Accident Analysis: A major icing foul-up

In February 2013, an Embraer Phenom 100 in commercial service—flown by two qualified pilots—flew into conditions not considered all that challenging to the capabilities of the aircraft and crew. However, it was severely damaged on landing. Thankfully, the crew and single passenger were unharmed, but a mere handful of seconds likely separated them from serious injury or death.

Thanks to the survival of all aboard and the presence of a cockpit voice data recorder (CVDR), the facts of the flight are clear. At 7:38 a.m. local time the crew departed on a 400-nautical-mile leg from Kortrijk-Welvelgem Airport (EBKT) in Belgium to Germany’s Berlin-Schönefeld Airport (EDDB). The captain had 4,500 hours of flight time with 800 in type, and was current in type, with 30 landings the Phenom in the past 90 days. What’s more, his work schedule didn’t give any reason for fatigue to have been an issue. He had more than 36 hours of rest preceding the accident flight.

The first hint of increased risk to the flight comes with the co-pilot’s details. She was a relatively new, 22-year-old commercial pilot with only 260 total hours. She earned her type rating in the Phenom 100 just more than three months earlier and had accumulated 32 hours in type, most of which was in the month preceding the accident.

The German Federal Bureau of Aircraft Accident Investigation (BFU) noted that: “The BFU is of the opinion that the CVDR recording indicates that the work relationship between the very experienced PIC and the co-pilot, who was at the beginning of her flying career, resembled more a relationship between flight instructor and student pilot.” Adding the workload of flight instruction to the demands of a commercial flight with additional pressures can only increase the demands on the pilot in command.

The climb, cruise, and descent phases of flight proceeded normally, and during a descent through Flight Level 200 into Berlin-Schönefeld Airport the ATIS reported winds from the east at 7 knots, visibility 3 miles in mist, with a ceiling of 1,400 feet broken, and a temperature of zero degrees Celsius. Of significance was ATIS remark that there was “moderate icing reported below 3,000 feet.”

The Phenom 100 requires that before landing the crew must decide which of two sets of landing speeds will be used: one for approach and landing in icing conditions, and one for use only if the “entire wing, including unprotected areas and areas behind the wing deicing boot, are free of ice accretion.” Because the landing speeds for icing conditions are more than 20 knots greater than those for ice-free conditions, the runway distance requirements increase significantly if landing in icing conditions.

The Phenom 100 requires that before landing the crew must decide which of two sets of landing speeds will be used: one for approach and landing in icing conditions, and one for use only if the “entire wing, including unprotected areas and areas behind the wing deicing boot, are free of ice accretion.”The NTSB cited this fact in its investigation of a 2016 fatal Phenom 100 accident in Maryland. Ruling the probable cause to be “the pilot’s conduct of an approach in structural icing conditions without turning on the aircraft’s wing and horizontal stabilizer deice system, leading to ice accumulation on those surfaces… result[ing] in an aerodynamic stall,” the NTSB discussed the possibility of the pilot’s concern with this increased landing distance as an explanation for not activating the wing de-icing system. Turning this system on increases the airspeeds for stall warning and stick-pusher activation so much that it’s impossible to fly a safe approach at normal, non-icing approach airspeeds.

In this case, there was more than 10,000 feet of landing distance available on EDDB’s Runway 7L, the landing runway. This was more than adequate for a safe landing out of an approach using recommended approach airspeeds; there was no incentive not to use the higher approach and landing airspeeds with the ice-protection system activated. Despite the absence of any downside to using the icing airspeeds, the crew elected to use non-icing airspeeds for landing. So, they calculated and set a landing reference speed (VREF) of 96 KIAS.

Descending through the tops of the clouds at 3,000 feet msl, the captain (and PIC) activated the engine anti-icing system, stating “adding the engines because it is negative again, they have said it, moderate icing below three thousand.” The wing anti-ice system—which inflates de-ice boots on the wing and tail leading edges—was left off, and its activation was not discussed by the crew.

A faster than normal speed was maintained during the approach per ATC’s request. Descending through 1,000 feet above field level, the captain reported the airport in sight and shut off all icing protection. He later recounted to the BFU that he visually inspected the wing at that time and did not see any ice accretion. The BFU noted in the accident report that with only a quick look at the wing “it is possible that he could have overlooked the milky-white ice accretion on the silver-grey deice boots.” It also said, “It cannot be ruled out, that he did not take a look at all.”

Airspeed gradually decreased until reaching 94 KIAS, just under the calculated VREF, at 50 feet above field level. Speed further decreased to 90 KIAS at approximately 30 feet above the field, when the aircraft began an uncontrolled roll to the left, reaching a bank of 30 degrees before the left wing struck the runway. The outboard 2 meters of the left wing bent upwards 10 degrees from the impact, followed by the right main landing gear striking the ground hard enough to fracture and drive parts through the upper wing surface.

Airport officials photographed the aircraft within a few minutes of the accident and documented one-quarter to one-half inch of mixed ice on both wings’ and horizontal stabilizers’ leading edges. Not surprising, it was determined the accident cause was “an abnormal flight attitude during the flare…due to ice accretion,” and that the crew “under known icing conditions…did not activate the wing and horizontal stabilizer de-ice system.”

So why would an experienced PIC allow the flight to reach such an end? The relationship characterized by the BFU as that of a flight instructor and student undoubtedly contributed. The report notes that when asked by ATC to maintain a higher than normal speed on approach, the PIC had requested the co-pilot to maintain the assigned speed nine times. Furthermore, the BFU noted that during the descent and approach phases of flight the crew did not complete any of the required checklists at the time mandated by their operating procedures—instead, every one was completed later than required.

That the captain’s attention was divided was also evidenced by late radio calls to ATC. Even though the landing clearance “had been issued about three minutes earlier…the PIC was still in communication with the controller regarding the landing clearance” when the aircraft reached minimums of 200 feet above field level, according to the BFU. Such division of attention between coaching the inexperienced co-pilot and maintaining awareness of the state of the aircraft likely contributed to the lack of awareness of the icing situation on the aircraft’s wings.

More important, though, is the crew’s attitude toward activating the wing de-ice system—one contrary to the explicit requirements of the Phenom’s airplane flight manual (AFM). The AFM is clear that the wing de-ice system is to be proactively turned on anytime there is visible moisture with a temperature below 5 degrees Celsius. Those conditions existed during the descent, and as such the presence or absence of icing on the wing was, in practice, irrelevant.

The PIC tellingly noted to the BFU that “there are discussions among pilots whereby the system should only be activated at ice accretion of at least 0.25 inch.” Originating in now well-discredited theories of “ice bridging” (the formation of ice in a shell around the de-ice boot), this operational procedure is in direct contradiction to the requirements of the aircraft’s AFM.

Yet the accident pilot was not alone in operating in such a manner. The BFU found by analyzing the cockpit voice data recorder information from the previous winter that on a half-dozen occasions, various pilots for the operator did not activate the wing and horizontal stabilizer deice boot system in accordance with the AFM when meteorological conditions dictated doing so. The investigators felt “the knowledge level of other EMB-500 pilots of the operator regarding the wing and horizontal stabilizer deice system and the SWPS [stall warning protection system] varies and is partially inadequate.”

A flawed operational methodology of a critical safety of flight system, based on antiquated and disproven theories, appears to have combined here with the task saturation that flight instruction can cause, in a near-tragic manner. Consider that with a roll rate of 15 degrees per second, had the stall occurred as little as four seconds earlier, the aircraft would have struck the surface in a knife-edge or inverted attitude, significantly changing the survivability of the event.

Pilots must constantly reexamine their knowledge and update their procedures when best practices change, and be especially cautious when they find themselves operating contrary to a manufacturer’s established procedures. Also, pilots and operators should not ignore or downplay the very real addition of workload that instruction and mentoring add to flight.

Neil Singer is a Master CFI with more than 8,500 hours in 15 years of flying.