EIA Form 923 – Power Plant Operations Report¶
Source URL |
|
|---|---|
Source Description |
The EIA Form 923 collects detailed monthly and annual electric power data on electricity generation, fuel consumption, fossil fuel stocks, and receipts at the power plant and prime mover level. |
Respondents |
Electric, CHP plants, and sometimes fuel transfer termianls with either 1MW+ or the ability to receive and deliver power to the grid. |
Records Liberated |
~5 million |
Source Format |
Microsoft Excel (.xls/.xlsx) |
Download Size |
332 MB |
Temporal Coverage |
2001-2024 |
PUDL Code |
|
Issues |
PUDL Database Tables¶
We’ve segmented the processed data into the following normalized data tables. Clicking on the links will show you a description of the table as well as the names and descriptions of each of its fields.
Background¶
Form EIA-923 is known as the Power Plant Operations Report. The data include electric power generation, energy source consumption, end of reporting period fossil fuel stocks, as well as the quality and cost of fossil fuel receipts at the power plant and prime mover level (with a subset of +10MW steam-electric plants reporting at the boiler and generator level). Information is available for non-utility plants starting in 1970 and utility plants beginning in 1999. The Form EIA-923 has evolved over the years, beginning as an environmental add-on in 2007 and ultimately eclipsing the information previously recorded in EIA-906, EIA-920, FERC 423, and EIA-423 by 2008.
As of 2019, the EIA-923 Form is organized into the following schedules:
Schedule 2: fuel receipts and costs
Schedules 3A & 5A: generator data including generation, fuel consumption and stocks
Schedule 4: fossil fuel stocks
Schedules 6 & 7: non-utility source and disposition of electricity
Schedules 8A-F: environmental data
Download additional documentation¶
Data available through PUDL¶
Some form of EIA-923 data stretches all the way back to 1970. PUDL coverage starts in 2001. Before 2001, the data like that contained in the EIA-923 was collected via EIA-867, EIA-906, and EIA-920. Prior to 2008 fuel receipts and costs data were reported in EIA Form 423 (which in turn replaced the earlier FERC Form 423). If you’re interested in this earlier data, get in touch with us!
Monthly interim EIA-923 data are periodically integrated into PUDL as well. Incomplete year-to-date data are excluded from the annualized tables to avoid confusion. We have not yet integrated tables reporting fuel stocks, data from Puerto Rico, or EIA-923 schedules 6, 7, and 8.
Who submits this data?¶
Respondents include all all electric and CHP plants, and in some cases fuel transfer terminals, that have a total generator nameplate capacity (sum for generators at a single site) of 1 Megawatt (MW) or greater and are connected to the local or regional electric power grid.
Selected plants may be permitted to report schedules 1-4B monthly and 6-8 annually so as to lighten their reporting burden. All other respondents must respond to the Form in its entirety once a year.
What does the original data look like?¶
Once the respondents have submitted their responses, the EIA creates a series of spreadsheets that reflect themes within the form. These spreadsheets have changed over the years as the form itself evolves. They are accessible on the EIA website as downloadable ZIP files categorized by year. The internal data are organized into excel spreadsheets. To gain greater insight into year-to-year nuances of the form, we recommend downloading multiple years of EIA-923 ZIP files and comparing both the Form and the Form Instructions files.
Notable Irregularities¶
File Naming Conventions¶
The naming conventions for the raw files are confusing and difficult to trace year to year. Subtle and not so subtle changes to the form and published spreadsheets make aggregating pre-2009 data difficult from a programmatic standpoint.
Protected Data¶
In accordance with the Freedom of Information Act and the Trade Secrets Act, certain information reported to EIA-923 may remain undisclosed to the public until three months after its collection date. The fields subject to this legislation include: total delivered cost of coal, natural gas, and petroleum received at non-utility power plants and the commodity cost information for all plants (Schedule 2).
Net generation & fuel consumed reported in two separate tables¶
Net generation and fuel consumption are reported in two separate tables in EIA-923: in the core_eia923__monthly_generation and core_eia923__monthly_generation_fuel tables. The core_eia923__monthly_generation_fuel table is more complete (the core_eia923__monthly_generation table includes only ~55% of the reported MWh), but the core_eia923__monthly_generation table is more granular (it is reported at the generator level).
Data Estimates¶
Plants that did not respond or reported unverified data were recorded as estimates rolled in with the state/fuel aggregates values reported under the plant id 99999.
Boiler Fuel Primary Keys¶
The core_eia923__monthly_boiler_fuel table has several sneaky primary keys and duplicate rows.
The main primary keys for the table are: plant_id_eia, boiler_id, energy_source_code, prime_mover_code,
report_date. There are some rows that also differ based on associated_combined_heat_power, due
to mid-year retirement of units that are assocated with combine heat and power systems, and
operator_name, due to lenient standards for string columns (the all have the same operator_id
value). We drop both the associated_combined_heat_power and operator_name fields from the final
normalized table, causing duplicate rows. Luckily, these rows don’t provide any conflicting information.
Because they are the same plant, when one row contains an NA value, the other contains a numeric value.
We can easily drop duplicates based on which rows contain NA values with no duplicate value reconciling
necessary.
There are still more duplicate rows with identical qualitative plant information. Luckily, none of these duplicates contain conflicting information either. All duplicate rows have at least one row containing solely NA and 0 values.
To address both issues at once, we drop all the duplicate rows with NA or 0 values in the non primary key columns. One side affect of this is that duplicate rows where both rows contain NA and 0 values will both get dropped. This leads to gaps in the data where certain months are missing. These values can be assumed to be 0 or NA.
Boiler Fuel Years¶
The core_eia923__monthly_boiler_fuel table reports all months in a given year, even if there is no data. At present, we haven’t truncated the data after the most recently integrated month, so you will see all months.
Fluctuations in row count between each quarterly update are therefore due to changes in primary key quirks as described above.
PUDL Data Transformations¶
To see the transformations applied to the data in each table, you can read the
docstrings for pudl.transform.eia923 created for each table’s
respective transform function.