Adding Chilled Water System (CHWS) to Temoa's model

[robfairh]

This is a tentative model that should close #128. A decision should be made before merging this PR. I am working on two models:

a) Models the CHWS as one technology (10_uiuc.sql):

Pros:

• Everything is modeled as one technology, making the input file simpler.

Cons:

• Need to define the table StorageDuration. For UIUC's case, the storage lasts around 12 hours, but the units of the simulation are years.
• Not clear what values the table CapacityFactorTech should take.

b) Models the CHWS as two technologies (10b_uiuc.sql):

Pros:

• Easier to define the table CapacityFactorTech.
• Table StorageDuration no longer necessary.

Cons:

• Needs to define two technologies and one extra commodity.

I would personally choose the second model because the implementation seems more straightforward.

[robfairh]

For the second model (10b_uiuc), I separated the charge and discharge of CHW into two technologies: CHWSC and CHWSD. I consider that the CHWS on campus only works in the summer and intermediate seasons, but not in the winter. Hence, the capacity factor is:

• 0 in the winter
• 1 in the summer
• 0.5 in intermediate seasons.

Finally, I consider that the capacity factor of CHWSC is:

• 0 during the day
• 0, 0.5, or 1 during the night according to the season.

For CHWSD the capacity factor is:

• 0 during the night
• 0, 0.5, or 1 during the day according to the season.

[samgdotson]

My concern is simply that Temoa will just not use the CWS at all. So we will probably have to force it by defining a MinActivity for each CWS tech, and this activity should match historical usage. So if UIUC discharges 20 ktons of CW every year (which stored some percentage of the total demand, right?) That percentage should be roughly constant. Does that make sense?

Also, Temoa does secretly have an hourly-ish resolution under the hood. You can see when defining a CapacityToActivity because that's based on the number of hours in a year (1 MW produces 8.76 GWh if running at 100% for a year).

To me, this means we should use the single technology in 10_uiuc.sql. It will reduce the number of assumptions we have to make (by reducing the degrees of freedom). StorageDuration should work. I like the CapacityFactorTech workaround. When we increase the time resolution we can specify more clearly when the CWS is working. But we may not need it.

1) Try adding a MinActivity that reflects historical values. 2) Give a StorageDuration

[robfairh]

Okay, the conversation IRL that we had with @samgdotson favored the first model. Here is the model's diagram:

And the figure scenario10_elc_generation:

The behavior of the CHWS is what we were expecting, the energy stored doesn't change throughout the years.

This PR is ready for review. The assumptions of the model are in data_processing/cws_power.ipynb. I didn't want to delete the previous analysis in the notebook, I would leave it there. Also, let me know if the formatting I used in the notebook is acceptable.

[samgdotson]

Looks good, the only change I would make is to the CapacityToActivity table.

The "capacity" isn't the TOTAL capacity of the chilled water. It's a single unit of capacity for a given technology. E.g. for IMPWIND the capacity to activity isn't 8 MW * 8760h = 70.080 GWh, it's 8.76 GWh because the capacity for wind is measured in MW. Similarly CHWS capacity is measured in MW, so the C2A table should also say 8.76