Service · Factory execution · FAT
Cleanroom systems FAT witnessing in China
A cleanroom module can look immaculate and still fail on the things that decide contamination control: bypass around an intact filter, a cascade that collapses when a door opens, an interlock that proves out electrically while the door package leaks, an alarm that annunciates locally and never reaches a trend. Sinospect witnesses cleanroom-systems FAT at Chinese suppliers against ISO 14644-3 test methods and your URS — and writes down, before anyone signs, exactly which results the factory can own and which remain a site-acceptance item.
- Systems covered
- Modular envelope, airlocks and pass-throughs, AHUs, terminal HEPA housings and FFUs, room pressure controls, PLC/HMI/BMS logic
- Test methods
- ISO 14644-3 — airflow, air-pressure difference, installed filter system leakage, airflow visualisation, recovery and containment leak
- Regulatory context
- EU GMP Annex 1, FDA aseptic processing guidance, WHO sterile GMP guidance, NMPA GMP; GB 50457 and GB 50591 on the Chinese build side
- Engagement output
- Witnessed FAT report with pressure and airflow maps, leak-scan results, categorised punch list and a shipment decision
Quick answers
What does cleanroom systems FAT witnessing in China cover?
Document freeze before the witness date, mechanical and architectural verification against the contamination-control intent, witnessed airflow and installed-filter leakage testing where the module allows it, a full controls and interlock challenge, and a release decision that separates what the factory proved from what remains a site-acceptance item.
| Subject | Verifiable at the supplier's factory | Belongs to site acceptance |
|---|---|---|
| Component identity and build | Panel, door, pass-through and housing build quality; traceability of major components | Nothing — this is a factory item |
| Air system performance | Fan rotation, VFD response, and representative airflow and leakage where the module is fully assembled | Final room classification and performance under real heat load and occupancy |
| Pressure cascade | Control logic, transmitter behaviour, alarm setpoints and delays on the module as built | Integrated room-to-corridor cascade with site doors and real building leakage |
| Controls and monitoring | Interlock logic, alarm annunciation, local-to-HMI value agreement, software revision | Final BMS and EMS integration, and live upstream and downstream interfaces |
| Utilities | Whether the factory test ran on the site's electrical and compressed-air basis or a temporary one | Performance under the real utilities themselves |
A FAT signature releases the equipment for packing and shipment. It is not site acceptance, and it should not start the warranty.
Which cleanroom defects does a generic inspection miss?
The ones that are invisible in a photograph of a clean-looking build: bypass leakage around an intact filter, velocity non-uniformity between adjacent filters, roll-up and reverse-sweep zones only visible under visualisation, door hardware that passes electrically while failing to restore the cascade, and alarms that annunciate locally but never reach a trend.
| Failure mode | Why a visual check passes it | What the witness does instead |
|---|---|---|
| Bypass at filter frame, gasket or housing | The filter is intact and its certificate is valid | Full face-and-frame scan against the upstream challenge, with retest after any repair |
| Velocity non-uniformity across adjacent filters | Every individual reading sits inside the range | A grid map with at least one point per filter or FFU, not one anemometer reading |
| Airflow roll-up and reverse sweep | Nothing about the build looks wrong | Visualisation at rest and in operation, including simulated interventions, with video retained |
| Cascade collapse on door opening | The steady-state differential is correct | Trend capture through disturbance, and recovery behaviour after the door closes |
| Door package undermining pressure control | The interlock proves out electrically | Self-closer force, latch and seal compression, and swing direction against the protected room |
| Alarm annunciated but never recorded | The alarm sounds when challenged | Setpoint, delay, local display, HMI value and historian record challenged as one chain |
01 · Scope
When cleanroom FAT needs a contamination-control reading, not an inspection
Cleanroom systems are the case where a general inspection is least useful, because the acceptance-critical properties are not properties of objects — they are properties of behaviour under disturbance. Pressure cascades interact: one unstable room disturbs its neighbours, and the drivers are sensing accuracy, exhaust and supply airflow accuracy, duct impedance and envelope airtightness rather than anything a walk-round would show. Filter leakage is a question about frames and gaskets, not media. Interlocks are a question about the mechanical door package, not just the logic. Witnessing this well means arriving with the URS, the P&IDs and the approved protocol, and knowing which of the ISO 14644-3 methods the module can honestly support.
It becomes useful when
- Modular cleanrooms, AHUs, FFUs, terminal HEPA housings, pass-throughs or airlocks are being built to a project URS in China.
- The application is sterile pharma, biologics, medical device or another environment where a contamination excursion is a regulatory event, not a quality note.
- A payment milestone or shipment release is tied to the FAT, and the contract does not distinguish it from site acceptance.
- The supplier's FAT protocol proposes single-point readings where the method calls for a grid, or omits visualisation and leak scanning entirely.
- Calibration certificates, filter certificates or as-built markups are being promised for later rather than presented at the witness.
- The site team commissioning the facility is not the team that specified it.
02 · Verification
What the witness controls at the supplier's works
Control point
Document freeze before the witness date
Approved FAT protocol, latest issued-for-manufacture drawings, panel schematics, P&IDs, sensor and I/O lists, filter certificates, calibration certificates, software revision list and as-built markups required before travel — missing calibration is treated as a valid blocker, not paperwork to send later.
Control point
Mechanical and architectural readiness
Panel flatness and coating continuity, glazing and gasket integrity, sealant finish, ceiling sealing against the void above, and the geometry that decides cleanability: recesses, ledges, sliding-door tracks, penetrations and drain treatment. Airlock and pass-through construction checked against the contamination-control intent, not just against the drawing.
Control point
Airflow and installed filter leakage
Airflow rate in non-unidirectional zones and velocity distribution in unidirectional ones, measured on a defined grid with at least one point per filter or FFU — for unidirectional installations, ISO 14644-3 measures velocity approximately 150 to 300 mm from the filter face. Installed filter scanning covers face and frame against the upstream challenge, with a downstream reading at 0.01% of that challenge treated as a significant leak and any repair followed by a re-scan.
Control point
Pressure cascade and visualisation
Differential pressure verified only after airflow and balancing results are acceptable, captured as a trend through steady state and disturbance rather than a handheld snapshot. Smoke visualisation demonstrates no ingress from lower to higher grade and no transfer over operators or equipment into the critical zone, at rest and in operation, with video retained as evidence.
Control point
Controls, interlocks and monitoring fidelity
Critical transmitters, setpoints, alarm delays, local displays, HMI values and historian records challenged as one chain; airlock and pass-through interlocks proven including failure behaviour and local warning; control voltage, phase rotation, E-stop and fail-safe damper and fan response verified against the site electrical basis rather than a temporary factory arrangement.
03 · Evidence
Evidence captured during the witnessed cleanroom FAT
- Baseline and factory conditions
- Frozen document set with revision numbers, plus the calibration register and filter certificate log as they stood at the witness.
- Actual factory conditions recorded: power source and voltage, compressed-air source and quality, aerosol method and upstream challenge, ambient conditions.
- Software revision and every temporary bypass or substitution used during the witness, named rather than summarised.
- Measured performance
- Room or module airflow maps and velocity grids, not a single averaged number.
- Leak-scan results by filter position, with repair and re-scan history where applicable.
- Pressure trends through steady state and disturbance, and airflow visualisation video at rest and in operation.
- Controls and disposition
- Alarm and interlock challenge records showing setpoint, delay, annunciation and what the historian actually captured.
- Punch list categorised by consequence: shipment blockers, pre-energisation blockers, site-acceptance items and documentation closure items with dates.
- FAT-to-SAT transfer package: open-items register, as-built drawings, software backup, preservation and packing record, and the reassembly notes the site team will need.
04 · Deliverables
Deliverables issued
Deliverable
For · Project engineering and the qualification teamProtocol review and readiness note
Comments on the supplier's FAT protocol, the methods it under-scopes, and whether the documents and calibration pack justify travelling.
Deliverable
For · Commissioning, qualification, quality and external reviewersWitnessed cleanroom FAT report
Test step, acceptance criterion, observed result, deviation and disposition — with airflow maps, pressure trends and leak-scan results attached, not summarised.
Deliverable
For · Engineering and qualityVisualisation and photo evidence pack
Retained smoke-study video, annotated photographs of build geometry and the conditions under which each test ran.
Deliverable
For · Procurement and the supplierCategorised punch list
Category A to D by consequence, so the shipment decision is separable from the documentation backlog.
Deliverable
For · The site commissioning teamFAT-to-SAT transfer package
What was proven at the factory, what was not, and the carryover risks the site team inherits — issued so nobody has to reconstruct it later.
05 · Risks reduced
Risks this control closes
Risk
A passed factory demonstration is treated as a qualified cleanroom.
How the witness closes it
The report separates factory-verifiable results from site-only items explicitly, and the FAT signature is scoped to shipment release.
Risk
A valid filter certificate is accepted in place of an installed leakage scan.
How the witness closes it
Face-and-frame scanning against the upstream challenge is witnessed on the installed assembly, with re-scan after any repair.
Risk
The cascade is proven with one handheld reading while everything is calm.
How the witness closes it
Differential pressure is captured as a trend through disturbance, with recovery behaviour and control-loop stability observed near setpoint.
Risk
The test has already been run, or the unit already packed, before the witness arrives.
How the witness closes it
Readiness is confirmed before travel, and a pre-run or pre-packed condition is recorded as a material weakening of the witness rather than accepted quietly.
Risk
The workshop builds from an earlier drawing revision and the as-built never catches up.
How the witness closes it
The document freeze fixes revisions before the witness, and panel wiring, tagging and as-built markups are checked against that frozen set.
Risk
Factory-only utilities hide a mismatch with the site electrical or compressed-air basis.
How the witness closes it
The actual power, control voltage and air source used during the test are recorded, and any divergence from the site basis is raised as a carryover risk.
06 · Background resources
Related
- Resource · Sourcing cleanroom and facility systems from ChinaThe upstream sourcing guide: specification, certification documents and the payment structure this FAT sits inside.
- Resource · FAT vs SATThe boundary in general terms: what each test verifies and what each signature commits the buyer to.
- Service · Factory acceptance test in ChinaThe parent FAT service: scope, control points and deliverables for any equipment class.
- Resource · Factory acceptance testing guideThe full FAT guide: sequence, evidence and the release decision.
- Resource · Site acceptance test checklistThe site-side checklist the carryover items from this FAT feed into.
- Sector · HealthcareHow Sinospect supports pharmaceutical and medical-device buyers sourcing facility systems from China.
07 · Questions
Frequently asked questions
Can a cleanroom be qualified at the supplier's factory?
No, and any offer to do so should be read carefully. What a factory FAT can prove is build quality, component identity, control logic, interlocks, alarm behaviour, documentation completeness and — where a module is fully assembled — representative airflow and leakage performance. Final room classification, the integrated cascade with site doors and building leakage, and performance under real utilities, heat load and occupancy all belong to site acceptance and qualification. The witness scope states which is which in writing.
What does an installed filter leakage scan actually prove?
That the filter, its frame, its gasket and its housing form a continuous barrier as installed — not that the filter media meets its rated efficiency, which is what the manufacturer's own certificate covers. The distinction matters because the common defects are bypass paths: knife-edge or gel-seal mismatch, uneven gasket compression, housing distortion, frame twist. A valid filter certificate and a failed installed scan are entirely compatible, which is why the scan is witnessed and repeated after any repair.
Why is a single handheld pressure reading not enough?
Because cascade failures are usually dynamic rather than static. A room can sit correctly at its setpoint and still collapse when a door opens, hunt around the setpoint because the control loop has no usable authority near its operating position, or read correctly on a transmitter whose range is far too wide for a differential of ten to fifteen pascals. The witness requires trend capture through steady state and disturbance, not a number taken once while everything is calm.
Which standards does the witness work against?
ISO 14644-3 supplies the test methods — airflow, air-pressure difference, installed filter system leakage, airflow direction and visualisation, temperature, humidity, recovery, containment leak. For sterile applications the context comes from EU GMP Annex 1, the FDA aseptic processing guidance, WHO sterile GMP guidance and China's NMPA GMP, while GB 50457 and GB 50591 govern how Chinese suppliers design and build pharmaceutical cleanrooms. These are the frameworks the witness verifies against; Sinospect is not an accreditation body under any of them.
What should the FAT signature commit us to?
Release for packing and shipment, and nothing more, unless the contract deliberately says otherwise. Tying warranty start, retention release or final payment to a factory demonstration transfers the leverage you will need during commissioning to the moment you have the least information. The recommended structure is simple: Category A punch items closed before shipment, Category B before energisation, retention held to site acceptance, warranty running from site acceptance.
08 · Getting started
What to send — start with what you have
The URS and the supplier's FAT protocol are the two documents that decide what the witness can realistically prove. Everything else can follow.
URS or performance specification
The requirement the module is being built against — including the values it sets where the design codes would otherwise supply defaults.
Approved drawings and P&IDs
The frozen baseline for the mechanical, architectural and control verification.
FAT protocol or ITP
The supplier's proposed test content, so under-scoped methods are found before travel rather than on the day.
Applicable GMP or ISO basis
Which regulatory context governs, so acceptance is read against the right framework.
Site utility standards and payment-milestone wording
What the module will actually run on, and what the FAT signature is contractually committing you to.
Partial information is fine. Sinospect states which ISO 14644-3 methods the module can support and which cannot be honestly run at the factory, before any commitment.
Book an independent witness for your cleanroom FAT
Send the URS or specification, the supplier's FAT protocol if it exists, and the expected test window. Sinospect replies within one business day with the proposed witness scope, the factory-verifiable and site-only split as it stands today, a quotation and the earliest feasible attendance.