feat: powerflow snapshot figure

[danielolsen]

Purpose

Add plot of powerflow snapshot to the codebase.

What the code is doing

plot_powerflow_snapshot is the main function, aggregate_plant_generation and plot_arrows are helper functions.

Testing

Tested against the notebooks that @victoriahunt used to generate figures for the Macro Grid report, the results are nearly identical (small differences in plot order, some line thicknesses, etc.). I figure that those small details can be ironed out in parallel with the main code review.

Usage Example/Visuals

Using a CSV of demand center locations, we can reproduce the USA powerflow snapshots with e.g.:

>>> from powersimdata.scenario.scenario import Scenario
>>> from postreise.plot.plot_powerflow_snapshot import plot_powerflow_snapshot
>>> import pandas as pd
>>> plot_powerflow_snapshot(Scenario(1270), '2016-11-02 22:00:00', b2b_dclines={"from": range(7), "to": [7, 8]}, demand_centers=pd.read_csv("loadzone_centers.csv", index_col=0), legend_font_size=20)

Original notebook:

Reproduction:

To reproduce the Texas figure, which features a zoomed-in area, different scalings, and four arrows per DC line:

>>> plot_powerflow_snapshot(Scenario(1270), '2016-11-03 00:00:00', x_range=(-11.9e6, -10.2e6), y_range=(2.7e6, 4.4e6), figsize=(800, 800), b2b_dclines={"from": range(7), "to": [7, 8]}, demand_centers=pd.read_csv("loadzone_centers.csv", index_col=0), pf_width_scale_factor=0.002, bg_width_scale_factor=0.00125, circle_scale_factor=(1 / 3), legend_font_size=20, num_dc_arrows=4)

Original notebook:

Reproduction:

Time estimate

30 minutes. By line count, it's a bit heavier than we'd usually like for a PR, but a lot of that is due to documentation and code formatting. The code itself is fairly straightforward I think.

[rouille]

Should we add the loadzone_centers.csv file to the data folder and rename it something like loadzone_center_usa_tamu_model.csv?

[danielolsen]

Should we add the loadzone_centers.csv file to the data folder and rename it something like loadzone_center_usa_tamu_model.csv?

That's a good idea. Default behavior remains the same (don't plot), but we have these data available if the user wants to use them?

[rouille]

Should we add the loadzone_centers.csv file to the data folder and rename it something like loadzone_center_usa_tamu_model.csv?

That's a good idea. Default behavior remains the same (don't plot), but we have these data available if the user wants to use them?

In fact the data frame could be part of the powersimdata.network.usa_tamu.constants.zones module and hence retrievable vie the the model_immutables attributes of a Grid object

[BainanXia]

@danielolsen @rouille I would suggest we handle the csvs in the same way as we do in the shortfall plot, i.e. make the plot function taking a dataframe as inputs and use the csv file as an example input in the demo notebook. The current demand center csv is generated by my simple back of envelope algorithm with many manual adjustments and should not be tied to the usa_tamu model.

[danielolsen]

Demo locations and a script are added.

[rouille]

I would add here that :func:postreise.plot.projection_helpers.project_branch can be used to obtain the coordinates of a branch from its (lat, lon).

[danielolsen]

Done.

[rouille]

In other plotting routines, we pass a pandas.Timestamp/numpy.datetime64/datetime.datetime as hour to avoid any issue related to weird string formatting when locating the indices. So instead of passing "2016-11-02 22:00:00",, the user would pass pd.Timestamp(2016, 11, 2, 22)

[rouille]

Also, if houris a pandas.Timestamp/numpy.datetime64/datetime.datetime then later you can use _check_date_range_in_scenario to make sure the query hour is in the scenario

[danielolsen]

Done.

[rouille]

You can check that the scnario is in analyze state here using _check_scenario_is_in_analyze_state

[danielolsen]

Done.

[rouille]

Very nice. You are now a bokeh expert, congratulations!