First, thank you for the clean and consistent data with easy access. It is very user friendly, and the EMI staff on the forum provide really helpful context and responses.


I am a researcher looking for historic time series datasets of actual/realized generation at power plants and generation facilities. The more granular the better! From browsing the wholesale datasets and also the forum, there seem to be a couple of possible datasets that would meet my immediate need. I also understand there is a planned revamp of the generation data backend at some point in the future (maybe 2019?).

Could you please confirm/correct these 3 assumptions and provide some guidance if there are better options?

(1) The Generation MD dataset has grid-injection values at power plant POCs, so the values are essentially 'Net Generation of Plant X' (total generation at plant minus power used and lost before the connection). The values are 'into-grid import', and not 'net of into-grid import and out-of-grid export'.

(2) The Unit-level Generation IR dataset does not explain what the values mean, but they seem to be 'Gross generation at Unit A of Plant X'. There are far fewer plants covered by this dataset than the the Generation MD dataset because fewer plants report unit-level generation. The values are measured on-site by the power producer and are not measurements at a POC.

(3) Both of these datasets are measured ~actual generation figures, and are not specifically bids or agreements to supply this this much power during a trading period.


I have noticed the Generation MD and sum of Unit-level IR data are close in magnitude for any trading period, but they are not exactly the same. While I would prefer to have a single source (and semantic meaning) for the generation values, the additional unit disaggregation might be useful if it can be made into comparable values with the plant-level data.


Thanks in advance.


Logan Byers


Thank You