A 10,000 m³/h fan at 1,500 Pa and 0.6 overall efficiency draws about 6.9 kW, around 61 MWh a year in a cleanroom. Every 100 Pa removed at design stage saves about 0.46 kW for the life of the plant. A pressure budget per AHU in the design basis is where it starts.
Air in a ceiling void has never passed a HEPA filter. If the void sits above room pressure, leaks bring it down past light frames and HEPA collars. Keep one void at or below the lowest room under it, seal duct joints, and measure room-to-void pressure at commissioning.
Raising a 6 × 5 m room from 3.0 m to 3.5 m drops 1,800 m³/h from 20 to 17.1 air changes. The in-operation particle count barely moves, because steady-state concentration depends on generation and supply airflow, not volume. Recovery takes about 17% longer. Restoring 20 ACH needs 2,100 m³/h.
ASHRAE Guideline 14 treats a model as calibrated to monthly data when NMBE is within ±5% and CV(RMSE) within 15%. Before a model justifies a retrofit in a running plant, it should meet that bar against 12 months of metered data, including the wet season.
Write "high-efficiency filtration" and three vendors can price three different filters. "H14 to EN 1822, installed leak test to ISO 14644-3" fixes the filter by a lab test at MPPS and the installed condition by an aerosol scan in the room. The housing then needs its own line, so the scan is physically possible.
NFPA 110 Type 10 restores power within 10 seconds. Controllers, EMS and network gear reboot long before that, leaving gaps in the record when room pressures are disturbed. Size a UPS for the loads that must not blink, keep fans on the generator, and prove it with a black-start test.
ASHRAE 0.4%, 1% and 2% design conditions are exceeded about 35, 88 and 175 hours a year. For a room with a validated RH limit, choosing the percentile is choosing how many excursion hours you accept. Use the dehumidification dew point column, choose per room, and write the choice into the design basis.
A particle counter on UPS can stay powered through a power dip and count nothing if its vacuum pump sits on the normal supply. Network switches in risers, transmitters on AHU field panels and unsynchronised clocks leave the same kind of gap. Trace the supply of every item between the sensor and the historian.
A 3 m cleanroom at 100 W/m² and an 8 K supply difference needs about 12 ACH for cooling. If cleanliness sets 40 ACH, the other 28 are dilution air. Moving it through the central AHU costs fan energy; a recirculation loop or FFUs can move it at lower pressure.
A probe with a 3 cm opening moving at 5 cm/s sits over any point for 0.6 seconds; triple the speed and it is 0.2. Frame leaks give thin, diluted plumes that a fast pass misses. Calculate the scan rate for the probe and instrument, overlap the strokes, and scan the perimeter separately and slowly.
A URS line such as at least 10 Pa above a Grade C corridor sets the airflow offset, the door leakage allowance, the pressure control range, the alarm limit and the OQ acceptance criterion, when the URS and validation strategy are written in parallel with the design.
With a ring main unit, a faulted section of a 22 kV loop is isolated by switching instead of waiting for the repair. Inside: ring load-break switches, a transformer feeder and interlocked earthing switches, with a sealed utility metering cubicle alongside. The plant stays dark until the switching is done, so plan the cleanroom restart.