At 3:40 a.m. on a freezing January morning, a Beech D-95A Travel Air crashed 8 miles short of a northern Michigan runway. The National Transportation Safety Board’s finding of probable cause explained that “the pilot failed to maintain a proper glidepath and obstacle clearance”—which was inarguable—and cited as contributing factors “dark night, icing conditions, flight into known adverse weather and conditions conducive to pilot fatigue.”
The pilot, 45 years old, was a 10,000-hour ATP and CFI-I, qualified as pilot in command on a number of single- and multiengine types operated by his Part 135 employer. The afternoon before the accident, he had been notified at 4:30 p.m. that some cargo needed to be delivered to Huntsville, Alabama. He departed at 5:30 and arrived at Huntsville four hours later. At the FBO, he complained of being cold because, he said, the airplane’s combustion heater had stopped working shortly after takeoff.
The pilot’s employer said he had told the him he could stay overnight in Huntsville if he wished, and return the next day. The pilot had replied he would see how he felt after landing and decide then. Apparently, for whatever reason, he chose to return home. A mechanic serviced the heater, which then seemed to operate normally. A lineman brought the pilot some food from a nearby restaurant, the tanks were topped off, and the pilot took off for the return flight a little before midnight.
At 3:24 a.m., the pilot—then at 9,000 feet and 50 miles from his destination—reported to Minneapolis Center that he was turning slightly left to intercept the localizer for Runway 1. The controller cleared him to descend at his discretion to 3,200 feet; crossing altitude at the final approach fix, 5.8 nm from the runway, was 2,900 feet.
The last radar return from the airplane came when it was at 3,900 feet and 22 miles out. A little later, the controller reported to the pilot that radar contact was lost and inquired whether he was established on the localizer. The pilot replied that he was: “We’re just getting some pretty good moderate ice here at 3,200.” The icing type was rime, he reported; the temperature was -10 degrees C. Center provided a frequency change, and the pilot acknowledged.
He was not heard from again.
The Travel Air struck a 60-foot tree about 7.8 nm from the runway threshold. The impact point was about 120 feet above the published touchdown zone elevation. The main wreckage, with the landing gear down, came to rest 380 feet beyond the initial impact and was partly consumed by fire. About halfway between the initial point of impact and the main wreckage, the fiberglass nose cone—behind which the heater was installed—was found lodged in a tree, 25 feet above the ground; it bore traces of smoke and burn damage. The heater itself, found with the main wreckage, also had traces of smoke and burn damage, presumably not related to the fire that had consumed the fuselage and cockpit.
Those traces, although tantalizing, had no role in the NTSB’s analysis of the accident. No clear explanation was found, hence the meaningless probable cause, which amounts to saying that the accident was caused by the plane hitting the ground. The list of contributing factors, however, provides some food for thought.
The weather at the destination airport at the time of the accident was around 1,200 overcast with 6 or 7 miles of visibility. The pilot would have expected to break out at around 2,300 msl. The level, thinly forested terrain below him would have been completely dark, but he would have seen the town and airport ahead. The situation below the clouds, however, would have been a classic setup for a certain type of visual illusion: Approaching distant lights over dark terrain, the pilot believes himself to be higher up than he actually is. Remaining on the glideslope, obviously, provides absolute protection.
Read More from Peter Garrison: Aftermath
Three airmets were active for the area and time of the accident, but only one of them was significant: occasional moderate rime or mixed icing in clouds between the surface and 4,000 feet. Moderate icing is defined in the Aeronautical Information Manual as a “rate of accumulation such that even short encounters become potentially hazardous and use of deicing/anti-icing equipment or diversion is necessary.” The Travel Air was equipped with prop and windshield anti-ice, deice boots, and a heated pitot tube and was approved for flight into light-to-moderate icing conditions—but not for continued flight in moderate icing. Possibly, the icing was relatively heavy; a Part 135 pilot based in northern Michigan must encounter a fair amount of icing in winter, and the report of “pretty good” icing, making allowance for professional pilots’ proclivity for understatement, may have meant “quite a lot of icing indeed.”
At the time of the icing report, the airplane must have been about 15 miles from the airport. Less than 40 seconds later, the pilot left the frequency. The airplane then traveled about 5 miles before it crashed. Assuming that the pilot was slowing from 160 knots to 100 or so, that would have taken between two and three minutes. What happened during those minutes, we do not know.
It’s possible that the pilot—this being his home field—knew the terrain below him was flat and conditions at the airport were visual, and decided to get out of the icing by dropping down below the overcast. He might then have proceeded visually toward the runway and fallen prey to the illusion of height. The role of fatigue is uncertain, as usual, but it’s not implausible that a tired pilot might be more prone to visual illusions than a fresh one.
The NTSB’s mention of “flight into known adverse weather” is worthy of comment. If there was a known possibility of icing between the surface and 4,000 feet, any ILS approach to the airport would have involved flight into known adverse weather. Is this all the NTSB means? Probably—they tend to pick their contributing factors from lists. But if the accident had not occurred, the routine ILS approach would not have been considered an unreasonably risky operation.
What is striking, however, is the timing of the pilot’s descent. He chose to enter the layer of known icing long before he needed to. He had to cross the FAF at 2,900 feet; there was no reason to be at 3,900 feet 22 miles from the runway, and at 3,200 15 miles from it. He could have slowed to approach speed at 5,000 feet and begun a 1,000-fpm descent onto the glideslope when he was 8 miles from the runway.
I mention this as a general principle and not because I believe that the rime-icing encounter was the direct cause of the crash. From the fact that the landing gear was found extended, we can infer that the icing did not become an emergency. A pilot whose airplane is struggling to carry a load of ice would not further burden it by putting his gear down sooner than he needs to. On the contrary, the fact of the gear being down suggests the pilot felt his landing was assured—felt so comfortable, in fact, that he forgot about his altimeter.
