A commissioning index written at design stage already carries fourteen named integrated system test scenarios, each bound to a primary failure domain, each structured on the same thirteen mandatory fields, and all fourteen at the highest criticality weight [MODELED]. The currently executable count of those fourteen is zero [FACT]. A later procurement file names fourteen scenes of its own. The heat-rejection chapter names fourteen measured items. A machine closed set sits beneath them with thirty test cases across six levels, every one recorded as not executed and ready for review [MODELED]. Serial numbers do not make those four lists the same fourteen. No acceptance criterion in the set carries a value. What every criterion does carry is a named owner, and that is what makes the set useful this early.
Fourteen IST Scenes Are Named and None Is Executable
The index is not a commissioning plan and not a report. It is an index and a rule-set, written before the procurement package existed. Publishing it does not establish a commissioning system. Every joint scene is blocked by an unfrozen design input, not by missing commissioning paperwork. That is why a named scene is not a test that has been run.
Naming a scenario forces the design to state which failure it claims to survive. The set spans black start, islanded steady-state operation, feeder loss with battery bridging and staged generator loading, generator failure to start or synchronise or accept load, return transfer and recharge, liquid-cooling leak and isolation, fire interlock conflict with the power and cooling interlocks, loss of cooling to the UPS room, carrier and route failure, out-of-band management during an outage, hardwired interlock independence under supervisory communications loss, and single-point failure of shared primary-side plant. Writing that list is itself a design act: a failure nobody is willing to name as a scenario is a failure the design has quietly decided not to claim.
The rule that every scenario be physically injectable removes the comfortable middle ground. A failure that cannot actually be injected is recorded as an analysis item rather than written up as a test, so the index cannot accumulate tests that will never be run. Two further rules follow from these tests running on real equipment and possibly on real IT load. Abort criteria are written before success criteria, and holding the protected supply takes precedence over completing the test. Order is controlled as well, because the baseline sequence scenario must precede the failure-branch scenarios under the same load condition or the curves cannot be attributed. The index also records failure as an admissible result — a failure-branch scenario that does not pass is a design input, not a contractor breach.
Baseline Is Unjudged and the Asymmetric Feeder Is Blocked
Two of the fourteen carry a disposition before anyone has been to site. The baseline sequence scenario can only be recorded as executed-but-unjudged while the Engineer of Record's transient criteria are undelivered, and one of its own criteria compares detection, start, first-stage and fan ramp times against a thermal ride-through of either 187 or 59 seconds depending on an unresolved branch [MODELED], with all six electrical timing quantities behind that comparison unobtained [MODELED]. The asymmetric feeder scenario is blocked outright: the load bias allocation it exists to verify has never been frozen, and the arithmetic premise behind the roughly 0.196 MW of overshoot on the smaller feeder is itself recorded as incomplete [MODELED].
Forty-four external decisions sit behind the set. None has been obtained. Two of them are first-priority letters, not self-sufficient unlocks. The other forty-two, however far they are advanced, cannot independently unlock any joint scene.
Limits and open items
Confirmed: the currently executable count of the fourteen joint scenes is zero; every scene is blocked by an unfrozen design input rather than by missing commissioning paperwork; publishing the index does not establish a commissioning system. [FACT]
Study values, not thresholds: the fourteen named scenes; the thirteen mandatory fields; the thirty unexecuted cases across six levels; the 187 or 59 second ride-through pair; the six unobtained electrical timings; and the 0.196 MW feeder overshoot. None is a pass criterion. [MODELED]
Open: the Engineer of Record's transient criteria; the supplier time constants; the maximum secondary supply temperature; the battery state-of-charge abort threshold; the load bias allocation; the approved load-shedding priority table; the owner's written authorisation, which for any scenario carrying real IT load requires the IT side to agree as well; and the forty-four external decisions. No test has been executed, witnessed or accepted. No evidence package exists, and no Engineer of Record or commissioning authority has approved this scope. [HOLD]
Ask Which Design Claim Has No Scenario That Can Disprove It
A blocked scenario reads like a programme problem and is not one. Nothing about the asymmetric feeder test is difficult; it is a breaker opened at a real switchboard. What blocks it is that the allocation it would verify does not yet exist as a frozen decision, and the block surfaced only because someone wrote the test long before it could be run. That is the argument for drafting commissioning scenarios during design rather than after procurement — the test set is the cheapest instrument available for finding decisions nobody has made. The question worth putting to a design at this stage is which of its own claims currently has no scenario capable of disproving it.
© 2026 K&K Data Service Inc. All rights reserved. Reproduction or republication is permitted only with clear attribution to K&K Data Service Inc. and a working hyperlink to the canonical URL of this article. Excerpts must preserve the technical context, maturity labels, assumptions, and limitations. No excerpt may imply project approval, field validation, certification, or endorsement that the original article does not state.
Source: K&K Data Service Inc., “Integrated System Test Scenarios Are Named Before the Plant Is Frozen,” https://www.kkdatasvc.com/lab/rfq-vendor-review-and-commissioning/ist-scenes-have-names-before-the-plant-is-frozen/.
Discuss This Constraint. If your project record shows a different result, or the same failure domain under another name, we want to see it. Email inquiry@kkdatasvc.com or use the contact page.