An archetype-in-neighbourhood framework for modelling cooling energy demand of a city's housing stock

As hot days are getting hotter and more frequent, urban dwelling is expected to increase cooling energy use in current and future climate. The applicability of dynamic building simulation in estimating cooling loads of a city’s housing stock can be limited due to lack of fine-grained on-site current and future weather inputs. For predicative modelling of residential cooling energy demand to aid a city's energy supply planning resilient to excessive heat conditions, it requires cooling energy demand projection based on a relational account of (1) the thermal-environmental interaction between housing stocks and urban microclimate conditions, (2) the city dwellers' cooling energy use behaviour, and (3) the city’s climate projections. In this paper, we introduce an ‘archetype-in-neighbourhood’ framework to meet these requirements. Combining empirical urban data modelling and EngeryPlus model calibration, this framework was developed to obtain statistically a maximal cooling energy demand model of a city’s housing stock during yearly hottest periods. We applied the framework to multiple datasets selected from Seoul’s open urban data sources for the period of 2014-2017 (2014 being the earliest year of data availability, 2017 being the end of the study period), including metered electricity use data of 659 apartment buildings (51,351 households) sampled from 18 city districts. The results show that maximal month cooling energy demand (MMCD, kWh/m2) of Seoul’s housing stock can be expressed as a regression function of two determinants: (1) the city's average outdoor temperature during the hottest month period (Tex, ˚C), and (2) estimated indoor cooling temperature set-point (Tin, ˚C) of the city’ housing stock during the same period. Through a k-fold (k=4) validation, the current regression model (2014-17) was evaluated to have an overall coefficient of determination R2=.969. Assuming no housing stock renovation, we applied the model to generate scenarios of maximal month cooling demand in future years according to some of the highest summer temperatures projected for Seoul (RCP8.5 2045, RCP4.5 2047, MM5 2071-2100). We conclude this paper with a brief discussion of the implication for cooling energy supply planning and further work to extend the applicability of this new framework to housing stock adaptation planning and design.