Port linearize_gencost and pmin modifications from REISE.jl #701
Conversation
|
Is powersimdata/design/generation/ a good location for the |
I agree. The original idea of |
powersimdata/scenario/execute.py
Outdated
| def _adjust_pmin(self): | ||
| mi = self.grid.model_immutables | ||
| plant = self.grid.plant | ||
| pmin_factor = mi.plants["pmin_as_share_of_pmax"] | ||
|
|
||
| def _scale(x): | ||
| factor = pmin_factor.get(x.type) | ||
| return x.Pmin if factor is None else factor * x.Pmax | ||
|
|
||
| plant.Pmin = plant.apply(_scale, axis=1) | ||
|
|
||
| # set Pmin to 0 for generators that are off or profile based | ||
| profile_resource = list(mi.plants["profile_resources"]) | ||
| plant.loc[plant.type.isin(profile_resource), "Pmin"] = 0 | ||
| plant.loc[plant.status == 0, "Pmin"] = 0 |
There was a problem hiding this comment.
Maybe out of the scope of this PR, two high level questions:
hydrois one ofprofile_resources, thePminof which will be set 0 at the end, however thepmin_factorofhydrois grid dependent .- We will need to build user interface to pass user specified
pmin_as_share_of_pmax(either plant based or plant type based or plant type * zone based) in the future, right?
There was a problem hiding this comment.
- resources with profile have all
Pminset to 0 in theplantdata frame but it is not used in the optimization. We did that for backward compatibility since this is what was hard coded in REISE.jl. - If we want a finer resolution on
pmin_as_share_of_pmax, we can have either an extra column in theplantdata frame or a dedicated data structure than can be passed to the engine. Either way we will need a user interface to specify the values
There was a problem hiding this comment.
- True. Just pointing out such entries are modified twice in the current process.
- Agreed.
| def test_linearize_gencost_pmin_equal_pmax(): | ||
| plant = mock_grid_gc.plant.copy() | ||
| plant.Pmin = plant.Pmax | ||
| grid = MockGrid({"plant": plant.reset_index().to_dict(), "gencost_before": mock_gc}) | ||
| actual = linearize_gencost(grid, num_segments=3) | ||
| assert_frame_equal(expected_all_equal, actual, check_dtype=False) |
There was a problem hiding this comment.
Thanks! We thought about adding such tests a while ago as in the very first version of implementation, dividing by zero wasn't avoided.
Is the ramp module essential? Especially once (if?) |
We will need that when scaling existing plants as well as adding new plants without specifying ramp rates until we have a better solution. |
It can be done in a separate PR. I am fine with having the |
Based on the current package structure, I think having those in |
jenhagg
force-pushed
the
jen/linearize
branch
from
a1b54c9 to
152c5b5
Compare
Purpose
To facilitate using a grid.pkl and remove the need for mapping between a case.mat/grid.mat/etc, we move grid modifications previously done in REISE.jl to the scenario preparation process.
This PR should be merged at the same time as Breakthrough-Energy/REISE.jl#187, after which we should update the existing grid.mat files. I believe we also need to update the ramp30 values in plant.csv as discussed in this issue #680 and release a new dataset on zenodo.
What the code is doing
Scale plant Pmin values and linearize cost curves the same way we were doing in REISE.jl - after we apply the change table, and just before we upload the grid.pkl to the server (or, container, or wherever). I also ported the linearize_gencost tests from julia since those will be unused/deleted.
Testing
Ran a scenario in a container and reloaded it after it finished for some cursory spot checking.
Time estimate
20 min