Ice Contamination Accidents On Takeoff Reach Epidemic Proportions

Aviation may be the only mode of transportation where frozen water can lead to a fireball. The case at hand involves the Nov. 28 crash on takeoff at Montrose, Colo., of a Challenger 601 twinjet in which icing is the leading culprit.

Three of the six persons aboard were killed, including the pilot in command and the flight attendant. Others were injured, including the copilot and NBC television sports executive Dick Ebersol.

The crash of the Jet Alliance aircraft takes on special importance for at least three reasons:

1) The accident is part of a larger pattern of ice-contaminated airplanes crashing on takeoff; indeed, more than 750 deaths in these takeoff accidents suggest that harsh lessons are repeatedly going unlearned. The Ebersol crash provides yet another example of the hazard inherent in even a barely perceptible film of frost or ice on the wings. Operators and aircrews may not be sufficiently alert to the suddenness with which such ice can kill – at a time when the airplane is most vulnerable trying to build up speed and lift to get airborne;

2) The crash occurred just 19 days after the NTSB criticized the Federal Aviation Administration (FAA) for its “unacceptable” response to recommendations coming out of a 1994 icing crash, so the latest disaster adds a deadly “we told you so” weight to the NTSB’s frustration (see ASW, Nov. 15);

3) The crash involves a prominent television executive, whose 14-year old son died in the crash, suggesting intense media scrutiny of the National Transportation Safety Board (NTSB) investigation.

Similar scenarios

The Ebersol crash has certain elements in common with a previous accident. On the morning of Jan. 4, 2002, another Challenger 604 crashed on takeoff at Birmingham International Airport in the UK. The airplane had sat on the freezing tarmac overnight, accumulating a layer of hoarfrost on the wings estimated at some 1-2 mm thick.

According to the UK’s Air Accidents Investigation Branch (AAIB) report, the captain asked the copilot, who was the flying pilot, about the situation (as captured on the cockpit voice recorder, CVR):

Commander: “Got a (unclear) frost on the leading edge, on there, did you look at it?”

Handling pilot: “Huh?”

Commander: “D’you (unclear) that frost on the leading edge – wings?”

Handling pilot: “Did I feel ’em?”

Commander: “Yeah, did you – all check that out?”

Handling pilot: “Yuh.”

The AAIB concluded that the crew’s discussion of the icing situation was “ineffective.”

“The discussion on icing initiated by the captain did not adequately address the issue or arrive at an appropriate conclusion,” the AAIB said.

During taxi to the runway, the crew carried out the pre-takeoff checklist. As the AAIB report recounted, when the anti-ice checklist item was reached, the handling pilot remarked, “We may need it right after takeoff.”

The AAIB report said, “This response seems to embody only a token acknowledgement of the de-icing problem – as something that could be left until later.”

The airplane was not de-iced before taxing to the runway for takeoff. As the AAIB noted, all other aircraft that had been parked overnight at Birmingham, and scheduled for morning flights, were de-iced. They departed safely. Immediately after getting airborne, the accident aircraft rolled sharply to the left and struck the ground inverted (possibly the result of the way the airplane had been parked, one side getting early morning sun and the other remaining in shadow, leading to one wing stalling, the other not). On impact, fuel tanks ruptured and the aircraft slid to a halt on fire. All five aboard were killed.

The Ebersol crash has certain elements in common. First, it involves the same airplane, which features a supercritical wing. This type of wing, found on many jets, is particularly vulnerable to lift loss from leading edge contamination. Second, the airplane carrying Ebersol and his two sons may have accumulated a thin layer of frost on the wings. It had landed at Montrose about an hour (and maybe as little as 35 minutes) before, descending through fog, which could have built up a layer of frozen water on the wings and engine fan blades. The airplane sat on the ground in cold temperature conditions before the Ebersols boarded. Snow was falling. In conditions of snow hitting and freezing, the situation would not be dissimilar to the frost accretions in the Birmingham accident.

As in the Birmingham crash, the aircraft was not de-iced before takeoff. The brief time on the ground may have been a factor in the pilots’ decision to forego de-icing. The crew may not have been sufficiently aware of the acute danger posed by a thin layer of hoarfrost or a smattering of frozen snow on the wings. In the waning light of the late afternoon, any frost or ice accumulation on the wings might not even be noticed, at least not without a walk around and physically running one’s ungloved hand across the wing leading edge – ice detection by Braille, as it were.

In the Ebersol crash, Capt. Luis Polanco, reportedly hired by Jet Alliance last August, was from the Dominican Republic. According to preliminary accounts, the bulk of his flying experience was in tropical climes. He was killed in the crash. Co-pilot Eric Wicksell, now hospitalized in critical condition with burn injuries, was from Daytona Beach, Fla. The icing training they received and the operator’s icing awareness program are sure to be examined closely by the NTSB. The airplane was equipped with a cockpit voice recorder (CVR), and pre-takeoff discussions between Polanco and Wicksell may shed light on their state of icing awareness and high elevation operations. What one can get away with at sea level might not apply at a h;igh density altitude (Montrose is at 7,500 ft. above sea level). The elevation sucks performance – and Polanco had opted for departure off the 2,500-ft. shorter runway, as the longer one was still being swept.

According to a witness, the Jet Alliance airplane appeared to twist perpendicular to the runway, getting airborne briefly, then skidding off the end of the runway and bursting into a fireball.

De-ice before departure

An icing-related crash on takeoff can involve a combination of factors. Depending upon whether the leading edge or the entire upper surface of the wing is contaminated with ice, stall speed can increase anywhere from 10 percent to 35 percent. What this means is a clean (uncontaminated) wing that would normally stall at 100 knots will stall when contaminated at 135 knots. For aircraft designed to take off at about this speed, the danger is obvious – angle of attack (AOA) increases as the pilot rotates for takeoff, and into that AOA range where the wing stalls with a dramatic reduction inn lift. If one wing has more ice than the other, it will stall first, leading to a sudden roll-rate at rotation.

Moreover, any significant ice sliding off the inner root of the wings on take-off can be ingested by the rear-mounted engines, as can ice being shed from the side and top of the fuselage or even accumulated in the bottom of the engine inlet. The moment of rotation is also the point at which any flexing of the wings is most likely to cause upper surface ice to break loose and be immediately sucked into the engines. The result is engine surging and a massive lost of thrust. Sluggish takeoffs, black engine smoke, alarming noises from the engines, and shaking preceding nose-drop – these characteristics are mentioned in so many of the witness accounts of these accidents.

In other words, when ice contamination is present, lift and thrust may not be trustworthy at the critical moment of takeoff.

Add one more factor: a wing with a hard leading edge. That is, one without leading edge devices (e.g., slats). Hard wings can be more vulnerable to contamination from thin ice and even from insects splattering and sticking on the leading edge. One study by experts Walt Valarezo and Frank Lynch found that “allowing initial ice formations of a size required for removal by presently proposed de-icing systems could lead to maximum lift losses of approximately 40 percent for single element airfoils.”

The advantage of multi-element airfoils (those with leading as well as trailing edge devices) is that the stalling AOA is delayed (and it wouldn’t then matter that the margin to stall buffet is a much narrower band of AOA).

Some NTSB officials have been concerned for some time that the growing popularity of regional jets (RJs), which feature hard wings for their weight-saving and simplicity, could portend a repeat in the rash of icing-related crashes involving the early model DC-9, which featured a hard wing (later models were fitted with leading edge devices). Manufacturer Bombardier‘s [BBD] popular RJ is derived from the Challenger. While the 50-passenger model features the same hard wing, leading edge slats have been fitted to the stretched variant.

No stall margin

A November 2003 study by Canada’s Institute for Aerospace Research examined the effects of ice and slush on a hard wing. The study examined the effect of ice contamination as measured by increasing coverage of the wing chord (the distance between the leading edge to trailing edge of the wing). The study found that if 30 percent of the upper surface chord is just slightly contaminated, stall speed increases 13 percent. That number is significant. The Canadian study pointed out that “modern transport aircraft are required by certification to have a stall speed margin at takeoff of about 13 percent, meaning that the aircraft’s safe takeoff speed must be 1.13 times higher than its 1G stall speed for the clean wing takeoff condition.”

“In other words, the wing would stall if the hypothetical aircraft attempted takeoff at the normal specified speed,” the report said. (ASW note: U.S. standards, 14 CFR Part 25.107(b)(2) require not less than a 1.15 margin above the IG stall speed.)

Roughness present on only one wing would result in stall and the sharp banking observed in numerous icing related takeoff crashes. The study found that aileron control (to recover from uncommanded roll) was “marginal.” The expression might be an understatement when considering that a downgoing aileron raises a wing by effectively increasing its angle of attack (i.e., in this case pitting it more deeply into the stall regime).

“Thus, a roughness-induced premature stall on only one wing would be compromised by a loss of aileron effectiveness on that wing,” the study said. The mid-range of the various degrees of roughness studied might be described as roughly equivalent to 100-grit sandpaper.

The Canadian aerodynamicists also explored contamination of 2 percent and 15 percent of the wing chord. At 15 percent chord coverage, stall margin was reduced by 10 percent. At 2 percent chord coverage, which is to say basically contamination limited to the leading edge, the effect was still pronounced – a 2� to 4� reduction in stall angle of attack, localized premature airflow separation and “significant loss of lift.”

The Canadian study looked particularly at the localized accretion of slush on the wings. “No data are available defining the conditions under which slush adheres to the wing during a takeoff run,” the report said. The evaluation found significant degradation of wing performance when contaminated with a 0.9 mm layer of slush (about 1/32 of an inch).

The airflow separation along the leading edge is particularly important with respect to the supercritical wing. For better cruise performance, these wings feature an increased leading edge radius and reduced curvature of the upper wing surface. Given this shape, the AAIB report of the 2002 accident at Birmingham said, “The flow separation from the leading edge produces a more abrupt loss of lift than for the thicker airfoil types … typical of previous, lower-speed airfoils.”

Furthermore, the AAIB observed, “Contamination on the leading edge was particularly significant.”

“Combined contamination on the leading edge and the wing and flap upper surfaces causes only slight reductions in peak lift and stalling angle compared to contamination on the leading edge alone,” the AAIB observed. What all this meant was that an uncontaminated wing designed to stall at 13.1� AOA would stall, as did the Birmingham accident aircraft, at 7.8� AOA – 5.3� lower than a clean wing.

Among its recommendations coming out of the Birmingham accident investigation, the AAIB called for a system that would directly monitor aerodynamic surfaces and warn the crew of a potentially hazardous condition. Such a direct measure of lift performance has been developed by Iowa-based AERS Midwest, Inc.. Called the System for Onboard Lift Analysis (SOLA), AERS Midwest president Steven Palmer said in a 1999 interview that the installation would be ideal for ice detection (see ASW, March 15, 1999).

Other actions to mitigate the hazard come to mind:

Flap scheduling. To increase the safety margin, have the flaps extend slightly at the point of rotation, and increase the rotate speed to 1.25Vs (e.g., 124 knots versus 112 knots). Flap extension from 20� to 30� would lower the stall speed. The flap could retract back to 20� automatically as the landing gear retracted. The idea is to reduce the possibility of one wing prematurely reaching an unknown (but reduced) stalling angle due to undetected contamination.

Slush deflection. Fit Teflon-strip nose gear mudguards and/or water deflectors. The idea is to minimize the spray of slush from the nose gear onto the wings. In addition, chined nose tires could be fitted. A chined tire features a deep center concave. It would channel the spray directly backward instead of spraying it outwards, contaminating the wing leading edges with semi-frozen slush. Regulatory authorities require that all roughness be removed from the wing prior to takeoff. However, the airplane can pick up taxiway slush and more of it on takeoff.

Detergents. Detergent washing of laminar flow wings on sailplanes is known to reduce stalling speed by a significant amount in rainy weather (by preventing the formation of draggy droplets on the wing surface). In winter, dry detergents could be regularly applied to the forward third of the wing chord, top and bottom, in order to reduce the tendency of slush to hit and stick in freezing weather. In summer, the same application would tend to reduce insect adhesion.

Priority of effort could focus first on airplanes with hard wings.

(For the AAIB report on the Birmingham accident, see http://www.dft.gov.uk/stellent/groups/dft_avsafety/documents/pdf_avsafety_pdf_0 30576.pdf; for the Canadian report on upper surface roughness, see http://www.tc..gc.ca/tdc/publication/pdf/14100/14180e.pdf)