Convert 2W transformer ratings/flows with the system base

[luke-kiernan]

Fixes #74, plus a related AreaInterchange fix in a separate commit.

_make_per_unit! leaves rate_a/rate_b/rate_c/pf/qf in system-base pu, but make_transformer_2w! converted them to natural units with the circuit's winding base (SBASE1-2). That scaled ratings and flows by SBASE1-2 / SBASE. Now they're converted with sys_mbase, as make_line! already does. Impedances and the magnetizing shunt are unchanged, since they really are device-base.

Compatibility

• No API change. make_transformer_2w! is internal and gains a sys_mbase argument, matching make_line!.
• Output only changes for a 2W transformer with SBASE1-2 ≠ SBASE (and SBASE1-2 ≠ 0). MATPOWER input and modified_14bus_system.raw are unaffected.
• The one fixture that does discriminate, synthetic_v35_transformer_discriminators.raw, had tests pinning the old behavior and labeling it "right". They now expect the raw RATE11 MVA: T1 is 40 instead of 20, and T2 is 64 instead of 51.2. Those two transformers already act as the regression case from the issue, so I didn't add a new fixture.
• This diverges from PSCB's oracle on such cases. Any oracle-parity check there will need that field fixed or excluded.

Second commit: AreaInterchange flow

The same kind of inherited scaling bug. The PSS/E manual defines PTRAN as "MW comprising this transfer", and _make_per_unit! never touches interarea_transfer, so power_transfer is already natural MW. read_area_interchanges! multiplied it by sys_mbase anyway: on a 100 MVA system, 25 MW came out as 2500 MW. The new fixture synthetic_v35_interarea_transfer.raw is the first to contain an actual inter-area transfer record, and its test replaces the one that pinned the inflated value. Output changes for any PSS/E case with a nonzero PTRAN.

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[Copilot]

🟢 Approval recommended

The conversions match the parser’s unit conventions, with only an optional regression-coverage improvement remaining.

1 open finding

• Expand T1 discriminator tests for rating_b, rating_c, pf, and qf · New

What changed in this PR

Corrects unit conversions in the OpenAPI reader to address #74 while preserving device-base impedances.

Changes:

• Converts two-winding transformer ratings and flows using the system base.
• Preserves inter-area transfers in natural MW.
• Updates regression expectations and adds an inter-area transfer fixture.

File	Description
test/​test_openapi_transformer_discriminators.jl	Corrects unequal-base rating expectations.
test/​test_openapi_dc_shunt.jl	Tests natural-MW inter-area transfers.
test/​test_openapi_branch.jl	Updates transformer assertions and explanatory comments.
test/​fixtures/​synthetic_v35_interarea_transfer.raw	Adds a 25 MW transfer case.
src/​openapi/​dc_branch.jl	Removes redundant transfer scaling.
src/​openapi/​branch.jl	Uses the system base for transformer ratings and flows.

🧠 Review effort: Balanced

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.