paper submission instructions for icece 2004 - CiteSeerX

10 downloads 2661 Views 871KB Size Report
The resulting DC power is then fed into an inverter model with 3 free parameters to calculate ..... WebSite. Exchange and Storage of Virtual 3D City Models, URL:.
Published in Applied Energy 93 (2012) 413–421

Large scale integration of photovoltaics in cities Aneta Strzalka*, Nazmul Alam**, Eric Duminil*, Volker Coors**, Ursula Eicker* University of Applied Sciences Stuttgart Schellingstr.24, 70174 Stuttgart, Germany Corresponding author: [email protected]

Abstract For a large scale implementation of photovoltaics (PV) in the urban environment, building integration is a major issue. This includes installations on roof or facade surfaces with orientations that are not ideal for maximum energy production. To evaluate the performance of PV systems in urban settings and compare it with the building user’s electricity consumption, three-dimensional geometry modelling was combined with photovoltaic system simulations. As an example, the modern residential district of Scharnhauser Park (SHP) near Stuttgart/Germany was used to calculate the potential of photovoltaic energy and to evaluate the local own consumption of the energy produced. For most buildings of the district only annual electrical consumption data was available and only selected buildings have electronic metering equipment. The available roof area for one of these multi-family case study buildings was used for a detailed hourly simulation of the PV power production , which was then compared to the hourly measured electricity consumption. The results were extrapolated to all buildings of the analyzed area by normalizing them to the annual consumption data. The PV systems can produce 35% of the quarter’s total electricity consumption and half of this generated electricity is directly used within the buildings. Keywords: urban photovoltaic potential, CityGML model, shadow effect

1

Introduction

Photovoltaic systems are ideally suited for decentral electricity supply in urban structures with a large electric energy demand. Germany is currently a key player in PV installations worldwide and has 17 GW installed capacity. However only a very small fraction of the installed capacity is building integrated. Especially building roofs represent large resources of areas for the conversion of solar energy [1]. The motivation to install photovoltaic systems with crystalline or thin film technologies can be economic (feed-in tariff and own consumption), based on the desire for backup power or uninterrupted power supply solutions, visual properties or environmental benefits [2]. However, electricity produced by photovoltaic systems is still more expensive than other renewable energy resources. Therefore, for PV systems it is of special importance to lower energy losses and to achieve a high reliability and very long lifetime [3]. Especially in case of the implementation of PV-systems at city scale, the economical aspect should be considered. Feed-in-tariffs (FIT) have proven to be the most effective incentive program for renewable technologies; countries, which have adopted FITs, have the largest growth rates in renewable energy technology deployment [4]. Half of the world’s PV installations are supported by FITs [5]. The success of FITs enabled e.g. Germany to reach its goal of having 12.5 % renewable energy supply in 2007 and 20% today [6]. Building integration is supported in many countries by increased feed-in tariffs. The contribution of PV power to cover decentral energy consumption depends not only on building size but also on different consumer types. Some countries already provide higher FITs, if the PV electricity is directly used by the consumer. The estimation of solar potential and the evaluation of energy demand for specific end-use activities are the basic steps to determine the best policies for large-scale deployment of

1

different solar energy applications [7]. There is still a lack of studies regarding the impact of different user behaviours. Several studies have shown that occupant behaviour plays a prominent role in the variation in energy consumption in different households [8]. Nomenclature Qel PV prod PV feed-in

electricity consumption [kWh] PV power production [kWh] PV power feed-in [kWh]

Abbreviations BIPV Building Integrated Photovoltaics CityGML City Geography Markup Language EnBW Energy Baden-Württemberg FA Feature Analyst GIS Geo-information System LOD Level of Detail

1.1. Background and motivation Today, different methodologies for the estimation of PV potential at an urban scale are proposed. Nowak et al. [9] uses statistical information to estimate the available building stock combined with assumptions to correct the surfaces obtained for architectural suitability for solar utilization. Defaix et al. [10] developed a refined method to estimate the surfaces and potentials for all 27 EU member states. In his work, the technical potential for Building Integrated Photovoltaics (BIPV) starts with floor areas and population data available from public databases. The floor area together with the number of floors is used to calculate the ground floor area. The technical potential of BIPV is then calculated using an irradiance database and the technical parameters of PV-systems. In further work [11], a methodology for estimating the rooftop solar feasibility on an urban scale for public buildings in the city of Phoenix was developed. This methodology, which uses AutoCAD and Sketchup to prepare the aerial images from Google Earth, can be then applied to the whole city. Also the work of Taguchi and Kurokawa [12] uses aerial images to determine suitable roof areas at urban scale. A more advanced methodology was developed by Wiginton et al. [13], who created a five-step procedure for estimating the total roof PV potential. This procedure involves: 1) geographical division of the analyzed region, 2) sampling using the Feature Analyst (FA) extraction software, 3) extrapolation using roof-areapopulation relationships, 4) reduction for shading and other uses and orientation and 5) conversion to power and energy outputs. Feature Analyst is an advanced feature extraction program, which exi sts as an extension to ArcGIS. This extension uses Orthophotos or aerial images as an input. Regarding the FA, none of the previous works has studied roof area quantification for PV deployment. In Germany, a method using geo-information systems [14] was developed, where the suitable roof areas are obtained from laser scanner and plan view data. Special algorithms are able to identify the necessary data, like outer form, inclination, orientation of each roof. Several authors, like Gadsden et al. [15], Izquierdo et al. [16] or Kraines et al. [17], have also applied GIS techniques to estimate the PV potential of building roofs in urban areas. These techniques were applied to a single building, group of buildings (e.g. 396 dwellings by Gadsden et al. [15]), but not to the large-scale region. Thus, there is a lack of research works, which consider the urban scale in the PV potential analysis; most of the m rely on single building or small scale areas. Furthermore, all above mentioned methodologies are not based on 3D city models. A recent development in the estimation of PV potential in urban areas is the use of 3D city models. For the purposes of urban scale simulation, it is important to achieve a good compromise between modeling accuracy, computational overheads and data availability [18]. Especially in case of 3D city models the maximum level of detail is mainly restricted to the detailed roof structure , which is a very important for the

2

estimation of PV potential of building roofs [19]. Joachem et al. [20] uses full 3D information for both feature extraction and solar potential analysis using LIDAR point clouds. A very good data basis for the automatic detection of best fitting roof surfaces for photovoltaics in terms of energy performance and integration possibilities is CityGML [21], [22]. CityGML enables to describe 3D city and landscape models including geometry, semantic, topology and appearance. It is a multifunctional model, which can be used for geospatial transactions, data storage, database modelling and provides a basis for 3D geospatial visualization, analyzing, simulation and exploration tools. Calculation of roofs area and tilt as well as shadow effects based on roofs structures and neighbourhoods and also the attribute extraction such as geographical position and orientation of building surfaces are possible based on CityGML models. Carrión et al. [23] proposed a method for the estimation of the energetic rehabilitation state of buildings using CityGML. It is possible to calculate geometric characteristics by using a library. Afte r the calculation, results for every building are written to new generic attributes within the CityGML file. The approach of using geometric properties, calculating geometric characteristics and writing generic attributes is also suitable for the PV analysis. Wen et al. [24] have pointed out solar collectors as one of the dynamic features in 3D city models as an energy system. GTA Geoinformatik Gmbh 1 has produced a solar map for showing the potential roofs for solar energy also using CityGML models. CPA Systems Gmbh2 has a Java based software component called SupportGIS/Java3D, which is based on a 3D-information system using CityGML. It has the ability of continuously updating information of the official geobasis data. At the University of Applied Sciences Stuttgart (HFT), a 3D management framework based on CityGML model has been developed. This framework can be used in different scenarios, e.g. urban planning [25], but also to provide data to the SolarCity3D engine in the photovoltaic potential analysis scenario [26]. Also Baumanns [27] presents a decision tree for a refined solar energy plant potential estimation on roof areas using CityGML. This approach is seen as a contribution for decision makers and private households to estimate the return on invest of solar energy plants. While accurate 3D modelling is thus gaining importance, the connection to accurate photovoltaic system simulation including mutual shading of buildings in urban settings is still quite weak [28]. However, shading is one of the major loss mechanisms in photovoltaic energy production, which was shown as early as during the German “1000-Roofs-Program” of the 1990ies [29]. PV-systems integrated into the built environment are frequently subject to partial shading resulting from the roof-landscape itself, other buildings located in the proximity of the array, minor obstacles such as antennas, lighting protection masts and electric poles. Shading of a single cell within a PV-module leads to a reverse bias operation of the cell, which may results in hot-spots and potential breakdown of the shaded cell. Castro [30] programmed a computer tool for simulating the electrical behaviour of shaded PV -modules and determining the performance loss. Here, shading of the beam radiation on to a surface is calculated by ray tracing techniques. Diffuse irradiance reduction by buildings has been analyzed by Quaschning [29], using surface polygons for all surrounding buildings. These methods are not yet applied to an urban scale and it is not possible to solve it with a 2D approach. The methodology for shadow effect calculation proposed in this paper is advanced as it is based on 3D-model in CityGML format. Also the weak point of the methodologies developed until now is the lack of validation process at urban scale. According to Matthews et al. [31], the success of the model development process depends on its validation, but this is very often neglected due to the difficulty in obtaining good data sets . In the work of Jin and Otanicar [32], a validation process for 932 government and commercial buildings (of 2800 buildings in total) of the analyzed area is available. In this work, PV energy potential and measured user influenced electricity consumption are compared, based on average annual electricity consumption data, measured from 2006 to 2008. The result showed that the renewable energy generation source could cover 10% of the total electricity consumption of all the buildings within the analyzed area (including also the residential buildings). In this comparison, the own consumption ratio is not considered; therefore our work is of benefit in this issue due to the availability of hourly measured electricity consumption data, which can be compared to the hourly produced PV electricity generation. 1 2

GTA Geoi nforma ti k Gmbh. Sol a r Potenti a l Ana l ys i s . Berl i n: ISPRS; 2010. CPA Sys tems Gmbh. Support GIS-Sol a r. Berl i n: ISPRS; 2010.

3

In summary, the work presented in this paper uses innovative 3D geoinformation technology including shading algorithms to simulate the photovoltaic energy potential and compare it to measured cons umption values to obtain the possible own consumption ratios in urban areas. 1.2. Case study area. The case study area of Scharnhauser Park (SHP) is an urban conversion and development area of 150 hectares in the community of Ostfildern on the southern border of Stuttgart. This area is a former military area converted to a mixed residential and commercial town quarter with 7000 inhabitants. About 80% of the heating energy and 35% of the electricity demand of the whole area is supplied by renewable energies. The main portion of the heating and electrical energy is delivered to the buildings by a wood -fired cogeneration plant. A small portion of electricity is delivered by existing roof top PV -systems. For all buildings in SHP, annual electricity consumption data are available and additionally hourly electricity consumption data for one case study building were monitored over one year. This building, which is a multi -family house with 12 apartments and a gross heated area of 1688 m², serves as a case study building to validate the developed methodology.

2

Methodology

The first step of the developed methodology was to organize a map in DXF-format and laser scanner data for the buildings of the analyzed residential district. The DXF-map of the analyzed area originates from the city of Ostfildern and includes the building contours. For the case study building, detailed architectural plans were available from the company Siedlungswerk Stuttgart. The laser scanner data for the area of Scharnhauser Park are obtained from the Land Survey Office Baden-Württemberg. The point density is of 4 points per m² with a vertical resolution of 0.2 m [33]. Airbone LIDAR (Light Detection and Ranging) data establish the digital surface models and the filtering approaches enable to separate the bare ground from natural objects, which cover the topographic surface. Large objects like buildings are a special challenge for filtering approaches [34]. Global horizontal irradiance and outside temperature have been measured by a local weather station. This data has been used as input for the hourly PV power simulations. In order to calculate the hourly own consumption ratio for the case study building, measured hourly electricity consumption data were used. These data have been monitored in the time period between October 2009 and September 2010 using smart meters of the company Energy Baden-Württemberg (EnBW). These smart metering systems have been installed in almost all flats of the case study building. Furthermore, measured annual electricity consumption data for all buildings for the year 2005 were used, which have also been obtained from EnBW. Additionally, the information about the number of floors for each building, which originate from the city of Ostfildern was used to estimate the own consumption ratio. In order to determine the PV potential of the building roofs on a city scale, a flow chart was developed, which includes the whole process from the data acquisition and their preparation, through the PV and shadow calculation until the validation of the results, as shown in the Fig. 1.

4

Case study building

Whole district SHP

Detailled plan view data

DXF + Laser scanner data + aerial image

3D city model (LOD 1)

manually

PV INSEL simulation model

Calculation: Shadow effect for each building

Simulation hourly PV-power production

∑ all buildings

Simulation annual PV power production

Extraction PV-suitable roof area

Hourly measured electr. consumption

GeoMedia Professional, incl. GeoMedia Grid

PV INSEL simulation model

Excel

PV-coverage ratio of the electr. consumption for each building (%) PV-coverage of the electr. consumption for whole district (%)

Calculation own consumption per hour Calculation annual own consumption ratio (%)

Extraction PVsuitable roof areas

%-fraction as own consumption

Access: Annual measured electr. consumption for each building Access: Annual measured electr. consumption for whole district Access: Number of floors for each building

Correlation between own consumption ratio and the number of floors

Fig. 1. Process of estimating the PV power potential at urban scale.

The flow chart shown in Fig. 1 presents a methodology for a two-tier process of the PV potential analysis. On the left side of this flow chart, only one case study building is analyzed in details. On the right side, all buildings in SHP are considered. 2.1

Estimation of PV-suitable roof areas

For the case study building, the PV-suitable roof area was manually extracted from detailed architectural plans. For all other buildings of SHP, laser scanner data and the DXF-map have been employed. On the basis of these materials, by using a Geographic Information System (GIS) the detection of the roofs with optimum conditions for producing solar power was possible. The intersection of the laser scanner data and the building contours [35] enabled to estimate the roof type, its inclination, orientation and size (see workflow in Fig. 2). Laser scanner data and ALK map (building contours)

Subtraction of the average of ground and vegetation points

Aggregation of the laser scanner data for the respective building

Classification of the flat and inclined roofs on the basis of the distribution of the vegetation points

Calculation of the average of the vegetation points and its standard deviation for the respective building

Aggregation: Total area of the roofs on the basis of the ALK – building contours

Fig. 2. Work flow of the analysis .

The height of each point was calculated by the subtraction of laser scanner data of first pulse (building + vegetation points) and last pulse (ground points) measurement. Then, the relative average height of all

5

points within each of the building contours was calculated, which is seen in the Fig. 3. These values are then used as a basis for the development of a 3D city model of SHP.

Fig. 3. Classification of the buildings by average building height in [m] (cut-out from GIS).

An important step was the classification of the building roofs of SHP regarding their type, by considering the distribution of the high points in order to divide them into flat and inclined roofs. The classification showed that 80% of the roofs are of flat type. After the automatic calculation of the areas of these roofs using the standard functions of the GIS system used, these areas were classified as PV-suitable. The analysis of the inclined roofs happened to be more complicated. Not only the size, but also the orientation and the inclination of the roofs had to be calculated using additional GIS-applications. A special raster calculator enabled to filter only the roof areas, which have an orientation of 145° to 215° (south is 180°) and the roofs with optimal inclination (between 15°-45°) in order to minimize the reduction of the solar irradiance. Lower (flat roof) or higher inclinations (>70°) reduce the solar gain by about 20%. The results of the calculation of the orientation for some building roofs are shown in Fig. 4.

Fig. 4. Result of the calculation of the roof orientation.

In SHP about 20% of the roofs are inclined and could be potentially suitable for PV-installation. Nevertheless, these roofs were all classified as not suitable, because they include windows, dormer windows and chimneys, which was analysed manually by viewing the aerial photographs of these roofs. 2.2

3D city model

The next step of the proposed methodology is the generation of a topologically consistent 3D city model on the basis of the given building footprints and the measured building height [36]. Hereby, the average building height, estimated from laser scanner data, as described in 2.1, was used. This model, as shown in the Fig. 5, is a simple block model, with the Level of Detail 1 (LOD1), which is used for the shadowing simulation.

6

Fig. 5. Topologically consistent 3D block model of SHP.

3

Photovoltaic yield simulations.

The next step of the process is to use the extracted surface areas for the calculation of the PV power production. The photovoltaic system simulation for the case study building was carried out using the commercially available software environment INSEL83, to which some of the authors still contribute in development and where all the source code is available to the authors. The photovoltaic modules are simulated using the one-diode current voltage equation with 6 free parameters, which have been derived from the manufacturer’s information. For each time step, the maximum power point is iteratively determined for the given irradiance. Module temperatures are simulated in each time step to account for temperature related voltage drops. The resulting DC power is then fed into an inverter model with 3 free parameters to calculate partial load performance. Details of the photovoltaic simulation algorithms can be found in the block module handbook of the INSEL8 software. Using this simulation tool, the calculation of the hourly PV power production for one case study building was performed. The suitable roof area of the case study multi-family house of the residential area SHP is of 230 m² (flat roof). The PV-simulation was done for a roof generator with 15 kW peak power and a tilt angle of 25° facing south, using as input hourly irradiance and temperature, which was measured on the weather station placed in SHP. There are 74 Suntechnics 210F mono-crystalline silicon modules. The modules are connected to one SMA SMC 10 000 and one SMA Sunnyboy 5000 inverter, with respectively 5 strings with 10 modules and 4 strings with 6 modules. The annual PV energy produced is 15,27 MWh for this particular installation, which amounts to 980 kWh/kWp and a performance ratio of 85%. In order to calculate the own consumption ratio on an hourly basis, the measured electricity consumption was subtracted from the simulated PV power production at each time step to obtain the remaining feed-in PV power. The minimum feed-in value is zero.

PV feed in,i  PVprod ,i  Qel ,i

(1)

The summation of the hourly data resulted in the annual own consumption ratio.

  PV

8760

Qown 

i 1

prod ,i

 PV feed in ,i 

8760

 PV i 1

prod ,i

(2)

In the next step, a correlation between the own consumption ratio and the number of floors has been calculated on the basis of the available data for the case study building. Here, the measured specific 3

INSEL 8 i s devel oped by doppel i ntegra l GmbH, www.ins el .eu

7

90

40

80

35

70

30

60

25

50 20

Case study building

40

15

30

10

20

PV-coverage ratio in %

Own consumption energy ratio in %

electricity consumption (for 4 floors of this building) was extrapolated to different numbers of floors (from 1 to 8). The own consumption ratio was calculated for different numbers of the floors with a fixed given PVroof installation, as seen in the Fig. 6 (blue line).

5

10 0

0 0

1

2

3

4

5

6

7

8

9

Number of floors Own consumption ratio

PV-fraction of electr. consumption

Fig. 6. Correlation for the case study building.

This correlation is used as an input for the calculation of the own consumption ratio of each building of the analyzed area SHP. The red line represents the PV-coverage ratio of the annual measured electricity consumption. In addition to the number of floors, different levels of electricity consumption have to be taken into account (see Fig. 7). If the electricity consumption increases from 10 to 35 kWh/m²a, the own consumption increases from 5 to 30% for a single storey building, from 35 to 75% or a 4 storey building, and from 55 to 97% for an 8 floor building.

Own consumption energy ratio in %

120

100

80

60

Case study building 40

20

0 1

2

3

4

5

6

7

8

Number of floors 10 kWh/m²a

15 kWh/m²a

20 kWh/m²a

25 kWh/m²a

30 kWh/m²a

35 kWh/m²a

Fig. 7. Correlation for the case study building divided into different values of electricity consumption

Electricity consumption in kWh/m²a

The own consumption ratio depends strongly on the consumption level, i.e. the user. This has been analyzed for individual flats in the case study building, where the electricity consumption varies between 15-35 kWh/m²a, as seen in the Fig. 8. 40

35 30 25

20 15 10 5 0

Flat Annual value

Average_value

Fig. 8. Annual electricity consumption values for separate flats of the case study building ( Oct. 2009-Sept. 2010)

8

4

Shadow effect.

In order to consider shadow for direct radiation a model has been developed , which determines the exact shadow projected onto each of the roof surfaces of the analyzed area. The method includes several steps like sun angles detection, potential surface filtering, surface subdivision, sun’s ray calculation, potential shadow caster filtering, shadow calculation and surface regeneration. The CityGML model, where each of the surfaces consists of a set of polygons, was taken as the basis for the e stimation of the shadow effect. Only the roof surfaces and façades were taken into account. The shadowing calculation for the area of SHP in hourly interval was done for December 21, which is the worst day of the year concerning shading with the lowest sun height angle. For detecting shadows a surface subdivision approach has been developed, where a single shadow can be detected. According to Fig. 9, the first step is to read these sets of polygons. The second step is triangulation of each polygon. Then, to achieve a fine resolution each triangle is further triangulated. The middle point of each side is connected and thus the triangle is divided into four smaller triangles, the process is repeated until the length of the smallest side is larger than the desired resolution. Then , the centroid of the triangle is measured and a line towards the sun’s direction is calculated representing the sun’s ray. The next step is to look if the sun’s ray intersects any of the surfaces. For this purpose it is checked if the line intersects with any of the triangles found in the second step. If any inters ection point is found then the triangle can be declared as a shaded triangle and joining the shaded triangles together results in a shadow polygon on any faceset. The process may face problem with thin triangles. So, for triangles with very narrow angles triangulation can be done by dividing the triangles according to the longest side. Thus the problem with thin triangles can be avoided and a high quality result can be obtained. Centroids of final subdivided triangles are assumed as the target points for which shadow will be calculated. Whether the triangle will be in shadow or not will be determined by this point. For each target point, a distant point is measured in sun’s direction at a minimum distance and above the top most point in the city model. For each point, the whole city model is divided into four quadrants, divided by the north-south and east west axis. The quadrant, which contains the sun, is marked as active quadrant and surfaces, which have at least one vertex in this area, are selected as potential shadow caster surface. Surfaces, which are below the target point, are further filtered from the selection by comparing the elevation or height of each vertex of the surface with the target point. The procedure has to be applied to every time step of the simulation due to the changing solar position. Diffuse radiation reduction through shading is not yet considered.

Fig. 9. Workflow of the estimation of the shadow Fehler! Verweisquelle konnte nicht gefunden werden..

9

4.1

Shadow calculation This is the main step, where shadow is calculated. A line-plane intersection check is performed here. A line can be expressed as: (

)

(3) (

Where, target point

) and distant point

(

)

A surface can be expressed as: (

)

(

)

(4)

Where, subdivided triangle of potential shadow caster surface ( ); k=0,1,2, which represents three points of the triangle. At intersection point the point on line will be equal to the point on surface so by solving the equation in matrix form: [

]

[ If shadowed.

[

][ ] ] and

(5)

, then the line and the triangle intersects and the point is marked as

If the line lies upon the surface or parallel then will be linearly independent. In this case, if line lies upon the surface then the point must be also marked as shadowed. To determine if the line lays upon the surface a further line-line intersection check is performed for each side of the triangle with ( ) and ( ) the sun’s ray by solving equation of two lines. If two lines are then and can be obtained by solving the equations for x and y. If the value for and also satisfies the equations for z then there is an intersection. Then it is checked if the intersection point lies within the lines. If all requirements are fulfilled for intersection then the point is marked as shadowe d point. 4.2. Surface generation Only the surfaces where at least one subdivided triangle is shadowed are considered for surface regeneration. This step is only necessary when a visual output for real time shadow is required for an instance of time. If the calculation is carried out for any longer time period like and hourly or minutely shadow calculation then this step might be excluded. Neighboring subdivided triangles with same shadow status are joined together to form shadow and non-shadow region. These regions are further merged with other neighboring region with similar shadow status. The whole process produces result for an instance of time. To get hourly or minutely shadow calculation this has been repeatedly applied and the result has been presented in a tabular form.

5

Results

5.1

Case study building Fig. 10 shows the comparison between the hourly measured electricity consumption, hourly simulated PV-power production and hourly calculated own consumption for one summer week. The total own consumption ratio for this week was 48%; in comparison the own consumption ratios for spring and winter weeks have values of 60% and 66%.

10

Fig. 10. Comparison of electricity consumption, solar power production and own consumption (summer week).

On the basis of hourly data of simulated PV-power production and measured electricity consumption for the time period between October 2009 and September 2010 an annual own consumption energy ratio of 56% has been calculated, i.e. 44% of the produced PV energy is exported to the grid. The solar fraction with directly used electricity corresponds to 25% of the total building electricity consumption. The solar fraction based on the annual ratio of generated PV energy to electricity consumption is higher and amounts to 44 %. The simulation of the shadow effect showed that no shadowing occurs on the case study building roof during the shortest day of the year. 5.2

Whole district

The classification of the roofs, which was done within the GIS-system using laser scanner data allowed to divide all roofs of SHP into suitable and non-suitable for PV-installation as well as for the roofs, which have already installed PV-systems, as seen in the Fig. 11. 3%

43%

54%

suitable roof areas

Fig. 11. Classification of the roofs in the urban area Scharnhauser Park.

As seen in the Fig. 11, 54% of all roofs in SHP can be used for the installation of PV -systems. The calculation of the PV-power production for the case study building was extrapolated to all suitable roofs of SHP. The annual energy balance simulation showed that about 35% of the total measured electricity consumption of the whole district SHP, which was 10700 MWh (year 2005), could be covered by the electrical energy produced from PV-modules. Taking into consideration the measured electricity consumption of only the buildings with suitable roof areas, the PV -coverage is of 54% (see Table 1). In case, when the own power consumption ratio (from Table 1) was considered, only 17 % of the total electricity consumption of the district SHP could be directly covered by the PV power, and 26 % when taking into consideration only the electricity consumption of the buildings with suitable roof areas (Nr 2 in Table 1). The remaining 18% (total district) or 28% (buildings with suitable roof areas) are fed into the electrical grid.

11

Table 1. PV-coverage ratio depending on the assumptions made.

Buildings considered

Whole district

Buildings with suitable roof areas

Nr

Calculation method of the PV production

PV coverage ratio [%]

PV coverage ratio [%]

1

PV power production (tota l )

35

54

2

PV power production (own consumption)

17

26

On an individual building level, the percentage of the PV -coverage of the electricity consumption strongly depends on the building size and the user behaviour and varies from 5 to 100%. There are also many building, where the solar power produced is 2-10 times higher than the annual electricity consumption. The distribution of the PV coverage ratio of the annual electricity consumption for all residential buildings with PV-suitable roof areas is shown in the Fig. 12. The two distributions show the ratio of annual energy production to consumption and the hourly power balances. For similar building types, the ratio of PV production to own consumption can vary by a factor 2.

Fig. 12. Distribution of the PV-fraction of the measured annual electricity consumption when only annual energy balances are considered (total) and when hourly production and consumption are balanced (own consumption).

Considering only the buildings with PV-suitable roof areas, the shadowing simulation showed that 97% of the total roof area of these buildings is completely unshaded over the whole year. 5.3

Visualization

Finally, the GIS-system was employed to visualize and publish the results of this analysis. For data protection issues, only average values of the electricity consumption for similar building types are shown in the Fig. 13

12

Fig. 13. Average measured electricity consumption per building group [kWh/m²a].

5.4

Error analysis

Elberink & Vosselman, [37] discuss an approach of target based graph matching with both complete and incomplete laser data for 3D building reconstruction. Major problems are missing data features, deflected or absorbed laser pulses, missing laser segments, intersection lines, trees or cars, and occlusions etc. Uncertainties in our analysis are due to the lack of laser scanner data in case of buildings built after the year of 2002 (the height value was assumed to 7.5 m). This could lead to underestimate the building’s height and therefore influence the shadowing effect to the neighbour buildings. Some more errors can occur due to missing data of the number of floors for several new buildings, which have been built after the year of 2006 (the data about the number of floors have been recorde d in 2006). The lack of this information can influence the calculated own consumption ratio.

6

Conclusions

The paper uses geo-information systems, 3D city model and advanced extraction algorithms combined with PV system simulations to quantify potential rooftop solar photovoltaic deployment at an urban level. The developed methodology is a few steps semi-automated process, which consists mainly of the extraction of roof surfaces and orientations based on airborne laser scanner data and of the simulation of the solar power production using the modular simulation environment INSEL. Additionally, a general concept has been introduced for using 3D city models to calculate shadow effects in a solar potential analysis. Here, a special tool was programmed to determine the shaded parts of each suitable roof for PVinstallation. The analysis of the PV own consumption ratio was based on a case study building, for which detailed measured hourly electricity consumption values were available. On the basis of hourly data of simulated PV-power production and measured electricity consumption an own consumption energy ratio of 56% has been calculated. The annual own consumption represents about 25% of the annual measured electricity

13

consumption for the case study building. A correlation between the own consumption ratio and the number of floors was extracted from the case study building and finally extrapolated to all buildings of the urban district with PV-suitable roof areas. The ratio of PV own consumption and the total measured electricity consumption of 10700 MWh of the whole district was 17%. The total PV energy production (feed-in and own consumption) is 35% of the total electricity consumption of the district, i.e. about half of the locally produced PV energy is exported to the grid. The shadowing simulation showed that only 3% of the total roof area of the buildings with PV-suitable roof areas is partially shaded.

Acknowledgments We would like to thank the project RegioEnergie and the EU-Project POLYCITY (REF EC: TREN/05FP6EN/S07.43964/513481/), for funding part of this research. The authors would like to thank Hugo Ledoux and Martijn Maijers from TU Delft for their help generating the topologically correct 3D city model. Special thanks also to Prof. R. Kettemann, Prof. D. Schröder, and Mr. Arefi for the support by preparing the laser scanner data for our PV-analysis.

References [1] Wittmann H, Bajons P, Doneus M, Friesinger H. Identification of Roof Areas Suited for Solar Energy Conversion Systems. Renewable Energy; 1997; 11:1:25-36. [2] Tereci A, Schneider D, Kesten D, Strzalka A, Eicker U. Energy Saving Potential and Economical Analysis of Solar Systems in the Urban Quarter Scharnhauser Park. Proceedings of ISES Solar World Congress; 2009; Johannesburg, Africa. [3] Erge T, Hoffmann VU, Kiefer K. The German experience with grid-connected PV-systems. Solar Energy; 2001: 70:6:479-487. [4] Pietruszko S. Feed-in tariff: The most successful support programme. In conference record of the 2006 IEEE 4th World conference on photovoltaic energy conversion; 2006; 2, p. 2524-2527. [5] Renewable Energy Policy Network for the 21st Century (REN21). Renewables Global Status Report: 2009 Update. Renewable energy policy network for the 21st century; 2009, Paris, REN21 Secretariat. [6] Peters R, & Weis T. Feeding the grid renewably: Using feed-in tariffs to capitalize on renewable energy (primer). The Pembina Institute 1 (23); 2008. [7] Tse ROC, Gold CM, Kidner D. 3D City Modelling from LIDAR Data. [ed.] Oosterom, et al. Advances in 3D Geoinformation Systems. Delft; Springer; 2007:161-175. [8] Branco G, Lachal B, Gallinelli P, Weber W. Predicted versus observed heat consumption of a low energy multifamily complex in Switzerland based in long-term experimental data. Energy and Building 36; 2004, p. 543-555. [9] Nowak S, Gutschner M, Ruoss D, Togweiler P, Schoen T. Potential of building integrated photovoltaics. Report IEA-PVPS T7-4; 2002. [10]Defaix P, van Sark W, Worrell E, de Visser E. Technical Potential for Photovoltaics on Buildings in the EU-27. Elsevier Editorial System for Solar Energy. Manuscript Draft. [11]Bryan H. Methodology for estimating the rooftop solar feasibility on an urban scale. Conference Proceedings of the SOLAR 2010. [12]Taguchi S, Kurokawa K. PV Resource Survey for Urban Areas by means of Aerial Photographs. Tokyo University of Agriculture and Technology. [13]Wiginton LK, Nguyen HT, Pearce JM. Quantifying rooftop solar photovoltaic potential. Computers, Environment, and Urban Systems; 2010, doi: 10.1016/j.compenvurbsys.2010.01.001. [14]Klärle M.SUN-AREA - Ein Beitrag der Fernerkundung gegen den Klimawandel. In: Schiewe, J., Michel, U. (Eds), Geoinformatics paves the Highway to Digital Earth (On the occasion of the 60th

14

birthday of Professor Manfred Ehlers); Osnabrück, Institut für Geoinformatik und Fernerkundung (IGF); 2008:8:55-60. [15]Gadsden S, Rylatt M, Lomas K. Putting solar energy on the urban map: a new GIS-based approach for dwellings. Solar Energy 2002; 74:5:397-407. [16]Izquierdo S, Rodrigues F, Norberto A. A method for estimating the geographical distribution of the available roof surface area for large-scale photovoltaic energy-potential evaluations. Solar Energy. 82; 2008, p. 929-939. [17]Kraines SB, Wallace DR, Iwafune Y, Yoshida Y, Aramaki T, Kato K, et al. An integrated computational infrastructure for a virtual Tokyo: Concepts and examples. Journal of Industrial Ecology 1; 2001, p. 35-54. [18]Robinson D, Haldi F, Kämpf J, Leroux P, Perez D, Rasheed A, Wilke U. 2009. CITYSIM: Comprehensive micro-simulation of resource flows for sustainable urban planning. Proceddings of 11th International IBPSA Conference, Building Simulation 2009, Glasgow, Scotland. [19]Strzalka A, Bogdahn J, Eicker U. 3D City Modelling for Urban Scale Heating Energy Demand Forecasting. ASHRAE HVAC&R Research Journal on “Indoor air quality, ventilation and energy conservation in buildings”; 2010; 17:4:526-539. [20]Joachem A, et al. Automatic roof plane detection and analysis in airborne LIDAR point clouds for solar potential assessment. 7, s.l. : MDPI, July 2, 2009; Sensors; 9: 5241-526. [21]CITYGML WebSite. Exchange and Storage of Virtual 3D City Models, URL: http://www.citygml.org/1523/ (access 28th April 2010); 2010. [22]OGC Standards Website: OpenGIS® City Geography Markup Language (CityGML) Encoding Standard; URL: http://www.opengeospatial.org/standards/citygml (access 28th April 2010) [23]Carrión D, Lorenz A, & Kolbe TH. Estimation of the energetic rehabilitation state of buildings for the city of Berlin using a 3D city model represented in CityGML. In T. H. Kolbe, G. König, & C. Nagel (Hrsg.), ISPRS Conference: International Conference on 3D Geoinformation; 2010: XXXVIII-4, p. 31-35. [24]Wen W, Kjems E, Bodum L, & Kolar J. Dynamic features in a 3D city model as an energy system. In T. H. Kolbe, G. König, & C. Nagel (Hrsg.), ISPRS Conference: International Conference on 3D Geoinformation. XXXVIII-4; 2010, p. 73-78. [25]Knapp S, Bogdahn J, Coors V. Improve public participation in planning processes by using web based 3D-models for communication platforms. REAL CORP 2007; Proceedings / Tagungsband: ISBN: 978-39502139-3-5: (Print) Editors: Schrenk,v. Popovich, Benedikt [26]Eicker U, Strzalka A, Schulte C, Bogdahn J, Schumacher J, Coors V. Large Scale Integration of Photovoltaic in Cities. ICSU 2010, 1st Intern. Conference on Sustainable Urbanization; 2010, p. 774-780. [27]Baumanns K, Löwner MO. Refined estimation of solar energy potential on roofs areas using decision trees on City GML-data. EGU General Assembly; Vienna: 2009, p. 14044. [28]Alam N. Shadow effect on 3D City Modelling for Photovoltaic Cells. PhD Research Proposal, GISr Report; 2010; 56. [29]Quaschning V. Simulation der Abschattungsverluste bei solarelektrischen Systemen. ISBN 389574-191-41996; Verlag Dr, Köster Berlin. [30]Castro FG. Performance Decrease in Photovoltaic Modules due the Shading Effects. Report of the Master Course “Sustainable Energy Competence”; 2007, Stuttgart. [31]Mathews, EH, et al. A Novel Thermal Simulation Model and its Application on Naturally Ventilated Desert Buildings. Building and Environment; 1997, 32 (5), p. 447-456. [32]Jin Jo H, Otanicar TP. A Hierarchical Methodology for the Mesoscale Assessment of Building Integrated Roof Solar Energy Systems, under Review; 2010. [33]Strzalka A, Md. Zahid Hasan S, Eicker U. Potential of Roof Top PV-systems for Supplying Electricity in Residential Area Scharnhauser Park. Proceedings of ICDRET ’09 in Dhaka, Bangladesh; 2009; 8386: ISBN 978-984-33-0616-6. [34]Hahn M. Extraction of spacious buildings from airbone LIDAR data. Research project at the Stuttgart University of Applied Sciences; 2007.

15

[35]Kettemann R. Preisgünstige Höheninformationen auf Knopfdruck. Einsatz von Laserscanner-Daten des Landesvermessungsamts. IngBlatt;2004;5. [36]Ledoux H and Meijers M. Extruding building footprints to create topologically consistent 3D city models. In: Urban and Regional Data Management, UDMS Annual 2009; Krek A, Rumor M, Zlatanova S, Fendel E (Eds); CRC Press. [37]Elberink SO & Vosselman G. Building Reconstruction by Target Based Graph Matching on Incomplete Laser Data: Analysis and Limitations. Sensors 2009; 9(8), p. 6101–6118.

16