In the early morning hours of the 5th of May 2017, Air Cargo Carriers flight 1260 crashed at Yeager Airport in Charleston, West Virginia. There were two souls on board, the two crew members, who were both killed in the crash. The NTSB took over two years to release their final report.

The aircraft was a Short SD3-30 Variant 200: a turboprop transport aircraft designed as an inexpensive and rugged aircraft capable of short-field landings. The unpressurised aircraft was designed to seat 30 passengers for short-range regional flights. Air Cargo Carriers operate throughout the US, Canada and the Caribbean; in 2017, their fleet consisted of 26 Short aircraft after the crash (twenty-five 360s and one 330), making them the primary users of Short aircraft. All of the aircraft have been converted to be used strictly for cargo. The cockpit voice recorder was removed from the 330 when it was converted for cargo use. The aircraft was in good shape and the documentation showed that it was within weight and balance limits.

The two members of the flight crew had very different levels of experience.

The captain had worked as a bush pilot in Alaska before coming to Air Cargo Carriers in 2015. He held an airline transport pilot certificate and had over 4,300 hours flight experience, with 3,700 of them as pilot-in-command.

This was the first officer’s first professional pilot job. She’d always loved flying and had previously worked as a flight attendant at another airline, where she always complained that the flights were too short. She took her ATP written exam before she even started flying, determined to make her dream come true. She was four weeks out from her check ride for CFI training when she received an offer from Air Cargo Carriers and had to make a decision whether to take the job as a cargo pilot or finish her training to look for a job as a flight instructor. She longed to fly a big airplane, said her brother, and she took the job at Air Cargo Carriers. She started her ground school and training at Air Cargo Carriers in September 2016 and built up 650 hours, with 214 total hours as pilot-in-command. Her flight time in the Short 330 were all as second-in-command.

Although she had some trouble adjusting to a schedule of night flights, she loved the new job. She told her brothers and her boyfriend how excited she was and they said that just she seemed happy to be flying. But there were some signs that her dream job wasn’t quite what she expected to be.

Her brother recalled some of her stories:

During the flight, [the captain] had pointed to something on the ground that she couldn’t see. He took control of the aircraft from her and dropped 1000 feet. He recovered it again, but the tower had asked about the altitude. She said the captain had told him it was turbulence.

She also mentioned … that there were times when he would say things like “I flew in Alaska, I got this” and she would look over and see a mountain “right there.”

Both flight crew were based at Yeager Airport in Charleston, West Virginia (CRW) and had flown together over the past six months; at Air Cargo Carriers, pilots were paired permanently rather than rostered; a captain and first officer were considered to be partners.

Air Cargo Carriers did not have a specific training course for CRM (Cockpit Resource Management); the chief pilot said it was a continuing process and reviewed throughout. Pilots are given an end of indoctrination “week 1” test which is a pass/fail exam. Failing this exam means removal from the training. They do not keep a copy of the individual results.

The director of training explained their training on CRM in an email.

I can’t say that I always use the same examples, but as we go through the PowerPoint presentation on CRM I will give personal stories about situations that have come up and how I used appropriate CRM to deal with those situations. Some stories are purely hypothetical, but I try to include as many real scenarios as possible, whether they are mine or someone has told me of a time when they used good CRM. Some stories have poor CRM practices and we discuss how things could have been handled differently and how the outcome may have changed if they were. Some stories have very good CRM and we discuss how things might have worked out differently if it weren’t for the proper use of CRM.

The chief pilot confirmed in the same email exchange that the only electronic device approved for use in the cockpit is the IPAD EFB (Electronic Flight Bag) and that this regulation was also a part of the training.

In January 2017, four months before the accident, the first officer had the following conversation with a friend via SMS.

The flight crew met up at Louisville International Airport at just past midnight the day before the flight. They departed at 04:50 and arrived at Charleston’s Yeager Airport at around 6:30 in the morning. The captain used a CPAP (a breathing device to combat sleep apnea) which showed to have been in use from 07:12 to 15:11. His girlfriend remembered that he got up and cooked dinner for them and started getting ready for work around 21:00. The first officer’s morning schedule isn’t known but she sent a text message around 13:30 and spoke to her brother on the phone for some time that afternoon. She cooked dinner and then spoke to her brother again before leaving to meet the captain at the airport.

They flew to Louisville, arriving just before midnight, ready for their scheduled cargo flight from Louisville back to Charleston.

It was 05:41 on the 5th of may when they departed Louisville on an instrument flight rules (IFR) flight plan and climbed to their cruising altitude of 9,000 feet. They picked up the automatic terminal information service (ATIS) information November with the 05:54 weather for Charleston: wind from 080° at 11 knots, 10 miles visibility, scattered clouds at 700 ft above ground level (agl), and a broken ceiling at 1,300 ft agl.

However, just before 06:30, the broken cloud ceiling dropped from 1,300 feet above ground level to 500 feet. The weather observer edited the special weather observation to note valley fog, a few clouds at 100 feet and an overcast layer at 500 feet. These observations are automatically sent to ATC however the tower controller, who expected the weather to change again in a short time, decided that it wasn’t worth updating the ATIS for this special observation. He finished his shift around 15 minutes later and didn’t brief the arriving tower controller that the ATIS wasn’t current. At 06:54, directly after the accident, the weather was recorded as wind from 230° at 3 knots with ten miles visibility, few clouds at 100 feet above ground level and an overcast ceiling at 500 feet above ground level.

A new ATIS recording should be made for any new official weather, whether or not there is any change in values. Air traffic controllers are expected to inform flight crew if either the ceiling or visible is below basic minima if that information was not included in the ATIS. But neither of these things happened.

The first officer contacted Charleston approach at 06:37. The controller told them to expect localiser approach to runway 5, which has a minimum descent altitude (MDA) of 1,320 feet above mean sea level or 373 feet above airport elevation, with a minimum visibility of 5,500 feet runway visual range.

The controller told them to expect localiser approach to runway 5, which has a minimum descent altitude (MDA) of 1,320 feet above mean sea level (amsl) with a minimum visibility of 5,500 feet runway visual range. The first officer, presumably on the request from the captain, asked for the VOR-A circling instrument approach instead.

The VOR-A instrument approach is a circling approach because the final approach segment is more than 30° different from the alignment of the runway (runway 5 in this case). This approach has a minimum descent altitude (MDA) of 1,600 feet amsl and minimum visibility 1 mile upon crossing the FOGAG waypoint. At that point, the flight crew should not descend further until the runway is in sight and can be reached “at a normal descent using normal maneuvers”, where in this situation, a normal descent would be about 500 feet per minute.

If the flight crew does not have visual contact with the runway by the time that they have reached the missed approach point, then they follow the missed approach procedure which in this case is to initiate a climbing left turn to 3,000 feet above mean sea level.

At Air Cargo Carriers, the standard operating procedure was to use the localiser approach if it was available. It is a straight-in approach aligned with runway 5, which is great if you were coming from that direction. They weren’t. It seems likely that the first officer requested the VOR approach because it lined up nicely with their flight path, whereas the localiser 5 approach meant a slightly longer flight to the airfield.

When the first officer contacted the Charleston tower controller, she initially stated that the flight was on a visual approach and then corrected herself to say that it was the VOR-A approach. The tower controller now on duty agreed and cleared the flight to land. About two miles out from the FOGAG waypoint, the aircraft descended below the minimum-step-down altitude by 120 feet . An alarm sounded in the tower and the controller contacted the first officer to issue a low-altitude alert. The first officer responded to say that they were at 2,200 feet and getting down to 1,600 feet (the minimum descent altitude).

The VOR-A circling instrument approach had a minimum descent height of 773 feet above the ground, compared to the localiser 5 approach which allowed a descent to 373 feet above the ground (technically speaking, above the elevation of the runway threshold). On that night, the clouds were at 500 feet, which meant that the crew would be unlikely to make visual contact with the runway using that approach. But although the weather observer had noted the valley fog and low cloud over the airfield, the tower controller had not added it to the ATIS recording. About fifteen minutes later, his shift ended and he didn’t think to mention it to the next controller. The ATIS information still said that there was a broken cloud ceiling at 1,300 feet above ground level, which meant that the VOR-A circling instrument approach with a minimum descent of 773 feet was fine.

The tower controller replied that the alarm might have been triggered by the descent rate, which was somewhere between 1,300 and 2,000 feet per minute.

The aircraft levelled off at 1,600 feet. A witness on the ground saw the Short 330 as he was driving down a hill about two miles east of the approach end of runway 5.

From the investigation:

He stated, “it was a beautiful sight” [and] the airplane was “hugging” the bottom of the clouds, “just grazing the bottom of the cloud or fog deck.” He stated he saw the landing lights and the shadow of the airplane perfectly as it headed towards him, then it passed directly overhead. He thought the airplane was on the VOR-Alpha approach.

About half a mile from the displaced threshold of runway 5, presumably at the moment when the runway became clearly visible, the aircraft suddenly entered a steep left turn, descending at 2,500 feet per minute. The Short 330 reached a 42° bank angle as the captain attempted to line up with the runway. Then the aircraft pitched nose up. The descent rate reduced to about 600 feet per minute, likely caused by the captain pulling up hard as they raced towards the asphalt. However, it was too little, too late.

The aircraft’s left wingtip struck the runway 5 centreline with the aircraft in a 22° bank and a 5° nose-down attitude and an airspeed of 92 knots. The left-landing gear and left propeller crashed into the pavement and the left wing separated from the aircraft as it slid off the left-side of the runway and down a hill, coming to rest in the trees.

The crash was captured on CCTV (if reading in email, you may need to click through to the post to see the video):

There was no chance of survival.

When investigators arrived, they found the aircraft scattered around the runway and the rough ground.

The flight deck was damaged and distorted. Both power levers were found at the “Flight Idle” position and both fuel levers were found at the “Ground” position. The left propeller lever was found in an aft position, and the right propeller lever was full forward. The flaps handle was in the 35° detent. Both control columns were broken at the base and had partial movement in the forward/aft and left/right directions. Both control wheels were jammed in a full-right-turn position. Both rudder pedal assemblies were deformed, and the left pedals were missing. The control column bases, control wheels, and rudder pedals remained coupled and moved together.

Why had the flight crew tried such a messy last-minute manoeuvre rather than just go around? The crew did not appear to be under any time pressure and there was no reason to believe that the captain had been fatigued. However, investigators found that the captain was known to have some performance issues, including a checkride deemed ‘unsatisfactory’ based on poor instrument flying. Although the captain had thousands of hours of flight experience, the investigation found evidence that he was not comfortable with instrument flying. It seemed that he felt more confident that he could pull off an unstable approach than the approach on the localiser.

The first officer should have called for a missed approach but she was a lot less experienced than the captain and, based on interviews with her friends and family, she was hesitant about speaking up as she was new. The company had not shown any signs that they would support a less experienced pilot complaining about her captain, a captain to whom she was assigned to as a partner.

Besides, this wasn’t the first time he’d flown into Charleston like this. The flight crew had landed at Charleston seventeen times over the past month. On three separate VOR approaches, the captain had descended below the minimum descent altitude while in instrument conditions.

Also, both flight crew consistently turned late onto final and at a lower altitude than recommended by the airline’s standard operating procedures. Airport security footage videoed the base and final leg of the flights for runway 23. When the NTSB reviewed these visual approaches, they found that the final approach, whether flown by the captain or by the first officer, lasted just 12 to 39 seconds, compared to the 75 seconds for the final leg as required by Air Cargo Carrier’s standard operating procedures.

This is a clear example of procedural drift. Procedural drift is the mismatch between textbook compliance and how procedures are followed in the real world. Procedural drift in some form (descriptions vary) appears as a primary or contributing cause in the majority of aviation accidents. Simple shortcuts and willingness to take risks slowly erodes the safety nets put in place. When a pilot in a rush skips a checklist without issue, it becomes easier to skip it next time. When a second pilot accepts this as reasonable, both will soon begin to skip the checklist or in other ways become apathetic to standard operating procedures. If management does not appear to care and training appears simply to be ticking the box, procedural violations are not seen as a big deal.

In this case, a pilot with 4,300 hours was paired with a newly qualified pilot with 650 hours. Once flying together, they received little or no external input. This is an environment which makes it easy to lose sight of the importance of specific procedures. Over time, the captain started to take short cuts in order to focus on other issues that they see as more important — in this case, a stabilised approach was not seen as important as landing the aircraft quickly. The first officer initially expressed concern but over time, she too began flying into their home airport low and with a late turn to final.

These changes to established procedure continued to drift, until there was a glaring gap between the standard operating procedures and reality.

At Air Cargo Carriers, once a pilot was assigned a base and started flying, there was “no real scheduled oversight” according to the director of training. “Annual recurrent, and that was it.” It was hard, he said in interview, to know what they were doing.

However, it wasn’t that difficult to know where the captain’s weak spots were.

A review of the FAA’s complete airman file for the captain found several notices of disapproval going back to 1999. Most recently, during the ATP certificate checkride, on July 22, 2016, the captain received a notice of disapproval due to excessive deflection of both the glideslope and localizer for an instrument landing system approach, repeated glideslope and sink rate warnings from the ground proximity warning system, and his subsequent failure to initiate a go-around. He passed the practical reexamination 3 days later.

At airlines where various crew are paired together based on a roster, each crew member gets to learn from different people, particularly important in a case like the first officer’s, where she had low hours and not a lot of experience. But also, the captain might well have been corrected by another, more experienced first officer. If the first officer had flown that same unstabilised approach with another more skilled pilot, her approach would have been corrected. It could also have given her the opportunity to report the concerns that she had shared with her brothers and her friends.

By varying pilot pairings, there are multiple chances to catch procedural drift from standard operating procedures before they become ingrained.

The chief pilot at the airline was asked about the pros and cons of permanent crew pairings after the accident and conceded that a negative aspect was that people could pick up bad habits. The airline did not have any formal program for reporting or documenting safety issues or irregularities. For the first officer to highlight the issues with her captain, she would have needed to phone or e-mail either the director of training or chief pilot. Such reports were not logged or tracked; the only operational oversight was the annual recurrent training. The chief pilot said that he normally checked in with new-hire pilots after the first week on the job to see how they were doing — however, no one checked in with the first officer after she started. The chief pilot said that he assumed if pilots were having difficulties, they would reach out.

Probable Cause:

The National Transportation Safety Board determines the probable
cause(s) of this accident to be:

the flight crew’s improper decision to conduct a circling approach contrary to the operator’s standard operating procedures (SOP) and the captain’s excessive descent rate and maneuvering during the approach, which led to inadvertent, uncontrolled contact with the ground.

Contributing to the accident was the operator’s lack of a formal safety and oversight program to assess hazards and compliance with SOPs and to monitor pilots with previous performance issues.

Air Cargo Carriers learnt the hard way that CRM is not just about how two pilots interact with each other in the cockpit. It is not simply a case of giving the right theoretical answer when presented with an example with an obvious conclusion. CRM starts at operations. Pilots should be paired to avoid steep authority gradients and everyone at the company needs to support a healthy safety culture. Pilots must have trust in their management and they must feel secure that they are expected to report possible safety concerns. Standard procedures need to include following up with flight crew to ensure that possible issues are highlighted early. And when such issues arise, they must be documented.

References:

• National Transportation Safety Board Aviation Accident Final Report DCA17FA109
• NTSB Operations/Human Performance Interview Summaries
• Excerpts from NTSB Email Correspondence with Air Cargo Carriers
• Air Cargo Carriers CRM PowerPoint
• Procedural Noncompliance: Why Pilots Don’t Always Play By The Book
• FlightGlobal: NTSB gives details of 2017 Shorts crash as video surfaces