Cooling coils pay for dehumidification in chiller kWh; desiccant rotors mostly in reactivation heat, often above 100 °C for silica gel rotors. Compare them by listing the heat the site already has and its temperature. Very low dew points need desiccant regardless, and rotor outlet air still needs cooling.
Wind-driven rain past a plain louvre soaks the first filter stage of a fresh-air AHU. A 10,000 m³/h intake needs about 1.4 m² of louvre at 2.0 m/s, or 2.8 m² at 1.0 m/s, and the slower louvre admits less water. Add a drained plenum, a cleanable screen and an ISO Coarse first stage.
AHU quotations show price, rarely running cost. EN 13053 velocity, fan power and heat recovery classes, declared at design airflow, make offers comparable on energy. Without them the tender is judged on capital cost and the pressure drop of a small casing shows up later in fan energy.
A peak-load spreadsheet sizes equipment but cannot see the humid part-load hours, around 5,000 in a southern Viet Nam wet season, where cleanrooms overcool and reheat. An hourly model runs all 8,760 hours against the site weather file and the real control sequence, and gives an energy target its baseline.
Record the as-found reading before any adjustment. A room pressure transmitter reading 2 Pa high on a 10 Pa limit can hide months at 9 Pa. An out-of-tolerance as-found result triggers an impact assessment on every record since the last good calibration.
A filter change record needs the isolation details, how the old filter was bagged, a frame and gasket check, ΔP before and after, and the leak-test decision. A terminal HEPA that is replaced or disturbed gets an installed leak test before the room returns to use.
ISO 29463 labels encode efficiency: ISO 35H is at least 99.95% and ISO 45H at least 99.995%, the integral efficiencies of EN 1822 H13 and H14. ISO 40H sits between them with no EN 1822 name and passes twice as many particles at MPPS as 45H. Name the accepted equivalent in the specification.
Humiley engineers have a web-based AHU selection platform with one psychrometric engine and an ErP compliance module, so a junior and a lead get the same coil from the same design point. Reviews can then question the inputs: the governing design condition, latent load, fresh-air fraction and the RH limit.
A general note saying "all lines to slope" loses to a ceiling void. If the BIM model shows clean-utility lines level, clash detection routes everything else around them and there is no height left on site. Model real slopes, low-point drains and high-point vents, and include them in the clash check.
Water needs about 2,450 kJ/kg to evaporate, so every kg/h leaving a mopped floor or wash bay adds about 0.68 kW of latent load. It arrives after production, when the cooling coil unloads. Treat the cleaning shift as its own dehumidification design case.
An AHU design copied from Ho Chi Minh City to a site near Hanoi can miss the heating coil and the humidifier. Use ASHRAE's heating dry-bulb column to size preheat, the humidification dew point column to size the humidifier, and the dehumidification columns for the cooling coil.
EN 1822 classifies a filter at its rated airflow. Above it, pressure drop rises roughly in proportion to media velocity, diffusion capture falls and the most penetrating particle size shifts smaller. A filter classed H14 at nominal flow may not reach H14 efficiency at 120% of it.