A pressure alarm tells you the cascade has failed. A trend tells you it started failing weeks earlier.
The typical differential between adjacent grades is 10–15 Pa, and Annex 1 guidance asks for at least 10 Pa. Most monitoring systems alarm somewhere near that floor, with a delay so door openings do not set it off. That catches the failure. It says nothing about the drift.
Take a room pair commissioned at 14 Pa that now sits at 11 Pa. No alarm, no deviation. Something has changed all the same: supply filters loading and cutting airflow, a door closer going soft, a return grille half-blocked by a trolley, a damper that moved, a new penetration nobody sealed. Each of those is cheap to fix now and expensive to explain after the alarm.
A trend makes the drift visible. Plot the daily median for each room pair against its commissioned value, and review it weekly. The median ignores door-opening spikes that the alarm delay is there to filter. A steady fall of a few pascals over a month is the signal to go looking.
The monitoring SOP should say who reviews the trend, how often, and what size of drift triggers a walk-down, as well as setting the alarm limits.
Our maintenance service uses condition monitoring and IIoT trending for slow drifts like this one. If your cascade only has alarms today, talk to an engineer: contact@humiley.com or +84 28 7777 6668 (Viet Nam office).




