Development of a Bayesian calibration framework for archetype-based housing stock models of summer indoor temperature

Abstract

Archetype-based housing stock models of summer indoor temperature can support the development of policies to manage the climate change-driven increase in cooling demand and heat-related health impacts. Calibration can reduce the performance gap of such models, however, work on this topic is limited. Motivated by the growing importance of this underexplored research area, this paper introduces a framework for the Bayesian calibration of archetype-based housing stock models of summer indoor temperature. The framework includes data-driven procedures to classify dwellings into homogeneous groups and specify prior probability distributions. To demonstrate its application, an established bottom-up model based on EnergyPlus was calibrated using data collected from 193 dwellings monitored during the 2009 4M survey in Leicester, England. Post-calibration, the root-mean-square error reduced from 2.5°C to 0.6°C and remaining uncertainties were quantified. The application of this modular framework may be extended to models of energy use and other indoor environmental parameters.