feat: analyze binding constraints (ramp, pmin, pmax) #121

danielolsen · 2020-06-11T17:20:08Z

Purpose

This is a re-creation of #58, to analyze binding generator constraints. Thanks to the MATReader framework, we can now re-open this PR!

Validation

New integration tests:

python
>>> from powersimdata.scenario.scenario import Scenario
>>> from postreise.analyze.generation import binding
>>> scenario = Scenario('87')
SCENARIO: base | WesternBase_2016_noHVDC_Final_2019Sep

--> State
analyze
--> Loading grid
Loading bus
Loading plant
Loading heat_rate_curve
Loading gencost_before
Loading gencost_after
Loading branch
Loading sub
Loading bus2sub
--> Loading ct
>>> binding_pmin = binding.pmin_constraints(scenario)
--> Loading PG
>>> binding_pmin
                     10390  10391  10392  10393  ...  13457  13458  13459  13460
UTC                                              ...                            
2016-01-01 00:00:00  False  False  False  False  ...  False  False  False  False
2016-01-01 01:00:00  False  False  False  False  ...  False  False  False  False
2016-01-01 02:00:00  False  False  False  False  ...  False  False  False  False
2016-01-01 03:00:00  False  False  False  False  ...  False  False  False  False
2016-01-01 04:00:00  False  False  False  False  ...  False  False  False  False
...                    ...    ...    ...    ...  ...    ...    ...    ...    ...
2016-12-31 19:00:00  False  False  False  False  ...  False  False  False  False
2016-12-31 20:00:00  False  False  False  False  ...  False  False  False  False
2016-12-31 21:00:00  False  False  False  False  ...  False  False  False  False
2016-12-31 22:00:00  False  False  False  False  ...  False  False  False  False
2016-12-31 23:00:00  False  False  False  False  ...  False  False  False  False

[8784 rows x 2527 columns]
>>> grid = scenario.state.get_grid()
>>> plant = grid.plant
>>> binding_pmin.sum().groupby(plant.type).mean()
type
coal          4033.089744
geothermal    0.000000
hydro         0.000000
ng            5680.429158
nuclear       0.000000
solar         4636.910486
wind          879.777778
dtype: float64
>>> binding_pmax = binding.pmax_constraints(scenario)
--> Loading PG
>>> binding_pmax.sum().groupby(plant.type).mean()
type
coal          4566.455128
geothermal    8784.000000
hydro         148.001399
ng            3082.960986
nuclear       8784.000000
solar         343.907928
wind          995.860082
dtype: float64
>>> binding_ramp = binding.ramp_constraints(scenario)
--> Loading PG
>>> binding_ramp.sum().groupby(plant.type).mean()
type
coal          136.429487
geothermal    0.000000
hydro         0.000000
ng            0.000000
nuclear       0.000000
solar         0.000000
wind          0.000000
dtype: float64

Explanation: each function generates a boolean dataframe of the same dimension as PG. We can perform some groupby analysis to see that, for the Western 2016 base scenario:

Coal, NG, and solar generators are, on average, at their Pmins ~50% of the hours of the year
Nuclear and geothermal generators are at their Pmax 100% of the time (since their power is effectively free), Coal and NG generator ~50% of the time, solar/wind/hydro rarely. Another interesting insight here: because we have a single cost curve segment, NG and coal are either at their Pmin or their Pmax 98% of the time.
The only type of generators hitting their ramp limits in this scenario are coal generators. Because with the MATReader object we are getting the grid parameters as they're passed to the OPF problem, we don't get false positives on e.g. hydro hitting its ramp limit (see feat: analyze binding constraints (ramp, pmin, pmax) #58) when in reality we are setting this to effectively INF within REISE.