[BurhanAbdullah]

Purpose

Existing aggregate voltage-violation metrics can obscure whether a disturbance is localized to a few buses or widespread across the network, and whether violations are brief or persistent. This contribution adds complementary spatial-temporal diagnostics for analyzing voltage-risk patterns in PowerAgentBench's Level 2 dynamic benchmark.

Changes and scope

Adds three files under benchmarks/dynamic/level2/:

• spatial_temporal_risk.py — standalone metrics for spatial voltage-violation exposure, affected-bus fractions, upper-tail violation severity, and longest contiguous sampled violation duration.
• test_spatial_temporal_risk.py — synthetic unit tests covering risk patterns, time sampling, and input validation.
• SPATIAL_TEMPORAL_RISK.md — documentation of the metrics and their manual use.

The implementation supplements existing diagnostics; it does not modify the benchmark's current scoring path or simulator execution.

Validation

Tested revision:92473e156c1bdce343fa1dd441d5c9e37da5ae33

CI status:All five GitHub checks passed on the current PR, as shown in the PR Checks section.

Test scope:The contribution includes synthetic unit tests for the new metrics. Passing CI does not imply that a full transient-stability simulation was run or that the metrics have been validated against real benchmark trajectories.

Validation not performed: No empirical comparison on representative real simulation trajectories or assessment of changes to agent rankings is claimed.

Benchmark and compatibility impact

No changes to existing benchmark scores, generator-speed instability detection, simulator execution, or dependencies are intended. The new metrics provide supplementary diagnostic outputs and do not alter existing benchmark results. No reference simulation outputs were regenerated.