Within capacity scaling, detect zones collisions and add ability to specify target area type

[danielolsen]

Purpose

When attempting to refactor the quick bite script to calculate shortfall for future scenarios, @BainanXia and I found a bug: if there are two target names which both include the same zone, the demand from that zone will be arbitrarily assigned to one or the other. E.g. if we include Texas and El Paso, Texas will by default look up the state of Texas which includes El Paso as one of its zones. We would like to detect this condition and return an error to the user to ask them to be more specific.

This can create a new challenge, where in some scenarios when we say Texas we mean just the interconnection, with El Paso specified separately. In this case, we would like to be able to optionally specify an area_type for each target in the input CSV.

What is the code doing

• in load_targets_from_csv(), we add another optional column area_type with a default value of np.nan.
• in _make_zonename2target() we refactor the input parameters to expect a dataframe, not just a list of strings: this is so that we can check for a non-nan entry in the area_type column when looking up which zones are in each area, and if we find one then we pass that along to the area_to_loadzone() lookup. This also allows us to remove the enforced_area_type input, since these are enforced target-wise via the dataframe. We also detect the case when there is a zone collision, and raise an error message.
• in add_resource_data_to_targets() and add_demand_to_targets(), we update the calls to _make_zonename2target().

Validation

If we pass a CSV with both Texas and El Paso, we get an error message: ValueError: zone in two target areas: El Paso in {'Texas', 'El Paso'}.

If we correct this by specifying an area_type of 'interconnect' for Texas, then we get a new error, demonstrating pre-existing checks: ValueError: Targets do not cover all load zones. Missing: {'East Texas', 'Texas Panhandle'}. When we add these, or when we remove El Paso (since in this case we do want to include it with the rest of Texas) the process runs as before.

Time to review

15 minutes. This is a relatively low priority, but will help us avoid mistakes when analyzing future scenarios.

[BainanXia]

This is consistent with our discussion. Thanks!

[danielolsen]

@jon-hagg I believe I've addressed all of your comments, please let me know if you think the latest version looks good.

[jenhagg]

Couple ideas, mostly for readability

• I think we can define duplicates using a set comprehension, e.g. {x for x in ...}
• We could also define zone_sets_list = [z for s in zone_sets for z in s]

[danielolsen]

Good call! I updated those comprehensions for more readability, as well as the error_areas comprehension.

[jenhagg]

@jon-hagg I believe I've addressed all of your comments, please let me know if you think the latest version looks good.

Thanks, looks good! I put a couple more comments if you're interested, but I'll go ahead and approve.