Reproducibility
These three experiments are fully deterministic and use the bundled DC powerflow solver (no external service required). Each can be reproduced from a fresh account in under a minute.
Experiment 1 — Single-run sanity check
Goal: verify the run pipeline end-to-end on the smallest case.
| Preset | [Demo] Light load relief — case5 |
| Case | case5 |
| Action | scale_all_loads value=0.9 |
| Mode | simulation (DC solver) |
| Expected feasibility | feasible |
| Expected violation_improvement | ≥ 0 |
Experiment 2 — Batch-of-3 stability
Goal: confirm queue + concurrency control on a small batch.
| Presets | All three [Demo] presets |
| Concurrency | 3 (default) |
| Expected completed runs | 3 |
| Expected failed runs | 0 |
| Expected total time | ≤ 30s |
After loading, open /batches, create a batch including the three demo presets, and click Run Batch. Watch progress at /system-status.
Experiment 3 — Validation suite
Goal: verify the system itself (parser, evaluator, reproducibility, batch stability) is healthy.
| Route | /validation |
| Categories | parser · evaluation · reproducibility · batch_stability |
| Expected pass rate | 100% |
Open /validation, click Run All Tests, and wait for green badges across every category.