Isolator Qualification and Smoke Studies for Aseptic Processing

C1.1  What each study type actually demonstrates

Requirement Annex 1 §4.15 states the core requirement plainly:

“Airflow patterns within cleanrooms and zones should be visualised to demonstrate that there is no ingress from lower grade to higher grade areas and that air does not travel from less clean areas (such as the floor) or over operators or equipment that may transfer contamination to the higher grade areas… Airflow pattern studies should be performed both at rest and in operation (e.g. simulating operator interventions).”EU GMP Annex 1 (2022), §4.15. Source

Practice Practitioners and standards bodies split that single sentence into two named study types, each answering a distinct question:

• Static (at rest) — no personnel present, nothing moving. The question is whether the room or barrier system's engineering design produces correct unidirectional airflow and cascade direction in the first place: does clean air actually sweep toward the critical zone and away from it toward lower-grade areas, with no stagnant pockets or turbulence, before a single person or process variable is introduced. A facility that fails static conditions has a design problem, not an operator problem.
• Dynamic (in operation) — the same space, now with simulated production activity: gowned personnel performing real interventions, equipment in its normal running position, doors opening and closing as they would on a live batch. The question is whether that same correct airflow survives contact with an actual process. A facility can pass static and still fail dynamic — the room's design was sound, but a specific intervention, a specific piece of equipment, or a specific operator's positioning disrupts first air in a way the empty-room test could never have shown.

Practice Neither result substitutes for the other, and a facility that runs only one has not actually answered Annex 1's question. Static alone proves the room can work; it says nothing about whether it does work once people are in it. Dynamic alone, run without an established static baseline, gives you no way to tell whether a disturbance you observe came from the intervention you just simulated or from a design flaw that was there all along.

C1.2  Isolators and RABS get their own explicit call-outs

Requirement Annex 1 does not leave barrier technology to the general rule in §4.15 alone — it names isolators and RABS specifically:

“Airflow pattern studies should be performed at the interfaces of open isolators to demonstrate the absence of air ingress… The background environment for RABS used for aseptic processing should correspond to a minimum of grade B and airflow pattern studies should be performed to demonstrate the absence of air ingress during interventions, including door openings if applicable.”EU GMP Annex 1 (2022), §4.20(i)(c) and §4.20(ii). Source

Requirement The pattern across both clauses is consistent with Arc B's framing of a transfer mechanism or a glove system: a barrier is only as good as the boundary conditions that have actually been tested. For an open isolator, that boundary is the interface — the RTP, the mousehole, the open point where the isolator meets the room. For a RABS, which by definition has a less complete physical barrier than a closed isolator, it is specifically the door-opening event, because that is the moment the barrier's main defense (physical separation) is temporarily suspended and airflow becomes the only thing still protecting the critical zone.

---

C2.1  Objective, scope, and the floor plan

Practice A protocol that will actually hold up starts by naming, specifically, what it is trying to demonstrate — "verify unidirectional airflow is maintained at the stopper bowl during the stopper-bowl refill intervention," not "verify airflow is acceptable in the filling suite." A vague objective produces an ambiguous result: reviewers and operators disagree afterward about what the study was even supposed to show. Good protocols pair that objective with a floor diagram of the room or isolator showing the position of every intervention, the smoke source, and every camera — decided and documented before execution, not improvised on the day.

C2.2  Choosing a smoke source that does not lie to you

Requirement ASTM E3379-25a, the current standard guide covering critical-environment airflow visualization, specifies what the tracer itself has to be:

generators should produce “constant, neutrally buoyant, and non-toxic tracer particles.”ASTM E3379-25a, Standard Guide for Critical Airflow Visualization. Source

Practice "Neutrally buoyant" is the load-bearing phrase. A tracer that rises because it is warmer than the surrounding air, or falls because it is a heavy aerosol or a fog rather than a true smoke, does not show you the room's actual airflow — it shows you its own physics superimposed on the room's airflow, and the two are easy to mistake for each other on video. This is one of the most common, and most avoidable, ways a smoke study's evidentiary value gets undermined: the room's airflow may have been genuinely fine, but the tracer chosen to visualize it moved on its own terms, and a reviewer — internal or regulatory — has no way to tell the difference after the fact from a recording alone.

Practice Common generator types include titanium tetrachloride smoke tubes, glycol- or glycerin-based fog machines tuned to a fine, low-volume output, and dry-ice or liquid-nitrogen-chilled water vapor. Each has a documented failure mode if used carelessly — too much output masks the very turbulence the study is meant to catch, too little is invisible on camera — which is why Practice a preliminary dry run, before the study is executed for record, is standard practice: it lets the team tune smoke density and confirm the tracer behaves neutrally in that specific room, that specific day, under that specific HVAC load, rather than discovering a visibility problem only after the recording is already the official record.

C2.3  Simulating interventions honestly, including the worst case

Practice A dynamic study is only as good as how faithfully it reproduces what actually happens on a production shift. That means using the actual gowned operators (or a representative range of operator heights and reach, since a six-foot operator and a five-foot operator do not create identical airflow disruption), the real equipment in its real position, and — critically — every intervention the SOP allows, not just the routine ones. An operator jam clearance, a stopper-bowl refill under time pressure, a dropped component recovery: these are exactly the moments Arc D's enforcement cases will show going wrong, and a protocol that only rehearses the easy interventions has not actually tested the boundary the regulation cares about.

C2.4  Acceptance criteria written before the smoke, not after

Practice Criteria set in advance typically require: visible unidirectional airflow sweeping away from exposed product and critical surfaces; rapid, complete dissipation of any turbulence introduced by an intervention, with first air re-established promptly; no visible air movement from a lower-grade area into a higher-grade one at any interface, door, or transfer point; and no instance of smoke reversing direction back over a critical surface after an intervention concludes. Practice A protocol should also say, explicitly, what happens if a criterion is not met — root-cause investigation, corrective action, and re-test — rather than leaving that decision to be worked out in the moment a failure is observed on camera.

---

C3.1  What Annex 1 actually requires you to keep

Requirement The requirement is short and easy to under-read:

“Video recordings of the airflow patterns should be retained.”EU GMP Annex 1 (2022), §4.15. Source

Practice That sentence carries more weight than its length suggests. It means the video itself — not just the protocol, not just a written summary of what was observed — is the retained GMP record, subject to the same document-control and retention expectations as any other qualification evidence. A facility that summarizes a study in a report and discards the underlying footage has not actually met this clause, even if the report describes a successful outcome, because the record that would let a later reviewer verify that description no longer exists.

C3.2  Filming so the recording can actually do its job

Practice A recording only functions as evidence if a viewer who was not in the room can follow it. Established practice includes: multiple camera angles positioned so no intervention is obscured by the operator's own body or equipment; a black background and supplemental lighting where the room's own finish or ambient light would otherwise wash out the tracer; audible or on-screen narration identifying which intervention is being performed and when, so the recording is self-documenting rather than requiring a separate log to interpret; and, where an isolator or barrier's curtain or physical divider blocks a clean line of sight, filming from inside the space the study is actually characterizing rather than only from outside it.

C3.3  Review is a second, independent step — not a formality

Practice The camera operator and the person conducting the intervention are, by convention, not the people who make the final call on whether the study passed. A separate quality reviewer — someone who was not managing the live execution — watches the retained footage against the protocol's written acceptance criteria and produces an independent assessment. This separation exists for the same reason independent second review exists anywhere in a quality system: the person running the study has an interest, even an unconscious one, in seeing a pass. Practice A disciplined review also does not stop at airflow. A trained reviewer watching smoke study footage is, incidentally, watching gowning technique, aseptic manipulation, and general behavior inside the critical zone — and a competent review process captures observations in those areas too, even though they were not the study's primary objective.

---

C4.1  The recurring failure signatures

Practice Across published inspection findings and industry guidance on airflow visualization, the same handful of problems recur often enough to be worth naming individually:

• Non-neutral tracer behavior. Smoke that rises, falls, or drifts on its own thermal or aerosol properties rather than tracking true room airflow (Module C2.2). On video this can look like a passing study — the smoke moves, the room looks "clean" — while actually showing nothing reliable about the airflow underneath it.
• Turbulence or reflux at a boundary. Smoke that reaches a critical surface, an interface, or a door opening and then visibly reverses or eddies back toward it, rather than sweeping cleanly away. This is precisely the ingress scenario Annex 1 §4.15 and §4.20 are written to catch.
• Incomplete intervention simulation. A study that rehearses only the easy, planned interventions and omits the awkward ones — the jam clearance, the dropped-component recovery, the reach across a taller piece of equipment — has not tested the conditions under which first air is actually most at risk (Module C2.3).
• Smoke density miscalibrated for the camera. Too little tracer is invisible on the retained footage even if airflow is genuinely fine; too much obscures the operator's hands and the intervention itself, defeating the point of filming it at all.
• A study that documents the wrong condition. Footage labeled or treated as representing actual operating conditions when it was, in substance, closer to an idealized or at-rest condition — for example, a slow, careful, single rehearsal pass standing in for what a real, time-pressured production intervention actually looks like.