A stairwell 10 m tall, 10 K warmer or cooler than the building around it, sets up about 4 Pa between its top and bottom.
A door near the top or bottom sees up to half of that against a 10–15 Pa cascade step, and the AHU has no part in it.
The physics is buoyancy. Warm air in a vertical shaft is lighter than the air around it, so the shaft pushes out near the top and draws in near the bottom, with a neutral plane somewhere between. The pressure is gravity times height times the difference in air density. At 10 m and 10 K the density difference is about 0.04 kg/m³, so 9.81 × 10 × 0.04 ≈ 4 Pa.
Stairwells, lift shafts and vertical service risers all behave this way. When one opens onto a clean corridor, the corridor's pressure relative to the shaft changes with the floor, with the weather and with whether the stair door is open. On a lower floor the shaft may draw air out of the corridor; higher up it may push into it. A cascade balanced on a mild afternoon reads a few pascals off another day, and nobody can explain the drift.
Three things help: an airlock or lobby between the shaft and the clean corridor, a shaft that is itself conditioned or pressurised, and pressure sensors referenced to a space that does not move with the shaft.
Humiley designs pressure cascades for multi-storey plants with the shafts drawn in, since the AHU controls cannot see them.
Does your cascade drift with the weather or with the time of day?



