This is a reimplementation of the electricity maps EU emission factors estimation method from this python script, which is similar to the method described by Unnewehr et al. [1]. In comparison to the python script, it is supposed to enable easier manual matching of generation units with their respective ETS emission data, and hopefully calculate heat emissions a bit better (see below for a description).
Example output of this tool is in "data/[year]/output" of this repository. Of course, since Entso-E regularly publishes corrections for old data, output will most likely differ a bit when downloading currently published Entso-E data and running the tool.
To start preprocessing data for one given year, some input data is needed:
- The "data/[year]/entsoe_unit_generation" directory needs to contain raw Entso-E unit generation data for the year in question, namely 12 zip files "[year]_[month]_ActualGenerationOutputPerGenerationUnit_16.1.A.zip". These can be downloaded from Entso-E sftp server after registration, from the server's "zip" directory.
- The "data/verified_ets_emissions" directory needs to contain the ETS "verified emissions" xlsx file. It needs to be renamed to "verified_emissions.xlsx", and can be downloaded from EU ETS, in "Documentation -> Phase [I-IV] -> Reports -> Verified Emissions for [year]". Files from previous years can probably be used as long as they include data for the year in question, but newer files may include corrections for past years so it is recommended to always use the most recently released file (releases seem to happen in April).
After the 13 files are placed inside the corresponding directories, data
preprocessing can start: cargo run --release -- preprocess <year>. This will
process the raw data and create two csv files in "data/[year]/preprocessed", one
for electricity generation per generation unit, and one for ETS emissions and
free-of-charge ETS allocations in the given year.
Note that raw input data does not appear to be available under a free license, so the generated csv files in the "preprocessed" dir may not be freely redistributable, as they're basically just copies / aggregated values of parts of the original "raw" files.
In addition to data from the preprocessing step, the "Energy statistics - cooling and heating degree days (nrg_chdd)" database from Eurostat is needed in the "data/degree_days" directory. It's a file called nrg_chdd_a.tsv. The file is freely redistributable, can be downloaded from the Eurostat energy database, and includes data from past years, so the most recent release is recommended.
The generation and emissions data sets have no common identifier for generation units / power plants, so generation units need to be matched with their respective emissions data in order to estimate emission factors. An automatic matching is attempted based on unit names. In addition, a file "data/[year]/manual_matches.csv" needs to be created for each year, although they can be largely the same between different years with only minor updates. The file can be used to manually map electricity generation data to emission data. The file should look like this, with each line defining one match:
generation,emission,settings,comment
Irsching 4,Kraftwerk Irsching Block 4,,
GKM AG DBEnergie|GKM AG TNG|GKM AG Amprion,Grosskraftwerk Mannheim,,
,,,This is a general comment
IKS Schwedt SE1 Block 1|IKS Schwedt SE2 Block 2,,,"only combined refinery + power plant ETS data is available"
eic:11WD2HANN5C---1X|eic:11WD2HANN5C---2V,GKH - Gemeinschaftskraftwerk Hannover GmbH,,
Plock B01,id:PL-0391-05:455,,
Rya KVV,Rya Kraftvärmeverk,plausible-emission-factor-range:200-600,"uses some biomass"
The generation field lists one or more generation unit names from the
"data/[year]/preprocessed/powerplant_generation.csv" file, separated by "|". The
emission field lists zero or more ETS data installation names from the
"data/[year]/preprocessed/powerplant_emissions.csv" file, also separated by "|".
The settings field may contain some settings to override default behavior,
separated by "|". Currently supported:
plausible-emssions-factor-range:[min]-[max]. Default is300-3000. If the emission factor calculation result for this plant is outside of the given range, the plant is ignored during country-level data aggregation.
The optional comment field may contain some text to explain the line. If the
generation and emission fields are empty, the line is ignored – so that's a
way to add general comments to the file.
In the example above, the "IKS Schwedt" line has an empty emission field. This
causes the generation unit(s) of that line to be ignored in the emission factors
calculation. No automatic matching to emissions data will be attempted for these
generation units.
Sometimes names for generation units or emissions data sets are too generic. For example, the Hannover power plant consists of two blocks, and their generation unit names in the Entso-E dataset are "Block 1" and "Block 2", respectively. In a case like this, the generation unit should be specified as "eic:[EIC of the generation unit]" as demonstrated above. Similarly, generic ETS record names like "ELEKTROCIEPŁOWNIA" for the Plock power plant (and other power plants in Poland) should be specified as "id:[permit_id]:[installation_id]". The value to append after "eic:" or "id:" is listed in the "powerplant_generation.csv" and "powerplant_emissions.csv" files, respectively.
Some matches are difficult to figure out, but this process usually works: Geocoordinates of generation units can be found based on their name and/or EIC using a search engine. Next, find the matching ETS installation on the map at euets.info. The installation name is displayed in a tooltip when hovering the mouse cursor over over the dot on the map. If the installation name is generic and there are multiple installations with that name, clicking the dot will show details about verified emissions and allocations per year, which can be used to find the correct match and reference it via "id:[permit_id]:[installation_id]".
Once preprocessing is done and a "manual_matches.csv" was created, emission
factors can be estimated: cargo run --release -- <year>. This will create
three csv files in "data/[year]/output":
- The estimated emission factors and related data for all relevant power plants that were successfuly matched to emissions data and passed some plausibility testing.
- Ignored power plants along with the reason why they were ignored. Depending on what exactly went wrong, it's often possible to fix this by adding a line to manual_matches.csv.
- Emission factors and other data aggregated at the country level, grouped by fuel type.
An attempt is made to estimate emissions caused by heat production, based on the number of free-of-charge ETS allocations. The basic idea is described by Unnewehr et al. [1, p. 5] and was previously used by Hermann et al. [2, p. 151]. These heat-related emissions are subtracted from the total emissions of each plant before calculating the emission factor for electricity generation. Details are described below.
Currently, the code only handles years 2020-2025.
Power plants receive free-of-charge ETS emission allowances if they provide heat for district heating or industry. This process is called "(free) allocations". The total number of allocations (corresponding to metric tons of CO2) per year for each power plant is published by ETS, along with the verified emissions for each year. The number of allocations depends on the yearly average of heat provided by the power plant in the "baseline period" 2014-2018. This baseline is multiplied by the "heat benchmark" number specified by the EU to calculate the so-called "preliminary allocation". The preliminary allocation is reduced by up to two different factors to calculate the actual allocations:
- For "privileged" and "non-privileged" heat: Multiplying by the "linear reduction factor", which was 0.8782 in 2020 and is reduced by 0.022 every year. This is the "beta" factor in this code base.
- For "non-privileged" heat only: Multiplying by the "carbon leakage exposure factor", which was fixed to 0.3 for the 2020-2025 time period. This is the "gamma" factor in this code base.
If some industry is determined to be at risk of shutting down because of high ETS/CO2 costs, and moving production outside of the EU, that is called "risk of carbon leakage". The factor of 0.3 is not applied for these "high carbon leakage risk" use cases, so these industries receive more free-of-charge allocations and hopefully stay competitive globally.
To determine the amount of heat provided by combined heat and power (CHP) plants from the number of free-of-charge ETS allocations, the share of "privileged" vs. "non-privileged" heat provided by each CHP plant needs to be calculated first. This happens in the preprocessing step. The method is like the one used in the source code of [3], but only years 2018 and 2019 are considered. The resulting "sigma" value is the share of "privileged" heat (at high risk of carbon leakage) provided by the CHP plant. For example, a value of sigma=0 might indicate that the CHP plant only provides heat for district heating, a very common situation, which is not at risk of carbon leakage. A value of sigma=0.5 might indicate that half of the heat provided by the plant is delivered to industry at risk of carbon leakage, and the other half is used for district heating.
The sigma calculation is possible from free-of-charge allocations data pre-2020. In that time period, heat provision had the "linear reduction factor" (beta) applied to its preliminary allocation like today (using a slightly different formula). However, the gamma factor was not fixed to 0.3, but was going from 1.0 to 0.3 over a few years. So if a power plant had its number of yearly allocations decrease only very slightly over the years, that indicates "privileged heat", so sigma=1.0. For power plants where the number of allocations decreased rapidly every year, that indicates sigma=0. By observing the reduction of the number of allocations from 2018 to 2019, a sigma value is estimated for every relevant ETS record in the preprocessing step [3, source code].
It is no longer possible to estimate sigma from 2020+ data, because the gamma factor was fixed to 0.3 for "non-privileged" heat and is still fixed to 1.0 for "privileged" heat. So these estimates might be out of date for some power plants, resulting in incorrect heat calculations.
Knowing the number of free-of-charge allocations and estimated sigma for each plant, the procedure described above for calculating allocations is applied in reverse to get the estimated number of preliminary allocations. Using the ETS heat benchmark, the yearly average for heat provided by each power plant in the baseline period 2014-2018 is calculated from the number of its estimated preliminary allocations. To account for colder/warmer winter temperatures in the current year, the "heating degree days" of each country (as provided by Eurostat) for the baseline period vs. the current year are used to scale the heat provided in the baseline period up or down to estimate the amount of heat provided in the current year.
This is obviously not a great estimation for the actual amount of heat provided in a given year. The CHP plant might have been offline for extended maintenance / upgrades / fuel switch / ... in a given year, reducing the amount of heat provided. And homes might have been upgraded with better insulation, reducing the need for heat from the CHP plant. Local heat-consuming industry might have shut down since the baseline period. And so on. However, it might be the best estimation possible.
Finally, knowing the amount of heat and electricity generated in the current year as well as the total verified emissions, the emissions are split between electricity generation and heat generation using the "efficiency method" [4, p. 6]. Currently, efficiency for heat generation is assumed to be 80% and power generation 35%, as recommended for US CHP plants [4, p. 9].
[1]: J. F. Unnewehr, A. Weidlich, L. Gfüllner and M. Schäfer, "Open-data based
carbon emission intensity signals for electricity generation in European
countries – top down vs. bottom up approach," Cleaner Energy Systems, vol.
3, 2022, doi:
10.1016/j.cles.2022.100018.
[2]: H. Hermann, F. Matthes and V. Cook, "Die deutsche Braunkohlenwirtschaft.
Historische Entwicklungen, Ressourcen, Technik, wirtschaftliche Strukturen
und Umweltauswirkungen", May
2017.
[3]: T. N. Schubert, G. Avenmarg, J. F. Unnewehr and M. Schäfer, "Generation
and emission data for main power plants in Germany (2015 - 2021)," Zenodo,
2022, doi: 10.5281/zenodo.7316186.
[4]: "Allocation of GHG Emissions from a Combined Heat and Power (CHP) Plant."
GHG Protocol.
https://ghgprotocol.org/sites/default/files/2023-03/CHP_guidance_v1.0.pdf
(accessed Jun. 5,
2023).
Licensed under either of
- Apache License, Version 2.0 (LICENSE-APACHE or http://www.apache.org/licenses/LICENSE-2.0)
- MIT license (LICENSE-MIT or http://opensource.org/licenses/MIT)
at your option.
Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.