A chilled-water reset driven by valve position alone can raise supply water by 2 K, save chiller energy in the dry season and leave dehumidifying coils unable to reach the dew point in the wet season. Lock the reset out on outdoor or supply-air dew point, retest it in the wet season and list the dehumidifying coils first.
ISO 14644-3 lists eleven supporting cleanroom tests and gives a method for each, from air pressure difference to segregation. It says they may not all be required and that tests should be selected in a manner agreed between customer and supplier. The test specification has to name the tests, the occupancy state and the acceptance criterion before the work is priced.
New chilled-water pipework carries mill scale, weld slag and swarf that settle in control valve seats and coil tubes. Bypass coils, valves and chillers, flush the mains, circulate the cleaning chemical, rinse to the agreed limits and dose inhibitor before the coils are opened. Then keep the water circulating until handover.
An airflow diagram shows every room's supply, return and exhaust on one sheet. For each room, supply minus return minus exhaust must leave a leakage that matches the pressure cascade, and the room flows must add up at each AHU. Five checks at the desk find errors that otherwise appear during balancing.
A variation is a change to the terms of a marketing authorisation. In the EU system, Type IA changes are made and then notified, Type IB changes are notified before implementation, and Type II changes need approval first. The class follows the change rather than its cost, so ask regulatory affairs before the purchase order.
A lead engineer's review time should go to engineering questions. Before sending a drawing, check the title block and revision against the register, name the design basis documents, match every tag and unit to the schedules, attach the calculation for each sized item, and do your own red-line pass on a printed copy.
A room pressure display showing 12.3 Pa has a resolution of 0.1 Pa. The transmitter behind it may be accurate to ±1% of a 100 Pa span, which is ±1 Pa, before drift. The last digit shows movement well and absolute value poorly. The calibration certificate, with its stated uncertainty, tells you which digit can carry a decision.
While a door stands open the room cannot hold pressure, so the controller's integral term keeps growing with the damper already at its limit. When the door closes, the controller must unwind that sum before the damper moves, and the room overshoots. Anti-windup, output limits or a door-contact hold in the controller cure it.
Two swapped sensor cables pass the checks made from the desk. A point-to-point loop check finds them: identify the device by tag, follow the wiring to its I/O channel, apply a known input at the device, confirm value, unit and tag on the graphic, trend and alarm, then sign the loop sheet.
A mechanical engineer can catch electrical problems on the single-line diagram: a fault level at each board, breaker and busbar ratings at or above it, generator-backed loads on their own section, harmonic loads and filters marked, earthing and metering shown. The short-circuit study still does the proving.
Most calibration certificates are filed unread. Check that the laboratory is accredited for this measurement, that tag, serial number and calibrated points match the instrument and its range, read as-found before as-left, compare the uncertainty with the tolerance, and look for traceability and the decision rule behind any pass statement.
A capture hood throttles the terminal it measures and assumes an even flow across its sensing grid, so on a swirl diffuser it can read well off the true airflow, in either direction. Name the method per terminal type in the protocol, use back-pressure compensation, and check the hood against a duct traverse.