Discussion

A principal limitation of UBEM on the basis of building typologies is that the descriptions and parameters of exemplary buildings originate from a statistical analysis of real-world buildings. The derived informations only represent a simplification of reality and might deviate from the actually constructed buildings. On the other hand, even though there are initiatives to capture a virtual representation of cities, such as CityGML [23], the generated models lack detailed information on construction and building systems for every individual building. The main obstacle is that data are simply not available in a digital format. The application process for construction permits is still mostly paper-based. As a result, gathered informations are not transferred to a digital repository. Data protection laws contribute to this situation. There are two possible remedies. One is used in this study: building typologies. The other possibility requires the full commitment of the local municipality with the purpose to digitalise their archives and conduct building surveys. Although the emerging trend towards building information models (BIM) will facilitate this process, the costs of such undertaking are currently prohibitive. Regarding the definition of refurbishment, the reference studies lack statistical analysis about the extent and effectiveness of refurbishment which limits the comparability of results with refurbished buildings. They were nevertheless included to describe in a more comprehensive way the building stock.
The inclusion of residential buildings with mixed usages (e.g. offices, eateries) is a priority for the future project development. However, overcoming this limitation requires changes in the building modelling approach because multiple thermal zones, at least one for each floor, would be necessary. Another problem is to gather data for validation of mixed usages. The incorporation of mixed usages and non-residential buildings (e.g. schools, retail shops, public buildings) would render UBEM more realistic. The lack of building typologies for non-residential buildings will pose an obstacle. On the other hand, energy saving ordinances and other standards might serve as starting point.
Another worthwhile extension would be the incorporation of behaviour-based load profiles [47]. More heterogeneous occupant behaviour could be simulated and might affect the differences between the simulation results and reference studies for smaller buildings. The simulation results of smaller buildings suffer from the lack of more extreme occupant behaviour, like wealthy residents, and underestimate their energy demand comparing to the reference studies. The reference studies might also be biased towards an over-representation of wealthier residents because their data originate from the issuance of EPCs. The different behaviour profiles would be derived and distributed on the basis of local subgroups with their distinct socio-economic features. Stochastic load profiles proved to be better than standard load profiles [16]. The shading of buildings should also receive attention. Previous studies on UBEM neglected the heat exchange between adjacent buildings. The proposed FMI-based model has the potential to consider these factors in future versions.
Furthermore, the lack of more localized information on the progress of refurbishment, conditioned cellars and roofs affects the capability to assess the environmental impact of the local building stock. More engagement from the local authorities would ameliorate the situation. Although time-series of energy demand of individual buildings are not gathered by public utilities, except in the rare cases of smart-metering projects, they could release district-wise measured data from their supervisory control and data acquisition (SCADA) systems. Such data would help to better understand the temporal distribution of energy demand and offer a better approach for validation. One reason that this has not yet happened might be that public utilities often combine local distribution and generation of energy. Because of European competition law, the release of such data needs to benefit all competitors and not just another subsidiary.