Warranty terms usually decide which chiller and AHU work goes to the OEM, and the split is rarely revisited. Keep compressor, refrigerant and firmware work with the OEM, bring filters, cleaning, trending and first response in-house, and file every service report in the site CMMS under change control.
At 20 °C, 30% RH means a dew point of about 1.9 °C and 45% RH about 7.7 °C. A common AHU must dry all its air to the lowest requirement and reheat the rest. Grouping rooms by dew point, or treating the dry room locally, removes that penalty.
Annex 1 (2022) expects an aseptic process simulation to imitate routine processing with worst-case conditions. An APS run right after a filter change and rebalance tests a different process. Schedule maintenance away from APS runs and record the HVAC state for each run like the interventions.
Two H14 filters from a new maker at 150 Pa initial pressure drop, fitted beside four originals at 110 Pa on one duct branch, take less air than their neighbours while the room total barely moves. Compare initial pressure drop before ordering, then change the whole set or retrim and measure every terminal.
At LOD 300 a BIM element has size, shape and location; LOD 350 adds its interfaces with other systems: connections, supports and clearances. Clash detection before LOD 350 finds ducts in beams but misses hangers, flanges and maintenance access, which then surface on site.
A room held at 45% RH ±2% is really at 47% when its sensor has drifted 2% low, and the BMS still shows 45%. Tight bands need sensor accuracy and reference accuracy tighter than the band, calibration near the setpoint, evidence-based intervals and as-found records.
ICH Q10 expects changes to be evaluated after implementation. Closing a change the day the part is installed skips that step. Define the effectiveness check when the change is approved: the data that shows the objective was met, a period long enough to include the original conditions, and the result that closes it.
After an AHU failure, QA decides on product from one report. It needs a timeline built from BMS and EMS timestamps, trend exports for every affected room, the impact on rooms and batches, and a CAPA with owners and dates. All four can be drafted before the day shift starts.
Desiccant rotors lose capacity slowly: worn sector seals let wet air cross over, a slipping drive slows the rotor, dust and solvent vapour foul the media. Measuring outlet dew point, airflow and reactivation temperature against the commissioning record shows the trend before a −60 °C dry room does.
"OK" on a FAT sheet records a judgement. "118 Pa, acceptance ≤ 150 Pa", with the gauge ID and calibration date, records a measurement that IQ, maintenance and an inspector can use years later, and shows when a pass was close. It is how the Excellence value looks on a test bench.
Three BIM model reviews at about 30, 60 and 90% of detail design work when each has the right people: QA and production for layout and flows, maintenance for services and access, QA and validation for sampling points and test ports. A comment after a review closes becomes a change request.
The pressure drop at which a supply AHU's final filters are changed should be the dirty-filter allowance the fan was selected for at design stage, checked at FAT and quoted in the maintenance task. Raise it later and room pressure drifts down inside its alarm band, visible only on the trend.