FREM Article Series: Modeling the Expansion of Ground-Mounted & Roof-Mounted Photovoltaic Systems

The development of the photovoltaic stock in Germany is picking up speed again after years of fluctuating additions. Roof-mounted photovoltaic (roof-PV) systems and ground-mounted photovoltaic (ground-PV) systems are an essential pillar for achieving the climate targets, which are becoming ever closer depending on the time horizon (climate neutrality in 2050 in Europe or 2045 in Germany). Grid operators are already facing increasing requests for ground-PV of ever-larger output classes. But also for the area of roof-PV, a boom is to be expected in the coming years, since some federal states already prescribe solar roof obligations for commercial and private new buildings, and in the run-up to the first conference of environment ministers in 2022, the presiding minister from Lower Saxony is pushing for a solar obligation on all new buildings. Questions such as which locations are suitable for ground-PV, or how the PV rooftop additions could develop based on scenarios, can be answered in our analyses across Europe with our PV-expansion models, which are described in more detail in this article.

Overview of the topics in the FREM article series:
  1. Development of an Energy-related Geo-database
  2. Regionalizations for the Network Development Plan
  3. The Wind Scenario Tool WiSTl
  4. Modeling the Expansion of Ground-Mounted & Roof-Mounted Photovoltaic Systems
  5. Weather Data
  6. Open Data

Ground-Mounted Photovoltaic Expansion

The scenario-based ground-PV expansion model was developed and refined within the scope of potential analyses for grid operators from Schleswig-Holstein, Saxony, and Bavaria. Grid operators are faced with the challenge of adapting to the increasing requests for new ground-mounted systems in terms of number and output and designing their grid planning accordingly. In particular, subsidy-free large-scale plants above 20 MW are becoming more and more common – these not only require potentially more extensive connection measures but are also not bound to the EEG land area framework.

In order to be able to develop the expansion probabilities of different regions on a scenario basis, the FfE has developed a multi-stage model, which is structured as follows:

  1. Identification of potential areas by geo-intersection
  2. Calculation of an area-wide evaluation grid
  3. Combination of potential areas and evaluation grid
  4. Distribution of power additions over the areas with the highest probabilities

Identification of Potential Areas by Geo-intersection

The beginning of any ground-PV analysis is a search for potentially eligible areas. Areas that are approved under the EEG (200 m margins around highways, railways, and disadvantaged agricultural land if the respective state has made use of the state opening clause) are considered, as well as all agricultural land for large, subsidy-free units. In both cases, the potential areas are identified via an exclusion procedure by subtracting all areas that are not eligible for ground-mounted photovoltaics (so-called exclusion areas) from the EEG land area or the base area of the area under investigation (see Table 1).

Exclusion areas
Highways National Parks
Biosphere Reserve (I+II) Nature reserves
Biosphere Reserve (III) OSM land use classes
Federal highways OSM Paths and waterways
FFH areas Rail
Air traffic Settlements in towns
Bodies of water Settlement open space
GHD and industrial areas SPA areas
Landfills Forests
Protected landscape areas Water protection areas (I+II)

After this intersection, a complete potential area dataset is available (see Figure 1).

Figure 1: Creation of the PV plant potential areas

Calculation of an Area-wide Evaluation Grid

In order to be able to evaluate the potential areas identified in the previous step, relevant evaluation criteria are defined in consultation with the grid operators as well as interviews with project developers. These criteria can be, for example, the spatial proximity to existing substations or overhead lines, land values as possible competition from agriculture, surface models to map the topography, or the proximity to existing plants to map synergy effects. The open structure of the model allows for interchangeable evaluation criteria as needed, thus ensuring a flexible model structure.

The selected evaluation criteria are then normalized and averaged to produce an area-wide grid that contains a value between 0 and 100 for each pixel, reflecting the suitability for ground-mounted photovoltaics at that location (see Figure 2):

Figure 2: PV potential on open spaces by suitability in Schleswig-Holstein. The highest suitability appears low at 39 % - however, this is due to the devaluation of individual factors after a previously performed sensitivity analysis, which reduces the geometric mean.

Combination of Potential Areas and Evaluation Grid

The area-wide evaluation grid is superimposed on the potential areas in the next step. Now, the average suitability can be specified for each area depending on the performance class (= size of the area).

Distribution of Additional Power Across the Areas with the Highest Probabilities

A dataset is now available that indicates the suitability of ground-PV for each potential area in the study area. However, since the available areas are far beyond realistic levels of additional installed power, the areas are sorted in descending order of suitability, starting with the best area. The respective scenario, which is developed together with the grid operators, determines the absolute amount of capacity additions by year and region (usually per state). These performance figures are now distributed over the sorted areas until as the overarching figure agreed upon with the grid operators has been reached.

This regionalization consequently allows statements about possible expansion paths in a federal state depending on different area settings and scenarios.

Roof-Mounted Photovoltaics

Decentralized power generation through roof-PV can significantly contribute to increasing electricity generation capacity. The category includes a wide range of different rooftops: from rooftops of homes with installations of a few kW to large-scale commercial rooftops. Owners are “prosumers” and include citizens acting as individuals or in energy communities or cooperatives, as well as businesses. The spatial distribution of installations is critical for forward-looking grid planning by grid operators.

The potential analysis presented here provides a Europe-wide spatial rooftop potential with a spatial resolution of 250 x 250 meters (in Germany, also 100 x 100 meters). The underlying model is based on an Open Data approach, does not require additional overflights, and includes the following steps:

  1. Preparation of a population and building grid
  2. Analysis of a high-resolution solar roof cadastre
  3. Conversion of the building polygons into a raster model
  4. Merging of regression parameters and raster building model

Figure 3 shows the flow of the model schematically.

Figure 3: Sequential calculation methodology.

Preparation of the Population and Built-up Grid

The starting point for the calculations is a population grid (Global Human Settlement Layer: Population) with a resolution of 250 x 250 meters and a built-up grid (Global Human Settlement Layer: Built-Up) with a resolution of 38 x 38 meters. The population grid is normalized in the first step to obtain the population density. The grid of the raster also serves as the target grid for the result. The built-up density is performed by neighborhood analysis per pixel in the next step. For this purpose, the sum of the built-up area is calculated in relation to the total area within a defined radius around a pixel. The resulting built-up density completes the raster as a function of the population density (see Figure 4).

Figure 4: Comparison between OSM building polygons (left) and development grid (right).

Analysis of Solar Roof Cadastre

A high-resolution solar roof cadastre (SRC) is evaluated to obtain regression parameters. The SRC distinguishes different house types, roof types, and their orientation. With the help of geo-operations, the building polygons, as well as specific information such as footprint, suitability area, and proportion of suitability area, are transferred to the previously generated target grid. The resulting data set is used to calculate regression parameters as a function of population density, building density, and proportionate suitability area. A further evaluation of the SRC generates a discount factor for unsuitable buildings. This includes, among others, protected historical buildings, but also other buildings defined as unsuitable.

Transfer of the Building Polygons into the Raster Model

In the following step, the areas of the OpenStreetMap building polygons of the respective area under consideration are transferred to the raster. This creates a raster dataset that shows the building footprints as a function of population and building density (see Figure 5).

Figure 5: Conversion from vector to raster format.

The regionalized PV potential can be determined with the help of the previously calculated regression parameters and the discount factor. Figure 6 shows the PV roof area potential in Berlin.

Figure 6: Potential grid section (kW/pixel at a power density of 0.15 kWp/m²).

Buildout of Potentials and Outlook

The determined potentials for ground-mounted and roof-mounted photovoltaics are then incorporated into the energy system analyses of the FfE together with the optimally configured wind onshore areas. In the Germany-wide or Europe-wide energy system studies, the linear optimization model ISAaR optimizes the remaining potentials (potential minus existing) in terms of costs, taking into account various scenarios.

Further information: