Change DK-DK1 and DK-DK2 parser to use energinet instead of ENTSOE

[silkeholmebonnen]

Issue

GMM-1507

Description

This PR changes the DK-DK1 and DK-DK2 parser to use energinet hourly values instead of ENTSOE. The reason for this is that ENTSOE currently does not have accurate values, and for DK-DK2 ENTSOE has a ~4 week delay on wind offshore, which makes up approximately 50% of DK-DK2's total production.

Preview

Double check

• I have tested my parser changes locally with uv run test_parser "zone_key"
• I have run pnpx prettier@2 --write . and uv run format in the top level directory to format my changes.

[VIKTORVAV99]

It does look like it fixes the wind issue but it also looks like almost all coal are now biomass and there is lower gas and oil production + a marginal change in solar.

New Parser:

 {'correctedModes': [],
  'datetime': datetime.datetime(2026, 2, 17, 12, 0, tzinfo=zoneinfo.ZoneInfo(key='UTC')),
  'production': {'biomass': 725.750951,
                 'coal': 0.005718,
                 'gas': 105.70942,
                 'hydro': 0.0,
                 'oil': 7.822502,
                 'solar': 151.76184,
                 'wind': 290.031814},
  'source': 'energinet.dk',
  'sourceType': <EventSourceType.measured: 'measured'>,
  'storage': {},
  'zoneKey': 'DK-DK2'}

Old Parser:

 {'correctedModes': [],
  'datetime': datetime.datetime(2026, 2, 17, 12, 0, tzinfo=datetime.timezone.utc),
  'production': {'biomass': 483.57, 'coal': 218.89, 'gas': 124.6, 'oil': 37.69, 'solar': 160.45, 'wind': 114.06},
  'source': 'entsoe.eu',
  'sourceType': <EventSourceType.measured: 'measured'>,
  'storage': {},
  'zoneKey': 'DK-DK2'},
 {'correctedModes': [],
  'datetime': datetime.datetime(2026, 2, 17, 12, 15, tzinfo=datetime.timezone.utc),
  'production': {'biomass': 481.43, 'coal': 207.38, 'gas': 117.71, 'oil': 35.5, 'solar': 152.29, 'wind': 115.48},
  'source': 'entsoe.eu',
  'sourceType': <EventSourceType.measured: 'measured'>,
  'storage': {},
  'zoneKey': 'DK-DK2'},
 {'correctedModes': [],
  'datetime': datetime.datetime(2026, 2, 17, 12, 30, tzinfo=datetime.timezone.utc),
  'production': {'biomass': 480.32, 'coal': 202.18, 'gas': 115.45, 'oil': 34.77, 'solar': 150.17, 'wind': 113.29},
  'source': 'entsoe.eu',
  'sourceType': <EventSourceType.measured: 'measured'>,
  'storage': {},
  'zoneKey': 'DK-DK2'},
 {'correctedModes': [],
  'datetime': datetime.datetime(2026, 2, 17, 12, 45, tzinfo=datetime.timezone.utc),
  'production': {'biomass': 476.24, 'coal': 200.0, 'gas': 114.13, 'oil': 34.34, 'solar': 144.95, 'wind': 107.62},
  'source': 'entsoe.eu',
  'sourceType': <EventSourceType.measured: 'measured'>,
  'storage': {},
  'zoneKey': 'DK-DK2'}

If our goal is forecast performance I'm not sure this is "better" as more stuff changes here and our external solar and wind forecasts for DK-DK2 from energidataservice.dk looks pretty good.

So honestly don't know if this is better or worse.

[silkeholmebonnen]

It does look like it fixes the wind issue but it also looks like almost all coal are now biomass and there is lower gas and oil production + a marginal change in solar.

If our goal is forecast performance I'm not sure this is "better" as more stuff changes here and our external solar and wind forecasts for DK-DK2 from energidataservice.dk looks pretty good.

So honestly don't know if this is better or worse.

I trust the data coming straight from energinet more than what is reported through entsoe. And the wind issue is a pretty big issue when it is often around 50% of production.

Regarding the forecast, I suspect that it might be because it has been trained when we had both onshore and offshore wind power available. But as you can see in the picture below this means that the gap is quite big from what we say is in the grid right now (because we dont include wind offshore), to what we forecast (which might include both).

[VIKTORVAV99]

I don't think we can say that just because it's coming from energinets platform it's more accurate as it's energinet that is reporting to ENTSO-E as well.

Had it just been a few mw differences they could just be rounding things differently and such but with this big of a change we need to be able to explain why if we change. Not just internally but also externally as it will have a big impact on the carbon emissions.

The wind issue is a problem but I don't want to replace it with another one.

[silkeholmebonnen]

I converted this to draft since Energinet is not clearly better than ENTSOE

[xaverdh]

The ENTSO-E data has been wrongly reporting biomass as coal for many years. This comes mostly from fuel switch in power stations, that are still reported as coal fired, even if they are using biomass. According to my information (news mostly, but I have been following this for some time), the only significant coal fired plant in Denmark today is Nordjylland in DK1.

The best approach would probably to reach out to ENTSO-E and get them to report correct data. Otherwise what kind of source / reference would you accept as authoritative?

[VIKTORVAV99]

It's not ENTSO-E that are reporting the data it's ENERGINET that are reporting data to ENTSO-E that simply re-publishes the data they get. This means we can't be sure which source is correct since it's ENERGINET that is reporting data for both platforms.

We can however ask ENTSO-E to reach out to ENERGINET and validate that the data is correct.

[xaverdh]

So according to this list: https://ens.dk/media/7199/download from https://ens.dk/analyser-og-statistik/data-oversigt-over-energisektoren
in 2024 the power stations with >= 100TJ generation from coal (kul_TJ) were:

• Fyns (converted to gas fired in 2024: https://greenpowerdenmark.dk/nyheder/kulstop-giver-mindre-co2-udslip-milepael-fyn-hele-danmark)
• Studstrupværket (converted to biomass, the last coal unit switched off in 2024)
• Nordjyllandsværket (in DK1, still running on coal till 2028)
• Esbjergværket (closed 2024: https://orsted.com/en/media/news/2024/08/orsted-shuts-down-its-last-coal-fired-heat-and-pow-14000900)