Correct Gross to Net Generation Bugs

[grgmiller]

Purpose

As I was inspecting the OGE outputs, I noticed some issues with the gross to net generation data that were not fixed by #370. Specifically:

• There were a number of plants where zero gross generation data was being reported in CEMS where nonzero net generation data was being reported in EIA-923. However, since we switched to only using GTN ratios, this meant that we were calculating zero net generation for all of these plants. This PR re-introduces the shift factors, and specifically applies them in cases where gross generation is 0, so that we do not inadvertently under-report net generation data. This effectively applies a flat generation profile to the data, but ensures that the annual total net mwh will match the data in EIA.
• Broaden the range of acceptable GTN ratios from 0.75-1.25 to 0.5-1.25. This is based on some updated analysis of the GTN data, and also based on an email conversation with EIA in which they indicated that the EIA net generation is likely more trustworthy than the CEMS gross generation, and so discrepancies in the data should favor the EIA values (a narrower range of acceptable ratio implicitly trusts the CEMS data over the EIA data). This broader range covers about 90% of the GTN ratios, so only filters out the highest and lowest 5%.

What the code is doing

All of the subplant shift code that was added was previously in the OGE codebase prior to #370. The only new code is l.624-633 in gross_to_net_generation, which explicitly applies this to instances with zero gross generation, rather than applying this in a hierarchy.

Testing

Ran the pipeline for 2018 and examined the logs and outputs

Where to look

• It's helpful to clarify where your new code lives if you moved files around or there could be confusion/

• What files are most important?

Usage Example/Visuals

How the code can be used and/or images of any graphs, tables or other visuals (not always applicable).

Review estimate

10 min.

Future work

What issues were identified that are not being addressed in this PR but should be addressed in future work?

Checklist

• Update the documentation to reflect changes made in this PR
• Format all updated python files using black
• Clear outputs from all notebooks modified
• Add docstrings and type hints to any new functions created