Swiss-PDGs

KO | EN

gitlite — search

#javascript #python #hacktoberfest #react #ai #typescript #llm #go #golang #android #machine-learning #rust #deep-learning #linux

★ 19 Open GitHub ↗

Synthetic low- and medium-voltage power distribution grids in Switzerland.

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Synthetic low- and medium-voltage distribution grid models for Switzerland
[![DOI](https://zenodo.org/badge/717508044.svg)](https://doi.org/10.5281/zenodo.15167588)

Welcome to our repository, where we share our synthetic geo-referenced models of Switzerland's low- and medium-voltage power distribution grids (PDGs). These models were obtained using the framework introduced by [A. Oneto et. al](https://doi.org/10.1016/j.segan.2025.101678), and stem from the meticulous analysis of publicly available data on power demand and transport infrastructure. With 879 medium-voltage grids and 34,920 low-voltage grids, our dataset represents all of Switzerland's power distribution landscape. We invite you to harness this rich dataset to explore, analyze, and contribute to a deeper understanding of large-scale power distribution systems.

**When using the data, the article must be referenced and acknowledged**

## Files provided
The datasets include three types of data: GeoJSON files, CSV files for MATPOWER simulations, and an Excel file for pandapower simulations, all located in the 'grids' folder.

Additionally, a supporting intermediate file is provided in the 'demand_shares' folder, containing the share of electrical consumption from residential and commercial buildings at a 100 m × 100 m resolution. This file excludes industrial consumption and can be used to simulate load profiles based on the residential and commercial composition of electrical demand.

### Geojson file
The dataset has two main folders with medium-voltage (MV) and low-voltage (LV) grids in the folder '\grids'. All the PDGs are described by two geojson files: a nodes file and an edges file. These geojson files can be read by software such as ArcGIS and QGIS, with which you can easily visualize all the components and their attributes. Moreover, you can read this data as GeoDataFrame using Python GeoPandas. For instance, if you have two files named 'nodes.geojson' and 'edges.geojson', you can access their information with the following code:
```python
import geopandas as gpd
nodes_gdf, edges_gdf = gpd.read_file('nodes.geojson'), gpd.read_file('edges.geojson')
```
In addition, the MV and LV folders have the following content:
- LV: there are nine subfolders inside, which contain the following number of grids,

| Urbanization type | Alps   | Jura  | Midlands  |
| :---         |    :---        |          :--- |          :--- |
| Periurban   | 2038   | 551    | 3510    |
| Rural     | 8264       | 1178      | 3323      |
| Urban     | 5316       | 375      | 10365      |

</div>

Note that an edge and a node file define every grid. Hence, in the cases when node files do not have a matching edge file, it means that these demand nodes are then connected to the MV infrastructure. For more detail, check Section 3 of [A. Oneto et. al](https://doi.org/10.1016/j.segan.2025.101678).

<p align="center">
<img width="500" alt="lvgrids" src="https://github.com/aeonetos/Swiss-PDGs/assets/101415556/309d8564-474e-4bf5-831b-67b1670a8485">
</p>
<p align="center">
Synthetic LV grids in Switzerland.
</p>

- MV: 879 grids.

<p align="center">
<img width="500" alt="mvall" src="https://github.com/aeonetos/Swiss-PDGs/assets/101415556/d165d1e9-e56c-486a-b6f7-ca3238edbee7">
</p>
<p align="center">
Synthetic MV grids in Switzerland.
</p>

### Pandapower data
For the simulation in pandapower, we provide excel files containing the information of the grid for all LV and MV. For instance, if you want to load and test a grid, you can use the following code:
```python
import pandapower as pp
lv_grid=pp.from_excel('_grid.xlsx')
pp.runpp(lv_grid)
```
### Matpower data
For the simulation in Matpower, csv files for the information of bus, branch, and generator are provided. To load and test the grid, you can use the following code:
```matlab
bus = readtable(bus_data.csv, 'ReadVariableNames', true);
branch = readtable(branch_data.csv, 'ReadVariableNames', true);
gen = readtable(generator_data.csv, 'ReadVariableNames', true);
mpc.version = '2';
mpc.baseMVA = 100;
mpc.bus = bus{:, :};
mpc.branch = branch{:, :};
mpc.gen = gen{:, :};
results = runpf(mpc)
```
## Web app for data visualization

We have developed a web app for visualizing and querying the data, which can be [accessed here](https://swiss-pdg.streamlit.app/). A snapshot of it is presented below for illustrative purposes.

<p align="center">
<img width="500" alt="swiss_pdgs_app" src="https://github.com/aeonetos/Swiss-PDGs/assets/101415556/81767306-f38e-4d98-89f2-df824ef223fa">
</p>
<p align="center">
Web app for power distribution grids visualization.
</p>

## Detailed description of the data files

The geometries of the geojson files are in the EPSG:2056 projection.

### MV files

All the MV grids operate at 20 kV with 25 MVA 110/20 kV transformers. For more details about the transformer, go [here](https://pandapower.readthedocs.io/en/v2.6.0/std_types/basic.html#transformers).

The line types of overhead lines (OHL) and underground cables (CS) are shown below. For more details, go [here](https://pandapower.readthedocs.io/en/v2.6.0/std_types/basic.html#lines).

</div>
 
Note that OHL 5 and 6, and CS 4, 5, and 6, are custom lines, that are defined for modeling purposes, as they allow us to place more than one line in a given edge geometry.

**Nodes**

- osmid: unique identifier for each node in the grid.
- x: east-west coordinate.
- y: north-south coordinate.
- el_dmd: active power demand in MW.
- lv_grid: the unique identifier of the MV/LV transformer connection. If the value is -1, the node is either an MV consumer or a node with no power demand.
- consumers: True if there is power consumption in the node, False otherwise.
- label: has two kinds of entries: 1) an identifier from OSM data in case the node was retrieved from that database, or 2) a string ‘intersect_X’, which identifies the projection of a load to its closest street point. 
- source: True if the node corresponds to the transformer location. False otherwise.
- voltage: The resulting voltage in p.u. from a steady state AC power flow analysis.
- geometry: Point objects used to map the nodes with GIS tools.

**Edges**

- u: the osmid of the start node of the edge.
- v: the osmid of the end node of the edge.
- key: all 0. This is used to define a MultiGraph, since the OSMnx library provides useful functions for processing street data, and works with MultiGraph.
- length: the length of the edge in km.
- OHL: True if the line is an overhead line. False if it is an underground cable.
- x: reactance of the line in Ohms.
- r: resistance of the line in Ohms.
- b: shunt susceptance in Siemens.
- s_nom: upper bound of the apparent power in the line, in MVA.
- line_type: identifier of the line type in the catalog. This identifier is used to access the type in the CS and OHL catalog, e.g., if the line type is 1 and the line is OHL, we should access the catalog of OHL in the identifier 1.
- load: the ratio of apparent power transferred in the line with s_nom in the steady state AC power flow analysis.

### LV files

All the LV grids operate at 400 V. The supplying transformer of the grid depends on its urbanization type as indicated in the table below. For more details about the transformers, go [here](https://pandapower.readthedocs.io/en/v2.6.0/std_types/basic.html#transformers).

</div>

The line types are shown below. For more details, go [here](https://pandapower.readthedocs.io/en/v2.6.0/std_types/basic.html#lines).

</div>

**Nodes**

- osmid: unique identifier for each node in the grid.
- x: east-west coordinate.
- y: north-south coordinate.
- el_dmd: active power demand in MW.
- Category: the type of grid, e.g., ‘Midlands-Urban’.
- consumers: True if there is power consumption in the node, False otherwise.
- label: has two kinds of entries: 1) an identifier from OSM data in case the node was retrieved from that database, or 2) a string ‘intersect_X’, which identifies the projection of a load to its closest street point. 
- source: True if the node corresponds to the transformer location. False otherwise.
- voltage: The resulting voltage in p.u. from a steady state AC power flow analysis.
- geometry: Point objects used to map the nodes with GIS tools.

**Edges**

- u: the osmid of the start node of the edge.
- v: the osmid of the end node of the edge.
- key: all 0. This is used to define a MultiGraph, since the OSMnx library provides useful functions for processing street data, and works with MultiGraph.
- length: the length of the edge in km.
- x: reactance of the line in Ohms.
- r: resistance of the line in Ohms.
- b: shunt susceptance in Siemens.
- s_nom: upper bound of the apparent power in the line, in MVA.
- line_type: number of cables that are placed in the given edge geometry. 
- load: the ratio of apparent power transferred in the line with s_nom in the steady state AC power flow analysis.

## References
When using this dataset, please cite the article that describes the model used for its generation:

[A. Oneto, B. Gjorgiev, F. Tettamanti, and G.Sansavini “Large-scale generation of geo-referenced power distribution grids from open data with load clustering”, 2025 in Sustainable Energy, Grids and Networks.](https://doi.org/10.1016/j.segan.2025.101678)

Swiss-PDGs

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README