Closed irm-codebase closed 3 weeks ago
With Global Energy Monitor, we have an increasing amount of retired power plants in the powerplants.csv
, which is useful for reproducing historical market outcomes in 2020 or other years. I think this is where the discrepancy comes from.
To get capacities for 2022, apply some filters:
import powerplantmatching as pm
powerplants = pm.powerplants()
query = "(DateOut > 2022 or DateOut != DateOut) and (DateIn < 2023 or DateIn != DateIn)"
powerplants.query(query)
There will also be some trimmings in #163 with very good alignment with country-level statistics (rooftop solar and utility-scale below 1MW not included):
Stats have been created with the ENTSO-E Python API.
from entsoe import EntsoePandasClient
from powerplantmatching.cleaning import gather_fueltype_info
import pandas as pd
client = EntsoePandasClient(api_key=ENTSOE_TOKEN)
start = pd.Timestamp("20220101", tz="Europe/Berlin")
end = pd.Timestamp("20230101", tz="Europe/Berlin")
kwargs = dict(start=start, end=end, psr_type=None)
def parse(c):
rename = {"GB": "UK"}
for n in range(2):
try:
print(c, n)
return client.query_installed_generation_capacity(rename.get(c, c), **kwargs).iloc[0]
except Exception as e:
print(f"Country {c} failed with {e}")
time.sleep(3)
return np.nan
stats = pd.DataFrame({c: parse(c) for c in powerplants.Country.unique()})
fueltypes = gather_fueltype_info(pd.DataFrame({"Fueltype": stats.index}), ["Fueltype"])
stats = stats.groupby(fueltypes.Fueltype.values).sum().unstack()
# https://de.wikipedia.org/wiki/Liste_von_Wasserkraftwerken_in_der_Schweiz?oldformat=true
stats.loc["CH", "Hydro"] = 17038
During some testing, I re-ran the default configuration:
plants = ppm.powerplants(update=True)
The resulting dataset is similar to the documentation, except for a big increase in Hard Coal facilities:
The seems like the statistics are from 2018(*?), but I doubt that Europe had such a massive increase in coal facilities in the last half decade... Maybe some double counting is happening due to changes in one of the source datasets?