Thermal Resilience During Prolonged Cooling Outages: A Multi-Zone Field Study of Disruption and Recovery in Arizona Apartments

Saroj Lamichhane

Increasing frequency and severity of extreme heat events, and growing dependence on air-conditioning systems emphasize the need to evaluate thermal resiliency of buildings during cooling outages. Existing assessments often emphasize isolated metrics or steady state conditions and may not adequately characterize cumulative thermal exposure, thermal buffering, and post outage recovery. This study evaluated high-resolution, multi-zone field measured data from two apartments in Arizona during cooling outages lasting up to 78 hours and subsequent recovery once cooling was restored. Cumulative thermal exposure was quantified using continuous degree-hours above 82°F, and delay component by time required to reach 85°F. Results showed that, immediately after the cooling outage, all the zones exhibited exponential behavior in temperature rise rate followed by harmonic model with a linear drift after the transient hour. Transient hours are determined to be 17 hours and 21.5 hours for 1-bedroom and 2-bedroom apartment respectively. The exponential model exhibited strong capability across all zones in both apartments with R2 values ranging from 0.8943 to 0.9833. The sinusoidal regression model exhibited strong capability across monitored zones with R2 values ranging from 0.656 to 0.975. Significant spatial variability was observed between zones and within zones. At the apartment level, the sinusoidal model produced R2 of 0.879 0.893, RMSE 0f 0.56°F and 0.44°F, and MAE of 0.48°F and 0.32°F for 1-bedroom and 2-bedroom apartment respectively. Both the apartments crossed 90°F threshold in at least one zone.  1-bedroom apartment accumulated more degree hours and recovered slightly slower than 2-bedroom apartment. Recovery scores are comparatively uniform across zones. Sensitivity analysis showed that weightage assigned in three different metrics didn’t change the ranking among zones and were low to moderately sensitive. The findings demonstrate that apartment average metrics can conceal important zone level vulnerability. The proposed indices should be considered exploratory and require further validation.