The fatal crash of N928JP occurred following the second consecutive RNAV approach to runway 20 at Burley Municipal Airport (KBYI) in Idaho. The aircraft impacted the top of a 100-foot-tall agglomerate stack over a potato processing plant. It was a Part 135 cargo run out of Salt Lake City under a UPS contract. The Cessna Caravan had a full load and was taking off at max gross weight. Spring snow showers were prevalent along the route. An AIRMET had been issued for moderate icing up to 16,000 feet. The pilot was the sole occupant. Earlier this year, the NTSB published the final report of the accident: “The pilot’s failure to maintain altitude during an instrument approach, which resulted in a descent below the approach path and impact with a vent stack. Also causal was the failure of the processing plant to correctly paint the vent stacks, which the FAA had determined to be a hazard to navigation due to their proximity to the landing approach path.”
Six years before the accident, the airport manager at Burley observed the Gem State Processing plant adding a row of six exhaust stacks directly along the final approach path to runway 20. The manager notified the FAA, who issued a ‘Notice of Presumed Hazard’ due to the obstructions penetrating the clearance surface to runway 20. The VASI for the runway was decommissioned as a result. Not long after this, Gem State had to increase the height of the stacks to comply with EPA standards. In what can only be described as the odd logic of bureaucracy, when Gem State refiled with the taller structures, the FAA improved its findings to a ‘Determination of No Hazard to Air Navigation.’
The FAA’s clarification to the NTSB regarding the downgraded risk assessment was muddy. While they found the structures to be obstructions, they nevertheless determined that it did not adversely affect operations to runway 20. There were a few caveats: the removal of the previously decommissioned VASI; painting the stacks high visibility orange and white; new takeoff minimums for runway 2; and an RNAV procedure with a rather steep 3.75-degree final glidepath (published with LNAV only minimums, the glidepath was advisory). A close-in fix (JAMID), only 1.6 NM from the runway, protected against early descent into the obstructions. The FAA stated: “As a condition of the No Hazard Determination…the stacks were required to be painted…and lit with red obstruction lights.” Though Gem State had, in compliance with FAA instructions, installed aircraft warning lights on each of the six structures, the NTSB was unable to inspect the beacon from the agglomerate stack. A YouTube personality had trespassed on Gem State property to investigate the crash. He filmed his exploits, which included handling the light. Later, he posted a second video from Georgia featuring a strikingly similar light. He denied to investigators that it was the light from the agglomerate stack. This resulted in the unusual inclusion of a Sheriff’s report in the accident docket. The trespass and stolen property case was ultimately dismissed due to the exceedance of the statute of limitations.
Given that the accident occurred during daylight, the status of the light was mostly irrelevant. More concerning was the fact that none of the six structures had been painted the prescribed aviation orange and white in the five-year period that separated the FAA instruction to do so and the crash. the day of the fatal accident, grey stacks
loomed across the Snake River.
When flown properly, the 3.75-degree glidepath to the runway provided 98 feet of clearance to the agglomerate stack. ADS-B data memorialized Brittney’s two approaches to the airport. The first adhered to the published vertical path. Airspeed fluctuated between 128 and 132 knots before peaking at 156 knots following capture of the glide path. The approach ended in a flyby that was captured on an airport surveillance video. The low pass was in line with company policy to accomplish a runway inspection prior to landing in such conditions. The tarmac was covered in a thin, white blanket. ASOS was reporting 1-mile visibility in light snow.
Initially, airspeed on the second approach varied between 114 and 134 knots. The aircraft leveled momentarily at 6,000’, approaching HIKLO (the final approach fix). This resulted in the aircraft being high on the glidepath (though an LNAV-only approach, the G1000 displayed an advisory glidepath that the flight director and autopilot could capture for a stable descent). The Caravan made a steep descent to recapture the glidepath, which was accomplished prior to JAMID waypoint. The rate of descent decreased, and airspeed deteriorated. The aircraft continued decelerating until the final ADS-B datapoint (0.6 nm from the runway) which depicted an airspeed of 85 KCAS. With flaps up, the minimum speed in icing conditions for the Caravan is 95 knots. Cessna’s Low Speed Awareness System (LAA) was designed to alert if airspeed decreased below 97.5 knots in icing conditions.
Mission-Oriented Risk
On a winter operations review worksheet for Gem Air (the Part 135 operator was unaffiliated with Gem State Processing plant), question 22 was tragically prophetic. “You are flying from Salt Lake to Burley. What course of action if Burley is snowing at the time of your arrival?” The pilot handwrote the answer. “Divert Twin Falls.” The day before the accident flight, the pilot had done just this. The reason for the diversion was not related to the steep RNAV approach but rather the lack of ground deicing equipment at the airport. The operations manager for Gem Air, clarified to the NTSB: “If it is snowing, the pilots will go to Twin Falls Airport [since they have] de-ice.”
The Caravan had an early reputation for being a handful in icing conditions. Cessna employed three items to rehabilitate the workhorse’s image. TKS weeping wings replaced the inflatable boots of early models (TKS was installed on the accident aircraft). Aircraft-specific cold weather training was developed to emphasize to pilots the absolute need to maintain airspeed margins while in icing conditions (Brittney had a certificate of completion for the program). The LAA system was developed to provide aural and visual cues when airspeed degraded. The efforts have yielded positive results. The early rash of Caravan crashes in the snow has trickled in line with the winter performance of other fleets.
Operationally speaking, a diversion to Twin Falls was no big deal. The UPS contract with Gem Air had a provision for just such an event. If it were snowing in Burley, the drivers would hoof it to Twin Falls to unload the Caravan. If it delayed the offload by an hour, so be it. Investigators had asked during interviews with Gem State personnel if there was any reason why the pilot would have felt pressure to complete the mission. The response among the pilots and managers was uniform. If it was snowing in Burley, you were supposed to go to Twin Falls. If there was any pressure, it was to divert. An hour’s delay was far better than stranding a turbine asset in the snow. With very few exceptions, deicing is required prior to operating a contaminated aircraft. During interviews with the NTSB, Gem Air personnel noted that the pilot had an important appointment in Salt Lake immediately following her scheduled roundtrip. There was concern that Brittney had felt pressure to land in Burley to expedite the return to Salt Lake City.
The Gem Air pilot would not be the first to succumb to such a compulsion. Two other accidents provide examples. The first, like the Caravan crash, occurred in the snow. A Part 91 flight in a Pilatus PC-12 out of Chamberlain, South Dakota, terminated in a low-altitude aerodynamic stall that resulted in a fatal crash that claimed the lives of 9 of the 12 occupants. If you know the PC-12, that last figure will make you blink. Though the aircraft can technically be configured with 12 seats, it rarely is (the accident aircraft was equipped with 9). The three excess occupants created problems for the NTSB when calculating weight and balance for the accident flight. Likely, they were sitting somewhere along the length of the aisle, a rather imprecise datum reference to determine the center of gravity. The aircraft was photographed and videotaped via cell phone both before and after it began taxiing. It displayed a noticeable slope front to back, which indicated a C.G. that was significantly aft of certification limits.
Not helping was the snow on top of the horizontal stabilizer. The pilot had arrived before the passengers to remove contaminants from the aircraft wings manually. A ladder was not readily available, so he was unable to clear the t-tail. The NTSB had this to say: “Performance analysis indicated that the accumulated snow and ice on the empennage did not significantly degrade…performance after takeoff. However, the effect of the snow and ice on the airplane center-of-gravity…could not be determined.” Pitch oscillations associated with instability due to the aft center of gravity ultimately resulted in a fatal loss of control.
A pilot of a Beechcraft Baron some years ago failed a similar test. When he filed a post-Memorial Day flight plan, he reported eight passengers. He confessed that his enroute cruise speed would be abnormally low due to landing gear that would not retract. The FSS specialist responded, “I thought the Baron only had six seats?” The response was, “Some extra kids showed up.” Calculations following the fatal crash determined that, despite having more passengers than seats, the aircraft was likely below max weight at takeoff. During the initial climb, the right engine failed. The Baron was unable to maintain altitude with the gear extended. The pilot did his job and maintained airspeed, but the airport was surrounded by forest. There were no survivors. Experienced pilots often succumb to risk creep due to a mission-oriented focus. We have a hard time accepting that throwing in the towel often represents the peak of professionalism.
During the fatal second approach into Burley, the Caravan inadvertently leveled off at 6,000 feet. A steep descent to recapture the glide path ensued. On both approaches, the pilot failed to reduce torque during initial descent, resulting in an acceleration well above normal approach speed. The resulting correction produced a continual deceleration for the remainder of both approaches. On the first approach, peak airspeed was 30 knots above the normal approach speed. The peak on the second approach was better controlled, but the deceleration curve occurred with the same intensity. The final recorded speed was 35 knots below normal approach speed and ten knots below minimum icing speed. Had there been a moderate or greater accumulation of icing, the aircraft would have likely stalled and spun. On this day, the abnormally slow speed produced an angle of attack that resulted in the agglomerate stack being below the glare shield. The grey exhaust stacks were out of sight until mere seconds prior to impac
A significant percentage of aircraft mishaps begin with the intentional decision to be procedurally non-compliant. This is a tough one because inflexible compliance with procedures can itself be dangerous. And even when it is not, regulatory fundamentalism can elevate trivial discrepancies into obscene inconveniences. In an early job interview, I was asked whether I would cancel a Christmas Eve flight with a dozen passengers onboard for an inoperative navigation light. Answers in interviews are often dishonest things. I went with the regulations. The interviewer was unimpressed. The real question is, where exactly is the line? It is a short answer if it involves snow or seatbelts. In other instances, it is more complex. One striking difference between an average and a great pilot is converting a marginal approach into a go-around. The average pilot plods along and considers the salvaged landing a success. The expert mutters disappointment while firewalling the throttles. There are many things in aviation that you’ll get away with until you don’t. It takes great discipline to come to peace with “no.” It might even save a life.
