Reproducible first study

Prepare

Use the publication-safe synthetic starter from the website when available, or create a private synthetic network. Do not start with customer data. Open it natively, verify circuit identity, active equipment, voltage base and load/generator units. Preserve the original and create an alternative for experiments.

Run

Open Studies and discover definitions/settings. Run readiness first. Select balanced AC power flow, exact network snapshot and explicit settings/case. Submit once, monitor the retained job and open Results only for that run. Record convergence, warnings, bus voltages, loading, losses, dispatch and nodal balance against the run’s own tolerance. Nonconvergence is a result to inspect, not permission to replace inputs silently.

Compare

Change one declared synthetic parameter in an alternative, save a milestone, rerun with comparable settings and compare exact-version receipts. Use history Before/After for model differences and study comparison for numerical differences; these are different comparisons. Confirm Main and approvals are unchanged.

AI walkthrough

Ask in Plan: “List the case, equipment and supported study settings for this exact version; explain limitations without edits.” In Build, explicitly request the bounded synthetic study. Inspect the actual typed tool calls and retained run. The assistant’s final text is not evidence that the solver ran. End by exporting a source-bound report to a consented destination.

Expected results

Use only the starter’s published proof/receipt for numerical expected values; do not memorize illustrative numbers from screenshots. Different solver/settings/input hashes require a new check.