Exploring indoor thermal comfort and its causes and consequences amid heatwaves in a Southern European city—An unsupervised learning approach

Abstract

This study investigates indoor thermal comfort during heatwaves in dwellings of the Southern European city of Pamplona, Spain. Utilizing K-means and Hierarchical clustering, it explores clustering patterns from occupants' survey responses (n = 189) to thermal comfort-related questions (i.e. day and night thermal sensation, thermal satisfaction and thermal preference) as well as causal links (i.e. indoor temperatures, building/occupant features) and consequences (i.e. sleep quality, heat-related symptoms) of such clusterings. Both unsupervised learning techniques coherently revealed two groups: the comfortable and uncomfortable clusters. Uncomfortable occupants coherently experience more sensation to heat, greater preference for cooler temperatures, and more thermal dissatisfaction, especially during daytime hours. Dwellings of comfortable occupants experience median indoor temperatures ranging 25.7–26 °C; dwellings of uncomfortable occupants 27.4 °C, with median temperatures above 28 °C during 15:00–23:00 and 23:00–07:00 periods. Discomfort or overheating—coherently expressed by the thermally uncomfortable cluster—is alleviated by multiple factors related to the presence of active cooling technologies in all rooms, and use of passive and low-energy cooling measures (e,g. fans); exacerbated by heatwave conditions. As coherently expressed by the uncomfortable cluster heat worsens the sleep quality of occupants (3 to 6-fold) and increases the likelihood of occupants to experience heat-related symptoms (10–19-fold). This study is particularly important to policymakers, as it sheds light, from dwellers’ first-hand experience in a Southern Europe city, on relevant factors that should be taken in consideration to allow them to cope better with heatwaves without compromising their comfort and health.