missing plants in EIA 923 monthly generation table

[mfripp]

I'm trying to pull power generation data from EIA form 923. But it seems that a lot of plants, especially nuclear ones, don't make it from the EIA website into the pudl table. I'm looking in core_eia923__monthly_generation_fuel on the pudl website and comparing it to the "EIA923_Schedules_2_3_4_5_M_12_2024_Final.xlsx" spreadsheet downloaded from the 2024 section at https://www.eia.gov/electricity/data/eia923/.

For example, generators with plant_id_eia 7722, 8036 and 8055 all appear in the EIA spreadsheet with positive net gen MWh values throughout the year, but none of them appear at all in the pudl table. Is this a known problem or am I doing something wrong?

[mfripp]

I just checked and these data are also available via the EIA API, e.g., this link brings up lots of data for plant 7722.

[mfripp]

Just saw the note on core_eia923__monthly_generation saying that "This is the most granular information we have about how much electricity individual generators are producing, but only about half of all the generation reported in the core_eia923__monthly_generation_fuel appears in this table due to the different reporting requirements imposed on different types and sizes of generators."

And also saw the note on core_eia923__monthly_generation_fuel_nuclear that "This data is originally reported alongside similar information for fossil fuel plants, but the nuclear data is reported by (nuclear) generation unit rather than fuel type and prime mover, and so has a different primary key."

And indeed many of the generators I'm interested in are incore_eia923__monthly_generation_fuel_nuclear or possibly core_eia923__monthly_generation_fuel, so I'll plan on using those instead.

I realize now that I was confusing the monthly_generation table with the monthly_generation_fuel table (even as I mentioned monthly_generation_fuel, I thought I was posting about monthly_generation). I see that the key distinction is that the monthly_generation table shows data by plant and generator, while the other two show it by plant, fuel and prime mover type, but not at the level of individual generators.

I also failed to notice that nuclear plant data is moved to monthly_generation_fuel_nuclear. The code I'm working with combines this with monthly_generation_fuel in some places but not others, and that was why the nuclear plants disappeared in some cases.

[aesharpe]

@mfripp thank you for documenting your process! I'm sure this will be helpful to others.

Do you have any suggestions for how we could improve the documentation to make this nuance more readily understandable for folks using PUDL? Also, if there a table configuration that would be more convenient for you as a user, let us know!

[zaneselvans]

@mfripp as you've sussed out, the way fuel consumption and corresponding net generation is reported in EIA is unfortunately complicated and non-uniform across different types of generators / respondents. These core_ tables are the building blocks that we use to create other more immediately usable wide-format out_ tables. The big one that currently contains our best estimates of per-generator values is out_eia__monthly_generators, but to get those estimates it necessarily makes some opinionated decisions about how to allocate fuel / net generation across boilers / generators / generation units.

We should add a new Methodology page to the documentation that explains what's going on in there.