Ground icing

The Phenom remained outside without any deicing treatment for 40 minutes before the pilot commenced the takeoff roll. Within 30 seconds of thrust levers being advanced for takeoff, the Phenom was in pieces, with both wings and engines separated from the airframe. It had briefly lifted off only to reach a maximum height of 14 feet above the runway while entering an increasingly steep left bank. Despite full opposite aileron input, the bank angle peaked at 60 degrees left wing down as the aircraft stalled and impacted the runway.

Two years later, a Bombardier Challenger 650 began its takeoff roll in Bangor, Maine. The crew had deiced the aircraft, after which a protective layer of anti-icing fluid was applied. Only 13 minutes elapsed from when the fluid application was completed to the beginning of the takeoff roll. Again, in less than one minute from advancing thrust, the aircraft was destroyed. In an eerie echo of the Phenom accident, the Challenger reached a peak altitude of roughly 15 feet while entering an increasingly steep bank unaltered by full opposite aileron deflection, ultimately striking the runway surface in a 75-degree bank.

Both accidents are perfect illustrations of how nonlinear the threat of icing can be. The Phenom’s aircraft manager stated after the accident that in the previous eight years, they had only had to deice the aircraft twice. Given the sequence of events preceding the accident flight, it’s not difficult to imagine there were previous occasions where the airplane was kept in a hangar, pulled out without any fluid application, and successfully departed in similar precipitation conditions.

The Challenger crew was recorded on the cockpit voice recorder (CVR) discussing holdover times (HOTs)—how long an anti-icing application is likely (but not guaranteed) to protect the airfoils from precipitation freezing to the surface. According to the NTSB preliminary report, “The pilot commented that it was ‘standard’ to have 14 to 18 minutes and that if the wait was more than 30 minutes, they would return to the ramp to deice again. The copilot concurred with the pilot.”

The pilot was correct. For the thick “Type IV” fluid that was applied, 30 minutes of holdover time would often be quite conservative; in many snowfall conditions, the holdover time would actually be measured in hours. Unfortunately, on the day of the accident, the combination of snowfall intensity and very cold (minus-16 degrees Celsius/ 3 degrees Fahrenheit) conditions resulted in a holdover time of only two to nine minutes. Sixteen minutes had passed between the start of application and the takeoff roll, well exceeding the maximum expected protection time for this specific flight.

In both cases, it is easy to envision that to the pilots, the scenario they faced during the accident flight felt like many they had encountered before. Consequently, they applied the same “procedure” (skipping deicing, or using a rule of thumb for holdover time) that resulted in a successful takeoff on each previous flight. Tragically, the smallest shift in ambient conditions on the days in question drove the results to different ends entirely.

Pilots may not be aware of how close to stall speed a jet is during liftoff—for light jets rotation speed (V_R) may only be roughly 10 knots over stall speed, and for Part 25 certified jets, takeoff safety speed (V₂) only needs to be 20 percent above stall speed. NASA has found that the smallest amount of contamination on the wing can cause lift to decrease by 30 percent; since stall speed varies as function of square root of maximum lift coefficient, a decrease in lift of 30 percent raises stall speed by approximately 20 percent. Flying published speeds, the pilot of a contaminated aircraft may find the airplane able to lift off, but unable to maintain controlled flight just feet off the surface.It is easy to envision that to the pilots, the scenario they faced during the accident flight felt like many they had encountered before.

Especially dangerous is when contamination has accumulated more on one wing than the other, as likely happened to both aircraft discussed. One wing produces more lift while the other wing is in a stalled state, causing the airplane to roll as it becomes airborne. The instinct to apply yoke input against the roll would lower the aileron of the stalled wing, increasing the local angle of attack at the wing tip and further stalling the wing. Further, full yoke movement in many jets will result in roll-spoiler extension, exacerbating the stalled condition.

A rigorous adherence to FAR 91.527, which specifies that no pilot of any aircraft may take off an airplane that has frost, ice, or snow adhering to (among other things) any wing or stabilizing or control surface, is the key to a safe departure in active icing conditions. In turn, the only way to have guaranteed compliance is through scrupulous observation of proper deicing and anti-icing procedures when departing in freezing precipitation.

Unfortunately, deicing is not as simple as spraying fluid on an airplane and heading out; an assortment of details needs to be considered by the pilot in command and coordinated with deicing personnel. What fluids will be applied and in what ratio (in some cases mixing fluid with water can increase the holdover time, in some cases it decreases it), where on the aircraft to spray and not, and how communication between the aircraft and deice crew can be conducted are some of the more critical, but far from only, items to be worked out before deicing begins.

In conjunction, a careful and conservative calculation of holdover time should be performed before the deicing starts, so that critical holdover minutes are not spent looking up data. Every autumn the FAA publishes a document of the holdover time tables for all types of fluid in use, along with supporting tables for proper calculation of snowfall intensity, and minimum-use temperatures for the various fluids.

Even better, several companies have created reasonably priced apps that automate the holdover time calculation. Able to pull current METAR information from the internet, the apps take tremendous time and potential for error out of holdover time calculations, representing a “must have” for pilots faced with departure in ground icing conditions.

In one of the best deals in aviation, NASA has produced a free online training course titled “A Pilot’s Guide to Ground Icing.” Using animations and accident reports, NASA has done an excellent job breaking the subject down into manageable bits, and throws in some great reference documents, to boot. Like many things in aviation, what is bewildering to the uninitiated resolves into a logical and straightforward process with study and operational discipline.

Neil Singer is a corporate pilot, FAA designated pilot examiner, and instructor in Embraer Phenoms and Cessna Citations. He has more than 10,000 hours of flight time.