Introduction

In response to climate change, the German Federal Government announced in 2010 a shift in the German energy policy: "Energiewende" (German for energy transition). This policy programme aims at replacing fossil fuels with renewable energy sources by 2050 [56,25]. In the aftermath of the Fukushima nuclear accident, the scope of the transition was broadened due to the abandonment of nuclear energy as bridging technology.
The housing sector plays a key role because considerable significance is attributed to the improvement of energy efficiency and the housing sector still has potential in this regard [65]. Therefore, the German state-owned development bank KfW launched an energy-efficient urban redevelopment programme to support integrated municipal strategies for improving energy efficiency and expanding the utilization of distributed energy sources [61,33]. The process of improving the sustainability of cities not only requires political will and financial resources, but also a profound understanding of the spatial and temporal patterns caused by energy demand and supply [21,4,14]. The intermittent nature of renewable energy sources complicates the process. The misalignment between the varying supply of renewable energy (e.g. wind and solar) and energy demand poses a daunting challenge [63].
The research to understand the spatial and temporal patterns of weather and its impact on renewable energy supply continues to receive strong support. But the planning of a sustainable energy system also requires that the demand side receives more attention with the aim to improve the coverage of energy demand with renewable energy supply [66,34]. Additional efforts have to be invested in tools capable to comprehend the dynamics of energy demand of each building. Although statistical urban energy models [29,30] are robust and accurately estimate energy use on the basis of the demographic and economic developments of the entire area, i.e., top-down approach, or on the basis of vintage and function of the individual buildings in the area, i.e. bottom-up approach [46], these models depend on measured data. Their applications in the assessment of changing conditions, such as policies, occupant behaviours etc., are thus limited. Statistical bottom-up approaches have the additional constrain of depending on typically quarterly or annually collected data [67,2]. This data basis restricts their applicability in the prediction of energy demand of individual buildings with a daily or hourly time step [13]. The relevance of this capacity lies in the evaluation of the outcome which individual or combined measures in the field of energy efficiency, storage and demand response have on the satisfaction of energy demand with distributed and renewable energy supply [50]. Because data availability is a widespread concern, statistical models offer little recourse, in particular for higher spatial resolution.
Urban building energy modelling (UBEM) contributes to address these challenges [36]. UBEM applies engineering methods in a bottom-up approach to perform building energy simulations for every building in the area of interest. The engineering methods refer to thermal models which are derived from mass and energy conservation laws. They form the underpinning of building energy models (BEM), which are used in the design and certification processes for high-efficiency buildings. The understanding of the energy demand patterns with an high spatial and temporal resolution under varying conditions (e.g. weather and behaviour of occupants) has the potential to quantify the impacts of new technologies and policies. In the short-term, a detailed demand assessment can inform demand response strategies and/or the need for energy storage in order to satisfy the demand with renewable supply. UBEM can furnish valuable insights about robust and long-term solutions in designing and sizing a distributed and renewable energy system [55], that must be able to satisfy peak loads. The uncertainties of climate change exacerbates the need to understand the consequences of different scenarios.
The prolificacy of this field underlines the emerging importance of such modelling frameworks. A common implementation of UBEMs relies on the building energy simulation engine EnergyPlus (EPlus) [10], which is developed by the US Department of Energy: Urban Modelling Interface (UMI) [49]/Boston UBEM [11] and City Building Energy Saver (CityBES) [26]. Other implementations apply simplified thermal models: CitySim [52], SimStadt [42] and City Energy Analyst (CEA) [17]. Furthermore, various literature reviews are available [50,36,58,27] and emphasize the importance to overcome the still existing shortcomings before general adoption can be considered.
Related to this field, but relying on an alternative, interesting approach, the load profile generator [48,47] is derived from a psychological model to produce detailed load profiles for different household types, which can be aggregated to neighbourhoods and municipalities. This behaviour-based modelling would be a suitable extension for UBEMs to mitigate the uncertainty originating from occupant behaviour [43,28].
Many existing implementations of UBEMs are theoretically capable to benefit from 3D city models with a Level-of-Detail 2 (LoD) [23] because they rely on CityGML [23], which is an open data model and XML-based format for the representation of virtual 3D city models. At LoD2, the buildings are three-dimensional, but their roofs are modelled with predefined shapes and not completely authentic. In practice, the 3D city models used in the corresponding studies, however, remain at LoD1 where the 3D model of a building is extruded from its footprint and its roof is consequently flat. Thus far, previous studies preferred simple 3D models or publicly available CityGML models. The reason is that the derivation of 3D city models is laborious. Further research efforts should focus to remedy this shortcoming with the intention to achieve higher fidelity to reality. This is important because roof sections are often completely or at least partially heated in Europe [40]. Regarding simplified thermal models, they lack the sophistication and versatility of EPlus. On other hand, EPlus-based UBEMs are constrained by labour-intensive approaches in the modelling of buildings and rigid simulation requirements. For instance, simulating different scenarios involves the tedious process of altering text files which serve as input. The weather conditions also needs to be defined in an external text file for an entire year. The reliance on proprietary software and data in some UBEMs is another factor that hinders wide-spread adoption due to license fees, lower degree of interoperability and constraints in usage [62].
To support wide-spread adaptation of UBEMs and advance their development, the proposed approach extends UBEM with Functional Mockup Interface (FMI). FMI is an open-source standard coupling independent models and allowing the exchange of data. This extension yields interactive capacities because the model can be supplied during the simulation with "live" weather and occupancy data or other scenario data. Moreover, the parameters of every surface in a building model can be individually modified (e.g. temperature) as well as their states can be retrieved (e.g. heat loss). Such interactive capacities can particularly benefit urban microclimate analysis and help to merge UBEM with their climate models. To the knowledge of the authors, this is the first UBEM study to derive detailed 3D models of buildings from a point cloud.
This study focuses on the detailed description of a FMI-based UBEM, which is referred as UBE-FMI, its comparison to reference studies and its potential application. The objectives are:

• To provide a FMI-based urban building energy model with the interactive capacity to change variables of individual model parts during the simulation. The impact of changing conditions (e.g. weather and occupancy) on the energy demand can be forecasted and peak loads can be identified with less effort.
• To derive and visualize a three-dimensional city model with LoD3 from a point cloud by exclusively utilizing public data and open-source software.
• To derive and visualize a three-dimensional city model with LoD3 from a point cloud by exclusively utilizing public data and open-source software.

As stated by Li et al.[36], the prevalence of using Typical Meteorological Year (TMY) in UBEM simulations underrates the potential benefits for operational applications in municipalities and their utilities. Because actual year-to-year weather data can considerably vary from one another and from TMY, UBE-FMI was conceived with the idea of facilitating scenario analysis of extreme and untypical climate conditions. UBE-FMI is also well suited for the analysis of urban heat islands for the reason that its FMI-based implementation can exchange data during the simulation between individual model parts like heat exchange between specific building surfaces and their immediate surrounding. This implementation remedies the rigidity of importing weather data from external text files (e.g. EPW file), which is a common feature in UBEM. UBE-FMI is a further step in overcoming the limitations mentioned in Davila et al. [11]: handling large amounts of data supplied by Light Detection and Ranging (LiDAR) to derive detailed building geometry; defining building parameters on construction, ventilation, heating and warm water systems. In this study, a scenario for a future low-carbon energy system in the German city of Wuppertal is constructed where the heating system is electrified, i.e., heat pumps, and a demand response strategy is introduced. The climate conditions are defined by building-specific microclimate data. The impact on energy demand and the potential energy savings are evaluated. The scenario highlights the functional enhancement of UBEM through dynamically alterable parameters in the building models due to the FMI-based implementation, and thus the potential extension of its application to operational planning in public utilities. An additional merit is the attempt to foster environmental awareness of stakeholders and the general public by having produced a web-interface [31], which is called Wuppertal WorldWind Environmental Monitor (WupperWWEM), that indicates the impact of residential housing under "live" weather conditions. The long-term intention with the development of UBE-FMI is to address urban microclimate and the optimisation of energy saving measures.