Sensitivity analysis for minimization of input data for urban scale heat demand forecasting

In the paper a methodology based on 3D building and city models is presented to calculate urban heat demand for different refurbishment scenarios. This methodology is validated on six case study buildings of the City of Essen in Germany. The influences of the availability and quality of data input regarding the geometrical and physical parameters on the accuracy of simulation models are analysed. Different CityGML Levels of Details (LoDs) of the building models as well as different sources of the physical parameters are tested in order to investigate the uncertainty of the methods used. In the first step, a semi-automated method with data pre-processing (FME software,) as well as simulation of the heat demand (INSEL8 software) is used. The results are compared with a fully-automated method implemented in the urban simulation platform SimStadt, whose development is ongoing in the project with the same name (www.simstadt.eu). This platform has a special module integrated, which allows an automatic data pre-processing. Both methods calculate heat demand based on monthly energy balance (standardised in Germany with the DIN V 18599, or in Europe with the ISO 13790). The calculation of the heat demand with different accuracy of the data input enables to make a statement about which parameters have the most influence on the results. Considering the difficulties in obtaining data available at a city scale this information is very useful for future reductions of the effort of data capturing. For example, the analysis showed that the geometrical Level of Detail can give up to 12% of error depending which of the LoDs are available for the analysed building. In the next stage, the methods tested first on the six case study buildings can be extrapolated for the whole City of Essen. This methodology could be even extended to other cities on condition that they have 3D city models available.