Energy performance of buildings: A statistical approach to marry calculated demand and measured consumption

In public debate, Energy Performance Certificates (EPCs) of buildings have been criticised for not reflecting the energy demand realistically. And indeed, measurement, as in energy bills, usually differs from the calculation, in particular, when simplified energy performance calculation models and standard specifications are applied, as in EPCs. Thus, energy-saving potentials of refurbishment recommendations and their cost-effectiveness tend to be over-estimated. Of course, this is not desirable. These effects were analysed in two sets of data, the Energy Performance Certificate Register for residential buildings in Luxemburg, run by the Luxemburg Ministry of the Economy (Lichtmeß, 2012) and a database gathered in the research project “Teilenergiekennwerte von Nichtwohngebäuden (TEK)” (Hörner et al., 2014a) funded by the German Federal Ministry of Economic Affairs and Energy. Multiple linear regression and error calculus were applied to study the gap between measurement and various calculation models in detail. A statistical procedure is proposed to estimate expectation value and variance of the future energy consumption of buildings in case of refurbishment, as a supplement to standard calculations in EPCs for example. Prerequisite is that for a sufficient number of buildings, data on both, measured energy consumption and calculated demand, are available.