Two identical AHUs.

One drifted.

AHU1 and AHU2 serve equivalent gallery loads on the same sequence of operations. Only one of them followed it. Finding out why took a functional performance test, not a guess.

THE SITUATION

The comparison that made it obvious

AHU1 and AHU2 condition effectively identical gallery spaces on the same sequence of operations, so their behaviour should be close to indistinguishable. Compared side by side they correlate well on humidification and economising — and diverge completely on dehumidification.

AHU1 had not entered dehumidification since mid-November 2022, and never entered neutral mode either. It sat in humidification almost permanently while zones above 53 % relative humidity went unanswered.

AHU1 never dehumidified

No dehumidification mode since mid-November 2022.

Supply dewpoint would not hold

Roughly a 20 °F oscillation around setpoint.

The galleries paid for it

Every downstream VAV inherits the air handler’s instability.

THE CHALLENGE

Four faults across two air handlers

Mode logic, setpoint reset, loop tuning and hardware — four separate layers, each of which will produce the same complaint downstream.

Mode criteria never firing

AHU1 stayed in one mode
No dehumidification since mid-November 2022 and no neutral mode at all, against an explicit SOO.

Dewpoint setpoint reset

The setpoint itself was wrong
AHU1’s dewpoint setpoint violated the SOO while AHU2, on identical rules, reset correctly.

Coil loop tuning

The loops hunted, they did not track
Supply dewpoint oscillating roughly 20 °F as the cooling and heating coil PI loops chased setpoint.

Humidification hardware

The equipment could not keep up
Humidifier refill faults, static pressure below setpoint at full fan, and unstable fan control.

FAILURE DOMAIN 01

Mode criteria never firing

AHU1 stayed in one mode
The sequence is unambiguous: dehumidify when any zone exceeds 53 % relative humidity, humidify when any zone falls below 47 %, hold neutral in between. AHU1 had not entered dehumidification and had never entered neutral mode, sitting in humidification almost permanently. AHU2, serving an equivalent gallery load on identical rules, moved between all three modes correctly. Comparing the two air handlers mode by mode isolated the problem to AHU1 within a single review — they correlate closely on humidification and economising, and diverge only where AHU1 fails.

FAILURE MODES

Bottom line: the SOO never reached the code

FAILURE DOMAIN 02

Dewpoint setpoint reset

The setpoint itself was wrong

Mode is only half the sequence; the dewpoint setpoint the air handler resets to is the other half. AHU1’s dewpoint setpoint was in open violation of the sequence of operations while AHU2’s followed it. In full humidification the supply dewpoint setpoint should be 63 °F, and in dehumidification 48 °F. AHU1 was reaching neither at the right time — so even a perfectly tuned coil loop would have been tracking the wrong target.

FAILURE MODES

Bottom line: right loop, wrong target

FAILURE DOMAIN 03

Coil loop tuning

The loops hunted, they did not track

Supply air dewpoint on both AHU1 and AHU2 oscillated across roughly a 20 °F band. The cooling and heating coil PI loops that throttle to hold supply dewpoint at setpoint were not tracking it — they were hunting around it. This is a controls problem rather than a design problem, and the data proves the distinction: the supply dewpoint does reach 63 °F at times, so the capacity is present. By design the air handlers are the primary conditioning equipment and the VAVs only trim, which means an unstable supply dewpoint is inherited by every box downstream rather than corrected there.

FAILURE MODES

Bottom line:a controls problem, not design

FAILURE DOMAIN 04

Humidification hardware

The equipment could not keep up

The AHU1 humidifier carried a water-level refill fault and its actual humidity output collapsed the moment it faulted.  The AHU3 humidifier showed the same status 3 refill fault. On the air side, AHU1 could not reach its supply static pressure setpoint even with the supply fan at 100 %, as downstream VAVs opened and the setpoint reset from 0.4 to 0.7 in. w.c. AHU2’s supply fan speed was highly unstable, its static pressure swinging between zero and setpoint as downstream dampers short-cycled. AHU1’s return fan status also did not read as continuously on.

FAILURE MODES

Bottom line: hardware faults surfaced late

THE SOLUTION

The MelRok approach

Verify the air handler against its own sequence — mode, setpoint and loop — every single day.

One platform across both air handlers

Mode criteria

Dehumidification, humidification and neutral criteria are evaluated continuously against live zone relative humidity, and every mode the air handler should have entered but did not is flagged.

Psychrometric checks

The dewpoint setpoint the air handler actually resets to is checked against the value the sequence calls for — so a wrong target is caught before the loop gets blamed for it.

Loop tracking

Supply dewpoint is scored against its setpoint continuously, separating a tuning problem from a capacity problem and proving which one you actually have.

Hardware & data hygiene

Humidifier faults, static pressure shortfalls and fan instability surface directly — and fault windows are excluded from analysis so the diagnosis is not contaminated by them.

ROOT-CAUSE ISOLATION

Proving where the fault was — not guessing

AHU1 would not dehumidify.

Force the high-humidity case

Max-zone RH was overridden to 60 % at the BAS. AHU1 entered dehumidification mode immediately and correctly.

Force the low-humidity case

Min-zone RH was overridden to 40 %. AHU1 entered humidification mode immediately and correctly.

Force the in-band case

With both inputs returned inside the band, AHU1 held neutral mode — neither humidifying nor dehumidifying.

Result: the mode logic was correct. The fault was in the calculated max-zone / min-zone relative humidity point feeding it — handed to the controls contractor as a specific, testable action item rather than a symptom.

WHAT CONTINUOUS COMMISSIONING CAUGHT

Faults found in the data — not in the galleries

AHU1 out of dehumidification

Five months without entering dehumidification mode, on an air handler whose twin followed the same sequence correctly. Caught by mode-criteria comparison, not by any built-in alarm.

A 20 °F dewpoint oscillation

The coil loops were hunting rather than tracking. Because the capacity was demonstrably present, the finding could be handed over as a tuning task rather than an equipment question.

A humidifier failing during an extreme event

The AHU1 humidifier’s refill fault was caught in the water-level data — the same day ambient relative humidity fell to an extreme and the galleries dropped with it.

Each of these was visible in the air handler data well before it was visible in the galleries.

THE RESULT

Both air handlers back in sequence

The controls contractor implemented dewpoint setpoint changes. 

BEFORE

ONE

of the two gallery air handlers was resetting its dewpoint setpoint according to the sequence. AHU1 had not dehumidified.

AFTER

BOTH

air handlers resetting dewpoint setpoint correctly — AHU2 holding 48 °F in dehumidification, AHU1 sitting inside the neutral-zone range.

10 DAYS

From corrected setpoint logic to verified conformance measured in the field.

NEUTRAL MODE

Entered correctly by AHU1 for the first time in the analysis record.

NO FAULTS

Humidifiers clear across the verification week apart from one transient event.

THE REAL VALUE

What continuous commissioning replaced

Stable air to every gallery

The air handlers are the primary conditioning equipment. Correcting them corrected the input every downstream box depends on, instead of trimming symptoms zone by zone.

Energy conservation measures

An air handler locked in permanent humidification, and coil loops hunting across a 20 °F band, are both energy spent to produce an out-of-spec result.

O&M labour and outside studies

A functional performance test turned "AHU1 will not dehumidify" into a named point to fix — the difference between a work order and an investigation.

THE OBJECTIVE, DELIVERED

Two identical AHUs. Both in sequence.

Continuous, design-intent-based diagnostics and monitoring-based commissioning — engineer-backed, with high-resolution IoT data and an AI-powered cloud on one platform. Monthly commissioning reviews and on-demand reporting included.