Description

On 4 March 2019, an Embraer 145 (N14171) being operated by CommutAir on behalf of United Express was on a scheduled domestic passenger flight from Newark to Presque Isle and attempting to land on runway 01 at destination in day IMC and in conditions equivalent to ILS minima. The first approach ended in a go around and the second ended in an off-runway crash landing which resulted in minor injuries to one of the pilots and two of the 28 passengers but also caused sufficient damage to the aircraft that it was declared a hull loss.

The subsequently written off aircraft in its final resting position. [Reproduced from the Official Report]

Investigation

An Investigation was carried out by the US National Transportation Safety Board (NTSB). Relevant data was recovered from both the FDR and CVR.

The Flight Crew

It was found that the 40 year-old Captain had a total of 5,655 hours flying experience which included 1,044 hours on type. She had first worked for CommutAir between 2013 and 2015 as a DHC8 First Officer and, after leaving to work for another operator for six months, had then rejoined CommutAir as a DHC8 First Officer. According to the Vice President of Flight Operations, a few months after this, she had “received a disciplinary letter from the company and agreed to forgo an upgrade to Captain and be monitored for 9 months”. A DHC8 command then followed but during training for an Embraer 145 Type rating in September 2017, she was twice placed by the Operator under “increased scrutiny” due to training failures, including a failed Proficiency Check and received a FAA ‘Notice of Disapproval’ “due to difficulties in performing steep turns and an engine failure takeoff” during this Check. The following month, she received an Embraer 145 type rating and was released on type as Captain shortly afterwards. Remedial training to address her check/training difficulties during type conversion did not occur until March 2018. The Investigation took the view that “her repeated training problems indicated an inadequate foundation for being a Captain, which CommutAir did not effectively address”. The accident flight was the Captain’s first after two days off and she reported that she had “slept well” and more generally “had no sleep disorders or issues”.

The 51 year-old First Officer had a total of 4,909 hours flying experience which included 470 hours on type obtained since he began working for CommutAir in May 2018. He had been off sick for a few days prior to the accident but had recovered and had commuted from his home in Palm Beach the previous day arriving at a hotel near Newark airport “about 0000 on March 4” and subsequently reported sleeping “for about five hours” rather than his normal 7 to 8 hours but was “still coughing” and had consequently used a Continuous Positive Airway Pressure (CPAP) machine which he had obtained after being diagnosed with “moderate obstructive sleep apnoea after a sleep study in 2012”. The Investigation concluded the First Officer “was likely fatigued on the day of the accident”. 

It was noted that First Officer’s most recent previous flight to Presque Isle had been five days earlier (27 Feb) and during the pre flight briefing prior to departure from Newark on the accident flight, he had mentioned that during the previous flight for which he had been PM, the runway 01 ILS LOC had been offset from the runway centreline when flying a visual approach. 

What Happened   

The flight north with the First Officer acting as PF was uneventful and Boston Centre cleared the flight for an ILS approach to runway 01 at Presque Isle before terminating radar service and instructing the flight  to call on the destination common traffic advisory frequency (CTAF) which the Captain then did. The applicable ILS minima were a DA equivalent to 200 feet agl and a minimum visibility requirement equivalent to 800 metres.

The Aerodrome Automated Weather Observing System (AWOS) gave a report timed at eleven minutes before the accident which included a surface wind of 060º/4 knots, visibility equivalent to 800 metres in moderate snow and freezing fog with a temperature of -3ºC and dew point of -4ºC but the remarks section included an indicated precipitation rate for the previous hour of “trace”. An observation fourteen minutes after the accident gave the visibility as equivalent to 1,200 metres in light snow with cloud SCT (scattered)/800 feet and OVC (overcast)/1300 feet with a temperature -1ºC and dew point -4ºC with precipitation during the previous hour still recorded as “trace”. 

A call from the aerodrome maintenance foreman then confirmed that the 2265 metre-long runway was clear of snow removal vehicles. The Captain made the 1,000 foot call and confirmed that the aircraft was stabilised and that she had “ground contact”. She then asked the First Officer if he wanted the airport lights turned on (these were crew-controlled) and received an affirmative response.

After the Captain’s 300 feet call, the First Officer stated that he was disconnecting the AP and the automatic calls “approaching minimums” and “minimums” followed procedural annunciations from the enhanced ground proximity warning system. About ten seconds later, the Captain was recorded saying “runway in sight, see it?” to which the First Officer responded with “yeah.....well I got somethin’ [that] looks like a runway up there” and a few seconds after that, the automatic ‘100’ feet agl call occurred. About 3½ seconds after this, the Captain stated “watch your speed” which was followed by “sounds similar to the stick shaker” and after a further 5½ seconds, the Captain called “go missed” twice and the First Officer complied. FDR data recorded a minimum of 169 feet agl before the climb commenced.

During interview, the Captain stated that during this first approach she had seen the approach lights but had also seen a tower that looked “very close” to the aircraft at the same time. The First Officer stated that when he transitioned from the instruments to looking outside, all he saw apart from an “antenna” was “white on white” and he could not see the runway. 

The Captain notified Boston Centre of the go around and of their intention to make another approach and the flight was instructed to climb to and maintain 3,200 feet. She then called the aerodrome maintenance controller and advised him that they intended to make a second approach and would advise doing so. The First Officer agreed with the Captain that if the second approach was not successful, they would divert to Burlington International, the nominated alternate.

Boston Centre subsequently commenced provision of radar vectors to intercept the runway 01 ILS and when this was communicated to the maintenance foreman with about 16 miles to run, he confirmed that the lights had “again been activated to high intensity”. When the flight was about 8 miles south of the Locator Outer Marker (LOM) Boston cleared the flight for the approach and on reporting established on the LOC, the flight was instructed to change to the CTAF.

The approach was stable and continued with the Captain reminding the First Officer to “keep the autopilot on until 200 feet agl”. At 200 feet agl, she told the First Officer to “get the autopilot off” and this occurred about six seconds later. Two seconds after manual flight commenced, the automated “approaching minimums” call occurred followed seven seconds later by the automated “minimums” call. In between these two automated callouts, the captain called “100 feet” and about 2 seconds after the “minimums” annunciation, she called “runway in sight twelve o’clock”. During post accident interview, the Captain stated that both LOC and GS needles had been centred at that time and it was also noted that there had been no discussion about any deviation during the approach. 

Twenty seconds after the Captain’s ‘200 feet’ call, the First Officer was recorded saying “I’m staying on the flight director ‘cause I don’t see it yet” to which the Captain said “stay in” several times over a five second period. The First Officer subsequently stated that he had not seen the antenna structure during the second approach “because the aircraft had flown over it by the time that he looked outside” and the Captain stated that she had seen it again but that the aircraft had “levelled off to clear the structure before continuing to descend”. 

Descent continued without the runway in sight and a couple of seconds before ground contact was registered, the First Officer stated “I don’t know what I’m seein’”. Ground contact followed with a maximum vertical acceleration of 3.35g recorded and after twenty seconds deceleration on the ground during which the Captain recalled that the aircraft had been “bouncing up and down a few times”, it came to a stop. The cabin crew stated that the landing was “rough and violent” with seat cushions and passenger belongings falling into the cabin aisle.

The final stopping position was within the snow-covered grassy area between the intended landing runway and the parallel taxiway 93 metres to the right of the runway centreline approximately 1,100 metres past the runway threshold. The aerodrome maintenance foreman estimated that at the time of the accident, around 20% to 25% of the runway surface, which had been subject to regular snow clearance prior to the approaches, would have been visible.

The Captain instructed the passengers to remain seated and the engines were shut down. It was apparent to both the Captain and the Cabin Crew that an immediate evacuation was not necessary and it subsequently occurred by ladder with the aerodrome RFFS in attendance. A snowplough made a walkway so that those on board would not have to walk through the snow after exiting the aircraft and all were subsequently then taken to the terminal by bus.

The extensive direct and indirect impact damage to the aircraft was fully recorded in the Investigation Report.

The left main landing gear lodged between left engine nacelle and the aircraft fuselage. [Reproduced from the Official Report]

Why It Happened

All the available evidence pointed to both approaches having followed the displayed ILS flight path and having been stable until the ILS DA had been reached. However, both approaches were continued below the applicable DA without the mandatory acquisition of sufficient visual reference to do so. The Investigation attributed this intentional continuation of the accident approach below minima to Confirmation Bias on the part of the First Officer. It was also considered that the First Officer’s two to three hour “sleep debt” prior to reporting for the accident flight and the fact that he was still coughing and had not made recent his CPAP would have induced a level of fatigue which was likely to have exacerbated the cognitive bias that he experienced during the flight.

In respect of command, it was considered that the Captain had “demonstrated poor judgment and decision-making when she instructed the First Officer to stay on the flight instruments as the aircraft descended below DA” on the accident approach and that by the time he had transitioned to looking outside the aircraft, there was not enough time to determine the aircraft position in relation to the runway. The evidence available beyond that directly about the accident flight was considered to have “revealed deficiencies regarding her piloting abilities” whilst employed by CommutAir and “her repeated training problems indicated an inadequate foundation for being a Captain, which CommutAir did not effectively address”.

However, in respect of the tracking of the aircraft on the ILS, FAA ADS-B data showed that the airplane had been to the right of the runway 01 centreline during both approaches and FDR data showed that the localiser and glideslope needles were “mostly centred with only small deviations consistent with normal piloting” in both cases. 

It was found that two days prior to the accident, another CommutAir flight making a night approach to the same runway had observed at 2500 feet on finals, their ILS LOC displays showed that they were on the runway extended centreline but visually they were actually to the right of it. Having corrected their track, they then noted “fly right” LOC indications. After landing the crew reported this localiser misalignment (an offset to the right of course all the way to the runway) to the Boston Centre controller who stated that he would “forward that information onward” and did so. However, no action subsequently followed. The applicable FAA Order stated that in the case of a reported navigational aid malfunction, a report from a second aircraft should be requested and if this confirms the malfunction “or in the absence of a second aircraft report”, activate the standby equipment or request the monitor facility to activate. The flight which reported the fault was the last scheduled IFR flight into Presque Isle before the accident flight and the responsible “National Airspace System Operations Manager” stated that he had assumed that the lack of a second pilot report was an indication that the reported localiser misalignment was no longer a problem and had therefore not notified the organisation that monitored the status of the ILS equipment involved. 

It was also found that as well as the accident flight, during the five days prior to it five other CommutAir pilots had “encountered issues with the ILS LOC at Presque Isle but none of them had submitted an ASAP". The CommutAir ‘Managing Director of Safety’ stated that “he did not know why the reports were not filed”. It was concluded that CommutAir had “missed an opportunity to make this information available to relevant company flight crew and employ strategies to mitigate any potential threat that the misalignment posed”.

Three days after the accident, the FAA conducted a flight inspection of the runway 01 ILS and confirmed that the LOC signal was out of tolerance by about 200 feet to the right of course. The depth of accumulated snow around the LOC signal transmitter was estimated at between 2 feet (60 cm) and 5 feet (1.5 metres) in the area of the LOC signal transmitter. After this snow had been removed, another FAA inspection six days later found that both LOC and GS signals were within the required tolerances and the ILS was returned to service the following day.

Airport operations personnel at Presque Isle “stated that problems with the localiser signal had occurred during previous winters”. Noting that the aerodrome is not able to monitor ILS signal integrity, they would only learn about localiser signal problems from FAA technical operations personnel conducting checks either routinely or in response to a reported problem. They noted (only) “after a failed check”, airport operations personnel would remove snow until the problem was resolved.

It was found that although the applicable FAA AC 150/5200-30D, ‘Airport Field Condition Assessments and Winter Operations Safety’ dated July 29, 2016 “stated that any snow or ice that affects the signal of an electronic navigational aid should be removed, the ILS guidance discussed snow clearance areas only for the glideslope”.

Having confirmed that the ILS LOC signal had been erroneously transmitting a track to the right of the runway extended centreline rather than along it as promulgated, it was found that the fact that the aircraft had tracked this erroneous LOC meant that the ‘antenna’ seen by the pilots would have been a 30 feet high Automatic Weather Observing Station (AWOS) mast carrying wind-sensing equipment. It was additionally concluded that damage found to the lightning detector on top of this mast was the result of contact with the accident aircraft as it flew over the mast during the second approach when, unlike the earlier approach, it had only been seen by the Captain.

The Probable Cause of the accident was determined as "the flight crew’s decision, due to confirmation bias, to continue the descent below the decision altitude when the runway had not been positively identified”. 

Two Contributory Factors to the accident were identified as:

1. the First Officer’s fatigue, which exacerbated his confirmation bias,
2. the failure of CommutAir pilots who had previously observed the localiser misalignment to report it to their company and to air traffic control before the accident.  

Safety Action taken by the FAA as a result of the accident was noted as including the issue of a revised AC 150/5200-30D which incorporated additional guidance for airport operators about snow removal around navigational aids. This specifically added advice in respect of the previously ignored ILS LOC to the effect that “the accumulation of large amounts of snow can change the surface area in front of the Localizer and consequentially may affect its radiated signal”. It also introduced a snow depth of 2 feet (60cm) as the trigger for an FAA ILS system specialist to start observing the condition of the LOC signal and a requirement to issue a NOTAM “when a determination is made that snow or ice accumulations jeopardise signal strength from the LOC or GS transmitters”.

The Final Report was published on 12 July 2022. No Safety Recommendations were made. 

Related Articles

• Controlled Flight Into Terrain (CFIT)
• Precision Approach
• Instrument Approach Procedure (IAP)
• Aerodrome Operating Minima (AOM)
• Confirmation Bias
• Visual References
• Fatigue