Add time-varying ORDC; port SC slacks

[acostarelli]

Port of PowerSimulations.jl#1629 (time-varying ORDC), adapted to the psy6 data model where a time-varying ORDC is a ReserveDemandTimeSeriesCurve whose variable is a CostCurve{TimeSeriesPiecewiseIncrementalCurve} (rather than a ReserveDemandCurve carrying a bare TimeSeriesKey).

• add_parameters.jl: service-level piecewise slope/breakpoint cost-parameter population for ReserveDemandTimeSeriesCurve (calc_additional_axes, get_max_tranches, _get_time_series_name, _add_parameters!, _param_to_vars).
• reserves.jl / market_bid_overrides.jl: drop the "not supported" error and route ORDC cost through the decremental block-offer path, reading per-timestep data from the parameter arrays (time-varying) or the curve (static); time-varying cost goes to the variant objective expression.
• services_constructor.jl: route time-series-backed ORDC to parameter setup.
• Broaden shared reserve / production-cost-expression methods to Union{ReserveDemandCurve, ReserveDemandTimeSeriesCurve}; update hybrid/storage disambiguators to keep dispatch unambiguous.
• test/test_services_constructor.jl: build, and build & solve, time-varying ORDC.
• Fix a stray end in hybrid_systems.jl that blocked precompilation.
• Point the IOM source at the ac/ordc branch (companion changes live there).

Thanks for opening a PR to PowerOperationsModels.jl, please take note of the following when making a PR:

Check the contributor guidelines

[Copilot]

Pull request overview

Adds modeling support for time-varying ORDC by treating it as a ReserveDemandTimeSeriesCurve with a time-series-backed CostCurve, routing its stepwise cost through the decremental block-offer path and populating per-timestep breakpoint/slope parameters.

Changes:

• Add parameter-container population for ORDC time-series breakpoint/slope data and route time-varying ORDC costs to the variant objective expression.
• Broaden reserve-demand-curve dispatch across reserves/objective/expression plumbing to include ReserveDemandTimeSeriesCurve alongside ReserveDemandCurve.
• Add tests that build (and build & solve) models with one or more time-varying ORDC services, and fix a precompilation-blocking stray end.

Reviewed changes

Copilot reviewed 9 out of 10 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
src/common_models/add_parameters.jl	Adds service-side breakpoint/slope parameter population for time-varying ORDC and tranche-axis sizing logic.
src/common_models/market_bid_overrides.jl	Routes ORDC through decremental block-offer constraints and adds per-timestep PWL data retrieval from parameter arrays for TS-backed curves.
src/services_models/reserves.jl	Broadens ORDC handling to TS curves; routes TS-backed ORDC cost into the variant objective expression; adds parameter processing hook.
src/services_models/services_constructor.jl	Hooks ORDC TS parameter setup into service construction when the ORDC curve is time-series-backed.
src/common_models/add_to_expression.jl	Broadens cost-expression dispatch to include ReserveDemandTimeSeriesCurve.
src/hybrid_system_models/hybrid_systems.jl	Updates ORDC-related method disambiguation for hybrids and removes a stray end that blocked precompilation.
src/energy_storage_models/storage_models.jl	Broadens ORDC-related reserve upper-bound dispatch to include ReserveDemandTimeSeriesCurve.
test/test_services_constructor.jl	New tests exercising time-varying ORDC build and solve paths (including varying tranche counts).
Project.toml	Points InfrastructureOptimizationModels source override to ac/ordc branch.
test/Project.toml	Points test environment InfrastructureOptimizationModels source override to ac/ordc branch.

Comments suppressed due to low confidence (1)

test/Project.toml:41

• Pinning InfrastructureOptimizationModels to a moving branch (rev = "ac/ordc") in the test environment reduces reproducibility and can cause CI drift. Prefer pinning to a commit SHA (or tagged release) that contains the required companion changes.

[sources]
InfrastructureOptimizationModels = {rev = "ac/ordc", url = "https://github.com/Sienna-Platform/InfrastructureOptimizationModels.jl"}
InfrastructureSystems = {rev = "IS4", url = "https://github.com/Sienna-Platform/InfrastructureSystems.jl"}
PowerSystems = {rev = "psy6", url = "https://github.com/Sienna-Platform/PowerSystems.jl"}
PowerFlows = {rev = "psy6", url = "https://github.com/Sienna-Platform/PowerFlows.jl"}
PowerNetworkMatrices = {rev = "psy6", url = "https://github.com/Sienna-Platform/PowerNetworkMatrices.jl"}

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 12 out of 13 changed files in this pull request and generated 1 comment.

[luke-kiernan]

On the "port former PNM methods" side, can someone explain to me why these things were removed from PNM in the first place? imo they'd belong better back there.

On the ORDC side: very little of the code I'm seeing is ORDC specific. 95% of it applies equally well to other PWL offer curves. That tells me we should be calling helpers back in IOM instead of copy-pasting. I had Claude code take a look:

1. _add_parameters! should be unified. The device ObjectiveFunctionParameter method already handles N components; pass the service as a 1-element collection.
2. _get_reserve_pwl_data should route through OfferDirection, like the other 3 consumers of the PWL offer abstraction (gen incremental, load decremental, and market-bid). Right now, it re-derives the TS/static split and hard codes Decremental*. Same for the block-offer var/constraint selection in add_pwl_term_delta!.
3. The hardcoded Decremental* types mean the reserve path silently can't ever be incremental; going through _slope_param(dir) makes that a one-line dispatch rather than a structural assumption.

[luke-kiernan]

On the "port former PNM methods" side, can someone explain to me why these things were removed from PNM in the first place? imo they'd belong better back there.

I did a search in the PNM repo: we haven't had a branch_admittance function in a long time, perhaps ever. The closest thing I see is some branch_admittance_from_to fields in the BranchAdmittanceMatrix struct, since renamed to use arc instead of branch. So I'm puzzled: why did the native-networkmodels work introduce that PNM.branch_admittance call? AI hallucinations perhaps?

edit: confirmed that there never was such a function. Now, what's the similar function in PNM that does/did exist...

[Copilot]

Pull request overview

Copilot reviewed 17 out of 18 changed files in this pull request and generated 3 comments.