As research into the causes and prevention of aircraft icing continues, the number one defence remains crew awareness of the dangers

Icing is not a new problem. The industry’s knowledge about aircraft icing, and the ice control systems in use today, has been in place since the 1950s. Flightcrew and airline operations personnel know, theoretically, that airframe and engine icing is dangerous. Yet aircraft accidents caused by icing continue to occur.

Twin Otter Big

As operators in the northern hemisphere prepare for another winter season, research continues into the mechanics of icing and efforts are under way to raise icing awareness, but no fundamentally new approach to the problem is on the horizon. David Hammond, head of icing research at the UK’s Cranfield University, says aircraft ice-protection systems have been “very stable” for a long time, and so have the certification requirements.

Serious icing incidents have almost always involved piston- and turboprop-powered aircraft because they have less spare power than jets to produce the necessary hot engine bleed air or electrical generator output, Hammond says. But jets represent an advance in the ability to cope with airborne icing because they have the power to get rid of ice or prevent its adhesion and not because the systems they use are fundamentally different.

When jets have crashed because of ice it has almost always happened at or soon after take-off, and the cause has been ice or slush accumulation on the airframe or engines while on the ground, combined with insufficient de-icing measures before taxiing out.

“Strange as it may seem, a very light coating of snow or ice, light enough to be hardly visible, will have a tremendous effect on reducing the performance of a modern airplane.” These words, spoken at a lecture in 1939 by Jerry Lederer, founder of the Flight Safety Foundation, were echoed in February this year in a US Federal Aviation Administration airworthiness directive (AD) that said: “Even small amounts of frost, ice, snow or slush on the wing leading edges or forward upper wing surfaces can cause loss of control at take-off.”

The AD mandates tactile as well as visual pre-flight checks for the presence of ice on the leading edges of aircraft wings and tailplanes and followed three fatal icing-related accidents since 2002 involving Bombardier Challenger and CRJ-series aircraft. A factor in all three was loss of control at or immediately after take-off. The first was the January 2002 take-off crash of a Challenger 604 business jet at Birmingham, UK. Since then two fatal accidents in icing conditions – both of which occurred in November 2004 and are still under investigation – have involved a Challenger 604 on take-off from Montrose, Colorado, and a Yunnan Airlines CRJ200 just after take-off from Baotou, China.

Mandating checks

The FAA later issued another AD mandating similar tactile checks for the Cessna 208B Caravan – a high-wing, single-turboprop where the Challenger/CRJ is a low-wing twinjet. The FAA was responding to a National Transportation Safety Board (NTSB) recommendation in December 2004, which cited six Caravan accidents in the previous two icing seasons and “nine events in the last few months”.

Bombardier has been working with Canada’s Transportation Safety Board and the NTSB on whether any design or operational technique changes are appropriate, but says all three parties have decided they are not warranted. Safety manager Jim Donnelly and CRJ customer liaison pilot Mike Lohmann point out that advice about pre-flight tactile icing checks – since called for by the UK Air Accidents Investigation Branch (AAIB) and the FAA – was already in its flightcrew operating manuals, but the company has reviewed the manuals to see if there was a clearer way of presenting the information.

Meanwhile, the primary thrust of Bombardier’s remedial action is to run programmes to raise icing awareness among not just flightcrews, but operations and airport personnel including de-icing operators. The company has“been on the road”, Donnelly says, pushing its icing awareness programme to all who will listen. There is a geographical difference in the degree of icing awareness among operators, says Bombardier, but the company will not reveal where the less-aware regions are.

It is easier to call for tactile checks than to carry them out on a large aircraft in freezing conditions, warns the UK Civil Aviation Authority. For example, for health and safety reasons, tactile checks must be carried out using gloves, says CAA regional manager operations Dave Prior. But this raises the questions of how effective such a check would be, and what kind of glove would do the best job while protecting the checker’s hands from extreme cold.

Developing effective icing detectors that would eliminate the need for tactile checks is also not as easy as it sounds, says CAA research project manager Paul Spooner. The sensors need a realistic detection threshold – probably a depth of ice of about 0.5mm – but this can be confounded by roughness and variable ice distribution. Hammond believes icing detectors could only be mandated widely if they were more reliable than tactile checks, and at present they are not.

Nasa-Turboprops-Icing big

Stall warning

Aircraft stall warning systems – like stick-shakers and stick-pushers – are calibrated to activate at a certain angle of attack, but the presence of ice on the wing can cause it to stall before that angle has been reached, Hammond says. On some turboprop aircraft, however, pilots can alter the stall warning speed if the presence of ice is detected, using an “ice-speed” button to reset the stall warning speed upward.

It seems the only reliable answer to aircraft icing on the ramp at present is “don’t think twice, de-ice” – the slogan of the CAA’s current icing awareness campaign. Prior says the history of icing accidents at take-off proves that the decision whether or not to de-ice – or to de-ice a second time if the aircraft does not take off quickly after its first de-icing – is critical. Both the CAA and Bombardier say continued work on improved de-icing fluid specification is needed, and a “Type 3” fluid is under evaluation.

Research at Cranfield – in association with the USA’s NASA and Italy’s CIRA – is aimed developing better mathematical tools for modelling – and therefore predicting – the behaviour of ice when it forms on an aircraft. Central to this is windtunnel research into the behaviour of large supercooled water droplets (SLD) when they hit an airframe, be it the wing or any other part of the aircraft.

Supercooled droplets are just below the 0°C freezing point, but have not yet changed from the liquid state into ice. Striking an airframe makes the change take place almost instantly, causing ice to spread over the forward edges of all surfaces, adding weight and degrading lift. This is the most dangerous source of ice formation during flight because it can erode an aircraft’s performance rapidly. And one of the characteristics of this kind of icing is that it builds up faster and more readily on thin or sharp leading edges than on thicker or more rounded ones.

Cranfield has a vertical windtunnel specifically for icing research, and both NASA and CIRA also have specialist icing tunnels, the most modern and largest of them being in Italy.

The ideal is to be able to predict the way in which ice from supercooled large droplets will form on an airframe. This entails working out the area it will cover, the shapes it will form, and the aerodynamic effects of the accretion.

Prediction tools

NASA says the goal of the multinational SLD research programme is “to develop validated tools for predicting ice growth, ice protection systems, and the effects of ice contamination for design, analysis and certification”.

Says Cranfield’s Hammond: “The basic thrust of our [ice-accretion modeling] research comes from the inexactitude of the science for detecting and removing ice.” One of the reasons more work is needed, he says, is because it is “very difficult” to produce SLD behaviour in a windtunnel that compares with the way the droplets act in the natural environment. Accuracy is essential to enable the formulation of a reliable mathematical model that will enable engineers and aerodynamicists to predict how a wing should be designed and what kind of anti- or de-icing system it should have.

Validation of the models is one of the tasks NASA’s Glenn Research Center carries out. Based in Cleveland, Ohio, NASA Glenn is the home of the agency’s icing tunnel, but also operates a pair of aircraft that carry out airborne research. One is a twin-turboprop de Havilland Canada Twin Otter, and the other a twinjet – a former US Navy Lockheed S-3 Viking. Both are used to fly in icing conditions, instrumented to validate the ground-based experimental and computational research as part of the multinational SLD programme.

The 1994 Roselawn, Indiana crash of an American Eagle ATR 72 underlined the need for SLD research, the aircraft’s control-surface aerodynamics becoming so badly affected by ice accretion that the crew lost control. The aircraft was flying a holding pattern, operating in icing conditions the NTSB described as outside the design specification of the aircraft. During the hold, a ridge of ice built up behind the wing leading-edge de-icing boots. The resulting aerodynamic effect caused an aileron hinge-moment reversal and wing drop from which the crew failed to recover.

The NTSB report on the Roselawn crash said the icing took place in conditions that subjected to the aircraft to impact by SLD, but concluded that knowledge of the behaviour of the supercooled droplets on impact was not sufficient to determine precisely what occurred. One of the many problems with investigating icing incidents that occur at altitude is that the ice formed often no longer exists when – or after – the aircraft hits the ground.

Much of what happened in the Roselawn crash was deduced by reproducing the conditions in the windtunnel, and validating the results in the air. The NTSB found that supercooled droplets of a certain size accumulated on the wing further back than had been anticipated at certification, hence the eventual FAA AD requiring the area covered by the leading edge de-icing boot to be extended further aft on the wing upper surface.


NASA says the ice build-up that caused the aileron hinge-moment reversal experienced by the ATR 72 at Roselawn can, in any turboprop aircraft type given the worst circumstances, cause elevator hinge-moment reversal if icing on the tailplane forms a ridge aft of its de-icing boot. That can cause the tailplane to stall, causing the aircraft to pitch nose-down and requiring a counter-intuitive movement of the elevator downward to unstall the tailplane, possibly combined possibly with an elevator hinge-moment reversal that may be too strong to overcome.

Engine icing is also a serious problem. Finnair says it is one of the relatively few factors that can cause simultaneous failure of all engines for the same reason. In jets the inlet and spinner – and the inlet guide vanes on engines that have them – are protected by anti-icing systems. But these measures alone are not enough, under certain circumstances: on a single day in December 1998 at Oslo Gardermoen airport, Norway, SAS Scandinavian Airlines suffered severe icing damage to five engines in its fleet there, and Braathens saw twice that number of engines badly damaged. The prevailing weather conditions were freezing fog and drizzle – classic conditions for airframe and engine ice accretion.

Airframes can be de-iced on the ground at the airport and they also have anti-ice systems, but engines are different. The only protection against ice that accumulates on engine fan and compressor blades is centripetal force, which throws ice off before it forms properly. But, Finnair warns in a presentation for pilots: “Unfortunately, no engine on the market is protected against [fan blade] icing at idle power [as used during taxiing]. Periodic run-ups are needed, but sometimes are impossible to carry out because of slippery taxiways and the risk of damage to aircraft behind you.” Now, for environmental reasons, pilots are forbidden to carry out run-ups during taxiing or to conduct the formerly recommended 15s power increase before releasing brakes for take-off. Finnair’s presentation concludes with the observation that “some days are no-go days”.

Narrow escape

Earlier, in December 1991, an SAS Boeing MD-80 had a narrow escape from a different consequence of wing icing. The aircraft had arrived from at Stockholm Arlanda airport, Sweden from Zurich, Switzerland in temperatures of 1°C and was parked on the ramp overnight. When it took off early next day in temperatures of -1°C, there was some clear ice on the wing upper surface, particularly near the wing root, despite de-icing.

When the wing flexed on rotation, ice broke off and entered both engines, damaging them and causing them first to surge and then to fail. The pilots carried out a forced landing on fairly flat ground and, although the aircraft broke into three sections, there was no fire and there were no fatalities. The accident report ruled: “The accident was caused because SAS’s instructions and routines were inadequate to ensure that clear ice was removed from the wings of the aircraft prior to takeoff.” Checks had been done on the leading edges only.

As a result of experiences like these, SAS Ground Services Norway is deploying a new kind of ground-based de-icing system at the airport. Made by Radiant Energy of Ontario, Canada, the InfraTek system uses infrared energy to disperse ice rather than the traditional method of spraying de-icing fluid over the airframe. Gardermoen is not the first to install the sytstem – New York Kennedy did last year and New York Newark installed it in 1999. But the SAS-operated system will be the first InfraTek installation at a European airport, and it will be capable of de-icing aircraft up to the size of a Boeing 757. Larger installations can handle 747s, says Radiant.

For all the research into ice-accretion patterns that is going on, the answer to the icing risk for now remains icing awareness by flightcrew and operations personnel. Formed by the FAA and European Joint Aviation Authorities, the Ice Protection Harmonisation Working Group is attempting to ensure that the awareness programmes and the research complement, rather than duplicate, each other. Within a month, the UK CAA says, it will be publishing a web-based icing training tool aimed primarily at business aircraft operators.

The consequences of a lack of icing awareness were illustrated as long ago as January 1982, when an Air Florida Boeing 737-200 leaving Washington National airport could not maintain height after take-off and crashed into the Potomac river. The probable cause was “the flightcrew’s failure to use engine anti-ice during ground operation and take-off, their decision to take off with snow/ice on the aerofoil surfaces of the aircraft, and the captain’s failure to reject the take-off during the early stage when his attention was called to anomalous engine instrument readings.”

Icing ignorance

The report provides clear evidence that the crew were not icing-aware, citing “the prolonged ground delay between de-icing and the receipt of ATC take-off clearance during which the aircraft was exposed to continual precipitation; the known inherent pitch-up characteristics of the Boeing 737 aircraft when the leading edge is contaminated with even small amounts of snow or ice; and the limited experience of the flightcrew in jet transport winter operations.”

While efforts are still under way, two decades on, to ensure icing awareness among flightcrews and operations personnel, the next challenge for the researchers, predicts Cranfield’s Hammond, will be to reduce the energy needed for ice-protection systems because of the hike in fuel prices.


These boots were made for de-icing

De-icing boots are made of flexible material attached to – and flush with – the leading edge of an aircraft’s wing. A continuous series of pulses of air bled from the engine compressors are routed to the underside of the boots, causing the surface to deform with each pulse, then revert to its resting position between pulses. This breaks off ice as it forms, and requires less energy than the electrically or pneumatically heated surfaces of anti-icing systems that prevent adhesion in the first place. Turboprop engines cannot generate sufficient spare energy to support either of the latter. Neither the de-icing boot nor direct-heating anti-icing systems are new – they existed in the Second World War.

More recently – in 2001 – NASA hailed a hybrid combination of thermal anti-icing for the leading edge and mechanical de-icing for the upper aerofoil surface behind it. The system, dubbed EMEDS (electro-mechanical expulsion de-icing system) was developed by Cox and Co of New York, and was subsequently certificated for the horizontal stabiliser of the Raytheon Aircraft Premier I business jet. NASA says EMEDS uses less energy than a fully thermal system. Behind the heated leading edge an actuator system “periodically deflects the [upper] wing skin” to break off ice forming there.

Another mature system is a leading edge of porous metal, with thousands of minuscule holes in it, through which de-icing fluid is forced at a pressure that can be varied according to the severity of the icing conditions. Austria’s Diamond Aircraft is promising full icing protection for its DA42 Twin Star four-seat piston twin this year using such a system, adding all-weather capability to the European instrument flight rules certification it achieved last December. This degree of protection is extraordinarily unusual in a piston-engined aircraft. The company says it had struggled with alternative icing protection systems before coming up with the solution. The titanium leading edge panels have 300 laser-drilled micropores per square centimetre, through which the fluid is expressed under pump pressure for de-icing, or released without pressure for anti-icing, giving “several hours” of de-icing or anti-icing in severe icing conditions.

Predicting and avoiding

The best way of countering the airborne icing risk is to avoid entering the airspace in which it occurs. The traditional way of doing this is for the pilot to use make use of upper level forecasts – a relatively crude tool – and the aircraft’s weather radar if it has one, as commercial transport category aircraft do. Radar is not a perfect indicator. It shows the presence of precipitation, although not necessarily the type nor its temperature. It could be rain, snow or hail. The forecast and the outside air temperature can provide a rough guide as to which it might be.

Hail can damage an aircraft and engines, and rain or snow can cause severe icing depending on its temperature and state. Supercooled large water droplets are most feared.

NASA is working on remote sensing systems, both ground-based and airborne, for locating icing in three dimensions and providing the information to pilots via an accurate, intuitive display showing its position relative to the aircraft (see below). The agency admits such a system is a long way from entering operation, referring to the concept at present as “notional”. As well as the need to determine an effective sensing technology, NASA admits that “currently, aspects of the structure of the icing environment are not well understood”, and it is carrying out research to learn more about it.


Source: Flight International