique environmental and social conditions of Sisal. Valuable insights were gained from fish farms in Celestún, a nearby village with successful communityled offshore aquaculture. Celestún’s approach, using smaller, manageable, and collectively funded cage, has proven to be both economically and socially beneficial. This makes it a relevant model for Sisal, though Sisal’s steeper coastal gradients and greater exposure to maritime forces require adaptations to ensure durability and long-term success. The research follows a multi-step methodology, beginning with interviews with local fishermen and experts to understand their needs and preferences for cage design and placement. These insights were integrated with environmental data on wave height, wave period, and current speeds collected through field measurements and the ERA5 reanalysis dataset. Using this input, an Multi-Criteria-Analysis (MCA) was conducted to determine the optimal offshore location for the fish farms. To determine the structural needs for fish cages under Sisal’s conditions, the research used ProteusDS simulation software [14] to model various cage dimensions, mooring tensions, and layout configurations. Key findings indicate that positioning the fish farms at 8 kilometres offshore is optimal for long term success. At closer distances to the coast, water quality decreases, resulting in higher maintenance requirements and compromised fish health. Greater distances increase installation costs and operational costs due to higher fuel demands. With the optimal location established, the research follows with the determination of key design parameters essential for the structural integrity of the fish cages near Sisal. An extreme value analysis of an ERA5 dataset was performed to estimate the 20-year return level for the wave height, resulting in a design wave height of 4.19 metres. This value was adjusted for local conditions using a scaling factor derived from the comparison between local and ERA5 data, resulting in an adjusted design wave height of 3.40 metres. A power-law regression was then applied to establish the relationship between wave height and wave period, estimating a design wave period of approximately 8.01 seconds corresponding to the adjusted wave height. For the current analysis, the 95th percentile of current speeds was examined, determining a maximum design current speed of 0.50 m/s near the surface. Furthermore, analysis of wave and current directions revealed that extreme waves predominantly come from the north to north-east directions (340° to 20°), while the strongest currents flow toward 70° and 250°, indicating eastward and westward flows. The optimal cage design determined through simulations includes a cage diameter of 12 metres and a net depth of 4.7 metres to withstand Sisal’s environmental forces. Additionally, distinct mooring-tension configurations were tested in the ProteusDS software, including Concept 1 (a single-cage setup), Concept 2 (a two-cage configuration with four mooring anchors), and Concept 3 (a three-cage arrangement with three anchors). Each concept required specific anchor weights and dimensions to endure the high wave and current forces at this location. Orientation adjustments were also incorporated to reduce tension, aligning each cage setup with different wave and current directions, thereby optimizing structural reliability. Future fish cage designs should include adaptive anchoring and precise orientation to enhance stability and involve the local community for sustainable, long-term success. The study concludes that, to achieve long-term viability, fish cages in Sisal must be affordable, easy to maintain, and capable of withstanding local environmental conditions. Future recommendations include deepening community involvement, implementing enhanced safety and resilience measures, and refining cost analysis to foster broad acceptance among local fishermen. By ensuring that the fish cages are both economically viable and environmentally sustainable, this project aims to secure a stable income for Sisal’s fishing community, thereby improving their quality of life while reducing pressure on marine ecosystems