Fix power sensor output for disabled appliances

[appleweiping]

Summary

• Return null power-sensor output when the measured source, shunt, load, or generator is disabled.
• Keep branch and branch3 power sensors governed by topology, including partially disconnected branches, and leave current-sensor behavior unchanged.
• Preserve reduced main-core component-container compatibility by guarding lookups for appliance types that the container omits.
• Re-enable the existing symmetric/asymmetric batch regression cases, add focused C++ coverage, and document how sensor-output energized is derived from the measured object.

Root cause

Power-sensor output only checked whether the measured object had a connected mathematical topology group. Status-only updates for shunts, loads, and generators do not rebuild that topology, so a disabled appliance could still produce energized == 1 and non-zero residuals. The output path now also checks the actual status of measured appliances while preserving topology-only semantics for branches, branch3 components, and nodes.

Because the measured terminal is selected at runtime, direct appliance lookups would otherwise be instantiated even for reduced component containers that omit unused appliance types. Compile-time type guards retain the previous topology-only behavior for those unreachable paths.

Validation

• uv run --frozen pytest — 1778 passed, 4 skipped, 7 xfailed; 97.13% coverage.
• uv run --frozen pre-commit run --files <changed files> — all applicable hooks passed, including REUSE, pytest, clang-format, and markdownlint.
• Focused C++ main-core build and CTest selection — 9/9 passed, including six symmetric/asymmetric sensor-output cases.
• The two affected dataset fixture families — 9/9 passed.
• The old implementation fails the disabled-appliance cases and cannot instantiate the reduced-container regression; both pass with this change.
• Read the Docs build #34120066 — passed.
• DCO — passed for every commit.

The temporary Python migration reference was removed after the maintainer resolved the C++ test-migration thread.

No public API, schema, generated file, or current-sensor implementation is changed.

Closes #1464

Assisted-by: OpenAI Codex

[Copilot]

Pull request overview

This PR fixes state-estimation power-sensor outputs so that sensors measuring disabled appliances (source, shunt, load, generator) return a null output (energized = 0 and zero residuals), while preserving topology-governed energized semantics for branch/branch3 power sensors and leaving current-sensor behavior unchanged.

Changes:

• Update main-core power-sensor output logic to additionally gate energized output on the measured appliance’s status() (for appliance terminal types).
• Re-enable previously xfailed regression datasets by updating expected outputs to reflect null residuals when measured appliances are disabled.
• Add C++ unit tests for energized/null-output behavior across appliance-vs-branch topology semantics and document the energized derivation in the user manual.

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
tests/data/state_estimation/dummy-test-line-into-itself/params.json	Removes xfail now that appliance-disabled behavior is fixed.
tests/data/state_estimation/dummy-test-line-into-itself/asym_output_batch.json	Updates expected residuals to zeros for null (energized=0) outputs.
tests/data/state_estimation/dummy-test-line-into-itself-minimal/sym_output_batch.json	Updates expected null-output residuals to zero for disabled measured appliance.
tests/data/state_estimation/dummy-test-line-into-itself-minimal/params.json	Removes xfail now that appliance-disabled behavior is fixed.
tests/cpp_unit_tests/main_core/test_main_core_output.cpp	Adds focused unit tests for power/current sensor energized semantics.
power_grid_model_c/power_grid_model/include/power_grid_model/main_core/output.hpp	Implements measured-object status gating for appliance-measured power sensors.
docs/user_manual/components.md	Documents how energized is derived and that null outputs have zero residuals.

---

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

[mgovers]

Hi @appleweiping, Something seems to have gone wrong with the base branch. I had to change the base branch due to some blocking things in the root, but the base branch was not automatically updated. Can you please re-open the PR and point the base branch to PowerGridModel:main?

[nitbharambe]

I see this PR got closed because underlying branch was deleted. That branch and PR is now merged.
@appleweiping would you please create a new PR where we can continue with the contribution?

[nitbharambe]

Nevermind, resolved! We will continue with review here.

[mgovers]

Hi @appleweiping,

It looks like compilation is failing for MSVC and clang-tidy for unreachable code and misc-include-cleaner reasons, respectively. If you need any help, please let us know and we'll have a look.

[TonyXiang8787]

@mgovers @appleweiping Does this PR only partially fixes problem?

With this PR, you indeed solve the problem with sensors on disconnected appliances. But for energized branch or branch3, if one side is disconnected and there is a sensor on this side, even after this PR, the sensor value will still be wrongfully calculated instead of getting a null output.

To fix that part also, you cannot use fall through here for all branch switch cases. Instead, you need to check the actual status of that side of the branch/branch3.

[appleweiping]

Thanks for raising this, Tony. I intentionally kept partially disconnected branch/branch3 sensors topology-governed based on the scope previously confirmed in #1464 and the earlier review guidance in #1468: when the component still contributes to the mathematical grid, the sensor remains energized.

This proposed terminal-specific behavior would change that agreed contract. @mgovers, could you please confirm which behavior should apply? If terminal-specific behavior is intended, should it apply only to power sensors, or to current sensors as well? I’ll wait for alignment before changing the implementation and tests.