#17: Multiple company failures behind the 2022 Guardian Flight crash in Hawaii

The crash off the coast near Kaupo was the second of three fatal Guardian Flight accidents in five years.

The NTSB released its final report in September on the December 2022 crash of a Guardian Flight King Air C90 which killed the pilot, flight nurse and paramedic. Operating as Hawaii Life Flight, the aircraft was enroute from Kahului Airport on Maui to Waimea, 21 minutes away on the island of Hawaii, to pick up a patient. It departed about 8:53 PM, flying IFR under Part 91 with a single pilot, as permitted by the company Operations Specifications with an operating autopilot. Thirteen minutes into the flight the vertical gyro failed, which failed the Electric Attitude Indicator and caused the autopilot to disconnect. From the report, here is what happened next:

The pilot did not notify ATC of the autopilot failure. He did not request any special assistance. He maintained communications, responding to directions to enter the instrument approach for Waimea as if there was nothing wrong. The last contact was at 9:13PM when the pilot was asked to verify he was flying direct “to the TAMMI”, the initial approach fix for the approach. He responded, “Uhh, 13GZ is off navigation here… we’re gonna… we’re gonna give it a try.” Approach acknowledged his response and then thirty seconds later the aircraft crashed into 6,400+ feet of water ten miles from Kaupo. It is only when the wreckage was recovered three months later, and the aircraft’s inflight Appareo Vision 1000 airborne image recording system (AIRS) was found intact, that investigators learned of the loss of the autopilot. The AIRS camera, mounted above and behind the pilot, provided direct images of the instrument panel and the pilot’s actions. Here’s a still from the recording (the NTSB removed the pilot):

The report notes that the pilot’s instrument flight history was problematic and that he “had six Notice of Disapproval entries in his FAA records. Of these six notices, three were in rotorcraft and three were in fixed-wing aircraft, each one in the culmination of multiple unsatisfactory training events.” He had 7,668 hours of total flight time, 6,000 of which was in helicopters and 615 in the C90. He had been with Hawaii Life Flight for three years and had six checkrides in that period, failing three of them and requiring additional training to meet satisfactory standards. (He also flew part time for Sunshine Helicopters, a local sightseeing company.)

What Guardian Flight management had to say

Guardian Flight has a Director of Operations, Director of Training, Director of Safety, two Chief Pilots, and six Assistant Chief Pilots. Before the accident the DO was to be downgraded to Assistant DO and a new DO to be hired; they postponed this decision as the accident was investigated. (The new DO was approved in April 2023.)

The chief pilots were divided by aircraft type; one for the C90 and PC-12 fleets and the other over the Lear 45, King Air 200 and Cessna 208 fleet which was mostly in Alaska. One of the six assistant chief pilots was based in Hawaii.

There are 1,600+ pages of interviews in the docket for this crash and I am still reading through them. The report notes however that the DO, CP, DT and DS (all of whom were based in Utah where Guardian Flight was headquartered at the time - it’s now in Texas), were all somewhat surprised when presented with the pilot’s training history. They all also did a good job of diverting the responsibility for the pilot’s hiring was with the Assistant CP in Hawaii.

The Director of Operations said it typically was the assistant chief pilots who reviewed pilot training and FAA records. He “could not say for certain” he was involved with the accident pilot’s hiring but when he reviewed his training records after the accident told investigators that “Looking back at that, I think there are definitely concerns, some of them more significant than others…that wasn't, that wasn't what I would consider a normal training record.” As to how the pilot was hired, he pointed to Hawaii:

“I think there was some, some missed [sic] on some communication. I think that there's a cultural drive, particularly in the islands, to take care of their own problems. And so, I think that, you know, looking at that, they're like, hey, we've got, you know, we've got the sim [simulator] here. We can train them, you know, we can go back.”

According to the C90/PC12 chief pilot, the initial selection for pilot interviews was conducted by Human Resources, who checked the paperwork and weeded out those who did not qualify, and then the assistant chief pilots use a standard form for the interviews. This question was then posed by an investigator:

Q. Understood. So, then, let's say they narrow it down to five individuals. At what point are they now -- are you getting involved? So, now they're notifying you like, hey, boss, this is the five we think are good to go. Now, you're looking at them.

A. Yes, I would say that for a majority of the people, I am now into a point where it is document review, notes from the interview, it is, like I said, the PRD review, and going that way. I will have -- if assistant chief pilot is not 100 percent this is a person we need to hire, I will set up phone conversations with applicants at the request or the recommendation of the assistant chief pilot.

He did not discuss direct involvement in hiring the accident pilot. He also noted issues with Hawaii:

“I would say that if I had to look at my footprint, I don't know that we operate anywhere that there's more community pressure on the operation. Probably because of the logistics involved with transporting a patient or anything like that. And we do as much as -- from my perspective, we do as much as we can from a company perspective to alleviate that. But the communities are so tight knit out there that I think that there is community pressure for flying at times.”

How this “pressure” impacted pilot hiring and training was unclear.

The Director of Training managed instructors in Utah and also “basically manage the new-hire training, the onboarding of new hires, as well as the recurrent training for our current pilots.” He also was involved in “…evaluating their single-pilot instrument proficiency in the simulator, whether or not they hit the airman certification standards. And then provide my feedback in a briefing and a recommendation to hire or not.”

When it came to the accident pilot, he reviewed his records after the crash and had this to say:

Q. And in doing so, was there anything that stood out to you?

A. Yes.

Q. What was that

A. Multiple unsatisfactory remarks, repeat unsatisfactory remarks on similar or same tasks, failed check rides.

Q. Did you know about that prior to the accident?

A. Very, very minimal information about it.

Q. Minimal means many things, so my apologies. But can you give me an idea of what you mean by minimal?

A. Yeah, so I had been informed several years ago when he was hired that he had struggled throughout training. I was a part of creating an additional panel, a plan when he was a new hire. And then he had subsequent failed check rides that either were not communicated or just got lost.

The Director of Safety discussed the Aviation Safety Assessment program and risk assessment forms for each flight and said they had a safety shutdown for Hawaii Life Flight after the crash. He did not discuss how a pilot who could not fly an aircraft without the autopilot was in the air alone at night.

Finally, the Assistant Chief Pilot in Hawaii remembered nothing remarkable about the accident pilot’s hiring. He was recommended by another pilot, his interview and qualifications were not remarkable. His failed checkrides, which were later corrected via additional training, were not remarkable. He did acknowledge when asked that they did not train at the company for a significant avionics failure.

Before the Kaupo accident

Four years before the crash in Hawaii, a Guardian Flight Beech 99 was enroute to Kake, Alaska, to pick up a patient when it crashed into Frederick Sound in January 2019. The single pilot, flight nurse, and paramedic were killed. In that accident, although the wreckage was also recovered from the water, the aircraft did not have the Appareo Vision 1000 onboard. The pilot, who was flying IFR at night, was in regular communication with Anchorage center until three minutes before the crash, which occurred about 6:11 PM. He was cleared for the RNAV runway 11 approach and then confirmed a change of frequency. The NTSB was unable to determine what caused the accident and the Probable Cause reads: “a loss of control for reasons that could not be determined based on available information.”

Without the Appareo Vision onboard the Kaupo aircraft, a similar Probable Cause would likely have been issued for that crash.

After the Kaupo accident

Two months after the crash in Hawaii, a Guardian Flight PC-12 was about ten minutes out of Reno, Nevada when it crashed near the town of Stagecoach in February 2023. The single pilot, flight nurse, paramedic, patient, and patient family member were killed. In that accident, which is still under investigation, the pilot was in regular communication with Oakland ARTCC as he flew IFR to Salt Lake City. At 9:08PM, five minutes before the crash, he acknowledged a directive to climb and maintain FL 250. At 19,100 feet he began a descent and did not recover. The NTSB’s Preliminary Report does not state if the Appareo Vision 1000 was installed in the aircraft. I would expect the final on this accident by early next year.

Patterns emerge

It is not difficult to see what these accidents have in common. Here is Kaupo:

Here is Kake:

Here is Stagecoach:

All of them were single pilot, all at night, all IFR, all crashed after descending right turns.

For Stagecoach, there is no information on the pilot’s flight time or training in the Preliminary Report.

The Kake pilot had 17,744 hours of total time, with 1,644 in the Beech 99, and had been with Guardian Flight for four years. As his body was not recovered, it is unknown if he suffered a sudden critical medical event which caused the accident. Like Kaupo, and Stagecoach, he was in constant contact with controllers and communicated nothing suggesting he was in distress. Investigators conducted no interviews for that accident and there is little information beyond flight time for the pilot. For example, whether or not the company thought Alaska and its training methods or “culture” were a problem is unknown.

What the FAA had to say

Guardian Flight’s FAA Principal Operations Inspector (POI) was interviewed in March 2023 for the Kaupo accident. Her interview occurred one month after Stagecoach but she was not asked anything about it, or the Kake crash. She became POI in early 2021 and most of her interview dwelled on the complexity of handling a company that is based in UT and operates all over the country, while she is based in CO. An interesting series of questions were posed about events or “trends” at Guardian. Here is her reply:

There’s -- you know, so, we’re trying to track trends. Like so, within -- I’m trying to think. So, one thing we recognize as a CMT…meaning certificate management team, okay? So, that’s the avionics, maintenance, I mean, as a team. We meet regularly, too, just -- they’re usually on monthly meetings, and we meet regularly, too, just so that way, we can discuss whether it’s risk or whatever. You know, and, yeah, we recognize that you can pull in all these occurrences, and pilot deviations, and stuff like that, and there’s not really --you can go and search data, right? You know. But the -- and within some of our systems, there’s like -- you know, I can see, okay, we had accidents, right? I can see that. I can see incidents. But the occurrences and pilot deviations don’t kind of line up in a reporting process that’s easy to data mine for us as inspectors, so we actually put that in as a part of our risk assessment, is that we recognize that we don’t have like a good catch-all, per se.

Trends are something I pay a lot of attention to when writing about a company. In my experience, it is rare that any accident occurs solely due to actions in the cockpit; that is simply the location of the last actions before a crash. Trends are found by studying the prevalence not just of incidents and accidents, but also events and occurrences, which are things like hard landings, etc. that don’t meet the damage requirement to be classified as incident or accident, plus altitude deviations, emergency declarations, runway incursions, engine shutdowns, etc. This FAA interview shows what I long suspected; not all POIs are collecting and processing trend data in a timely manner and are thus unable to effectively track it. Collecting this info is not the sort of work that requires a POI - qualified researchers and assistants with the FAA can do this for the inspectors. (I’ve been doing it for years but had to submit FOIAs to get it.) Based on what I have found here and elsewhere, I do not think the FAA is capturing the information that different parts of its far reaching system already knows.

They can’t see the forest for the trees.

The Kaupo Probable Cause

From the report: “Guardian Flight’s inadequate pilot training and performance tracking, which failed to identify and correct the pilot’s consistent lack of skill, and which resulted in the pilot’s inability to maintain his position inflight using secondary instruments to navigate when the airplane’s electronic attitude direction indicator failed, leading to his spatial disorientation and subsequent loss of control.”

The media is not reporting this story

The release of the Kaupo accident report was barely covered in the media and those who mentioned it failed to bring up the similarities with Kake and Stagecoach. In the brief reports I’ve read, I see no evidence of FOIA requests on Guardian to determine what else might have been going on between January 2019 in Kake and February 2023 in Stagecoach. With the Nevada final report looming, I know what info I would be tracking and requesting right now if I was writing for an industry publication. I know what I would want to learn more about.

If this kind of work is something you are interested in supporting, then becoming a paying subscriber to this newsletter is critical. We all know the major media is in trouble, and I can tell you that aviation media is not very interested in deep dives on Part 135 operations. I can do the work; I will do the work, but I need the funding to make it a viable concern.

(It will also help pay my lawyer because, yes, the Plaintiff is appealing the verdict in my anti-SLAPP lawsuit win.)

More soon on medevac accident statistics and what I’ve learned in reading FAA interviews in various accident reports (including this one). And if I can, more on Guardian Flight as well.