A qualification package arrives with a flawless visualisation video: steady framing, the medium descending from the terminal filter in even sheets, no turbulence. The room, however, is empty and no operator appears. Months later, during a media fill, an intervention on the dosing unit forces a forearm across open containers, and growth appears where nobody had chosen to monitor.
That video was not wrong: it was irrelevant, because it documented a state that never exists in production. An airflow visualisation study is not footage of an installation: it is an experiment answering a precise question about how air moves while people work. If that question is not in the protocol, no image quality recovers it. This article is part of the Cleanrooms & HVAC Systems track.
What depends on the study
The study does not end in an annex. Annex 1 §4.30 requires the air velocity of unidirectional systems and the measurement location to be justified in the protocol, and that location is credible only if someone has observed where the protecting air travels. Section 4.31 ties the microbiological sampling points to a risk assessment that also considers visualisation results. A generic video produces locations chosen for convenience, not for risk.
Regulatory framework: what actually binds
| Source | Status | What it actually binds |
|---|---|---|
| Annex 1 §4.15 | Applicable since 25 August 2023 | Airflow patterns must be visualised to demonstrate that air does not ingress from a lower to a higher grade, and does not travel from less clean areas, for example the floor, or over operators and equipment |
| Annex 1 §4.24–4.25 and §4.29–4.31 | As above | Airflow direction and visualisation among the mandatory qualification items; "at rest" and "in operation"; velocity and measurement location justified; microbiological points informed by visualisation |
| Annex 15, sections 3 and 4 | Operational since 1 October 2015 | Predefined criteria, deviations handled, requalification at justified intervals |
| FDA, Aseptic Processing Guidance | September 2004, nonbinding | Recommends in situ analysis of airflow patterns in the critical area under dynamic conditions, to demonstrate unidirectional flow and sweeping action away from the product; video as an aid |
| ISO 14644-3:2019 | Voluntary standard | Test method for airflow direction test and visualisation: the method, not the acceptance criteria |
The last line is the most misread. No universal acceptance criterion exists for a smoke study: ISO 14644-3 standardises how the test is run, not what counts as a favourable outcome. The protocol writes and justifies the criteria, tied to the real operations of the room [GUIDEGXP].
What the study has to demonstrate
The two demonstrations of §4.15
Section 4.15 asks for two distinct demonstrations [REQUIREMENT]. The first concerns interfaces: no air ingress from a lower grade towards a higher one, hence doors, airlocks and pass-boxes, subjects of the article on finishes and airlocks. The second concerns the room itself: air must not reach the higher grade from less clean zones, typically the floor, nor after passing over operators and equipment. Two criteria, two families of scenarios.
Static and dynamic answer different questions
The static study, with the room essentially "at rest" as defined in §4.29(i), verifies the geometry of undisturbed flow: supplies, returns, fixed obstructions, recirculations, short-circuits. It isolates design defects from the effects of activity, and a room that fails empty will never work full. It does not demonstrate what Annex 1 asks for, because the sources named in §4.15 exist only while work is under way: personnel are at once a particle source, an obstacle and a thermal source; machinery entrains air; doors change the air balance, as in the article on airflow and pressure cascades. The "in operation" state of §4.29(ii) assumes equipment running and the maximum number of people performing or simulating routine work. A study without interventions shows that the installation was built as designed, not that the product is protected while people work.
Worst case selection is a risk decision
No source states how many scenarios are needed. The selection is a QRM decision to document, and it starts from the process, not from the HVAC: process flow, the interventions foreseen in the SOPs, and the corrective interventions actually performed, rare ones included. Typical candidates are the intervention closest to exposed product, the most awkward posture, the largest obstruction above the critical point, and material transfers. A scenario discarded without a written rationale is the first question an assessor asks.
Skeleton of the protocol
The protocol follows the logic of Annex 15 and rests on seven sections.
- Purpose: rooms, critical zones and interfaces, with objectives traced to the two demonstrations of §4.15.
- Prerequisites: HVAC in the qualified state, final filter integrity, air volumes and differentials verified, instruments calibrated.
- Room state: occupancy and equipment configuration per scenario, per §4.29.
- Scenario matrix: location, activity, personnel, duration and rationale, traceable to process flow and SOPs.
- Medium and instrumentation: category, release, lighting, camera positions, compatibility documented.
- Roles: who executes, who handles the medium, who acts as process personnel, who observes, who approves.
- Criteria, deviations and deliverables: an acceptable outcome per scenario in observable terms, written before execution; video files tied to the protocol line.
Execution: medium, lighting, recording
The visualisation medium
The media in use belong to generic technology categories and are selected on three properties, not on the look of the result. Neutral buoyancy: a medium warmer or colder than room air tells the story of its own motion. Absence of momentum at release: a fast jet masks the weak phenomena that matter. Compatibility with product, surfaces and operators, assessed and documented before the study together with the subsequent cleaning.
The source belongs upstream of the phenomenon, not on the critical point, and release positions are recorded scenario by scenario: a result obtained by releasing elsewhere is not comparable. No official source fixes the flow of medium, the probe distance or the release duration: these are experimental parameters to define in the protocol and keep constant between repetitions [GUIDEGXP].
Lighting, recording and retention
Contrast is a validity condition: without grazing or backlighting on the plane of interest a backflow stays invisible, and the video documents an absence of evidence, not an absence of a problem. The frame must hold critical point, disturbance and operator together, or the causal relationship cannot be reconstructed.
The video is a qualification record: identified by protocol, scenario, date and operators, not alterable, readable throughout the retention period, which follows from the site document policy, not from a figure in a standard. Annex 11 §7.1 requires periodic checks of accessibility and readability of archived data, the logic also applied to BMS and EMS data. Re-shooting until one take looks pleasing, without retaining the attempts, is a data integrity problem.
Reading the video without concluding too early
Reading a video is subjective: every observation admits several explanations, some concerning the test method rather than the installation.
| Observation in the video | What it may mean | What to check before concluding |
|---|---|---|
| Stagnation above a horizontal surface near the critical point | Dead zone from an aerodynamic shadow | That the configuration is the production one; that release is upstream of the shadow; the supply volume |
| Rise from below towards the protected zone | Thermal plume or backflow from the return | Active thermal sources; return positions; momentum of the release |
| Ingress into the higher grade room as the door opens | Loss of direction at the interface | The differential at that instant; the opening sequence; the interlock in its qualified configuration |
| Passage over hands or forearms before the product | A path contrary to the second demonstration of §4.15 | That posture and intervention are those of the SOP, not a version softened for the camera |
| Diffuse turbulence in a unidirectional area | Mixing from an obstruction, operator movement or the release | Velocity and measurement location; release technique; whether it persists without activity |
| Dispersion too fast to be followed | Quantity, contrast or lighting inadequate | Lighting and framing: rapid dilution is not evidence of protection |
| No anomaly anywhere in the video | Correct flow, or unrepresentative scenarios | The matrix against process flow, SOPs and interventions performed; operators present |
Deviations, remediation and repetition
An observation contrary to the protocol criteria is a deviation to handle under Annex 15. There are four remediation routes, in this order: geometry, the position of equipment and obstructions; procedure, intervention technique and posture; air, supplies, returns or velocity; and physical separation. Changing the video is the fifth, and it is not a remedy.
After remediation the scenario is repeated under the same conditions and its downstream effects assessed: velocity measurement location (§4.30), sampling points (§4.31), SOPs and training. On periodic repetition precision is needed: visualisation does not appear in the minimum requalification list of §4.32, which lists classification, final filter integrity, air volume, pressure differential and air velocity, with a maximum interval of 6 months for grades A and B and 12 months for C and D [REQUIREMENT]. It is repeated when equipment, facility or processes change; any additional periodicity is a choice to justify in the CCS and in the requalification and troubleshooting plan.
Justifying the choices with QRM
ICH Q9(R1), Step 4 on 18 January 2023, is the methodological frame and contains no design parameters; Annex II.4 applies QRM to facilities, equipment and utilities, scope of qualification included. Section 5.1 treats formality as a continuum: the scenarios of an aseptic filling line deserve a structured tool, those of a grade D room do not. Section 5.3 acknowledges subjectivity and bias, the dominant risk here: whoever designed the flow tends to see confirmations, and whoever executes tends to produce a scenario that works. Criteria written before execution and an independent observer are the countermeasure; the outcomes feed the CCS (§2.4).
Worked example: Site Vega
Site Vega is a realistic but entirely fictitious example. An aseptic filling line in grade A with a grade B background is qualified with a study run with the line stopped: QA notes that no scenario contains an intervention and the matrix is not traceable to the SOPs.
The protocol is reissued from the interventions listed in the batch record. Three activities emerge that nobody had considered: replacing a component above the filling zone, recovering a fallen container, and manual loading during a jam. The repeat shows a stable backflow above the filling point during the component replacement alone. The correction does not touch the HVAC: it changes the intervention sequence and moves the tool rest out of the vertical above the product. The scenario is repeated and the SOP updated.
Evaluation matrix for the study set-up
Weighting is assigned by the team: no transferable weighting exists across sites [QRM].
| Criterion | Weight | Static only | Dynamic with simulated activities | Dynamic on a representative sequence |
|---|---|---|---|---|
| Representativeness against §4.15 | First demonstration only | Both, if interventions are real | Highest coverage | |
| Control of variables | High | Medium | Low | |
| Ability to isolate the cause | Excellent on design defects | Good with one variable per scenario | Limited | |
| Organisational burden | Low | Medium | High |
Levels of prescriptiveness
| Statement | Level | Source |
|---|---|---|
| Airflow patterns visualised for the two demonstrations; direction and visualisation part of qualification | [REQUIREMENT] | Annex 1 §4.15 and §4.25(iv) |
| Requalification within 6 months for grades A and B and 12 months for C and D, and after changes | [REQUIREMENT] | Annex 1 §4.32 |
| In situ analysis of patterns under dynamic conditions, video as an aid; 0.36–0.54 m/s at the working position | [GUIDANCE] | FDA 2004, nonbinding; Annex 1 §4.30 |
| Method for airflow direction test and visualisation, without acceptance criteria | [STANDARD] | ISO 14644-3:2019 |
| Neutral buoyancy, upstream release, contrast and framing | [GEP] | Good engineering practice |
| Worst case selection and any periodicity beyond change control | [QRM] | ICH Q9(R1) Annex II.4; CCS §2.3–2.6 |
| Criteria written before execution, independent observer, retention of attempts | [GUIDEGXP] | GuideGxP recommendation |
Operational checklist
- Derive the scenario matrix from process flow and real interventions, with a rationale for every scenario included and excluded.
- State occupancy and equipment configuration per scenario.
- Verify the HVAC prerequisites and record the compatibility assessment of the medium.
- Fix release position and technique and keep them constant between repetitions.
- Arrange lighting and framing that hold critical point and operator together.
- Approve the criteria before execution and appoint an independent observer.
- Identify every file uniquely and retain unsuccessful attempts.
- Handle any observation contrary to the criteria as a deviation and repeat the corrected scenario under the same conditions.
- Update sampling points, SOPs and training with the study outcomes, and link the conclusions to the CCS and the requalification plan.
Recurring mistakes and red flags
- A study of the empty room only, presented as evidence of product protection.
- Acceptance criteria attributed to ISO 14644-3 or copied from another site.
- A matrix not traceable to the SOPs, or postures agreed for the camera.
- Attempts discarded and not retained, or files without unique identification.
- A remedy applied to the HVAC when the problem is geometry or procedure.
If this way of separating requirement, guidance and good practice is useful in daily work, the discussion continues in the technical newsletter The Pragmatic GMP.
Key takeaways
- Section 4.15 asks for two demonstrations: no ingress from the lower grade, and no air path from less clean zones or over operators and equipment.
- The static study verifies flow geometry; only a dynamic study with real interventions answers the regulatory question.
- No universal criteria exist: ISO 14644-3 gives the method, the protocol writes the criteria.
- Worst case selection is a QRM decision to document, exclusions included.
- Visualisation is not in the minimum list of §4.32: it is repeated through change control.
References
- EudraLex Volume 4, Annex 1, applicable since 25 August 2023.
- EudraLex Volume 4, Annex 15 (2015) and Annex 11 (2011).
- FDA, Sterile Drug Products Produced by Aseptic Processing, September 2004.
- ISO 14644-3:2019, test methods.
- ICH Q9(R1), Step 4 of 18 January 2023.
