Doctoral Thesis
The Global South (GS) faces a dual challenge: severe housing shortages and heightened vulnerability to climate change. Over 1 billion people face cooling access risks, while more than 2 billion are expected to rely on inefficient cooling devices, driving up energy consumption and carbon emissions. Addressing this issue requires sustainable and equitable climate-resilient solutions for the built environment, particularly in Latin America, where, despite economic progress, significant housing deficits persist, and informal settlements continue to expand. This thesis contributes to this effort by integrating data science techniques—including surrogate modeling, transfer learning, multi-objective optimization, and multi-criteria decision-making methods—into the field of building performance simulation (BPS). The research is structured around four core chapters. The first two are systematic literature reviews that synthesize the state of the art in: (1) the potential of multi-objective optimization based on surrogate models for sustainable building design, and (2) the use of BPS tools for adaptation and mitigation assessments with future weather files. Expanding upon the findings from these chapters, the subsequent two chapters propose a simulation-based assessment focused on Latin American residential buildings. More specifically, the investigation considers the cities of Rio de Janeiro, Sao Paulo, Santiago, Bogota, and Lima. By employing Extreme Gradient Boosting (XGBoost) and Artificial Neural Networks (ANN) as surrogate models and integrating future urban weather files that account for both IPCC emissions scenarios and urban heat island (UHI) effects, this investigation efficiently explores design alternatives. Optimization algorithms, such as the Adaptive Geometry Estimation-based Multi-objective Evolutionary Algorithm (AGE-MOEA) and the Non Dominated Sorting Genetic Algorithm III (NSGA-III), then identify Pareto-optimal solutions that balance thermal comfort, energy demand, carbon emissions, and construction costs. Additionally, multi-criteria decision-making (MCDM) methods facilitate transparent trade-off evaluations. This work bridges architectural research and data science at a macro-level geopolitical scale, demonstrating the feasibility of climate-resilient building design in the Global South. Beyond its methodological contributions, the thesis provides practical guidelines for policymakers, industry professionals, and researchers seeking to develop more sustainable, climate-adapted building policies. Ultimately, it offers a scalable framework for transforming Latin America’s residential sector and encourages further exploration of cost-effective, high-performance building solutions for other regions in need.
THESIS LINK
Date: 28/04/2025
Person
- Alexandre Santana Cruz [author]
- Aline Calazans Marques
- Eduardo Grala da Cunha
- Leopoldo Eurico Gonçalves Bastos [contributor]
- Lucas Rosse Caldas
- Nathan Mendes
ResearchLine
- Architecture, Project and Sustainability
Course
- Doctor of Architecture PROARQ