Skip to content

Life Sciences

How to plan a smoke study

  • Life Sciences
  • 1 min read

Smoke video shot in one forty-minute take from the doorway shows an inspector that the room had smoke in it.

It does not show whether reaching over the stopper bowl is acceptable, because nobody decided beforehand what the smoke had to do.

EU GMP Annex 1 (2022) says airflow pattern studies should be performed both at rest and in operation, for example simulating operator interventions, and that video recordings should be retained. That describes a planned study with a record at the end, planned in this order:

  1. List the locations and interventions from the contamination risk assessment: each Grade A zone, each opening between grades, each intervention by name.
  2. Define the runs: at rest, with the equipment installed but not operating and no personnel present, and in operation, with each intervention performed as the SOP describes it.
  3. Fix the camera positions and the smoke source for every shot. The tracer should follow the air: low outlet velocity, close to neutral buoyancy.
  4. Write the script and the acceptance criteria. For each shot, say what the smoke should do and what counts as a failure, such as air rising from the operator's arm toward open containers.
  5. Review the footage with QA against those criteria, record a result per shot and archive the video, indexed by location and run.

Forty short clips with a criterion each take longer to film. They can be read by someone who was not in the room.

Industries

Related

More on this subject

Pass boxes: where the grade boundary leaks

A pass box sits across a grade boundary designed for at least 10 Pa. The interlock does not fix a worn door seal, a static box is made safe by the disinfectant contact time rather than by waiting, and a dynamic box needs its supply balanced against both rooms and its HEPA filter leak tested.

InsightRead

PW and WFI: sampling points are drawn on the P&ID

EU GMP Annex 1 (2022) clause 6.13 expects sampling at all outlets and points of use, worst-case locations considered, and a representative sample of manufacturing water every day. That only works if reachable sample valves, sub-cooler outlets, the loop return and flushing drains are on the P&ID before isometrics.

InsightRead