Data-driven characterization of cooling needs in a portfolio of co-located commercial buildings


Commercial buildings have growing cooling demands. Greater energy efficiency and flexibility are needed, especially in existing buildings that have a slow stock-turnover. We collect, analyze, and release a dataset on chilled water use from 119 co-located buildings in a warm-summer Mediterranean climate. Factoring out geography-driven differences, we observe a strong heterogeneity within and across different building types. The average estimated base cooling intensity at 18 degrees C varies from 0.50 to 4.4 MJ/m2/day across buildings, with the highest loads in healthcare and the lowest in residences. We find that simple, interpretable regressions can be used to model cooling load and provide data-driven benchmarks for comparing building performance. Over five years, these regressions explain over 70% of variance for buildings that collectively represent 85-94% of the portfolio’s overall cooling load. Consumption increases by 7.6-9.8% for every 1 degree C increase in mean daily outside temperature and drops on weekends, by up to 27% in offices.