Home »

Air Handler Failure

Air Handler Failure
Root Cause Investigation
Community Engineering Services, PLLC

 

(click to enlarge photographs) 

A product liability claimant retained Dan Robles to investigate the root cause of an air handler failure.  The failure actuated the smoke alarm system and the affected rooms sustained some smoke related damage.  No fire was recorded but extensive damage was evident in the air handler unit indicating a clear danger to occupants was present.

Air Handler Unit Fire:

DSCN1376DSCN1349

Operator report:

Initially, warm air was reported warm air entering the occupied area, but no fan pressure was evident.  The circuit breakers tripped and were reset by the occupant.  Instead of the circuit breakers actuating a second time, the space filled with smoke.

Summary:  

Upon interviewing occupant and after removal and inspection of the unit, the following conditions were found:

  1. Extensive melting of wires causing multiple short circuit conditions which likely bypassed safety systems.
  2. Fuse link and high temperature shut off circuitry specific to the heat kit were also damaged.
  3. The failed fan caused an over-heat condition that was did not cause the over-heat sensors to trip.  Instead, the wires burned on two of the 5 KW heating elements causing a short circuit that tripped one of two 60 Amp breakers.
  4. The third 5KW heating element remained in an over-heat condition slowly burning under the house.  Over-heat sensors did not trip, circuit breaker did not trip.
  5. In-house thermostat did not turn system OFF.
  6. A motor mount to the fan had broken causing it to be silently inoperable.  It appears that there was no vibration damping devices on this fan/motor assembly and over time the vibration caused a fatigue failure in the bracket.

Root Cause:

It is unacceptable that such a meltdown should occur allowing the system to become damaged to the extent shown in the following analysis.  The exterior of the unit could have easily ignited any combustible material in the vicinity.  However, The system did not in fact support flame despite possible repeated resetting of the circuit breakers.

It is more likely than not that the sequence of events leading to this condition occurred as a result of a fan mounting failure.  Upon inspection, the squirrel fan was lodged in the housing rendering he motor stalled in place.  One of the three motor mounts was broken in what appears to be a classic fatigue failure in the spring steel bracket.  Since there were only three mounts (instead of a more secure 4 mounts) the fan easily fell off center.  Further, where one would expect to see vibration isolating materials such as rubber pad or felt washers, there was a clear metal to metal transmission of vibration to the air handler housing.  Any number of harmonic conditions could have lead to the pre-mature failure of the motor mount.  As such, it is more likely than not the initial cause for the eventual destruction of the unit.

Analysis 

Broken Motor Mounts: One of three motor mounts broken at 45 degrees relative to the mounting surfaces.  Motor canted inward to housing.  Squirrel cage lodged in housing. Note absence of vibration isolating in left hand photograph.

DSCN1348DSCN1361

A failed motor mount caused the squirrel fan to fail.  No noise of a rubbing fan was evident at anytime.   At the orientation of the installation, the fan blade fixed to the siding such that no motion was possible.

Spring steel bracket likely suffered a premature fatigue failure due to manufacturing or heat treat process failure.  No damping devices were installed between the bracket and the housing.  At 1200 RPM, for example, may experience up to 1.7 million reciprocating cycles per day of operation if not isolated.

Three brackets instead of four created a condition of more catastrophic failure.  Four brackets would have provided aural notification of imminent system failure by likely causing the fan to chattering loudly.

 

Wiring Diagram (burned on edges) for heater insert show relationship of high temperature limit switches, fan relay, heating elements, circuit breakers, and thermostat.

DSCN1357

Wiring Diagram indicated that three hi-temperature limit switches should turn the unit off in the event of an over-heat condition.  The insulation on wires coming off the hi-temp sensors were burned through to exposure. Same black wires were fused together and possibly crossed current between them producing a possible electric bridge over-riding the switch.

 

Extensive burned and fused wires.  Leads to hi temp switched are fused.  Note extensive burning on relay switch possibly not allowing thermostat to turn system OFF – contributing to constant ON failure mode.  Extensive burned and fused wires and scorched bi-metal overheat sensors.  Continuity across limit switch suggests run away failure in the ON position.

 

DSCN1378

 

Conclusion:

The damage to the unit was too extensive to test for sequence of cascading failures leading to the severe overheat condition.  Interviews with the operator suggest that the circuit breakers were reset at least once, possibly more times suggesting poor judgement.  The presence of fused wires that may have bypassed safety circuitry suggests an over reliance on breaker circuits by the manufacturer.

However, the most obvious failure is the broken fan motor bracket.  Like most lower value appliances that are too bulky to transport over long distances, components are often shipped to a regional manufacturer who assembles packaged systems from a variety of components.

The assembly of the fan and housing may have occurred at any number of sources under a variety of quality control conditions.  The combination of an unbalance fan wheel, and fail prone 3-spoke bracket,  and the omission of vibration  damping devices may all have contributed to the failure of the fan assembly.

The unit was approximatly 15 years old and the bracket may have experienced over a billion cycles before failure.  Still, the possibility that a fire danger would be present is not acceptable if further design and quality improvements are made.

 

Submitted September 15, 2009

 

Daniel R. Robles, PE