A systematic approach for data analysis and prediction methods for annual energy profiles: An example for school buildings in Norway

Current research on energy supply systems and building energy demand presents positive impacts from the two sides, with potentials of combining top-down and bottom-up modelling. Mostly, the energy demand input has been employed directly from energy utility companies as a package of information, without considering energy use patterns regarding building type. There lacks a bridge between demand profiles on building stock functions and urban energy supply systems. Accordingly, this article proposes a framework that enables the prediction of annual energy profiles, applied to one educational building type on an hourly basis. The work consists of five steps: (1) getting energy information of 40 district heating (DH) supplied schools in Norway, (2, 3) processing data for getting the modified average hourly demand per m(2) and holiday breakpoints through a modified Z-Score, (4) energy forecast of DH and electricity load profiles through temperature moving average, correlation, and linear regression analysis, (5) validation of the predicted yearly profiles by three criteria, and further with the cluster methods for DH profiles. The results showed that the suggested methods for annual energy forecast were satisfying. The defined load profiles might represent the current energy demand of the Nordic school and the methods could be transferred to other building types. With energy analysis of typical building types, the proposed method enables the planners to better understand the energy needs for different building functions. (C) 2021 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).