Overheating risk reduction in UK dwellings.

Alrasheed, Mousa 2024. Overheating risk reduction in UK dwellings. PhD Thesis, Cardiff University. Item availability restricted.

Preview	PDF (Thesis) - Accepted Post-Print Version Download (7MB) | Preview
	PDF (Cardiff University Electronic Publication Form) Restricted to Repository staff only Download (93kB)

Abstract

Rising global temperatures are driving more frequent heatwaves in the UK, increasing overheating risk in the housing stock and compromising the health and comfort of occupants, particularly the elderly. Given the diversity and scale of the stock, assessing individual buildings for overheating risk is impractical. Instead, representative archetypes provide a scalable approach to capturing key dwelling variations, enabling large-scale investigations that are essential for understanding overheating patterns and developing effective mitigation strategies. However, the influence of methodological choices on archetype representativeness remains under-explored. To address this, a minimum segmentation frequency (MSF) approach was introduced to preserve feature diversity. A sensitivity analysis was conducted on archetype representativeness to investigate the influence of different MSF levels, variable counts and clustering metrics. Results showed that lower MSF values improved representativeness, and the choice of clustering metric impacted the optimal number of archetypes. The Davies-Bouldin index consistently identified more representative archetypes than the Calinski-Harabasz and Silhouette indices. Subsequently, a framework for archetype development was established, integrating geographical and temporal scales, computational cost and research focus to balance representativeness and simulation feasibility. Using the framework, building archetypes derived from English Housing Survey (EHS) data were developed to analyse overheating risk across regions and dwelling types. The developed archetypes were evaluated through dynamic thermal simulations, revealing consistent overheating patterns that align with observed overheating trends, demonstrating both the cooling potential of a passive measure (e.g., external shutters) and the overheating severity for different typologies. Simulated internal temperatures reflected monitoring studies, reinforcing their reliability. A Random Forest model demonstrated that the data variation within the developed archetypes is sufficient to reliably identify key drivers of overheating. The model demonstrated strong predictive iii performance, with R² values ranging from 0.79 to 1 for living rooms and 0.79 to 0.96 for bedrooms across both baseline and 2050 climate scenarios. The findings confirm the utility of the archetypes, derived from the suggested framework, for large-scale overheating assessments, providing a foundation for future research, policy and adaptive cooling strategies in a changing climate.

Item Type:	Thesis (PhD)
Date Type:	Completion
Status:	Unpublished
Schools:	Schools > Engineering
Uncontrolled Keywords:	1) Overheating 2) Passive cooling 3) Archetype 4) Building stock modelling 5) Minimum segmentation frequency 6) Representativeness
Date of First Compliant Deposit:	7 March 2025
Last Modified:	07 Mar 2025 11:44
URI:	https://orca.cardiff.ac.uk/id/eprint/176694

Actions (repository staff only)

Edit Item