Find your next learning step
Commissioning is not the end of validation. A queueing benchmark enters a changing service system: arrival mixes move, staffing rules change, sensors fail and operators adapt to recommendations. Post-commissioning monitoring asks whether the evidence that earned authority still holds. When a named acceptance criterion is breached, the response is not automatic retirement or quiet threshold adjustment. Authority is contained, the cause is investigated and the successor earns requalification against a deliberately refreshed evidence set.
Open the chapter contents · Queueing Theory root · Gold Standard of Education · How X Works
Chapter contents
Choose the route closest to the decision, or read from Chapter 1. The chapters remain open; these grouped routes are a navigable index.
Route 1 · Define the decision
Route 2 · Work the ledgers
Route 3 · Transfer the method
Route 4 · Check limits and evidence
Route 5 · Ask and continue
CHAPTER 1 OF 12 · DEFINE THE DECISION
1. Freeze the commissioned baseline before measuring drift
Record the accepted release, input contract, transformation code, service classes, decision thresholds, operating envelope, approval date and each criterion that supported authority. The baseline must describe the decision process, not only prediction accuracy.
Name the monitoring window and comparison rule in advance. A moving benchmark that is silently recalibrated after every surprise cannot show whether the commissioned claim still holds.
CHAPTER 2 OF 12 · DEFINE THE DECISION
2. Separate drift, defect and criterion breach
Drift is a change in inputs, relationships, outcomes or operator behaviour. A defect is a fault in data, code or implementation. A breach occurs when a predeclared acceptance or operating criterion is crossed. One event can involve all three, but the labels lead to different evidence and containment actions.
Ordinary variation should not trigger repeated emergency resets. Use decision-relevant limits, persistence rules and consequence tiers so the monitor is sensitive to harmful change without treating every fluctuation as loss of validity.
CHAPTER 3 OF 12 · WORK THE LEDGERS
3. Build a post-commissioning drift register
A hospital benchmark B2 has been commissioned for routine room-opening decisions but not automatic diversion.
| Signal | Baseline | Alert rule | Observed | Initial action |
|---|---|---|---|---|
| Urgent-arrival share | 18% to 24% | Above 30% for 3 weeks | 33% | Open scope review |
| Tail-wait error | Within 10% | Above 10% twice | 14%, 13% | Restrict urgent class |
| Room-choice regret | At most 3 units | Any case above 3 | 2.7 | Continue monitor |
| Missing-data mode | 100% safe drills | Any unsafe output | One stale feed | Disable affected feed |
| Operator override | Below 12% | Above 20% | 23% | Interview users |
The tail-wait criterion is breached and the arrival mix has shifted. The stale feed is also a defect, but it is investigated separately because fixing it may not explain the population change.
CHAPTER 4 OF 12 · WORK THE LEDGERS
4. Work a breach and requalification decision
The review board contains B2 to routine non-urgent staffing while a four-week locked requalification study runs.
| Requalification gate | Rule | Observed | Result | Authority consequence |
|---|---|---|---|---|
| Feed repair | No stale inputs | No stale inputs | Pass | Restore data path |
| Urgent-tail calibration | Within 10% | 8% | Pass | Eligible for review |
| Boundary regret | At most 3 units | 3.4 | Fail | Keep urgent restriction |
| Stress-week coverage | Two surge weeks | Two included | Pass | Evidence representative |
| Rerun reproducibility | Exact action sequence | Exact | Pass | Freeze candidate |
The successor remains useful for routine staffing. A revised threshold becomes a new candidate release and must be tested prospectively rather than edited into the accepted record.
CHAPTER 5 OF 12 · WORK THE LEDGERS
5. Contain authority before diagnosing cause
Disable only the affected decision path when roles are separable. Preserve safe fallback, manual review and unaffected service classes. Record the time, users, cases and outputs exposed between first breach and containment.
Then test plausible causes: arrival-mix shift, service-time change, new priority practice, sensor latency, threshold instability and operator feedback. Avoid selecting one explanation merely because it is easiest to repair.
CHAPTER 6 OF 12 · TRANSFER THE METHOD
6. Design requalification as a new evidence obligation
Requalification should specify the changed claim, representative strata, locked transformations, evaluation metrics, stopping rule, failure actions and approvers. Reuse historical commissioning data only for context unless it remains independent of the repair choices.
A corrected implementation may require verification alone, while changed relationships or thresholds require validation. If operating scope expands, qualification must cover the new consequence rather than borrowing authority from the old boundary.
CHAPTER 7 OF 12 · TRANSFER THE METHOD
7. Changed-context transfer: commuter-platform control
Move the method to a rail platform benchmark commissioned to open gates when passenger density rises. A timetable change increases transfer bursts, and camera latency appears during rain.
The operator contains automatic gate commands, keeps advisory counts visible and separates the latency defect from the new burst pattern. After camera repair, a locked wet-weather study still fails the rapid-burst criterion, so automatic authority remains suspended while ordinary advisory use continues.
CHAPTER 8 OF 12 · TRANSFER THE METHOD
8. Connect commissioning, retirement and drift governance
Start with successor-benchmark commissioning and dual-run acceptance. Use transport-failure escalation and benchmark retirement when evidence cannot safely support the current transport.
Return to post-handover drift and challenger refresh for the wider champion-challenger lifecycle. This article focuses on what happens when the accepted successor itself crosses a criterion.
CHAPTER 9 OF 12 · CHECK LIMITS AND EVIDENCE
9. Limits and common misconceptions
A monitoring alert is not proof of invalidity, and absence of alerts is not proof of continuing validity when exposure is low. Requalification is not routine recalibration hidden under a new name; the changed release and evidence must remain visible.
Restoring average accuracy does not necessarily restore the decision boundary, tail behaviour or safe failure mode that earned authority. Nor should every breach cause wholesale retirement when roles can be responsibly separated.
CHAPTER 10 OF 12 · CHECK LIMITS AND EVIDENCE
10. Practical evidence and release checks
Archive the commissioned baseline, criteria, monitor code, data contracts, alert history, exposure log, overrides, incident timeline, containment decision, root-cause hypotheses, defect repairs and affected-case review.
For requalification, retain the protocol, frozen candidate, independent data manifest, stratum coverage, every gate result, deviations, residual limits, authority decision, effective date, fallback, monitoring reset and next review.
CHAPTER 11 OF 12 · ASK AND CONTINUE
11. Frequently asked questions
Does one breach invalidate the whole benchmark?
Not necessarily. Apply the predeclared persistence and consequence rules, then contain the affected claim while checking whether other decision roles remain supported.
Can repaired commissioning data be reused?
It can support diagnosis and regression testing, but evidence exposed during repair should not be represented as independent confirmation.
When is the successor requalified?
When the defined gates pass on representative locked evidence, residual limits are accepted and the named authority explicitly restores the stated decision role.
CHAPTER 12 OF 12 · ASK AND CONTINUE
12. Sources and next reading
NASA’s 2024 Standard for Models and Simulations requires intended use, acceptance criteria, risk assessment, reviews and tracking of problems. The NASA implementation handbook approved 3 February 2026 develops lifecycle credibility, reporting and maintenance practices. These sources inform the governance framework; the hospital and rail cases are original teaching examples.
Continue through the Queueing Theory root, the Gold Standard of Education and How X Works.
Top · Contents · Previous · How X Works