Reorganize emissions info #121

[urwahah]

Fuel switching (and maybe load shifting later on) requires the grid emissions data to be available for loads to site energy calculations. This is a fairly significant revision to the architecture of energy.py so it would be better to review and implement this before I start changing the calculations :-)

To summarize the changes:
Currently:

1. run_loads_to_site calls loads_to_site_energy, which iterates through the equipment scenarios, and saves the site energy results as a temporary file
2. run_site_to_source passes those results to site_to_source, which iterates through the emissions scenarios for final source emissions

Proposed:

1. run_loads_to_site calls loads_to_site_energy
   a. loads_to_site_energy calls grid_emissions, which iterates through the emissions scenarios to return a dataframe containing grid emissions data
   b. loads_to_site_energy iterates over both emissions and equipment scenarios (nested) to calculate site energy
2. run_site_to_source passes the site energy results file (same as before) to site_to_source, which iterates through the emissions scenarios for final source emissions

[urwahah]

unrelated but i changed this because it was doing a cumulative sum of all emissions scenarios which didn't seem helpful to me. now it outputs the total emissions for each scenario - still summed up for all equipment scenarios, but better for a high level check i suppose

[t-kramer]

@urwahah Thanks for putting in all this work.

One thing I wanted to flag before this gets built on: the nested loop means site_energy.pkl grows from N_equipment × 8760 rows to N_equipment × N_emissions × 8760. So with 3 equipment × 3 emission scenarios that's already 9× more data and 9× more simulation runs, right? And it would be more with additional scenarios.

So please remind me: What was the reason we need the emissions data in the loop? Since sizing in the current code is driven purely by peak thermal loads (e.g. df["hhw_W"].max()), is the plan for fuel switching to affect sizing at all, or will it be a purely operational decision (i.e. which fuel to use at each hour)? If it's the latter, the fuel-switching logic could potentially stay in site_to_source() and avoid the data explosion entirely. We could just add some extra input columns (e.g. "potential gas" and "potential electric" per hour). Equipment simulation stays at N_equipment runs.

I might not be thinking this through properly here, so feel to explain again. Maybe even if we need the loop, there's an alternative way to reduce the "data explosion".

[urwahah]

the nested loop is definitely the issue with this change. i like the idea that we could avoid the extra runs by passing new 'potential' gas and electricity input columns to site_to_source. i don't think fuel switching should affect the AWHP sizing? but it might affect the boiler sizing, which is currently irrelevant to the calculations but i think in the future we want to report total area and peak electric demand. to solve that we could move the boiler sizing to site_to_source as well, since it's not needed for any further calculations in loads_to_site.

however, thinking ahead to load shifting, this will come up again if we want to include AWHP sizing strategies that are dependent on peak day emissions. we should discuss