Develop shapefiles for balancing authority regions and sub-regions

[danielolsen]

🚀

• Is your feature request essential for your project?

Describe the workflow you want to enable

I wish I had a shapefile that contained geometries for as many of the contiguous U.S. balancing authority regions and sub-regions as possible. The ideal set of regions and sub-regions includes all those present in the dataset at https://zenodo.org/record/4116342.

Describe alternatives you've considered, if relevant

• We currently have a rough county-to-BA mapping (see prereise/gather/data/BA_County_map.json), but this performs poorly when a county is split between multiple balancing authorities.
• There are datasets within the HIFLD data on Control Areas and Utility Areas, but they're mostly county-based as well and have lots of overlap.
• NREL has a map for utilities that may have better granularity, but their labeling leaves a lot to be desired when trying to aggregate back up to balancing authorities. https://maps.nrel.gov/solar-for-all/ (select Electric Utility from the Utilities tab on the left, and then you can download the shapefile).
• Some other scattered data are available publicly, e.g. NYISO sub-regions and CAISO sub-regions.

Additional context

It's most important to get the large balancing authorities and their sub-regions correct. Smaller balancing authorities can be merged with larger surrounding balancing authorities as necessary.

[limingzhou2004]

Is the question how to aggregate the utility zones to form finer control areas (BA) than HIFLD?

[limingzhou2004]

If a utility zone intersects with only one BA in the HIFLD, we can safely assign it. If more than one BAs, then there seems no reliable way to label that zone. One way is to search the neighbor zones fo the same company, and assign it based on the majority BAs.

[danielolsen]

The ultimate goal is a set of geometries (ideally non-intersecting, or minimally intersecting), one for each BA, that can be used for further downstream analyses. Downstream examples: using other data corresponding to defined geometries (e.g. data for census regions) to develop load models, defining which buses are in which load zones for use in OPF simulations (where we define a load shape for each zone and distribute the total zone demand across all buses classified as being within that zone), creating new sets of load zones (e.g. by looking at combinations of state geometries and balancing authority geometries to create more granular zones to get more control over how we build scenarios for simulation).

[limingzhou2004]

NREL's data are finer, but do not provide the relationship to BA or ISO zones. The utility company name seems not helpful. NYISO provides the town/village relationship to the ISO zones.
Should this type of data be of more help?

[danielolsen]

We have the shapes for the NYISO regions already, so we don't need to get them by looking at the town names. For other states though, we could potentially use something like that to generate a Voronoi diagram to approximate the boundaries.

[danielolsen]

We may be able to use a mapping of ZIP codes to utilities with this data set: https://data.openei.org/submissions/4042, and then map utilities to balancing authority areas.

EDIT: after further conversations, it seems that one utility may span several balancing authorities or sub-regions, so relying on utilities to build sub-regions may not work very well. This may be BA-dependent though, with some BAs having 'cleaner' sub-regions than others.

[limingzhou2004]

The newer version of BA_map is uploaded to dropbox, . We can disaggregate the subregional load by the census tract population. Another way is to use HIFLD retail service data. The latter provides the annual retail/whole sale power, which should be better than population for a disaggregation.

[limingzhou2004]

The excel file to organize the BA shape files is uploaded to the dropbox. Looks we got all control areas covered by Zenoko, except PJM_RECO. Some issues are listed in the readme tab.

[limingzhou2004]

The control areas have lots of overlaps. One CA can even be located within another CA. Although at the finest spatial resolution, they do not overlap but intertwined, we don't have the data to identify them, nor is it necessary. For the areas with heavy overlaps, we can define the spatial areas over multiple CAs, and aggregate the CA load to the defined areas. For example, we define FL as one area, and four CAs fall into FL with significant overlap amongst them, then we aggregate the four CA load into a single FL load. It doesn't give the perfect accuracy, but we are actually getting the best feasible spatial resolution.

[danielolsen]

I know #270 is merged, but I want to keep this Issue open for now until we finish #293, since we may identify further splits in our current regions/sub-regions during that effort.