Calculate capacity value

[danielolsen]

Purpose

Calculate the capacity value for one or more resources. This is defined by the difference between the average demand over the top N hours and the average net demand over the top N hours (with N hours selected independently in each case, since the set of hours may differ).
Closes #80.

What is the code doing

capacity_value.py defines the calculation function.

test_capacity_value.py tests the calculation function.

Time to review

Half an hour. Actual calculations are ~10 lines, the rest is input checking and tests.

[rouille]

nice trick

[rouille]

I don't understand why PyCharm complains about the type of the scenario parameter that you pass to the calculate_capacity_value function in your tests. It says 'Expected Scenario, got MockScenario instead'. It seems that type(class) is not type(obj), where obj is an instance of class.

[danlivengood]

Conceptually, I'm happy with this. Those tests might look counterintuitive (capacity value decreases with fewer hours), but I'm seeing similar results with data from Western and Texas. Once Ben's adjustments are addressed, I support Approving this PR.

[danielolsen]

@rouille I don't know what PyCharm's issue is. For our MockScenario, we redefine the __class__ attribute, and isinstance will evaluate as true for both Scenario and MockScenario, but the return of type is still the original type. I don't know if this matters for our purposes.

>>> from powersimdata.scenario.scenario import Scenario
>>> from powersimdata.tests.mock_scenario import MockScenario
>>> ms = MockScenario(grid_attrs={})
>>> type(ms)
<class 'powersimdata.tests.mock_scenario.MockScenario'>
>>> ms.__class__
<class 'powersimdata.scenario.scenario.Scenario'>
>>> isinstance(ms, MockScenario)
True
>>> isinstance(ms, Scenario)
True

[rouille]

Looks good. I agree with the calculation. All tests pass.

[rouille]

I think PyCharm looks at the docstring and see that the scenario argument should be a powersimdata.scenario.scenario.Scenario and not a powersimdata.tests.mock_scenario.MockScenario. We can't do much about it.

[danielolsen]

I added one more calculation method based on conversations with @danlivengood and @yixingxu and generalized the code a bit to be a little less repetitive, now that we have two different methods. The new method is also tested. If y'all liked the old one, you probably still like this too, but I'll leave this open until Monday night to give you a chance to register objections or extra approval.

[rouille]

Looks good. Second function makes sense. All tests pass.

[danlivengood]

Super nitpicky change request: This looks like a copy/paste from the NLDC function. However, I believe the return for calculate_net_load_peak is returning the resource capacity during the hours of net demand, not the differential.

[danielolsen]

Good catch, fixed!