Research for Global Powerplant Database project
The Global Powerplant Database project is working to build an open, global database of powerplant data. It is inspired by projects like CARMA, Enipedia, Ventus, and CoalSwarm.
A key challenge in powerplant data is estimating missing data, such as annual or monthly generation (in MWh). This repo contains code designed to test various research approaches to improving the estimation methodology for powerplant data.
Generation is likely to be correlated to fuel cost. Monthly fuel cost and generation data are available for (almost) all US powerplants from EIA, based on Form EIA-923 data.
Unfortunately, these data are structured inconsistently, and need to be restructured and aggregated to allow comparison. To do this, we first obtain complete data for several years from EIA-923. We then manually export ‘Page 4’ (generation data) and ‘Page 5’ (fuel cost data) to two separate CSV files.
The script process_eia_data.py takes these two CSV files as parameters, reads them, and builds a dictionary of powerplant objects with the plant id as the key. Monthly generation and fuel cost data are read and saved to attributes on the appropriate powerplant object.
The script then outputs a CSV file with each powerplant as one row, and monthly fuel cost and generation data as columns. Currently, the script works for data back to 2009, with the results saved in the EIA-data-processed directory.