91av

SHORTLY after leaving Rio de Janeiro on its way to Paris in 2009, Air France flight 447 flew over a band of thunderstorms. It was an unremarkable event that at first resulted only in light turbulence. But, as anyone familiar with the headlines knows, things got much worse. It ended in one of the deadliest air disasters in recent history, killing all 228 people on board. The wreckage took two years to retrieve from the bottom of the Atlantic Ocean.

When the black boxes were finally pulled out, multiple culprits were identified, including pilot error and faulty software. However, one apparently minor detail is beginning to take on new significance: ice blockage in a sensor called a pitot tube, which measures air speed.

That shouldn’t have happened at 35,000 feet. According to our understanding of weather, it is not possible for ice to form at this altitude.

Flight 447 is not alone. It is the grisliest example of a stealthy phenomenon that has been pulling planes out of the sky for the past half-century. Aerospace companies, researchers and government bodies have joined forces to try to understand a threat whose reach is only starting to become clear. As a picture has emerged, though, it has also become apparent just how hard it will be to keep aircraft safe.

Rumours of planes falling out of the sky for no apparent reason have been around since the 1950s. Some research was done back then, but seems to have been lost. It wasn’t looked into again until 1994, when American Eagle flight 4184 flew into “unknown icing conditions” over Indiana and ploughed into a soybean field at full speed, killing all 68 people on board.

“People didn’t know what was going on. There were wild theories.“

The US Federal Aviation Administration (FAA) launched an investigation, and subsequently identified 32 other incidents : a sudden engine shutdown at high altitude, weird sensor malfunctions, and a mysterious heavy “rain” spattering the aircraft’s windshield. And all despite clear skies, temperatures too cold for liquid water to exist, and with no radar indications of any weather anomalies.

“People just didn’t know what was going on,” says Walter Strapp, a cloud physicist with Environment Canada. “There were wild theories about water being wafted high up into the atmosphere.”

“The more we look for crystal icing as a culprit in crashes, the more we find it“

In most of these cases, pilots were able to restart the engines when they fell below 10,000 feet (3000 metres). The passengers survived, albeit badly shaken. In one case, the engine died entirely, but the pilot managed a deadstick landing.

The FAA’s investigation determined that the engine trouble was caused by a kind of icing – though one that doesn’t make much sense, because it affects only the hot parts of an aircraft.

In general, aircraft icing is well understood, happening when supercooled droplets of water strike the cold surface of the aircraft and freeze. This rarely catches pilots by surprise. “It’s pretty obvious,” says Dan Fuleki, who studies aircraft icing at the National Research Council Canada. “Pilots can see the ice building up on their windshield wipers.” This only happens below 22,000 feet, and freezing rain is also clearly visible on radar. What’s more, planes are equipped with sensors and their wings can spray de-icer.

None of that is true for crystal icing.

Unlike normal icing, crystal icing happens at altitudes where water should not be liquid. The culprits are plumes of crystals around 40 micrometres across, no bigger than grains of flour, which are invisible to weather radar that pick up normal precipitation.

Because they’re solid, these crystals bounce off the wings and other areas equipped with sensors and defences. But when they land on the warmest parts of the plane – such as the engine or pitot tube – they melt. This also happens in contact with the heated windscreen, where they can cause the weird “rain” sometimes reported.

Once a layer has partially melted, it accretes more crystals. When these layers accumulate in the engine, they can stall it, or break off in solid chunks that damage the engine (see diagram).

Ice build-up in the sensors leads to more insidious damage. On flight 447, the ice crystals blocked the pitot tube, which started giving false readings, showing the plane flying too slowly. This caused the autopilot to disengage, which along with other malfunctioning sensors confused the pilots to the extent that they went against their usual training. Thinking they were losing altitude, they angled the aircraft upward so severely that the plane stalled.

as the culprit, the more incidents fit the bill. A 2011 NASA report , concluding that since the 1980s, crystal icing was probably implicated in 140 other events. The frequency of the problem, they warned, is “alarmingly high”.

So it seems like high time to redesign aircraft engines to help them withstand crystal icing. That was why NASA joined a multinational programme in 2006 that included the National Research Council Canada, Airbus, Boeing and many others.

But before they could redesign the engine, they needed to understand how hot engines get iced up. That meant recreating the exact conditions in a lab at NASA’s Glenn Research Center in Cleveland, Ohio – although “lab” may be underselling a facility that can simulate the frigid temperatures, several-hundred-mile-an-hour winds and shockingly low pressures that an airliner withstands at 30,000 feet.

After years of work, and aerospace engineer Ashlie Flegel designed an array of “ice bars” that can minutely adjust the size and concentration of ice crystals spraying through an engine until they start to build up on the hot blades.

Things did not go as planned. One challenge became apparent when they turned on the ice-cloud generator. The crystals acted like a sandblaster, destroying the sensitive measuring equipment. And because the crystals also carry static electricity, even the surviving instruments misread wildly. “We had to design and build some more robust instruments,” Van Zante says.

Their work has revealed a complex pattern of ice accretion: ice crystals may shatter on first contact, for example, creating a cloud of smaller particles. Turning the lab data into a working model is going to be a major multi-physics headache. For now, they have scaled back their focus, with plans to build a simpler model that will assess the risk of icing.

“A full three-dimensional model is probably 10 years away,” Flegel says. After that, turning the model into an engine redesign will take at least another decade.

We need much more immediate help to keep planes in the sky, with some understanding of the weather conditions that lead to ice build-up in the engine.

Boeing had previously identified a possible association between crystal icing and so-called decaying convective storms. These are not showy, with thunder or lightning, but put a great deal of water into the air – up to three times as much as an ordinary rainstorm. We don’t fully understand them yet, but one thing that’s clear is that these storms require heat, which goes some way to explaining why reported incidents seem to have clustered loosely around warmer latitudes (see map).

Tom Ratvasky, also at Glenn, and his colleagues decided that the best way to get to the heart of the matter was to fly a plane directly into such conditions and see what happens. They equipped a jet with a suite of weather and moisture sensors and flew it through likely hotspots looking for trouble.

They found it. Deliberately flying the jet into so-called “high water ice conditions” (HWIC) quickly caused readings from the air temperature probe to jump from well below freezing up to zero – a sign that ice was building up on it.

Crucially, Ratvasky was able to confirm the association between HWIC and convective storm systems, and identify the mechanism: the strong updraught from the storm sweeps water vapour to much higher altitudes than normal. The effect does not seem to appear when the convection is strongest, but when the storm is dying down, for reasons we don’t yet fully understand.

The hope is that all the data gathered will yield a distinct signature that corresponds to HWIC. This will enable “nowcasting” – using satellite data to predict where aircraft are likely to encounter dangerous conditions. Researchers are also working on modifications to existing radar that allow them to detect crystal-icing conditions.

That’s a few years away, but at least it’s closer than a full engine redesign. What’s more, the team managed to establish a makeshift HWIC detection method: by pointing the aircraft’s weather radar downwards, you can see if you are flying over patches of heavy rainfall and should consider changing course.

Invisible killer

Even better, Fuleki and his team have cleverly hacked existing instruments to create two sensors. The first, a beer-can-sized device called the particle ice probe, is mounted outside the aircraft and detects the presence of small particles by the way they change the air’s electrical characteristics. It was originally designed to detect debris from the engine, but the team modified it to distinguish the particular signal of ice crystals.

The other device – an ultrasound ice-accretion sensor – . A series of dime-sized sensors sends ultrasonic pulses whose reflection changes as ice builds up. Both devices are advanced enough that Fuleki is now in talks about turning them into commercial products.

Even when planes get the new sensors and radar, however, we’re not out of the woods. For one thing, we are still not done tallying up the true toll of crystal icing. Speculation is building on its role in yet more unexplained crashes, such as that of Air Algérie flight 5017 in 2014, which killed 116 people.

We know about flight 447’s blocked pitot tube because of the flight recorder, but with some other incidents we may never know for sure. Sometimes there is characteristic physical damage, but aside from that, all traces of crystal icing tend to vanish below 10,000 feet, whether the plane survives or not.

There’s even a chance the problem could get worse. “The warmer, moister world predicted by climate change will have more convective instability, ” says Sue Gray, a meteorologist at the University of Reading, UK. “These systems will be more vigorous and more frequent.”

And according to a recent analysis weather could make the conditions in which crystal icing flourishes more frequent. In a presentation, Rory Clarkson, an engine specialist with the company, offered an inconvenient but undeniably safe answer: “Restrict operation during severe weather.”

This article appeared in print under the headline “Hot ice”