A sterile suite running for years reaches its requalification with four signals on the table: short, repeated differential alarms on one airlock, always in the same time band; a fan close to the top of its control range; a previous requalification passed with the airflow of one grade B branch at the low end of the criterion; two pieces of equipment added in the filling room under a change control closed without environmental testing.
None of these, on its own, opens a deviation. Together they describe a facility drifting from the configuration on which fitness for use was demonstrated. The question is not whether the system runs, but whether the tests on file still represent today's facility. The wider picture is in the Cleanrooms & HVAC Systems hub.
Regulatory framework: what actually constrains requalification
| Source | Status / date | What it actually constrains |
|---|---|---|
| EudraLex Vol. 4, Annex 1 | C(2022) 5938 final, applicable since 25 August 2023 | §4.32: minimum content and maximum intervals of periodic requalification, plus requalification after corrective actions and after changes. §4.24 separates qualification from monitoring |
| EudraLex Vol. 4, Annex 15 | 2015 revision, operational since 1 October 2015; concept paper of 9 February 2026 not adopted | §4.1–4.2: appropriate frequency, with periodicity and criteria justified; no figure in the text. §11: change control |
| EudraLex Vol. 4, Chapter 5 | Operational since 1 March 2015 | §5.21 technical measures against cross-contamination; §5.22 periodic review of their effectiveness |
| ICH Q9(R1) | Step 4 on 18 January 2023 | QRM methodology with no HVAC parameters; Annex II.4 applies it to facilities and utilities |
| ISO 14644-2 and -3 | Voluntary, 2015 and 2019 editions | Periodic reclassification and risk-based monitoring; test methods, not acceptance criteria |
The §4.32 minimum and what it does not contain
Periodic requalification under Annex 1 §4.32 includes, as a minimum, particle classification, integrity testing of the final filters, airflow measurement, verification of the pressure differential and, depending on the grade, air velocity testing. The maximum interval is 6 months for grades A and B and 12 months for grades C and D [REQUIREMENT]. Requalification is also due after corrective actions and after changes to equipment, premises or processes.
The §4.32 figures are a minimum and a time ceiling, not a recommended frequency: Annex 15 §4.1–4.2 requires periodicity and criteria to be justified [REQUIREMENT], and an area with a history of deviations or frequent changes may warrant a tighter interval. The justification is written in advance, not reconstructed afterwards.
The clean up test appears neither in the §4.32 list nor among the §4.25 qualification tests: it lives in §4.29(iii), where the return to at rest limits after a disruption must be achieved within a clean up period determined during qualification, with a guidance value of less than 20 minutes [GUIDANCE]. Including it in requalification is a site decision on a risk basis, not an obligation. Airflow visualisation, microbial testing, temperature and humidity remain in §4.25 and enter the scope when a change or a trend makes them relevant; the methods are in cleanroom qualification under ISO 14644.
Three obligations stay distinct. Periodic reclassification under ISO 14644-2 comes from a voluntary standard, has its own intervals and covers total particles only: it does not close §4.32. Routine monitoring under section 9 feeds the trends, but does not demonstrate the qualified state.
From routine data to the signal that opens an investigation
The trend that decides is not the average of the parameter but the distribution of events. Three data sets must be read on the same timeline: alarms by count, duration, time of day and room; deviations by cause and recurrence; maintenance by unplanned interventions. A component appearing in all three needs a change, not another calibration. For the comparison to hold, thresholds must be stable and traceable in the audit trail, as in BMS, EMS and environmental controls.
Diagnostic table: symptom, causes, deciding data
| Observed symptom | Plausible causes to rule out, in order | Deciding data | Typical misreading |
|---|---|---|---|
| Cascade not holding: an airlock differential collapses in recurring time bands | 1) door usage; 2) upstream make-up and extract balance changed; 3) damper drift; 4) envelope leakage; 5) transmitter | Continuous differential overlaid with door status and damper position, with the terminal airflow | Rebalancing the dampers: it offsets an upstream airflow loss downstream and moves the problem |
| Recurring short differential alarms in one room only | 1) threshold and delay inconsistent with real door dynamics; 2) unit or mode changeover transient; 3) instrument exposed to a jet or a door; 4) control loop instability | High-resolution trace of the alarmed events only; threshold, hysteresis and delay checked in the audit trail | Widening the alarm band: that is a change to the control system, not a calibration |
| Terminal filter pressure drop higher than the as-left value | 1) expected progressive loading; 2) saturated prefilters or changed airflow; 3) recirculation altered after an intervention; 4) pressure tapping drift | The reading taken with the terminal airflow and the history of that point, against the manufacturer's specification | Applying rules of thumb tied to a multiple of the initial value, which have no source; or reading pressure drop as proof of integrity, which only the leak test provides |
| Relative humidity drift concentrated in one season | 1) outdoor conditions beyond design data; 2) internal load changed; 3) outside air or recirculation altered; 4) sensor out of calibration or unrepresentative | Recorded outdoor condition against URS design data, with actuator demand: if saturated it is capacity, otherwise control or instrument | Calling a design capacity limit an instrument drift, or launching a retrofit without checking the sensor |
| Particle counts out of trend at one location only, others stable | 1) real activity at that point; 2) sampling technique and tubing; 3) background counts; 4) local source or pass-box seal; 5) airflow pattern changed after a layout modification | Repeat sampling at the same point with another instrument, with and without the suspected activity, against the current visualisation study | Closing it as measurement error without evidence, or turning a single event into a reclassification of the whole room |
| Clean up period worse than the value determined during qualification | 1) reduced airflow along the filter-damper-fan chain; 2) envelope leakage; 3) new equipment shielding supply or return; 4) test method not replicated | Measured airflow against the TAB as-left figure and the qualification value, with the original protocol | Diagnosing system degradation by comparing tests run at different points, aerosols or occupancy states |
| BMS reading does not match the field reference instrument | 1) different installation position; 2) configured scale and offsets; 3) calibration of both instruments; 4) signal damping | Loop check at several points of the scale, certificates valid on the test date, scale, offset and damping checked in the audit trail | Applying a software offset to align the readings: it cures the symptom, alters the data and bypasses change control |
The rule shared by all seven rows: rule out first what can be measured without touching the system. Instrument, configuration and method come before the component; the component comes before the design change.
Capacity retrofit and compliance retrofit
A capacity retrofit is about doing more: more airflow, more dehumidification, more transfers, more redundancy. It comes from a change in demand, not from a defect, and its success criterion is performance-based and belongs in the URS. The typical case is doors: rising interlock overrides signal an undersized material flow, not a discipline issue.
A compliance retrofit closes a gap between what the system does and what it must do: a cascade that does not hold in all operating conditions — between adjacent rooms of different grades Annex 1 §4.14 states a minimum of 10 Pa, explicitly qualified as a guidance value [GUIDANCE] — an airflow pattern that does not satisfy §4.15, an airlock without the separation of PAL and MAL required by §4.12. Here the criterion is demonstrated compliance and the timeline is driven by product risk. Confusing the two funds an expansion while a defect stays open. The options are in HVAC architecture for GMP cleanrooms.
Brownfield constraints are project inputs
In an existing facility, constraints are surveyed on site and written into the URS as requirements: technical space, clear ceiling void height, condition of the ductwork to be reused, spare capacity of plant rooms and switchboards, shutdown windows realistically obtainable, construction routes that do not cross classified areas in use. A constraint not surveyed comes back as a site variation, when the cost of change is highest. The discrepancy between as-builts and the field is itself a project input.
Change control and return to the qualified state
The sequence is fixed: impact assessment under Annex 15 §11, verification scope defined before execution, engineering verification after the work, requalification of the affected parameters, update of as-builts, procedures and CCS. Two temptations recur: declaring a change non-impacting to avoid testing, or repeating the whole qualification rather than arguing the scope.
| Change scenario | Question that sets the scope | Minimum expected verification | Where it is recorded |
|---|---|---|---|
| Terminal filter replaced with equivalent specification | Does air distribution change? | Integrity test of the installed filter, room airflow and differential | Change control with test results |
| Critical sensor replaced or set point changed | Do position, measuring chain or control arrangement change? | Loop check, traceable calibration, differentials of adjacent rooms | Change control and partial requalification |
| New equipment in a classified room | Do airflow pattern and loads stay qualified? | Airflow visualisation, classification in both states, T and RH | Requalification and CCS update |
| AHU overhauled or replaced | Does it touch capacity or only the failed component? | Functional and performance testing, then §4.32 on the served areas | Turnover and requalification |
| Layout or material routes modified | Do zoning, adjacencies or flow direction change? | Airflow direction, visualisation, differentials, transfers | Requalification and CCS revision |
Engineering verification always precedes requalification: balancing and functional testing bring the system to a stable, documented configuration, and only on that configuration is the GMP test run. Cascade sizing criteria are in airflow and pressure cascades.
Decisions to be justified through QRM
ICH Q9(R1) contains no HVAC parameters but governs three decisions. Formality as a continuum (§5.1) scales investigation depth and requalification scope to product impact. Risk-based decision-making (§5.2) requires the interval chosen within the §4.32 ceilings to be justified by area criticality, data history and change frequency [QRM]. Product availability (§6.1) enters shutdown planning: deferring a compliance retrofit is a risk decision to be documented as such.
Worked example: Site Vega
Site Vega is a realistic but fictional example. In a suite with a grade B filling room, the EMS has logged short differential alarms on the material airlock for months, in the busiest shifts; deviations record the same assigned cause eleven times, prolonged door opening; the door status sensor has been replaced twice in a year.
The investigation followed the order of the table: threshold and delay consistent with the URS, loop check confirmed, terminal airflow in specification but low, overrides rising with transferred volumes, by then double the design figures. The work was classified as a capacity retrofit, not as a CAPA: a dedicated MAL for outgoing materials, with the brownfield constraint of a ceiling void that ruled out ductwork on the shortest route. The following requalification covered those rooms with §4.32 plus airflow visualisation.
Levels of prescriptiveness
| Statement | Level | Source |
|---|---|---|
| Minimum content of requalification and requalification after changes | [REQUIREMENT] | Annex 1 §4.32 |
| Maximum interval of 6 months for grades A and B, 12 months for grades C and D | [REQUIREMENT] | Annex 1 §4.32 |
| Periodicity and criteria to be justified; no figure is fixed | [REQUIREMENT] | Annex 15 §4.1–4.2 |
| Clean up period with a guidance value of less than 20 minutes | [GUIDANCE] | Annex 1 §4.29(iii) |
| Minimum 10 Pa between adjacent rooms of different grades | [GUIDANCE] | Annex 1 §4.14, guidance value |
| Periodic reclassification does not replace requalification | [STANDARD] | ISO 14644-2:2015 |
| Rule out instrument, configuration and method before the component | [GEP] | Good engineering practice |
| Actual interval, scope after a change, extent of the investigation | [QRM] | ICH Q9(R1) §5.1–5.2 |
| Separate capacity and compliance retrofits; brownfield constraints in the URS | [GUIDEGXP] | GuideGxP recommendation |
Operational checklist
- Justify in writing the requalification interval of every area.
- Check that the scope contains the full §4.32 minimum list.
- State whether the clean up test is included and on what grounds.
- Freeze thresholds and delays and trace every change in the audit trail.
- Rule out instrument, configuration and method before touching the system.
- Define the verification scope before executing the change.
- Classify every retrofit as capacity or compliance.
- Survey brownfield constraints on site and write them into the URS.
- Complete balancing and functional testing before requalification.
- Update as-builts, procedures and the CCS to close the change control.
- Take trends, changes and planned retrofits into periodic review.
Recurring mistakes and red flags
- Attributing a numeric periodicity to Annex 15.
- Presenting an ISO reclassification certificate as evidence of requalification, or setting filter action thresholds on rules of thumb with no source.
- Declaring a change non-impacting to avoid testing, then executing it in a classified room in use.
- Using CFD to close an airflow pattern anomaly: it replaces neither qualification, nor visualisation, nor field testing.
- Closing the same deviation with the same action every time: recurrence says the root cause has not been found.
To turn trends, investigations and change control into documents that hold up in inspection, The Pragmatic GMP is the GuideGxP newsletter: short analyses on qualification, HVAC and contamination control, from its dedicated page.
Key takeaways
- §4.32 sets minimum content and a time ceiling; the actual frequency is justified under Annex 15 §4.1–4.2.
- Requalification, ISO reclassification and routine monitoring are three distinct processes.
- In troubleshooting, rule out first what can be measured without touching the system.
- Capacity and compliance retrofits have different success criteria and different timelines.
- Brownfield constraints are URS requirements surveyed on site, not site variations.
