A survey-based approach to estimate residential electricity consumption at municipal level in Germany
In the context of the German Energiewende (energy transition), energy system modelling is used to investigate possible future scenarios of the national energy system. These models depend on regionally disaggregated input data to ade-quately capture interdependencies in the energy system at high resolution. In Germany, official energy consumption statistics are only published at the national (AGEB, 2019a) and federal state (LAK, 2019) levels - there are no official statis-tics on residential electricity consumption with higher regional resolution. So far, energy system modelling has typically relied on specific consumption val-ues or constant per-capita estimators (see Beer (2012) and Hartel et al. (2017)) to approximate residential electricity consumption with a regional resolution be-yond that of federal state. The use of primary data on electricity consumption at household level has so far been limited. Yet, primary data, e.g. from the German Residential Energy Consumption Survey (GRECS), displays heterogeneity in household-level electricity consumption, which cannot be captured by constant per-capita distribution keys. This study aims to investigate whether integrating primary data into quantitative modelling of energy systems contributes to a more realistic representation of regional electricity consumption by accounting for het-erogeneity in household electricity consumption. A synthetic population is generated via the Iterative Proportional Fitting (IPF) al-gorithm based on primary data on household electricity consumption taken from the German Residential Energy Consumption Survey (GRECS), as well as re-gion-specific data at municipal (Gemeindeebene) level taken from the 2011 German census. Total residential electricity consumption at municipal level is then inferred from the synthetic population. Estimates of total residential electricity consumption were derived for 2011 for all municipalities of the German state Rhineland-Palatinate and evaluated against benchmark values from the Net-zentwicklungsplan 2030 (Fraunhofer ISI, 2017) of the same year, which is avail-able at the regional resolution of German urban and rural districts, referred to here as "counties" (Stadt- und Landkreise). The derived estimates achieved an R² of around 0.99 with respect to benchmark values. Overall, the estimates were 6.8% below the benchmark value for Rhine-land-Palatinate. It can be concluded that Iterative Proportional Fitting (IPF) con-stitutes a viable approach to integrate primary data into deriving regional esti-mates of residential electricity consumption at municipal level.