Battery storage cost constant version 2

[toddleif]

This new code allows for a battery storage cost constant to be included in the model. This feature was added in order to align REopt's battery cost modeling with NREL's Annual Technology Baseline (ATB). Changes to the code include:

• In the src/core/energy_storage/electric_storage.jl struct, the following fields were added: installed_cost_constant, replace_cost_constant, and cost_constant_replacement_year. All three of the added fields default to zero.
• The new binary variable binIncludeStorageCostConstant was added. This variable is indexed on p.s.storage.types.elec.

[adfarth]

@toddleif could you add the new inputs to this text in electric_storage.jl, starting line 33:

When modeling degradation the following ElectricStorage inputs are not used:
    - `replace_cost_per_kw`
    - `replace_cost_per_kwh`
    - `inverter_replacement_year`
    - `battery_replacement_year`

[adfarth]

@toddleif in general this is looking good! In addition to my in-line comments above, just a few high-level notes:

• Can you add tests to test/test_with_xpress.jl and runtests.jl that check that things are working as expected with your new addition? This could check, for example, some expected financial outputs, that you have manually calculated in Excel.
• I think we need to reconfigure the constraint to not be an indicator constraint, if possible. I suspect this is why so many of the hightests are failing. We'll want all tests to be passing before we merge.

[adfarth]

@toddleif could you please update the CHANGELOG.md file with a description of these changes? You can see other entries in there for examples. For this PR, you could list all 3 new inputs that are added and note that they default to zero. After you add that, you could update the PR description with the same information (should be a bit more detailed than what is currently the PR description). The PR description could also include a bit of reasoning for why we made this change (e.g. reference the ATB cost structure)

[adfarth]

@toddleif could you please add a brief description of this input to this markdown area?

[toddleif]

That's a great idea. I just added a description.

[adfarth]

@toddleif I just modified the description to be in-line with the inputs and to just give a short high-level description of what that input is. Could you review?

[adfarth]

@toddleif I think this is looking good! I left a few comments throughout, and have noticed just two more places where changes may need to be made:

• It looks like you'll need to make a few changes to proforma.jl. See e.g. line 71 "# calculate Storage o+m costs, incentives, and depreciation" (there may be other places in there that require updates too)
• outage_constraints.jl line 80 where I think the cost constant should be added to the "bigM" value being subtracted from the capital costs

@constraint(m,
            m[:dvMGStorageUpgradeCost] >= p.s.financial.microgrid_upgrade_cost_fraction * m[:TotalStorageCapCosts] - (
                p.s.financial.microgrid_upgrade_cost_fraction * p.third_party_factor * (
                    sum( p.s.storage.attr[b].net_present_cost_per_kw * p.s.storage.attr[b].max_kw for b in p.s.storage.types.elec) + 
                    sum( p.s.storage.attr[b].net_present_cost_per_kwh * p.s.storage.attr[b].max_kwh for b in p.s.storage.types.all )
                ) * (1-m[:binMGStorageUsed])  # Big-M is capital cost of battery with max size kw and kwh
            )
        )

[adfarth]

Also noting that the tests did fail previously (looks like they may have timed out) so I re-ran them on Tucker's last commit. The commit that I made was just a change to the Changelog, so only the documentation tests passed, which is misleading :)

[toddleif]

@toddleif I think this is looking good! I left a few comments throughout, and have noticed just two more places where changes may need to be made:

• It looks like you'll need to make a few changes to proforma.jl. See e.g. line 71 "# calculate Storage o+m costs, incentives, and depreciation" (there may be other places in there that require updates too)
• outage_constraints.jl line 80 where I think the cost constant should be added to the "bigM" value being subtracted from the capital costs

@constraint(m,
            m[:dvMGStorageUpgradeCost] >= p.s.financial.microgrid_upgrade_cost_fraction * m[:TotalStorageCapCosts] - (
                p.s.financial.microgrid_upgrade_cost_fraction * p.third_party_factor * (
                    sum( p.s.storage.attr[b].net_present_cost_per_kw * p.s.storage.attr[b].max_kw for b in p.s.storage.types.elec) + 
                    sum( p.s.storage.attr[b].net_present_cost_per_kwh * p.s.storage.attr[b].max_kwh for b in p.s.storage.types.all )
                ) * (1-m[:binMGStorageUsed])  # Big-M is capital cost of battery with max size kw and kwh
            )
        )

@adfarth I addressed these edits too, thanks for catching these as well!

[adfarth]

@toddleif If I'm understanding correctly, I think the following could work for avoiding the increased run time even when a cost constant isn't included:

• Move the if statementif p.s.storage.attr[b].installed_cost_constant != 0 || p.s.storage.attr[b].replace_cost_constant != 0 to reopt.jl. Within that statement, add a warning about binary vars and then call "add_elec_storage_cost_constant_constraints".
• Define the binary vars within "add_elec_storage_cost_constant_constraints" here. (see add_prod_incent_vars_and_constraints as an example)