Two recent accidents related to encounters with in-flight icing while operating an airplane not equipped or certificated for icing highlight some of the inherent risks associated with such encounters. In the first, a private pilot operating a Cirrus SR22 on an instrument flight plan encountered ice and elected to divert. He failed to capture the localizer on the first attempt, and during the subsequent radar-vectored maneuvering for a second attempt, appears to have become spatially disoriented and crashed. In the second, a commercial pilot operating a Beech G33 Bonanza on an instrument flight plan encountered icing and obtained a clearance to descend to 3000 feet. At this altitude, the pilot reported that he was below the cloud layer and he believed he was below the freezing level and that the ice was shedding. However, the airplane continued to descend until impact. Investigation at the site identified substantial chunks of ice conforming to the airfoil shape. No one survived either accident.
In a study of icing accidents that I presented as a paper for the American Institute of Astronautics and Aeronautics in 2006, I identified 142 events in which the pilot made the decision to land due to ice accumulation; in 84 of these, the decision was made before any aerodynamic consequences had been encountered. In only 23 of these 142 cases was a successful precautionary landing made. In the remainder, aircraft damage and/or personal injury or fatality was the result. This has to be tempered with the knowledge that there may be hundreds of cases in which a successful diversion was executed without any type of report generated, whether that be an official investigation or an ASRS report. That said, these results strongly suggest that the default strategy of simply diverting and landing when ice is encountered may not be as reliable as we commonly believe. In many of the cases that I have studied, the diversion ends with a hard landing and structural damage. In some, it results in an off-airport landing; in the worst cases, a loss of control.
In the first accident cited above, the role that icing played could not be determined, beyond triggering the decision to divert. It is likely that the pilot was not able to program the aircraft flight management system either properly or quickly enough in the developing situation. It remains possible that the effects of ice accretion hampered the autoflight system, and/or the manual handling characteristics of the airplane with ice accretion created a significant distraction. At the very least, he was not prepared to execute the diversion.
In the second case, the pilot reported outside temperatures of 34 degrees F and later 39 degrees F to air traffic control after he had descended below the clouds. The freezing level was forecast to be pretty much right at 3000 feet, and the surface temperatures at the nearest reporting site, 43 miles away, were dropping and recorded at 3 degrees C around the time of the accident. The presence of ice at the crash site hours after the accident seems to indicate that the pilot’s belief that he had descended to warmer air was probably incorrect, and the accuracy of his OAT gauge must be considered. Sadly, the Board did not address this issue in the report; such an investigation could have yielded important information. In any event, when reaching air that is marginally above freezing, the shedding of ice may take some time. Indeed, for many years Boeing has required icing penalties applied to landing and go-around performance if ice has accreted on the airframe at any time during the flight and the forecast temperature at the landing field remains at or below 10 degrees C. When he was barely one thousand feet above the ground, the G33 pilot told ATC that he was “doing okay right now” and “waiting for this ice to dissipate…” three minutes after initiating the descent and less than two minutes before he crashed.
The first step in dealing with an icing encounter is usually to try to find warmer air. To do so, the pilot has to be intimately familiar with the atmospheric conditions. Had the G33 pilot reviewed the skew-T charts, forecasts of winds and temperatures aloft, or the Aviation Weather Center’s freezing level graphical presentation (no record could be found of a weather briefing), he would have perhaps realized that there was very little room between the freezing level and the ground. This is particularly important because of the time it may take the ice to shed. With diminished performance and not much altitude, that time may exceed the time available before a landing is required, intentional or otherwise. Knowledge of the forecast air temperatures and freezing level will also serve as a check against the OAT gauge; a couple of degrees variation would not be surprising in either the forecast temperatures or the OAT reading.
If the ice cannot be shed, an immediate diversion is absolutely essential. Unfortunately, one of the standard tropes of aeronautical mythology, namely the idea that the effects of icing are cumulative, can grossly distort the urgency and the need for aggressive action and precise execution.
For decades, the FAA published a cartoonish diagram showing how the four forces (lift, drag, weight and thrust) incur degradations which, when added together, result in a “cumulative” effect. The idea is fundamentally correct, and I’ll discuss that in a moment. However the word “cumulative” suggests linearity, in other words proportionality. In combination with the existing terminology used to describe icing severity (trace, light, moderate and severe) “cumulative” can be interpreted to mean that a little bit of ice is probably okay, a little more is not okay, a bit more than that is a problem, and any more beyond that is a serious threat. Many pilots interpret their own experience within this paradigm, i.e., their airplane “can carry a lot of ice.” This is almost universally incorrect.
In approximately 90 events in my research, I was able to find estimates of ice thickness, either from the pilot or from the investigators on the accident site shortly afterwards. These estimates are highly subjective and not necessarily accurate for the actual time of the event. However, within that set of 90 events, the median of minimum thicknesses was one half inch, and the median of maximum thicknesses was five eighths of an inch. Bear in mind that the term median means that fully one half of the events exhibited lesser thicknesses.
We have known for decades that very small accretions of ice can have serious aerodynamic effects. The most commonly discussed effect is the loss of lift and increase in stall speed. However, the increase in drag is often poorly understood, particularly how it may literally skyrocket when the angle of attack is increased during the landing flare. These effects can be experienced with very thin ice accretions; their precise aerodynamic location on the wing, surface roughness, and even horn angle are crucial factors that cannot possibly be estimated by visual inspection during flight.
Even less commonly discussed is the loss of propeller efficiency. The prop is a thin surface; it is likely to accrete ice earlier than the wing. You can’t see the ice on the prop, but in the end it is still an airfoil and ice will destroy its ability to generate thrust no matter what the RPM or manifold pressure. In this way, these effects are additive (a word I much prefer over cumulative), but individually, none of these effects are linear. Aerodynamics is instead a nonlinear science. Turbulent flow is actually the mother of all nonlinear problems. Small changes in ice shape, roughness, angle of attack and/or load factor can have very non-proportional effects.
What often is linear are the degradations you experience with ice accretion up to the “cliff,” the unknown, almost impossible-to-predict point at which you encounter dramatic nonlinearity and the lift or drag curve simply goes straight south. This can further reinforce the idea of linear effects, particularly if you have never had the pleasure of sailing over the cliff. You notice a speed decay, but have no idea that the angle of attack associated with the next two knots of speed decay will result in a stall without warning. You’re doing fine on the approach, and have no reason to believe that you will fall out of the sky when you flare; but many pilots do. Even Cessna managed to prang a 208 when landing following a test flight with artificial ice shapes attached to the wing. (Meanwhile the idea that ice increases weight is simply absurd unless you are a Zeppelin commander; the amount of weight added by a good coating of ice is almost certainly less than the weight of the fuel you have burned since you took off.)
The next formidable problem to be faced when diverting due to ice accretion is the additional ice you will accumulate during the diversion. A little bit of ice can be quite dangerous, but more ice is never better. You may feel that you have made a good, proactive decision to divert; all of the cases I cited in the 2006 study fell into this category. By the time you actually get to your diversion airport, you may be in much more trouble than you had been to start with.
The next problem is setting up and executing the approach. Even in the large jets that I fly with a two-man crew, a diversion is a very task-saturated event. When diverting an ice-contaminated airplane, you are going to land an under-performing airplane at an unfamiliar airport, probably in very much less-than-optimal weather conditions. You will be aware that time is of the essence, and you may be prone to rushing. In many general aviation cases, you may be faced with a need to circle after executing the approach, and a circling maneuver with ice on the airplane can be a potential death sentence. One of the earliest comprehensive investigations of an icing accident involved a United Airlines DC-3 that crashed while circling to land at Chicago Midway Airport in 1940. Nothing has changed since, and many of accidents I have studied involve an attempted circle to land. Managing the angle of attack and the load factor is absolutely essential.
The key here is to be prudent, and assertive, in placing the airplane in a position from which a stabilized approach to landing can be made. It won’t do to be high, or sloppy on the localizer, or not configured until the last second. Note that a stabilized approach can be flown at higher speeds, and at any flap configuration, so you have options to fly a faster-than-normal approach due to ice. Indeed, you absolutely must be aware of and utilize the manufacturer’s recommended speed additives, and configurations, for ice on the airframe. The NTSB commonly publishes the section of the POH describing the minimum icing speeds, as a supplement to the report in which the pilot is quoted, from his or her statement, as having flown ten or fifteen knots slower than those minimum speeds. But the airplane has to be stabilized by the final approach fix, or one thousand feet during a visual approach.
That said, some marginal instrument approach procedures, even when flown perfectly, may leave you in a position from which substantial maneuvering is required to get to the runway. This opens you to the same threats as a full blown circle-to-land approach. Manage the load factor.
Most of us understand the idea of avoiding known icing conditions in airplanes not certificated for flight in icing. Most of us also realize that, if you are going to fly IFR in such airplanes, sooner or later you will encounter some ice. The upshot to this is that you must be thoroughly familiar with the exit strategies; know the freezing level and know your enroute alternates. Study the skew T’s (and learn how to read them, and consider how long it has been since that last sounding). Study the winds and temperatures aloft; study the graphical forecasts for icing and freezing level available on the Aviation Weather Center site or in whatever weather service you subscribe to. Study the approaches to your possible enroute alternates; know which ones you don’t want to try with ice and which you do.
As soon as you make your decision to descend to warmer air, or to divert, you need to be on your toes. Get way ahead of the situation and stay there. Know your automation and flight management system cold, and know how to make it do what you need it to do. As a general rule of thumb, it is wise to disconnect the autopilot as soon as you start to accrete ice. The autopilot will mask handling issues until it simply gives up and hands you a very squirrelly airplane. This means that you must be proficient in hand-flying, bearing in mind once again that you will be in a very high-workload environment. Everything you can do to reduce the workload, in terms of planning, review and familiarity with systems, terrain and airport facilities will be of enormous help.
Manage the angle of attack and load factor very cautiously. Go for the stabilized approach, and be prepared to land with power, as the drag rise in the flare may be quite significant. Put the airplane on the ground by the thousand foot markers. Be prepared to land off the airport while you still have control if the airplane just ain’t doin’ it. Always remember that the safety of yourself and your passengers is all that matters; never try to protect your airplane when performance is marginal and disappearing.
In January of 2016, a flight instructor giving instrument dual instruction to a PA-46 pilot executed a forced landing off the airport after encountering freezing rain following a practice missed approach. Despite correctly operating all of the aircraft ice protection systems, they were unable to maintain altitude enroute back to their home airport. They walked away.