A VAV Building Running 24/7: Overrides, No Static Reset, and a Failed Airflow Sensor

Fan power: 14 AHUs x 18.5 kW = 259 kW at 60 Hz. Excess hours: 168 h/wk actual vs 60 h/wk scheduled = 108 h/wk x 52 = 5,616 h/yr. Excess energy: 259 x 5,616 = 1,454,544 kWh/yr ≈ 1.45 GWh Cost: x $0.09 = $131,000/yr, plus 259 kW of demand at $12/kW-mo = $3,100/month.

Assumptions and the honest caveat: this is the upper bound - it assumes every AHU could stop. Some surely cannot (24/7 labs, server rooms, a night-operations floor, telecom closets). Those must be identified and scheduled onto their own program rather than lumped in.

Recoverable portion: with the building unoccupied and all boxes at 30% minimum, a static-pressure reset would let the fans fall to roughly 35–40% speed. Fan power follows the cube law: 0.375³ = 5.3% of full power, i.e. about 1 kW per AHU instead of 18.5 kW. Realistically, expect to recover 70–85% of the excess - on the order of 1.0–1.2 GWh/yr and $95,000–$115,000/yr, plus most of the demand. Even at 50% recovery this is the largest single item in the building.

Failure mode: a failed airflow sensor driving the box into a limit cycle.

The sequence repeats every few minutes:

1. The airflow station (hot-wire element fouled, pitot tubes plugged or disconnected, or the transducer drifted) reports 0 cfm.
2. The box controller believes it is far below its 30% minimum, so it drives the damper fully open trying to make flow.
3. The zone overcools.
4. Reheat stages fire 100% to hold setpoint; the zone warms; the damper modulates back; airflow still reads 0; the loop restarts.

The space stays 'comfortable' precisely because the reheat is masking the failure - which is why nobody reported it. Meanwhile the box is cooling and heating the same air simultaneously, and the fully open damper on two boxes is helping hold the 1.5 in. w.c. setpoint high for the whole AHU.

Verify:

• Take a handheld balometer reading at the box inlet diffuser and at the box. If actual flow is ~30% minimum while the BAS reads 0, the sensor is at fault, not the damper.
• Compare damper command vs. feedback; check the tubing for kinks/disconnection and the transducer output at zero flow.
• Pull the 4-minute trend of damper position, airflow, zone temp, and reheat stage - the square-wave reheat pattern confirms the loop.

Correct:

1. Clean/recalibrate or replace the airflow station; re-zero the transducer.
2. Configure a sensor-fail fallback in the box program (fixed damper position or BMS-commanded flow) so a dead sensor parks the box instead of driving it wide open.
3. Add an airflow-fault alarm (reported flow < 10% of minimum while the damper is > 80% open for 10 minutes).
4. Add a reheat/cooling conflict alarm (reheat energized while the zone is below cooling setpoint by more than 2°F).
5. Sweep the whole building for the same pattern - airflow sensors fail in batches when they come from the same install crew and vintage.

Diagnosis: a single badge reader is starting the entire AHU group - all 14 air handlers, at full static pressure - for the rest of the night. That is almost certainly a janitorial or delivery badge used repeatedly, with no expiry and no zone-level granularity. It is the direct cause of the 24/7 operation quantified in Task 1.

Specify these changes:

1. Scope the override to the zone served, not the AHU group. A loading-dock badge should condition the dock and adjacent corridor, not the 12th floor.
2. Time-limit every override: 1 hour default, 2 hours maximum, renewable only by a second badge swipe inside the active window.
3. Tie overrides to occupancy confirmation: require a badge or occupancy-sensor event in the requested zone before the override is honored, and auto-cancel if the zone goes vacant for 20 minutes.
4. Replace fixed 6 a.m. start with optimal start (learn the warm-up rate, start just early enough to hit setpoint at occupancy) - this also removes the temptation to leave units on overnight 'so the building is ready.'
5. Add static-pressure reset during occupied hours (reset from the most-open VAV damper), and a much lower unoccupied setpoint or fan-off at night.
6. Exception reporting: an automated weekly report to the chief engineer listing every override by badge, zone, duration and outside normal hours. Visibility alone usually cuts this by half in a month.

Trend to prove it:

Point	Frequency	Purpose
AHU supply-fan status and command (start/stop)	15 min	Prove scheduled shutdown actually occurs
Override events with badge ID, zone, duration	Event log	Attribute and police the residual overrides
Fan VFD speed and kW per AHU	15 min	Convert schedule compliance into kWh and $
Duct static pressure and reset setpoint	15 min	Prove the reset is active and tracking
Most-open VAV damper position per AHU	15 min	Confirms the reset is not starving a box

Measurement & verification tie-in: use these trends as the IPMVP Option B evidence - pre/post 15-minute interval kW on the fan VFDs, normalized by occupied hours and CDD, so the savings claim survives a mild winter and a tenant change.