# Playbook — NX Real Run (Hydrotech Beam)

## Objective
Run a **real** DOE using NX/Simcenter (not stub), with clean artifacts and traceability.

## Pre-Run Checklist
- [ ] In study folder: `...\studies\01_doe_landscape`
- [ ] Model directory contains expected `.sim`, `.fem`, `.prt`
- [ ] Syncthing healthy on both ends (no major errors)
- [ ] No unintended `*.sync-conflict-*` in active results root
- [ ] Command prepared with explicit `--backend nxopen`

## Recommended command
```powershell
python .\run_doe.py --backend nxopen --model-dir "<PATH_TO_NX_MODELS>" --clean --study-name hydrotech_beam_doe_phase1_real
```

## During run — sanity checks
- Runtime per solved trial should be non-trivial (not near-zero for all trials).
- Log should indicate NX solve flow, not stub messages.
- Iteration outputs should be created.

## Post-run validation
1. Check `results/doe_summary.json`
2. Check `results/doe_results.csv`
3. Verify:
   - trial count expected (typically 51 for baseline + 50 LHS)
   - solved/failed counts make sense
   - mass values are numeric for solved trials
   - no obvious synthetic patterns indicating stub run

## Quality gate reminder
Phase 1 gate (current):
- ≥5 feasible points
- ≥80% solve success

## Mandatory logging
Record the run in project docs using the run log template from `USER_GUIDE.md`.