Tutorial: Electricity-Only

Note

If you have not done it yet, follow the Installation steps first.

In this tutorial, we will build a heavily simplified power system model for Belgium. But before getting started with PyPSA-Eur it makes sense to be familiar with its general modelling framework PyPSA.

Running the tutorial requires limited computational resources compared to the full model, which allows the user to explore most of its functionalities on a local machine. The tutorial will cover examples on how to configure and customise the PyPSA-Eur model and run the snakemake workflow step by step from network creation to the solved network. The configuration for the tutorial is located at test/config.electricity.yaml. It includes parts deviating from the default config file config/config.default.yaml. To run the tutorial with this configuration, execute

snakemake -call results/test-elec/networks/elec_s_6_ec_lcopt_Co2L-24H.nc --configfile config/test/config.electricity.yaml

This configuration is set to download a reduced data set via the rules retrieve_databundle, retrieve_natura_raster, retrieve_cutout. For more information on the data dependencies of PyPSA-Eur, continue reading Retrieving Data.

How to configure runs?

The model can be adapted to only include selected countries (e.g. Belgium) instead of all European countries to limit the spatial scope.

countries: ['BE']

Likewise, the example’s temporal scope can be restricted (e.g. to a single week).

snapshots:
  start: "2013-03-01"
  end: "2013-03-08"

It is also possible to allow less or more carbon-dioxide emissions. Here, we limit the emissions of Belgium to 100 Mt per year.

electricity:
  co2limit: 100.e+6

PyPSA-Eur also includes a database of existing conventional powerplants. We can select which types of existing powerplants we like to be extendable:

  extendable_carriers:
    Generator: [OCGT]
    StorageUnit: [battery]
    Store: [H2]
    Link: [H2 pipeline]

To accurately model the temporal and spatial availability of renewables such as wind and solar energy, we rely on historical weather data. It is advisable to adapt the required range of coordinates to the selection of countries.

atlite:
  default_cutout: be-03-2013-era5
  cutouts:
    be-03-2013-era5:
      module: era5
      x: [4., 15.]
      y: [46., 56.]
      time: ["2013-03-01", "2013-03-08"]

We can also decide which weather data source should be used to calculate potentials and capacity factor time-series for each carrier. For example, we may want to use the ERA-5 dataset for solar and not the default SARAH-2 dataset.

  solar:
    cutout: be-03-2013-era5

Finally, it is possible to pick a solver. For instance, this tutorial uses the open-source solver GLPK.

  solver:
    name: glpk
    options: "glpk-default"

Note, that test/config.electricity.yaml only includes changes relative to the default configuration. There are many more configuration options, which are documented at Configuration.

How to use snakemake rules?

Open a terminal, go into the PyPSA-Eur directory, and activate the pypsa-eur environment with

mamba activate pypsa-eur

Let’s say based on the modifications above we would like to solve a very simplified model clustered down to 6 buses and every 24 hours aggregated to one snapshot. The command

snakemake -call results/test-elec/networks/elec_s_6_ec_lcopt_Co2L-24H.nc --configfile config/test/config.electricity.yaml

orders snakemake to run the rule solve_network that produces the solved network and stores it in results/networks with the name elec_s_6_ec_lcopt_Co2L-24H.nc:

rule solve_network:
    params:
        solving=config["solving"],
        foresight=config["foresight"],
        planning_horizons=config["scenario"]["planning_horizons"],
        co2_sequestration_potential=config["sector"].get(
            "co2_sequestration_potential", 200
        ),
    input:
        network=RESOURCES + "networks/elec_s{simpl}_{clusters}_ec_l{ll}_{opts}.nc",
        config=RESULTS + "config.yaml",
    output:
        network=RESULTS + "networks/elec_s{simpl}_{clusters}_ec_l{ll}_{opts}.nc",
    log:
        solver=normpath(
            LOGS + "solve_network/elec_s{simpl}_{clusters}_ec_l{ll}_{opts}_solver.log"
        ),
        python=LOGS
        + "solve_network/elec_s{simpl}_{clusters}_ec_l{ll}_{opts}_python.log",
    benchmark:
        BENCHMARKS + "solve_network/elec_s{simpl}_{clusters}_ec_l{ll}_{opts}"
    threads: 4
    resources:
        mem_mb=memory,
        walltime=config["solving"].get("walltime", "12:00:00"),
    shadow:
        "minimal"
    conda:
        "../envs/environment.yaml"
    script:
        "../scripts/solve_network.py"

This triggers a workflow of multiple preceding jobs that depend on each rule’s inputs and outputs:

In the terminal, this will show up as a list of jobs to be run:

Building DAG of jobs...
job                         count    min threads    max threads
------------------------  -------  -------------  -------------
add_electricity                 1              1              1
add_extra_components            1              1              1
base_network                    1              1              1
build_bus_regions               1              1              1
build_hydro_profile             1              1              1
build_electricity_demand        1              1              1
build_powerplants               1              1              1
build_renewable_profiles        4              1              1
build_shapes                    1              1              1
build_ship_raster               1              1              1
cluster_network                 1              1              1
prepare_network                 1              1              1
retrieve_cost_data              1              1              1
retrieve_databundle             1              1              1
retrieve_natura_raster          1              1              1
simplify_network                1              1              1
solve_network                   1              1              1
total                          20              1              1

snakemake then runs these jobs in the correct order.

A job (here simplify_network) will display its attributes and normally some logs below this block:

[Mon Jan 1 00:00:00 2023]
rule simplify_network:
    input: networks/elec.nc, resources/costs.csv, resources/regions_onshore.geojson, resources/regions_offshore.geojson
    output: networks/elec_s.nc, resources/regions_onshore_elec_s.geojson, resources/regions_offshore_elec_s.geojson, resources/busmap_elec_s.csv, resources/connection_costs_s.csv
    log: logs/simplify_network/elec_s.log
    jobid: 4
    benchmark: benchmarks/simplify_network/elec_s
    reason: Missing output files: resources/busmap_elec_s.csv, resources/regions_onshore_elec_s.geojson, networks/elec_s.nc, resources/regions_offshore_elec_s.geojson; Input files updated by another job: resources/regions_offshore.geojson, resources/regions_onshore.geojson, resources/costs.csv, networks/elec.nc
    wildcards: simpl=
    resources: tmpdir=/tmp, mem_mb=4000, mem_mib=3815

Once the whole worktree is finished, it should state so in the terminal.

You will notice that many intermediate stages are saved, namely the outputs of each individual snakemake rule.

You can produce any output file occurring in the Snakefile by running

snakemake -call <output file>

For example, you can explore the evolution of the PyPSA networks by running

1. snakemake resources/networks/base.nc -call --configfile config/test/config.electricity.yaml

2. snakemake resources/networks/elec.nc -call --configfile config/test/config.electricity.yaml

3. snakemake resources/networks/elec_s.nc -call --configfile config/test/config.electricity.yaml

4. snakemake resources/networks/elec_s_6.nc -call --configfile config/test/config.electricity.yaml

5. snakemake resources/networks/elec_s_6_ec_lcopt_Co2L-24H.nc -call --configfile config/test/config.electricity.yaml

To run all combinations of wildcard values provided in the config/config.yaml under scenario:, you can use the collection rule solve_elec_networks.

snakemake -call solve_elec_networks --configfile config/test/config.electricity.yaml

If you now feel confident and want to tackle runs with larger temporal and spatial scope, clean-up the repository and after modifying the config/config.yaml file target the collection rule solve_elec_networks again without providing the test configuration file.

snakemake -call purge
snakemake -call solve_elec_networks

Note

It is good practice to perform a dry-run using the option -n, before you commit to a run:

snakemake -call solve_elec_networks -n

How to analyse results?

The solved networks can be analysed just like any other PyPSA network (e.g. in Jupyter Notebooks).

import pypsa

n = pypsa.Network("results/networks/elec_s_6_ec_lcopt_Co2L-24H.nc")

For inspiration, read the examples section in the PyPSA documentation.