Assessment of the photovoltaic potential at urban level based on 3D city models: A case study and new methodological approach

1 Assessment of the photovoltaic potential at urban level based on 3D city models: A case study and new methodological approach Laura Romero Rodríguez P a,* P, Eric DuminilP b P, José Sánchez RamosP a P, Ursula EickerP b P. P a P Grupo de Termotecnia, Escuela Superior de Ingenieros, Universidad de Sevilla. Camino de los Descubrimientos S/N, 41092 Sevilla, Spain. P b PResearch Center for Sustainable Energy Technologies, Stuttgart University of Applied Sciences. Schellingstr. 24, 70174 Stuttgart, Germany. * Corresponding author: Laura Romero Rodríguez. E-mail: 12TU[email protected] ABSTRACT The use of 3D city models combined with simulation functionalities allows to quantify energy demand and renewable generation for a very large set of buildings. The scope of this paper is to determine the solar photovoltaic potential at an urban and regional scale using CityGML geometry descriptions of every building. An innovative urban simulation platform is used to calculate the PV potential of the Ludwigsburg County in south-west Germany, in which every building was simulated by using 3D city models. Both technical and economic potential (considering roof area and insolation thresholds) are investigated, as well as two different PV efficiency scenarios. In this way, it was possible to determine the fraction of the electricity demand that can be covered in each municipality and the whole region, deciding the best strategy, the profitability of the investments and determining optimal locations. Additionally, another important contribution is a literature review regarding the different methods of PV potential estimation and the available roof area reduction coefficients. An economic analysis and emission assessment has also been developed. The results of the study show that it is possible to achieve high annual rates of covered electricity demand in several municipalities for some of the considered scenarios, reaching even more than 100% in some cases. The use of all available roof space (technical potential) could cover 77 % of the region’s electricity consumption and 56% as an economic potential with only high irradiance roofs considered. The proposed methodological approach should contribute valuably in helping policy-making processes and communicating the advantages of distributed generation and PV systems in buildings to regulators, researchers and the general public. UKeywords Urban energy consumption; PV potential; Urban solar potential; Roof-top photovoltaic systems; Distributed Generation; 3D city models. 2 1. Introduction It is an undeniable fact that our present living standard strongly depends on electricity and other forms of energy. Urbanization has led to a high increase in energy use, with buildings being one of its largest contributors and playing a significant role on climate change. As part of the sustainability strategy in Europe, the Energy Performance of Buildings Directive (EU, 2010) and others such as the Renewable Energy Directive (EU, 2009) have defined a package of measures that sets the path for notable and long term improvements in the energy performance of Europe׳s building stock. Some examples are the introduction of Nearly Zero Energy Buildings (NZEB) or the obligation to utilize on-site renewable energy. In addition, the tendency of new regulations is to extend the system boundaries from a single building to the urban area, allowing the interaction between different energy flows. The new concept of distributed energy generation is becoming increasingly important, with the effect that the distribution network is evolving from a once passive power-consuming to an active power-generating part of the electric power system (Srećković et al., 2016). Among the different widespread distributed energy applications, there is a growing consensus that the deployment of photovoltaic (PV) systems in buildings is an attractive option. Analyses have shown that about 60% of the roof area in Europe is suitable for solar technologies (IEA, 2002; Weiss et al., 2010), which could be solar thermal (SRTHR) or photovoltaics. In this work the focus is on solar photovoltaics. However, in spite of the fact that the advantages for individual buildings have been studied, there is little understanding of the potential benefits of an urban scale implementation of such systems (Jo and Otanicar, 2011). Electricity production by PV is growing world-wide and grid-parity is a reality in many places, even in low irradiance countries such as Sweden (Molin et al., 2016). Solar radiation is a clean and abundant source of energy and PV is expected to contribute even more significantly in the future, since rooftops provide large areas suitable for solar energy exploitation. However, unlike the non-urban environment with little constraints to energy production, buildings have limitations on the available area, and many factors have to be considered such as construction restrictions or obstructions due to the surroundings. The better the knowledge about the PV potential and investment cost of a region, the easier it is to help policy-making processes, prevent future disparities between supply and demand, and communicate the advantages of building integrated systems to the general public (Freitas et al., 2015). Therefore, the first step for this approach is an analysis to determine the solar potential of regions, which might be a challenging task due to the complexity of the urban environment. Although a lot of research has been presented to measure the PV potential of buildings and plenty of studies have focused on the improvement of solar assessment by developing software and algorithms, 3D city models have not been made available in public domain on a full-scale yet. In order to estimate the PV potential, different approaches are applied, from simple estimations to airborne LiDAR (Light Detection and Ranging) technologies (Horváth et al., 2016). Depending on the scale and the level of detail required, some methodologies will be more appropriate than others. 3 In the last decade Germany has experienced a massive increase in constructed PV plants. However, only a small fraction of the installed capacity is integrated within buildings (Strzalka et al., 2012). There is a large disparity between regions, which motivates investigations of regional potentials that according to Mainzer et al. (2014) have not been done in earlier reports. 1.1. Aims and objectives The scope of this paper is to determine the PV potential at an urban scale, which might be highly beneficial for urban energy management considering different COR2R saving and investment approaches. Both the technical and economic potential are investigated and identified for each single building of the region according to its specific roof shape receiving solar radiation. The present study introduces an innovative tool for the determination of the PV potential at an urban and regional scale by using 3D city models: the Java-based SimStadt platform (SimStadt, 2016), which contains simulation models from the INtegrated Simulation Environment Language (INSEL, 2014), both developed at the Stuttgart University of Applied Sciences. In addition, a literature review regarding the different methods of PV potential estimation and available roof area reduction coefficients has been carried out. With the view of showing its full capabilities when dealing with PV potential analysis for whole regions, SimStadt has been used in this study to estimate the PV potential of the Ludwigsburg County in south-west Germany (state of Baden-Württemberg), in which every individual building was simulated (157724 buildings in total). The main purpose of this study is to determine what fraction of the electricity demand can be covered in both each municipality and the whole region, deciding the best strategy so as to reach that aim, the profitability of such investments and determining the optimal locations. An economic analysis and emission assessment has also been developed, as well as some insights into the uncertainty of the PV potential estimations. 2. Literature review 2.1. Review of methods for estimating the solar potential The literature review which has been carried out shows that there are many different methodologies which aim to determine the PV potential of a region, but as yet few methods for assessing urban scale impacts of solar energy system applications have been developed (Jo and Otanicar, 2011). One of the most important aspects which should be borne in mind is the scale, since the same techniques cannot be applied at local, regional or continental level. Additionally, it is necessary to know which data is available. Unlike Building Information Model (BIM) standards which serve as exchange support between different building tools allowing high interoperability, no comprehensively applicable model standard exists until now for Urban Energy Modelling (Nouvel et al., 2015a). That is the reason why developers had to start from the beginning and create their own data models. As it has been mentioned before, there are many different approaches when dealing with solar potential estimations. The study performed by Schallenberg-Rodríguez (2013) does a very complete methodology review and intercomparison. According to it, the main difference among the different procedures is the method used to determine the roof area: based on the ratio roof surface per capita, establishing a correlation between the population density and the roof area, or computing the total roof area of the target region. In (Li et al., 2015) the solar 4 potential in urban residential buildings is investigated at different levels of site densities, comparing the solar potential under different urban forms whose total available roof area was calculated with sample urban settings and weather data as the inputs. Other possible options are based on building typology through on-site data collection and visual inspection of a certain area (Horváth et al., 2016) or statistical calculation models which compute the total roof area through aerial object-specific image recognition (Karteris et al., 2013). Nevertheless, although the mentioned studies include very efficient and robust estimation models, they might not be replicable for the scope of the present study. On the other hand, the three most important roof-area estimation methods according to Melius et al. (2013) are the following: -Constant-value methods: they are a useful starting point for their speed, but they make very simplified rule-of-thumb assumptions such as the ratio of tilted versus flat roofs, the number of buildings with desirable rooftop orientations, or the amount of space obstructed by building components. The constants are then applied to the total building stock, determined from the Census for example. -Manual selection methods: rooftops with characteristics that appear suitable for PV are manually selected from sources such as aerial photographs and visually inspected for shading and building obstructions. They are more accurate, but very time-intensive and not easily replicable. -Geographic Information Systems (GIS)-based methods: used by the majority of analyses, they mainly use 3D models in order to determine the available rooftop area of a region, identify obstructions or assess shadow effects on buildings. They are much more accurate and replicable but computer-resource intensive. Our focus will be on GIS-based methods, since they can play a very important part in supporting decision making by tackling the urgently required energy transition (Ramirez Camargo et al., 2015). For very precise calculations the most appropriate option is 3D modeling and building simulation (Horváth et al., 2016). Nevertheless, this methodology might only be applied to small-scale regions such as a city or a county due to the fact that it is a timeconsuming and resource-intensive process (Kurdgelashvili et al., 2016). 3D city models have shown huge potentials in the field of city planning, and the number of cities represented is increasing exponentially, at the same time that the investment costs and time required to build these models is decreasing thanks to new data collection technologies such as LiDAR. Drones have also become a very efficient and low-cost solution. An example of study which makes use of 3D city models is the one presented by Singh and Banerjee (2015), which uses high-granularity land use data available in the public domain and GIS-based image analysis of satellite images. Conversely, Lukač et al. (2014) present a novel PV potential estimation over LiDAR data, taking into account the nonlinear efficiency characteristics of the PV modules and inverter. Other publications consider the time series analysis of supply and demand (Ramirez Camargo et al., 2015), assess the time-dependent annual electrical energy losses (Srećković et al., 2016), build a Digital Surface Model (DSM) from LiDAR data (Redweik et al., 2013), use ortho-imagery through cadastral data (Bergamasco and Asinari, 2011a) or do object oriented image analysis and GIS combined with remote sensing image data to quantify the available roof area (Jo and Otanicar, 2011). It should also be mentioned that the use of PV can help mitigate blackout 5 problems and assess the feasibility of rooftop PV in remote urban areas (Gautam et al., 2015). A very thorough and valuable GIS-based study was developed by Mainzer et al. (2014) for all municipalities in Germany. However, the statistical data was assumed to be homogeneous (no variation in typical building sizes between different municipalities for example), apart from the fact that the PV potential of non-residential buildings could not be assessed. In other studies such as (Srećković et al., 2016) or (Khan and Arsalan, 2016), even if GIS were used it was mainly to calculate roof areas, but not to compute solar production, which will be done in the present study. Therefore, the outcomes of this research should contribute valuably to the body of knowledge, since very few studies have used both detailed building and solar irradiance data to compute the PV production on specific sites (Schallenberg-Rodríguez, 2013). 2.2. Process of determination of the available roof area Once the 3D model of the region has been obtained, it is possible to know the total built area and the geometry of the buildings that shape it. However, many circumstances may lead to the reduction of the initial roof area. An extensive literature review has shown the great variety of different reduction coefficients used for calculating the available roof area of a region. Most studies focus on the determination of roof and facade areas, distinguishing between flat and tilted roofs (Kurdgelashvili et al., 2016; Mainzer et al., 2014; Melius et al., 2013; SchallenbergRodríguez, 2013) or between building types (Bergamasco and Asinari, 2011b; SchallenbergRodríguez, 2013). There seems to be an agreement so as to differentiate between architectural suitability and solar suitability (Byrne et al., 2015; Schallenberg-Rodríguez, 2013). However, these studies differ since their level of detail varies, and there is no common classification for their coefficients. Some of them give disaggregated factors (Bergamasco and Asinari, 2011b; Byrne et al., 2015; Izquierdo et al., 2008; Schallenberg-Rodríguez, 2013), while others show more global ones (IEA, 2002; Mainzer et al., 2014; Melius et al., 2013). In addition, unlike the publications made by Byrne et al. (2015) and Luque and Hegedus (2011) most of these studies do not consider the coefficients for the separation of the PV panels (GCR) or the Service Area (SA), necessary for maintenance operations. After gathering all the information from related studies, it was decided to use for this study the approach shown in the flowchart in Figure 1, which illustrates the way to calculate the utilization factor (UF). It should be noted that no previous study has used all of these factors at the same time, but only partially. This approach includes all the reduction coefficients which we consider as essential for our study and shows the calculation process for estimating the available roof area for PV purposes, after which calculations of the PV potential can be performed. Unlike previous publications in which these coefficients are applied to the aggregated results of a whole region, our study considers their application for each building individually, which increases the accuracy of the procedure. This is due to the fact that the 3D model allows us to know their characteristics, enabling us to apply different factors depending on the building that is being analyzed. 6 Figure 1: Flowchart of the available roof area calculation process. With a view to understanding the scope of each reduction coefficient, they are going to be briefly explained: -Construction restrictions (CRCONR): it refers to space already occupied by elements located on the roof, such as elevators, air extractors, chimneys, stairwells, water tanks, HVAC installations or windows. -Protected buildings (CRPROTR): this coefficient may be applied to buildings where for some reason no facility can be built on, due to historical considerations for example. -Shading effects (CRSHR): it considers the shadowing produced by the roof itself or by other buildings. -Service Area (CRSAR): necessary space for maintenance and access. At higher tilt angles the space freed up due to the spacing between the PV panels (CRGCRR) can be used (Byrne et al., 2015). -Orientation losses (CRAZR): it takes into account the relative amount of solar radiation which reaches the surface due to its azimuth. -Slope of the roof (CRSLR): it takes into account the relative amount of solar radiation which reaches the surface due to the slope of the roof. -Separation of the PV panels (CRGCRR): it considers the distance between the panels so as to avoid reciprocal shadowing. According to Luque and Hegedus (2011), shade on as little as 5-10% of an array can reduce its output by over 80%. -Ratio of PV panels (CRPVR): Ratio of the available roof area used to install PV panels. 7 -Ratio of SRTHR panels (CRSTR): Ratio of the available roof area used to install SRTHR panels. 2.3. Process of determination of the technical PV potential The PV potential is calculated in the way shown in Figure 2. Figure 2: Flowchart of the technical PV potential calculation process. -PV area (SRPVR): total available roof area used to install the PV panels [mP 2 P], determined by the SimStadt software (SimStadt, 2016). -Incoming solar energy (IRPVR): annual insolation in the PV modules surface [kWh/mP 2 P•year] also calculated by SimStadt. -PV modules efficiency (ηRefR): efficiency of the PV modules depending on the technology used. -Temperature and irradiance losses (ηRTHR): efficiency loss due to climate characteristics. This parameter is currently object of great interest in the technical community (Bergamasco and Asinari, 2011b). -Losses for orientation (ηRAZR): it takes into account the reflection losses due to non-normal incidence angle of the Sun’s rays (Li et al., 2015). -Performance ratio (ηRPRR): losses due to conversion efficiency of the inverter, cabling losses, dust on the panels and others. Electricity storage will not be considered in the present study. 3. Input data and simulation tools 3.1. Data model and weather processor For the modelling of the 3D building data, the Open Geospatial Consortium (OGC) Standard CityGML (CityGML, 2012) has been chosen. CityGML is an open, multifunctional XML-based data model, a flexible spatio-semantic data format which offers powerful methods for the evaluation of various analyses for city districts, whole cities or regions. A considerable advantage of CityGML in comparison with other 3D city model formats is that it specifies object modelling in four increasing Levels of Detail (LOD1, LOD2, LOD3 and LOD4), enabling the city model to adapt to local building parameter availability. The most simple building representation is LOD1, consisting in a rectangular block. LOD2 includes the full building geometry with varying heights of building parts and the roof shape, LOD3 a detailed façade geometry including doors and windows, and LOD4 the inclusion of indoor spaces. In 8 2014, the complete building stock of Germany was modelled with CityGML – LOD1, and some regions like Baden Württemberg or Saxony have already completed their 3D city model with LOD2 (Nouvel et al., 2015b). In order to generate the 3D city models, LiDAR, stereo air photo or digital cadaster enhanced with building information can be used. In particular, laser scanning methods which are often used nowadays allow an automatic generation of CityGML models of whole cities in a short time. On the other hand, analyzing the solar potential of a region requires local weather data, either hourly or monthly, in order to know the horizontal and diffuse radiations, ambient temperatures, etc. The quality of the solar radiation data depends on the source, including ground station measurements, satellite images or combinations of both types (Assouline et al., 2017).These data are imported into the SimStadt platform through a weather processor from different databases such as PVGIS (PVGIS, 2012), INSEL (INSEL, 2014), or by using Meteonorm weather files chosen by the user. 3.2. Urban modeling platform SimStadt Recently, urban simulation and 3D GIS have progressed considerably, but without notable interaction between them. With the purpose of taking both domains into account and supporting public authorities and engineering companies in the planning of the energy transition at urban scale, the urban energy simulation platform SimStadt (SimStadt, 2016) was developed by the Stuttgart University of Applied Sciences in the framework of a project funded by the German federal Ministry of Economic Affairs and Energy. Based on the open 3D CityGML models, its workflow-driven structure is highly modular and extensible, allowing for a potentially unlimited variety of urban analysis provided that the required data is available in the 3D model. Each workflow step has hypotheses, parameters and intermediate results which can be modified and assessed through the Graphical User Interface (GUI), enabling the user to create scenarios accordingly. In addition, if some information is not deducible or available at building level, such as building age necessary for heat demand calculations, default data are used from the building library. In the case of PV potential calculations all the required information is contained in the CityGML model, as only geometry data are used for the modeling. The start of the workflow in SimStadt is the virtual 3D CityGML model. It should be noted that SimStadt handles all LODs. Given the diversity of the quality of the 3D models, the next step would be the use of the healing module “CityDoctor”, required to check and correct the geometry of the model. Then, the data-processing allows for the completion of the model. After that, the energy simulations can be carried out. SimStadt has the capacity of obtaining hourly or monthly data in every simulation, although the results of the present study are annual given that the main goal is to estimate the annual PV potential of a region. The latest version of SimStadt can perform a variety of multi-scale energy analyses such as heating/cooling demand diagnosis, building refurbishment scenarios or photovoltaic potential. Other workflows are under way. Last of all, the results can be visualized in different ways with performance indices, graphs or maps, as well as being exported to a file. 9 Figure 3: Example of radiation map simulated using SimStadt to determine optimal PV locations in a municipality. Once the weather data are available, the radiation processor can compute the incoming irradiance on every building boundary surface, based on their geometry and the direct, diffuse and horizontal irradiances delivered by the weather processor. In the current version of the SimStadt platform, the user can select two different radiation distribution models: -INSEL model: based on the Hay sky model for diffuse irradiance calculation, requires INSEL and simulates solar irradiance on arbitrary surface orientations. Its execution time is fast and does not depend on the 3D model size. Shading and inter-reflections are not considered. -Simplified Radiosity Algorithm (SRA): it is coupled with the Perez sky model, and considers both shadowing and the reflection effects of the surrounding buildings. Its execution time depends on the 3D model size and the amount of simulated buildings. The current study will be based on the INSEL model without shading due to the large amount of buildings that will be analyzed. Shadowing effects will be approximated through a reduction coefficient (see section 5.1). 3.3. PV Potential analysis tool Regarding the PV potential tool included within SimStadt, the sequential workflow steps are shown in Figure 4. The input is the CityGML file of the region. Although it can also work with LOD1, LOD2 is preferable. LOD3 and LOD4 include more information and they could be interesting to analyze facades for example, but other data would be irrelevant for our purpose (such as internal partitions). The outputs of this tool are: irradiance, suitable roof area, nominal power and annual energy yield for every individual building. It is also able to perform the overall calculations, as well as show valuable graphs and 3D maps with data such as PV suitability in order to assess optimal locations. 16 Figure 8: Percentage of the electricity demand covered in each municipality by PV for the economic potential strategy. Regarding the aggregated values for the whole county, the results can be seen in Figure 9. The main characteristics of the region are summarized in Table 3. Figure 9: Percentage of the electricity demand covered by PV in the whole region for the two scenarios: technical and economic potential. Variable Result Total electricity demand of the region 1717 [GWh/year] Total population of the region 354551 inhabitants Total number of buildings simulated in SimStadt 157724 buildings Total roof area of the region 22.26 [km P 2 P ] Total available roof area of the region 11.14 [km P 2 P ] Average % of flat roofs: 16 % Average % of tilted roofs: 84 % Average surface to volume ratio of the buildings: 0.84 [m P -1 P ] Table 3: Summary of the characteristics of the Ludwigsburg County. If PV modules could be installed on all the available surface (technical potential), using waferbased silicon modules (scenario A) could cover 77 % of the electricity demand of the region. 17 On the other hand, if thin-film modules were used (scenario B with less efficiency), then only 51% could be achieved. It should be noted that the efficiency of the PV modules improves every year, so these percentages would increase accordingly. Conversely, taking the economic potential into account would result in lower payback periods, but the met electricity demand would be lower than that of the technical potential. Waferbased silicon modules would cover 56% of the electricity demand, while thin-film based modules would cover only 37 %. The summary of the obtained results for the region is shown in Table 4. Scenario Variable Result Description of the variable calculated by SimStadt RSCENARIO A 𝐸𝐸𝑃𝑃𝑃𝑃 𝑇𝑇𝑇𝑇𝑇𝑇ℎ𝑛𝑛𝑛𝑛𝑇𝑇𝑛𝑛𝑛𝑛 1318 [GWh/year] Technical PV potential 𝑃𝑃𝑃𝑃𝑃𝑃 𝑇𝑇𝑇𝑇𝑇𝑇ℎ𝑛𝑛𝑛𝑛𝑇𝑇𝑛𝑛𝑛𝑛 1642 [MW R p R ] Total technical PV nominal power 𝐸𝐸𝑃𝑃𝑃𝑃 𝐸𝐸𝑇𝑇𝐸𝐸𝑛𝑛𝐸𝐸𝐸𝐸𝑛𝑛𝑇𝑇 957 [GWh/year] Economic PV potential 𝑃𝑃𝑃𝑃𝑃𝑃 𝐸𝐸𝑇𝑇𝐸𝐸𝑛𝑛𝐸𝐸𝐸𝐸𝑛𝑛𝑇𝑇 1107 [MW R p R ] Total economic PV nominal power RSCENARIO B 𝐸𝐸𝑃𝑃𝑃𝑃 𝑇𝑇𝑇𝑇𝑇𝑇ℎ𝑛𝑛𝑛𝑛𝑇𝑇𝑛𝑛𝑛𝑛 872 [GWh/year] Technical PV potential 𝑃𝑃𝑃𝑃𝑃𝑃 𝑇𝑇𝑇𝑇𝑇𝑇ℎ𝑛𝑛𝑛𝑛𝑇𝑇𝑛𝑛𝑛𝑛 1087 [MW R p R ] Total technical PV nominal power 𝐸𝐸𝑃𝑃𝑃𝑃 𝐸𝐸𝑇𝑇𝐸𝐸𝑛𝑛𝐸𝐸𝐸𝐸𝑛𝑛𝑇𝑇 644 [GWh/year] Economic PV potential 𝑃𝑃𝑃𝑃𝑃𝑃 𝐸𝐸𝑇𝑇𝐸𝐸𝑛𝑛𝐸𝐸𝐸𝐸𝑛𝑛𝑇𝑇 744 [MW R p R ] Total economic PV nominal power Table 4: Summary of the results obtained by SimStadt for the two different scenarios and strategies. 6.2. Emission calculations Quantifying the potential COR2R emission savings due to the implementation of PV modules is another important outcome that may be inferred from this study. This way, we are able to evaluate for each strategy and scenario considered the amount of COR2R emissions avoided and the percentage of reduction compared to the initial situation, in which all the electricity is obtained from the grid. Table 5 shows the value of the COR2R emissions for the whole region not regarding any PV systems. Variable Result Description 𝐶𝐶𝐶𝐶2,𝑡𝑡𝐸𝐸𝑡𝑡 918814 [tCOR2R/year] Annual COR2R emissions. 𝐶𝐶𝐶𝐶2,𝑇𝑇𝐸𝐸𝑇𝑇𝑐𝑐 535 [gCOR2R/kWh] Coefficient of CO R2R emissions in Germany (Umweltbundesamt, 2016). Table 5: COR2R emissions ofR Rthe region in the initial case. To better understand the emission savings, a full Life-Cycle Assessment (LCA) would be necessary to evaluate the environmental impact of the PV modules. For simplicity, after reviewing related publications (Nugent and Sovacool, 2014; Peng et al., 2013) the present study will consider a COR2R emission coefficient of 50 gCOR2R/kWh for the PV electricity generation. The COR2R emissions avoided and the COR2 Remissions produced after PV implementation are calculated in the following way: 𝐶𝐶𝐶𝐶2,𝑛𝑛𝑎𝑎𝐸𝐸𝑛𝑛𝑎𝑎𝑇𝑇𝑎𝑎 =𝐸𝐸𝑃𝑃𝑃𝑃 ∙(𝐶𝐶𝐶𝐶2,𝑇𝑇𝐸𝐸𝑇𝑇𝑐𝑐 −𝐶𝐶𝐶𝐶2,𝑇𝑇𝐸𝐸𝑇𝑇𝑐𝑐,𝑃𝑃𝑃𝑃) 𝐶𝐶𝐶𝐶2,𝑝𝑝𝑝𝑝𝐸𝐸𝑎𝑎𝑝𝑝𝑇𝑇𝑇𝑇𝑎𝑎 =𝐸𝐸𝑃𝑃𝑃𝑃 ∙𝐶𝐶𝐶𝐶2,𝑇𝑇𝐸𝐸𝑇𝑇𝑐𝑐,𝑃𝑃𝑃𝑃 + (𝐸𝐸𝐷𝐷𝑇𝑇𝐸𝐸𝑛𝑛𝑛𝑛𝑎𝑎 − 𝐸𝐸𝑃𝑃𝑃𝑃)∙𝐶𝐶𝐶𝐶2,𝑇𝑇𝐸𝐸𝑇𝑇𝑐𝑐 18 Table 6 presents the results for the technical and economic potential of the two considered scenarios. SCENARIO Strategy CO R2R emissions avoided [tCO R 2 R /year] CO R2R emissions produced [tCO R 2 R /year] COR2R savings achieved [%] SCENARIO A Technical potential 639141 279674 70% Economic potential 464262 454552 51% SCENARIO B Technical potential 422907 495907 46% Economic potential 312288 606527 34% Table 6: COR2R emissions ofR Rthe region for each approach of the study. The results show the huge potential contribution of rooftop PV to the reduction of the COR2R emissions (and therefore other pollutants). 6.3. Economic feasibility Another purpose of the present study was to develop an economic analysis of the implementation of PV modules in the region regarding the proposed strategies, so as to assess their feasibility. For the calculations, it is assumed that 30% of the PV production of the region will be used for self-consumption (IEA-PVPS, 2016), while the remaining 70 % will benefit from the feed-intariffs devised by the government. In addition, it will be considered that maintenance of the systems would annually incur additional costs of 4% of the corresponding investment. The total investment costs 𝐶𝐶𝑡𝑡 [€] were estimated through the total nominal installed power 𝑃𝑃𝑃𝑃𝑃𝑃 [kWRpR], and the annual savings 𝐴𝐴𝑠𝑠 [€/year] (by avoiding the electricity costs) were identified and calculated in the following way: 𝐶𝐶𝑡𝑡=𝑃𝑃𝑃𝑃𝑃𝑃 • 𝐶𝐶𝑆𝑆𝑆𝑆𝑠𝑠𝑡𝑡𝑇𝑇𝐸𝐸 𝐴𝐴𝑠𝑠=𝐸𝐸𝑃𝑃𝑃𝑃 • �𝐹𝐹𝑆𝑆𝑇𝑇𝑛𝑛𝑐𝑐 •𝐶𝐶𝑇𝑇𝑛𝑛𝑇𝑇𝑇𝑇 +�1− 𝐹𝐹𝑆𝑆𝑇𝑇𝑛𝑛𝑐𝑐�•𝐶𝐶𝑐𝑐𝑡𝑡�− 𝐶𝐶𝑡𝑡•𝐹𝐹 𝐸𝐸 The chosen factors which were applied for the calculations are shown in Table 7. Variable Result Description 𝐶𝐶𝑇𝑇𝑛𝑛𝑇𝑇𝑇𝑇 0.22 [€/kWh] Electricity price per kWh (Experience value). 𝐶𝐶𝑐𝑐𝑡𝑡 0.1231 [€/kWh] Feed-in tariff for small PV facilities in Germany (Bundesnetzagentur, 2015). 𝐶𝐶 𝑆𝑆𝑆𝑆𝑠𝑠𝑡𝑡𝑇𝑇𝐸𝐸 1280 [€/kWp] Average price for the installation of 1 kWp PV (ISE Fraunhofer Institute for Solar Energy, 2016). 𝐹𝐹𝑆𝑆𝑇𝑇𝑛𝑛𝑐𝑐 30 [%] Percentage of the electricity used for self-consumption. 𝐹𝐹 𝐸𝐸 4 [%] Annual percentage of maintenance costs. Table 7: Listing of economic indicators and their parameters for the PV potential of the region. 19 After extracting the required variables from SimStadt and applying the chosen economic indicators, the results were obtained for each proposed strategy and scenario (see Table 8). SCENARIO Strategy Energy yield [GWh/year] Nominal power [MWp] Total Investment Costs [M€] Total Annual Savings [M€/year] 𝐸𝐸𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃𝑃𝑃 𝐶𝐶𝑡𝑡 As SCENARIO A Technical potential 1318 1642 2101 116 Economic potential 957 1107 1416 89 SCENARIO B Technical potential 872 1087 1391 77 Economic potential 644 744 953 60 Table 8: Economic results for each proposed strategy and scenario. Several findings can be deduced from the presented results. As can be seen, the economic potential of both scenarios would translate into much lower necessary investment costs for the implementation of the PV modules compared to the technical potential strategies. Nevertheless, it would also mean less PV yield, annual electricity savings and emissions reduction. The PV economic expectations could be enhanced through technological innovations allowed by economies of scale. This would make the projects more profitable, and attract new investors. The removal of administrative barriers by the governments themselves and incentives to persuade the population about the usefulness of PV systems on buildings should be emphasized in some countries, in order to allow for a widespread implementation of these promising solutions as far as sustainable development and energy conservation are concerned. 6.4. Uncertainty of the method Acknowledging the uncertainty of PV potential estimation methods is a major point for further research, since it is not frequently present in many studies. Limited input data and the use of default values are important sources of uncertainty, as well as simplifications and hypotheses. The variations of solar radiation also have to be taken into account. Depending on the question, either long term average weather files or weather data for a specific year under consideration should be used. In the case of PV potential estimations all the required information is geometry data, contained in the 3D model. The higher the Level of Detail, the more accurate the PV estimations. As an example, in order to evaluate the variations of using different LOD’s the aggregated results of each municipality regarding the technical PV potential have been compared to what the results would have been if modeled in LOD1, which considers all roofs to be flat. The results are shown in Figure 10. 20 Figure 10: Percentage of difference between the PV potential yield considering LOD2 and LOD1 models. As is apparent, in every case a LOD1 model would underestimate the PV potential of the region. The explanation lies in the fact that although considering only flat roofs with a south facing tilted PV generator would mean higher specific radiation, the module separation to avoid shading would reduce the useful roof area by a factor of 0.46, as well as other reduction coefficients which are more restrictive in flat roofs. The potential roof area is always underestimated in the LOD1 model. The installed module area in the region under the assumption of only flat roofs is between 6.67 % and 13.34 % lower than for the LOD2 roof structure with predominantly tilted roofs. In fact, the two municipalities in Figure 10 with a higher percentage of difference are the ones with a higher average tilt angle of their roofs. Nonetheless, the differences are rather small, so in case of having a LOD1 model the results can be expected to be accurate enough. It should also be noted that using reduction coefficients in order to assess construction restrictions in roofs or the influence of trees and buildings is another important source of uncertainty. An increase in the level of detail of the 3D models which includes this information could replace in the future these reduction coefficients with more accurate values for each individual building. With regards to the validation of the results, researchers have frequently little information about the accuracy of their estimates (Melius et al., 2013). In order to validate our results, the outcomes obtained by Mainzer et al. ,2014 (whose study considers the PV potential of all the regions in Germany) have been consulted. In the Ludwigsburg County area, they obtained a value of technical potential greater than 1000 MWh/kmP 2 P and 1000-4000 kWp/kmP 2 P for the region. The results of the technical potential in our study for Scenario A are 1443.5 MWh/kmP 2 P and 1799.4 kWp/kmP 2 P, and for Scenario B 955.8 MWh/kmP 2 P and 1191.0 kWp/kmP 2 P. Therefore, the results are quite consistent with the ones obtained by them. As stated in (Freitas et al., 2015), it is expected that as further and more sophisticated solar maps and further and more diverse installation case studies are published, an interactive dialogue between these two research areas will lead to model validation and improvement. 21 7. Conclusions This paper proposes to use 3D urban data models based on the CityGML standard to analyze the photovoltaic potential on an urban and even regional scale. The simulation methodology is based on a building by building roof surface analysis and irradiance simulation and carefully revises reduction factors for the energy yield determination, applying them for each building separately. Realistic strategies and scenarios for PV implementation were developed in a case study region in Germany. Economic calculations have also been performed so as to analyze the feasibility of the required investments. According to the results obtained, it is possible to achieve high rates of electricity demand covered by PV in many municipalities (even more than 100% for low density municipalities, which means an electricity surplus). Within the entire region with 34 municipalities investigated, PV systems could generate 77 % of the electricity consumption by using all available roof space, producing a total of 1318 GWh/year through the installation of 1642 MWp, thus reducing the CO2 emissions noticeably. Conversely, 56% of the electricity demand could be produced if only roofs with enough insolation and a minimum surface area for an economically feasible PV installation are used. To realize the economically viable PV installations and reduce the electricity related CO2 emissions by 51%, the estimated investment per capita is around 4000 Euros or a total of 1416 million Euros in the County. In conclusion, if properly designed these PV systems could significantly decrease primary energy consumption and emissions, reaffirming their usefulness and the important role they can play in the near future. 8. Future work During the development of this research work, some future directions have been identified which could result in more precise PV potential calculations. First of all, due to the large amount of buildings and the required computational time of more sophisticated radiation processors, the Hay model was used in this study to analyze all the involved municipalities. Since the interaction between buildings was not taken into account, a shadowing reduction factor was considered, taken from the literature review. However, this reduces the accuracy of the procedure which makes use of precise geometry building models. In the future, the use of tiling strategies that are currently under development will reduce the required computational time, consequently making the calculations feasible for the SRA radiation model, which considers the influence between buildings and is already implemented within SimStadt. This improvement will also help to make the large scale more valuable. Additionally, the present study has only considered roof surfaces, but it could be extended to facades. The importance of further research regarding the uncertainty of the PV potential estimations should also be highlighted. Last of all, the progress in the field of 3D modelling will ensure in the future that models with higher LODs are available, thus increasing the accuracy of the PV potential analyses. 22 UReferences Assouline, D., Mohajeri, N., Scartezzini, J.-L., 2017. Quantifying rooftop photovoltaic solar energy potential: A machine learning approach. Sol. Energy 141, 278–296. doi:10.1016/j.solener.2016.11.045 Bergamasco, L., Asinari, P., 2011a. Scalable methodology for the photovoltaic solar energy potential assessment based on available roof surface area: Further improvements by ortho-image analysis and application to Turin (Italy). Sol. Energy 85, 2741–2756. doi:10.1016/j.solener.2011.08.010 Bergamasco, L., Asinari, P., 2011b. Scalable methodology for the photovoltaic solar energy potential assessment based on available roof surface area: Application to Piedmont Region (Italy). Sol. Energy 85, 1041–1055. doi:10.1016/j.solener.2011.02.022 Bundesnetzagentur, 2015. <http://www.bundesnetzagentur.de>. Byrne, J., Taminiau, J., Kurdgelashvili, L., Kim, K.N., 2015. A review of the solar city concept and methods to assess rooftop solar electric potential, with an illustrative application to the city of Seoul. Renew. Sustain. Energy Rev. 41, 830–844. doi:10.1016/j.rser.2014.08.023 CityGML, 2012. Exchange and storage of virtual 3D city models. <http://www.citygml.org/>. Compagnon, R., 2004. Solar and daylight availability in the urban fabric. Energy Build. 36, 321– 328. doi:10.1016/j.enbuild.2004.01.009 Eicker, U., Nouvel, R., Duminil, E., Coors, V., 2014. Assessing passive and active solar energy resources in cities using 3D city models. Energy Procedia 57, 896–905. doi:10.1016/j.egypro.2014.10.299 EU, 2010. Directive 2010/31/EU of the European Parliament and of the Council of 19 May 2010 on the energy performance of buildings (recast). Off. J. Eur. Union 13–35. doi:10.3000/17252555.L_2010.153.eng EU, 2009. Directive 2009/28/EC of the European Parliament and of the Council of 23 April 2009. Off. J. Eur. Union 140, 16–62. doi:10.3000/17252555.L_2009.140.eng Freitas, S., Catita, C., Redweik, P., Brito, M.C., 2015. Modelling solar potential in the urban environment: State-of-the-art review. Renew. Sustain. Energy Rev. 41, 915–931. doi:10.1016/j.rser.2014.08.060 Gautam, B.R., Li, F., Ru, G., 2015. Assessment of urban roof top solar photovoltaic potential to solve power shortage problem in Nepal. Energy Build. 86, 735–744. doi:10.1016/j.enbuild.2014.10.038 Green, M.A., Emery, K., Hishikawa, Y., Warta, W., Dunlop, E.D., 2016. Solar cell efficiency tables (version 47). Prog. Photovoltaics Res. Appl. 24, 3–11. doi:10.1002/pip.2728 Horváth, M., Kassai-Szoó, D., Csoknyai, T., 2016. Solar energy potential of roofs on urban level based on building typology. Energy Build. 111, 278–289. doi:10.1016/j.enbuild.2015.11.031 IEA, 2002. Potential for building integrated photovoltaics. IEA-PVPS Task 2002, 2–4. IEA-PVPS, 2016. Review and analysis of self-consumption policies. Report IEA-PVPS T1-28:2016. 23 INSEL, 2014. A Simulation System for Renewable Energy Supply Systems. Version 8.2. <http://www.insel.eu>. ISE Fraunhofer Institute for Solar Energy, 2016. Photovoltaics report. Izquierdo, S., Rodrigues, M., Fueyo, N., 2008. A method for estimating the geographical distribution of the available roof surface area for large-scale photovoltaic energypotential evaluations. Sol. Energy 82, 929–939. doi:10.1016/j.solener.2008.03.007 Jo, J.H., Otanicar, T.P., 2011. A hierarchical methodology for the mesoscale assessment of building integrated roof solar energy systems. Renew. Energy 36, 2992–3000. doi:10.1016/j.renene.2011.03.038 Karteris, M., Slini, T., Papadopoulos, A.M., 2013. Urban solar energy potential in Greece: A statistical calculation model of suitable built roof areas for photovoltaics. Energy Build. 62, 459–468. doi:10.1016/j.enbuild.2013.03.033 Khan, J., Arsalan, M.H., 2016. Estimation of rooftop solar photovoltaic potential using geospatial techniques: A perspective from planned neighborhood of Karachi - Pakistan. Renew. Energy 90, 188–203. doi:10.1016/j.renene.2015.12.058 Kurdgelashvili, L., Li, J., Shih, C.-H., Attia, B., 2016. Estimating technical potential for rooftop photovoltaics in California, Arizona and New Jersey. Renew. Energy 95, 286–302. doi:10.1016/j.renene.2016.03.105 Li, D., Liu, G., Liao, S., 2015. Solar potential in urban residential buildings. Sol. Energy 111, 225– 235. doi:10.1016/j.solener.2014.10.045 Lukač, N., Seme, S., ZLaus, D., Stumberger, G., Zalik, B., 2014. Buildings roofs photovoltaic potential assessment based on LiDAR (Light Detection And Ranging) data. Energy 66, 598–609. doi:10.1016/j.energy.2013.12.066 Luque, A., Hegedus, S., 2011. Handbook of Photovoltaic Science and Engineering, Second. ed. John Wiley and Sons. doi:10.1002/9780470974704 Mainzer, K., Fath, K., Mckenna, R., Stengel, J., Fichtner, W., Schultmann, F., 2014. A highresolution determination of the technical potential for residential-roof-mounted photovoltaic systems in Germany. Sol. Energy 105, 715–731. doi:10.1016/j.solener.2014.04.015 Melius, J., Margolis, R., Ong, S., 2013. Estimating Rooftop Suitability for PV : A Review of Methods , Patents , and Validation Techniques. Technical Report NREL/TP-6A20-60593. Mohajeri, N., Upadhyay, G., Gudmundsson, A., Assouline, D., Kämpf, J., Scartezzini, J.L., 2016. Effects of urban compactness on solar energy potential. Renew. Energy 93, 469–482. doi:10.1016/j.renene.2016.02.053 Molin, A., Schneider, S., Rohdin, P., Moshfegh, B., 2016. Assessing a regional building applied PV potential – Spatial and dynamic analysis of supply and load matching. Renew. Energy 91, 261–274. doi:10.1016/j.renene.2016.01.084 Nouvel, R., Kaden, R., Bahu, J., Kaempf, J., Cipriano, P., Lauster, M., Benner, J., Munoz, E., Tournaire, O., Casper, E., 2015a. Genesis of the CityGML Energy ADE. Cisbat 2015 931– 936. doi:10.5075/epfl-cisbat2015-931-936 Nouvel, R., Mastrucci, A., Leopold, U., Baume, O., Coors, V., Eicker, U., 2015b. Combining GISbased statistical and engineering urban heat consumption models: Towards a new framework for multi-scale policy support. Energy Build. 107, 204–212. 24 doi:10.1016/j.enbuild.2015.08.021 Nugent, D., Sovacool, B.K., 2014. Assessing the lifecycle greenhouse gas emissions from solar PV and wind energy: A critical meta-survey. Energy Policy 65, 229–244. doi:10.1016/j.enpol.2013.10.048 Peng, J., Lu, L., Yang, H., 2013. Review on life cycle assessment of energy payback and greenhouse gas emission of solar photovoltaic systems. Renew. Sustain. Energy Rev. 19, 255–274. doi:10.1016/j.rser.2012.11.035 PVGIS, 2012. Photovoltaic Geographical Information System. <http://re.jrc.ec.europa.eu/pvgis/>. Ramirez Camargo, L., Zink, R., Dorner, W., Stoeglehner, G., 2015. Spatio-temporal modeling of roof-top photovoltaic panels for improved technical potential assessment and electricity peak load offsetting at the municipal scale. Comput. Environ. Urban Syst. 52, 58–69. doi:10.1016/j.compenvurbsys.2015.03.002 Redweik, P., Catita, C., Brito, M., 2013. Solar energy potential on roofs and facades in an urban landscape. Sol. Energy 97, 332–341. doi:10.1016/j.solener.2013.08.036 Schallenberg-Rodríguez, J., 2013. Photovoltaic techno-economical potential on roofs in regions and islands: The case of the Canary Islands. Methodological review and methodology proposal. Renew. Sustain. Energy Rev. 20, 219–239. doi:10.1016/j.rser.2012.11.078 SimStadt, 2016. <http://www.simstadt.eu/en/index.html>. Singh, R., Banerjee, R., 2015. Estimation of rooftop solar photovoltaic potential of a city. Sol. Energy 115, 589–602. doi:10.1016/j.solener.2015.03.016 Srećković, N., Lukač, N., Žalik, B., Štumberger, G., 2016. Determining roof surfaces suitable for the installation of PV (photovoltaic) systems, based on LiDAR (Light Detection And Ranging) data, pyranometer measurements, and distribution network configuration. Energy 96, 404–414. doi:10.1016/j.energy.2015.12.078 Strzalka, A., Alam, N., Duminil, E., Coors, V., Eicker, U., 2012. Large scale integration of photovoltaics in cities. Appl. Energy 93, 413–421. doi:10.1016/j.apenergy.2011.12.033 Umweltbundesamt, 2016. Entwicklung der spezifischen kohlendioxid-emissionen des deutschen strommix in den Jahren 1990 bis 2015. Weiss, W., Biermayr, P., 2010. Potential of Solar Thermal in Europe, Report of the EU-funded project RESTMAC. Wiginton, L.K., Nguyen, H.T., Pearce, J.M., 2010. Quantifying rooftop solar photovoltaic potential for regional renewable energy policy. Comput. Environ. Urban Syst. 34, 345– 357. doi:10.1016/j.compenvurbsys.2010.01.001