Modelling electric and heat load profiles of non-residential buildings for use in long-term aggregate load forecasts

• Published in: Utilities policy Vol. 58; pp. 63 - 88
• Main Authors: Lindberg, K.B., Bakker, S.J., Sartori, I.
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.06.2019
• Subjects: Electric load; Heat load; Load forecast; Load profiles; Regression model; Statistical analysis; Load profiles; Load forecast; Statistical analysis; Heat load; Regression model; Electric load
• ISSN: 0957-1787; 1878-4356
• DOI: 10.1016/j.jup.2019.03.004

Long-term forecasts of the aggregate electric load profile are crucial for grid investment decisions and energy system planning. With current developments in energy efficiency of new and renovated buildings, and the coupling of heating and electricity demand through heat pumps, the long-term load forecast cannot be based on its historic pattern anymore. This paper presents part of an on-going work aimed at improving forecasts of the electric load profile on a national level, based on a bottom-up approach. The proposed methodology allows to account for energy efficiency measures of buildings and introduction of heat pumps on the aggregated electric load profile. Based on monitored data from over 100 non-residential buildings from all over Norway, with hourly resolution, this paper presents panel data regression models for heat load and electric specific load separately. This distinction is crucial since it allows to consider future energy efficiency measures and substitution of heating technologies. The data set is divided into 7 building types, with two variants: regular and energy efficient. The load is dependent on hour of the day, outer temperature and type of day, such as weekday and weekend. The resulting parameter estimates characterize the energy signature for each building type and variant, normalized per floor area unit (m2). Hence, it is possible to generate load profiles for typical days, weeks and years, and make aggregated load forecasts for a given area, needing only outdoor temperature and floor areas as additional data inputs. •This paper investigates measured load profiles of more than 100 buildings.•A new bottom-up method to forecast hourly heat and electric specific loads per building type is proposed.•Panel data regression models are estimated for 7 different non-residential building types.•The estimated regression models show good significance and fit.•The load profiles can be aggregated for long-term load forecast on a regional or national level.