Collins Aerospace has announced a $250M manufacturing facility at Seletar Aerospace Park, and the coverage did what coverage does – square metres, capital expenditure, jobs. Not one line asked what the truck ride does to process qualification. Here is an uncomfortable observation from someone who has built the quality side of plants from bare concrete: the $250M is the least risky number in that release. Steel arrives, cranes leave, a building appears roughly on schedule. What does not arrive with the machines is their proof. In aerospace, an unproven process is an unqualified process, whatever the asset register says. A relocated plant is a new plant wearing a familiar name.

A relocated machine is a new machine

Capability does not survive a move. A press that ran Cpk 1.67 in the old hall sits in Seletar on a different floor – different flatness, a different vibration signature from its new neighbours, different humidity, different power quality. None of that shows on a nameplate. The physics changed, so the evidence has to be regenerated: capability studies re-run, gauge R&R redone with new appraisers in a new environment, every gauge and CMM recalibrated after transport, because calibration is a property of the instrument and its surroundings. Under AS9102, a change in the location of manufacture is an explicit first-article trigger. Full or delta FAI is a judgement call. The trigger has fired either way, part number by part number.

Special processes are worse. Nadcap approvals for welding, heat treatment, chemical processing and NDT attach to the facility, not the company logo. Move the cell, reopen the approval. On top of that sits AWS D17.1, the aerospace welding code – read the current revision twice before the movers arrive, because what it lets you carry over and what it makes you requalify is not obvious.

I lived the automotive version for years. At WITTE Automotive, and later in FOREAST audits, every transfer – tooling to another plant, production to a new line, a supplier relocation – was treated as a significant change under PPAP and IATF 16949. You notify the customer. You resubmit. The customer decides whether your old capability studies are worth the paper, and usually decides they are not. Same physics, different standard: AS9100 change management asks the identical question with more paperwork and higher stakes.

The serial number survives the move. The capability does not.

The quality system has a longer lead time than the building

Now the line item no budget captures: you cannot certify paper. AS9100 stage audits demand objective evidence from a system already running – records of real production, real nonconformances, real corrective actions raised at that address. A certificate can sometimes be extended administratively. Evidence cannot. The surveillance auditor wants months of process performance, and the clock starts with the first part made in the new hall. The building has a completion date. The evidence has a birth date.

When I built the QA/QC department for SNOP's greenfield plant – 900+ employees, everything from zero – sequencing was the whole game. Standards first: IATF 16949, control plans, PFMEAs for processes that did not physically exist yet. People second: quality engineers hired while the hall was still empty, embedded in process design so the mistake never got poured into concrete. Gauges ordered absurdly early, because a CMM needs months to install, temperature-stabilise and calibrate, and a gauge backlog at launch is a self-inflicted wound. Dashboards came last. That sequence delivered a 70% reduction in defect cost and zero critical customer escalations in the first quarter – not because we measured harder, but because we were measuring processes designed not to fail.

The ramp tests the system, not the steel

Escapes spike at rate. A process that coasted through pilot volumes develops cracks at takt – marginal fixtures, new operators, maintenance routines that have not yet learned the equipment's new voice. And quality is always the last department staffed, because it does not make parts and the finance system only notices the people who do. The arithmetic is not subtle. The all-in cost of a single escaped defect to an aerospace customer – containment, sorting, premium freight, the 8D hours, the visit – starts in the tens of thousands of euros and crosses six figures the moment a customer line stops. A senior quality engineer costs €80,000–€90,000 a year, loaded. I have signed 8Ds where containment alone exceeded the annual salary of the engineer nobody budgeted for. Ribbon-cutting scissors are cheap. Nobody holds a ceremony for a capability study.

Key takeaways

  • Treat relocation as the significant change the standards already say it is: re-run capability studies, redo gauge R&R with new appraisers, recalibrate every moved gauge and CMM, and trigger AS9102 first-article inspection – full or delta, part number by part number.
  • Schedule the evidence, not just the equipment. Certification and customer approvals run on records generated at the new address – build months of objective evidence into the ramp plan before anyone promises a shippable part.
  • Staff quality before production – control plans and PFMEAs written while the hall is empty are plans; the same documents written after launch are apologies.
  • Open a validation inventory on day zero: every asset, its qualification state, its expiry and its named owner, reviewed weekly like cash flow until the last re-approval lands.

Capability is a property of place, not of possession. Seletar will get its modern building, ready about when the press release promised. The proof arrives later – or not at all, unless someone owns it before the concrete sets. That is the sentence the release cannot contain: the real figure is not what was spent but what can be demonstrated, to an auditor and then to a customer, about a machine standing on ground it has never stood on before. On any new-facility project, the first document worth writing is the validation inventory. The $250M is the easy part.