Handover of a new suite almost always stalls at the same point: the system runs, the airflows are there, the differentials are visible on the BMS, but the document package from the construction site does not survive an IQ review. The as-built drawings do not show the routing changes agreed on site; the balancing report does not state the system condition under which measurements were taken; two calibration certificates had expired on the day of the test; the punch list has no owner and no impact classification. None of these is a technical failure: they are documentation-process failures, and each one turns a planned qualification into open deviations.
The site is run on finish-and-hand-over logic, while the cleanroom and its HVAC enter the GMP lifecycle the moment the first duct is installed: the information qualification will need is generated once and cannot be reconstructed. The wider picture is in the Cleanrooms & HVAC Systems hub.
Why construction quality is paid for during qualification
The effects appear at increasing cost. At tender stage, without a specification describing protection, cleaning and the turnover package, bids are not comparable: the bidder who includes that work loses to the one who leaves it out. In qualification, every missing piece of information becomes a test to repeat. In operation, residual contamination and imperfect sealing give a worse baseline. Over the lifecycle, without as-builts every change control starts with a survey.
Regulatory framework: what actually constrains site work
| Source | Status / date | What it actually constrains |
|---|---|---|
| EudraLex Vol. 4, Annex 1 | C(2022) 5938 final, applicable since 25 August 2023 | §4.23–4.25: cleanrooms and clean air equipment qualified with a methodology consistent with Annex 15, and qualification kept distinct from operational monitoring; §6.1–6.6: utilities lifecycle, pipework and ducting located to avoid recesses. Construction work itself is not regulated |
| EudraLex Vol. 4, Annex 15 | 2015 revision, operational since 1 October 2015; revision concept paper of 9 February 2026 not adopted | URS (3.2), DQ (3.3), FAT/SAT (3.4–3.7), IQ, OQ and PQ (3.8–3.14). The word commissioning does not appear in the text |
| ISO 14644-4 (2022), ISO 14644-3 (2019), ASTM E2500-25, ISPE Baseline Guide Vol. 5, PIC/S PI 009-4 | Voluntary; E2500-25 issued 15 March 2025; PI 009-4 effective 1 January 2021 | Design, construction and start-up (part 4); test methods without acceptance criteria (part 3); the specification – design – verification logic (E2500 and ISPE); PI 009-4 is inspector training material |
Clean construction: protection, dust control, cleaning
The clean construction protocol is a contractual document: it defines the construction phases, the boundary between dirty trades and areas already closed, access rules, permitted cutting and drilling techniques, and the method and frequency of progressive cleaning. Dust generated in one phase must stay out of the volumes that will become the inside of the system: that criterion orders the sequence of work.
Ductwork, AHUs and filters
Installed duct runs must be capped and sealed at the end of each shift: protection fails when a run is reopened for a later trade, not when it is first installed. Internal cleanliness of ductwork, verification method and acceptance criteria are contractual matters. No duct cleanliness class is imposed as a GMP requirement, so the level must be written into the specification, justified on a risk basis and verified with a method agreed before work starts.
The AHU is the most exposed item. If it is run to dry screeds, resins and paint it must operate with sacrificial construction filters: final filters are installed after fine cleaning, following a sequence stated in the construction programme and signed off at handover. An AHU used as a site fan loads return sections and dampers with dirt that no new terminal filter removes; the architectural implications are in HVAC architecture for GMP cleanrooms.
Sealing and penetrations
Penetrations are the most expensive defect to correct once ceilings are closed. What is needed is a room-by-room list of penetrations, a signed inspection before closure, and an explicit rule on who seals what when one trade follows another. The same applies to panel joints and coving: details in cleanroom finishes, airlocks and hygienic construction.
Instrumentation and wiring: installation decides qualification
A sensor installed somewhere other than the position stated in the URS produces data that is formally valid and substantially useless. Each position must be approved on drawings before installation, against three criteria: representativeness for the parameter to be controlled; the instrument's installation conditions, such as straight duct runs and pressure taps away from air jets and doors; accessibility for calibration. Control wiring needs segregation from power lines and tags consistent across P&ID, panel and BMS. A documented loop check proves the measurement chain is intact.
TAB: an engineering activity, not a qualification test
TAB (testing, adjusting and balancing) takes the system from installed to adjusted and documents its as-left configuration. It is an engineering activity with four method requirements, all verifiable on the report: a stated measurement method for each quantity; calibrated instruments identified by serial number, with traceable certificates valid on the test date; stated test conditions, that is system status, installed filters and their initial pressure drop, damper positions, doors closed; and a traceable record of who measured, when, with which instrument, the value found and the value left after adjustment.
The sequence runs from general to particular: mechanical checks and duct leakage tests, primary airflows at the AHU, branch balancing, terminal adjustment, room differentials following the design cascade, then iteration until airflows and pressures are stable together. They are coupled quantities: closing that loop is what the report must demonstrate.
Balancing tolerances are not universal and appear in none of the sources above: they are a contractual parameter, fixed in the URS and the specification. The relevant regulatory references are design constraints, not site tolerances: between adjacent rooms of different grade Annex 1 §4.14 gives a minimum of 10 Pa, explicitly qualified as a guidance value [GUIDANCE]; for unidirectional airflow §4.30 gives 0.36–0.54 m/s at the working position, unless otherwise scientifically justified in the CCS [GUIDANCE]. Cascade sizing is covered in airflow and pressure cascades.
Commissioning and qualification: two processes, two owners
| Aspect | Commissioning | Qualification |
|---|---|---|
| Nature | Engineering verification: system installed and working as designed | Documented GMP process: fitness for intended use |
| Document basis | Design specification, contract specification, manuals | URS, VMP, protocols per Annex 15 |
| Acceptance criteria | Engineering and contractual | From the URS and Annex 1 requirements |
| Process ownership | Engineering and contractor | Quality, with formal approval |
| Typical activities | TAB, functional and loop checks, punch list, turnover documentation | DQ, IQ, OQ, PQ, periodic requalification |
| Handling of shortfalls | Punch list with owner and date | GMP deviations and change control |
| Regulatory placement | Not defined in EudraLex; ISPE/ASTM literature | Annex 15 section 3; Annex 1 §4.23–4.32 |
The delicate point is data reuse. In the EudraLex framework, reuse of verification performed before formal qualification is formalised as FAT/SAT: Annex 15 §3.4–3.7 covers factory and site testing and the possibility of considering it in qualification, provided documentation is adequate and it is demonstrated that transport and installation have not altered performance [REQUIREMENT]. The expression commissioning leverage belongs to ISPE Baseline Guide Vol. 5 and ASTM E2500-25 [STANDARD]: in a European dossier it must be argued as FAT/SAT, never attributed to Annex 15.
Reuse is decided up front: a pre-approved protocol with written acceptance criteria, calibrated instruments with traceable certificates, identified and qualified executors, raw data retained and accessible to Quality.
Site risks and how they are demonstrated at turnover
| Site risk | When it arises | How it is prevented | How it is demonstrated at turnover |
|---|---|---|---|
| Construction dust in installed ductwork | Erection and finishing, with open ends | End-of-shift capping; separation of dirty trades | Dated photographic record and sample internal inspection |
| AHU used to condition the site | Drying of screeds and paint | Sacrificial filters; final filters after fine cleaning | Signed installation sequence and initial pressure drop |
| Unsealed penetrations | Services crossing walls and ceilings | Room-by-room list and inspection before closure | Signed sealing check-list |
| Measurements with out-of-calibration instruments | TAB phase | Certificate expiry checked before testing | Certificates valid on the date of the test |
| Unclassified punch list | At handover | Classification rule agreed in the specification | Items with impact and owner, reviewed by Quality |
The turnover package is a contractual deliverable
The handover package is no courtesy from the supplier: it is a deliverable listed in the URS for cleanrooms and HVAC systems, with content, format, timing and acceptance criteria, and payment tied to its acceptance. The Weight column is completed by the site according to area criticality.
| Package element | Minimum evidence | Produced by | Approved by | Weight |
|---|---|---|---|---|
| As-built drawings | Layouts, P&IDs and schematics updated | Contractor | Engineering | |
| Component certificates | Datasheets and certificates of the installed filters | Contractor | Engineering | |
| TAB report | Method, test conditions, as-left values, calibrations | TAB contractor | Engineering and Quality | |
| Functional and loop checks | Signed result for each tag and datapoint | Contractor | Engineering | |
| Clean construction records | Protection, cleaning, filter sequence | Contractor | Engineering | |
| Punch list | Classified impact, owner, closure date | Contractor | Engineering and Quality | |
| Traceability matrix | URS requirements, FAT/SAT tests, IQ/OQ tests | CQV | Quality |
Handover to qualification is a formal act: Quality reviews the open punch list, decides which items prevent IQ from starting, and freezes the documentation baseline. The tests that follow are covered in cleanroom qualification.
Decisions to justify with QRM
ICH Q9(R1), Step 4 on 18 January 2023, is methodology and contains no HVAC parameters; Annex II.4 nevertheless applies QRM to facilities, equipment and utilities, including the scope of qualification. Formality as a continuum (§5.1) grades site protection and documentation checks against the impact of the area. Subjectivity and bias (§5.3) explain why classification of blocking punch list items should be approved by people who did not do the work. The Annex I tools — FMEA, HAZOP, risk ranking and filtering — decide which checks are systematic and which are done on a sample [QRM].
Worked example: Site Vega
Site Vega is a realistic but fictitious example: an area with grade C and D rooms and unclassified support spaces. Three problems appear at handover: the TAB report states test conditions for only some branches and two instruments were out of calibration on the measurement date; protection on several duct runs, removed for late electrical work, was not reinstated; the punch list is not classified by impact.
The decision was not to reject the package outright. Measurements taken with out-of-calibration instruments were repeated on the affected branches; for the uncovered ducts a sample internal inspection was agreed, with the acceptance criterion written beforehand; the punch list was reclassified with Quality. IQ started late but with no open deviations, because the acceptance criteria were in the URS.
Levels of prescriptiveness
| Statement | Level | Source |
|---|---|---|
| Qualification follows Annex 15 and is distinct from operational monitoring | [REQUIREMENT] | Annex 1 §4.23–4.24 |
| Reuse of testing performed before qualification is governed as FAT/SAT | [REQUIREMENT] | Annex 15 §3.4–3.7 |
| Minimum 10 Pa between adjacent rooms of different grade | [GUIDANCE] | Annex 1 §4.14, guidance value |
| 0.36–0.54 m/s at the working position, unidirectional airflow | [GUIDANCE] | Annex 1 §4.30 |
| ISO 14644-3 gives methods, not acceptance criteria; commissioning leverage is ISPE/ASTM language | [STANDARD] | ISO 14644-3:2019; ISPE Vol. 5; ASTM E2500-25 |
| End-of-shift capping of ducts; final filters installed last | [GEP] | Good engineering practice |
| Duct cleanliness, balancing tolerances, extent of sample checks | [QRM] | Contract specification and documented risk analysis |
| The turnover package is listed in the URS as a contractual deliverable | [GUIDEGXP] | GuideGxP recommendation |
Operational checklist
- Make the clean construction protocol a contract document.
- Specify duct cleanliness, verification method and acceptance.
- Cap ducts at the end of each shift and after reopening.
- Prohibit running the AHU with final filters installed.
- Have the filter installation sequence signed off.
- Approve instrument positions before installation; verify tags before loop checks.
- Check calibration validity on the date of the test.
- Require method, test conditions and as-left values in the TAB report.
- Fix balancing tolerances and the tests to be reused as FAT/SAT in the URS.
- Classify punch list items by impact and have them reviewed by Quality.
- Freeze the as-built baseline before IQ; verify the package before final payment.
Common mistakes and red flags
- Presenting commissioning data as qualification already done, with no pre-approved protocol.
- Writing that Annex 15 allows commissioning leverage: it governs FAT/SAT and does not name commissioning.
- Specifications citing withdrawn parts of the EN 1822 series: E/H/U classification remains EN 1822-1:2019, the test methods are now in EN ISO 29463-2/-3/-4/-5.
- A flat punch list, with no impact and no owner: a container of work never closed.
- CFD used instead of field verification: it replaces neither qualification, nor airflow visualisation, nor field testing.
To turn these principles into documents that hold up in inspection, The Pragmatic GMP is the GuideGxP newsletter: short, operational analysis on qualification, HVAC and contamination control, from its dedicated page.
Key takeaways
- Construction is the first phase of the documentation lifecycle: data is generated once.
- Commissioning is engineering, qualification is GMP under Annex 15; reuse is argued as FAT/SAT.
- A TAB report is credible only if it states method, calibrations, conditions and as-left values.
- No duct cleanliness level is imposed by GMP: it is written into the specification and justified.
- The turnover package is a URS deliverable, with payment tied to acceptance.
