Impact of building geometry description within district energy simulations

An accurate quantification of the energy demand of a district serves as an important boundary condition for a correct assessment of the feasibility of district heating and cooling systems. In order to quantify the district’s energy demand, a considerable number of input parameters is required for each building. These include geometry and location, building envelope characteristics, heating, ventilation and air conditioning systems, renewable energy systems, building appliances and occupant behaviour. Due to the scarcity of input sources and the lack of an efficient methodology to process the input parameters on district level, current district energy simulations are often carried out using an archetype approach. In this approach only a limited set of representative buildings – archetypes – is used to model the district’s building stock. However, these archetype approaches fail to include the unneglectable variability in type and size of buildings that is characteristic for the existing building stock. Especially, on a smaller scale (~100 dwellings) the use of archetypes may therefore no longer be representative. The focus of this paper is on the influence of the geometrical representation of single-family dwellings within these district energy simulations. Using recent developments in geographical information systems (GIS), the variability in geometry can be introduced more accurately in district energy simulations. Nonetheless, detailed geometry descriptions – in the CityGML standard – are not yet available for all locations. Ideally, each building is characterised by its particular shape and defined in terms of surfaces that represent roof, ground floor or façade or parts of these. This corresponds to models with level of detail 2 (LOD2). In Flanders, GIS data contain a precise two-dimensional (2D) geometrical representation of the territory, including building footprints amongst others. These 2D files are combined with LiDAR data to derive the building ridge height and are then used to generate LOD1 CityGML models, in which buildings are represented by extruded volumes. However, the extrusion of the building’s footprint to its ridge height causes a significant overestimation of the building volume and its total envelope surface area. Within this context, this research quantifies the differences between four LOD1 models and a LOD2 reference model, as the LOD1 models fail to include particular roof shapes and usually also building extensions. Furthermore is analysed how LOD1 models could be enhanced in order to obtain a more accurate geometrical representation of the considered district. To establish this analysis, the traditional LOD1 models are first compared extensively to the LOD2 model through the evaluation of the ground floor area, the roof area, the external wall area, the total loss surface area and the building volume as geometrical key performance indicators (KPIs) as well as through the assessment of the peak power, the total energy use and overheating as energetic KPIs. FME is used to setup the CityGML models, TEASER to translate these CityGML models into IDEAS Modelica models and OpenIDEAS to run the district energy simulations in Dymola. Finally, this research presents guidelines for the generation of improved geometrical representations of single family dwellings within district energy models. To which extent should building models include roof shape and/or building extensions? Is e.g. a building model that distinguishes extensions from main volume but characterizes them both by a flat roof sufficient?status: publishe