PhD Oral Exam - Mostafa Saad, Building Engineering

When studying for a doctoral degree (PhD), candidates submit a thesis that provides a critical review of the current state of knowledge of the thesis subject as well as the student’s own contributions to the subject. The distinguishing criterion of doctoral graduate research is a significant and original contribution to knowledge.

Once accepted, the candidate presents the thesis orally. This oral exam is open to the public.

Abstract

Decarbonizing the built environment is a fundamental strategy for mitigating climate change, given the sector’s substantial contribution to global greenhouse gas emissions. Efforts to create low-carbon urban areas rely on reducing buildings’ carbon footprints, making energy-efficient retrofits an essential intervention. Building retrofit complexities are often present with multiple and contradicting objectives, requiring extensive computational simulations to evaluate a wide array of potential configurations.
Moreover, a holistic assessment of retrofit measures necessitates the incorporation of diverse and heterogeneous measures, which may expand the design space exponentially, resulting in exhaustive simulation approaches become impractical from both high computational and time aspects. This challenge is further compounded when addressing multi-building portfolios or projecting performance under future weather conditions, as each additional layer of complexity compounds the required computational effort. Consequently, the inherent complexity of building retrofit decision-making obstructs stakeholders’ ability to balance trade-offs and competing objectives. In this context, surrogate modelling emerges as a promising approach that imitates complex physics-based models, offering greater computational efficiency.

The principal objective of this dissertation is to investigate and propose an alternative modelling approach to conventional physics-based retrofit methodologies, with the aim of facilitating comprehensive building retrofit design and performance assessment. The dissertation tackles the following goals: 1) First, a building retrofit framework is established for a specific building typology in the Montreal region, which is followed by exploring the effectiveness of developing a set of surrogate models designed to replace physics-based models to assess the building retrofit modelling problem while ensuring a high degree of comprehensiveness. A multi-step methodology is developed to achieve careful feature engineering, a balance between accuracy and computational efficiency, and generalizability.

2) Secondly, an integration methodology for predicting building performance under future climate is introduced. In this process alternative feature engineering methods are compared and used to extract and identify robust predictive features, enriching the surrogate models’ capacity to estimate future building performance.

3) Given the prevalence of data scarcity in built environment research, addressing the accuracy of open-access data, archetypal data, and varying model complexities in representing existing buildings becomes evidently crucial. Such evaluations are essential for enabling surrogate model generalizability when data availability is limited. This goal accomplishes detailed accuracy analysis derived from multiple thermal zoning configurations, data sources, and modelling pipelines.

4) The final goal advances the scalability approach by proposing and evaluating a set of methodological workflows for integrating the optimal modelling pipeline with archetypal and geospatial data. These workflows facilitate greater generalizability and scalability, offering insights into the potential and limitations of a surrogate-based modelling strategy for multi-building retrofit intervention design.

The primary findings highlight the effectiveness of using a surrogate-based approach to assess building performance in retrofit scenarios. The enhanced computational and time efficiencies enable the deployment of low-cost, accelerated models that achieve acceptable precision, supporting decision-making in building transformations. Additionally, the research identifies prospective workflows that extend model predictions to future climate conditions and multibuilding contexts. Overall, this thesis introduces a layered methodological framework to improve building performance assessment using surrogate-based methods, emphasizing the identification of near optimal building retrofit configurations in multi-objective design contexts. By offering a computationally efficient alternative to traditional physics-based simulations, the proposed approach aims to facilitate informed and timely decisions for stakeholders involved in the transformation of the built environment.