Diagnostics of indoor environmental quality exposures and climate-infrastructure vulnerability in South African schools

The combined effects of climate change, infrastructure deficits, and energy poverty create a deteriorating feedback loop in South Africa, where poor indoor air quality and thermal conditions in school classrooms compromise the health, comfort, and cognitive performance of learners. This is amplified by a lack of long-term solutions with high-resolution data to quantify the environmental inequity between permanent (brick) and temporary (containerbased and prefabricated) classrooms, as well as a methodological gap in translating such data into actionable, low-cost interventions. These gaps are addressed through the development, deployment, and validation of an integrated, measurement-driven engineering framework that diagnoses climate-infrastructure vulnerabilities and informs evidence-based retrofit policies. A scalable, low-cost Internet of Things (IoT) sensing network was developed and deployed for 24 months across various classroom infrastructures, generating a novel public dataset of concurrent indoor environmental quality (IEQ) measurements. Observational analysis of this dataset quantifies the thermal and ventilation performance of each infrastructure type in relation to international standards. This provides the first quantitative evidence of worse thermal environments, characterised by more extreme temperatures and rapid fluctuations, in temporary infrastructures. In container classrooms, carbon dioxide (CO2) concentrations often exceed the thresholds associated with cognitive impairment, indicating poor ventilation. To diagnose the underlying causes, a novel framework is presented that applies a time-series, distribution-based global sensitivity analysis, specifically the Delta Moment-Independent Measure (DMIM), directly to measured data to derive effective thermal resistance-capacitance (RC) proxies without relying on physics-based parameters. The diagnostic framework successfully characterises brick classrooms as high-R, high-C structures with damped, delayed thermal responses. At the same time, temporary units are identified as low-R, low-C structures that respond sharply to external weather drivers. The framework’s infrastructure diagnostic utility is validated by its ability to characterise improvements in the RC dynamics of an uninsulated container classroom through a low-cost insulation retrofit. An exploratory analysis of household air pollution further demonstrates that thermal vulnerability drives occupant behaviours that degrade indoor air quality across socio-economic contexts. The core contributions of this dissertation form an integrated pathway from evidence to policy. This begins with an empirical foundation, consisting of a novel public dataset and scalable sensing methodology that produced it. An observational analysis of encountered IEQ conditions and their characterising against international benchmarks and metrics for comfort and cognition reports on exposures and how they differ among infrastructure types while accounting for the employed methods of heating, ventilation, and air conditioning (HVAC) in those classrooms and their impact on the indoor environment. A novel and model-agnostic framework for deriving RC proxies from measured data translates indoor temperature measurements into physical and actionable insights, providing information on the time-series influence of different climatic parameters on indoor temperatures, as well as the varying influences by infrastructure type. Finally, this dissertation translates these findings into tangible societal and scientific resources, including infrastructure-specific policy recommendations and a methodological enhancement to open-source scientific software. Through a replicable sensor-to-solution pathway, this dissertation transforms high-resolution data into policy-relevant insights and targeted engineering interventions to protect learner health and potential in resource-constrained environments.

Grantor dc:publisher

Stellenbosch : Stellenbosch University

Year dc:date.issued

2026

Author dc:creator

• Van der Walt, Rita Elise

Advisors dc:contributor.advisor

• Booysen, Marthinus J.
• Grobbelaar, Susara S.
• Rousseau, Pieter G.

Language dc:language.iso

en

Repository record dc:identifier.uri

https://scholar.sun.ac.za/handle/10019.1/135680

OAI identifier oai:identifier

oai:scholar.sun.ac.za:10019.1/135680