This type of icing does not appear on radar due to its low reflectivity. Neither airplane ice detectors nor visual indications reliably indicate the presence of ice crystal icing conditions. It is often undetected by the flight crew and has caused many high-altitude engine failures.

Many business jets have the capability to climb quickly into the mid-40 flight levels and cruise far above most weather. It can be tempting to sit back, enjoying the generally clear skies at these altitudes and taking relief that the weather below us can’t hurt us.

Unfortunately that “comfort zone” was temporarily burst on Nov. 28, 2005, when the dual-engine flame-out of a Beechjet rudely awakened the business jet community. Answers to important questions were not readily available in the immediate aftermath. Many of us wondered what could have caused two engines to simultaneously fail. Were these failures limited to the Pratt & Whitney Canada JT15D design, or could this happen to other engines?

According to NASA scientists Harold E. Addy Jr. and Jospeh P. Veres of the Glenn Research Center in Cleveland in “An Overview of NASA Engine Ice-Crystal Icing Research,” (NASA/TM-2011-217254, November 2011), “It is a problem whose frequency is alarmingly high…. Evidence indicates that engine icing incidents caused by ice accreting inside the core of jet-based engines have been occurring for over two decades.”

The Engine Harmonization Working Group (comprised of subject matter experts from within the aerospace industry, government and academia) has identified over 140 engine power loss events due to engine core icing since the early 1980s. Fourteen involved multi-engine flameouts. The pilots of the Beechjet event on Nov. 28, 2005, happened to be the unlucky ones who were unable to restart the engines, but they were able to successfully accomplish a dead-stick landing. Why haven’t you heard about the others? Those (thankfully) ended with a resumption of engine power at lower altitude.

One of the perplexing issues that confronted investigators during interviews with incident pilots was that the conditions at the time of the engine loss were “routine” and not perceived as a threat due to lack of airframe icing and only moderate turbulence. NASA’s Addy and Veres believe “It may well be that the difference between conditions that are hazardous and those that are not, is not discernable to the pilot.” The incidents further stymied investigators because they occurred outside of flight conditions that are currently defined by regulatory authorities as “icing conditions.”

The clues to solving the Beechjet mystery began with propulsion engineers who had worked a similarly mysterious problem in high bypass ratio turbofan engine powerless events in regional aircraft during the mid-1990s. The engines experienced uncommanded thrust reductions when cruising in the vicinity of major thunderstorms at altitudes between 28,000 ft. and 31,000 ft. All incidents occurred under IMC in cloud, at thrust levels between 90% and 100% max continuous thrust (MCT) with precipitation reported. The uncommanded thrust reduction manifested itself initially by a gradual decay in fan speed, an increase in turbine gas temperature (TGT), and a failure of the engine to respond to pilot commanded thrust changes. Thankfully, in these events, as the aircraft descended to a lower altitude (usually around 10,000 ft.), the ice blockage released, making the engine recovery process possible.

Addy and Veres point out that “The complex, aero-thermodynamics involved that permit ice to accrete inside the core of an engine in flight are not understood to a level that allows effective analysis and prevention or mitigation techniques to be employed in a robust manner.” Thus it took immense engineering and investigatory research to discover it was possible for ice to accumulate on the second stage stator inside the engine core passage without the presence of significant supercooled liquid water in the air.

Since each engine has its own unique design characteristics (centrifugal flow compressor versus axial compressor, just to name one of many differing variables), the actual mechanism for engine power loss may vary from type to type. Each engine’s overall stability is a balance between compressor stability, combustor stability and the fuel available for acceleration. The component with the least margin to cope with ice ingestion will be the weakest link. The common feature appears to be the initial accumulation of ice crystals on relatively warm surfaces in the forward part of the engine, followed by detachment into the airstream flowing through the engine core. In general, the types of engine power losses caused by ice crystal icing are categorized as engine surges and stalls, flameouts and engine damage.

Ice shed into the compressor can drive the engine into stall due to the combined effect of several mechanisms (lost inertial and heat energy to the ice), as well as the inefficiency of airfoils with ice on them. The chain of events begins with a sudden flow reversal (compressor surge) followed by engine rotor speed decay (engine “stall”) as airflow is reduced due to the presence of airflow separation in compressor stages. The combustor remains lit; however, due to lack of airflow, the EGT typically rises quickly.

Flameout occurs due to quenching of the combustor following the ingestion of ice. Engine damage happens when engine blades and vanes are impacted by shed ice. Minor blade-tip curl has occurred. Rare instances of blade release have occurred.

Are there any commonalities between the engine power loss events? An interdisciplinary team of Jeanne G. Mason (senior specialist propulsion engineer, Boeing Commercial Airplanes), J. Walter Strapp (physical scientist, Environment Canada) and Philip Chow (senior principal engineer, Honeywell International) evaluated data from 46 engine power loss or core damage events, which included nine different aircraft types (large transport and commuter jets) and eight different engine types. These are not all of the engine power loss events thought to be attributable to ice particles; however, insufficient event data was available for other events to draw significant conclusions. The data show that both older generation turbojet engines and newer generation high bypass ratio turbofan engines can be affected by ice particle icing. Their report titled “The Ice Particle Threat to Engines in Flight” was presented at the 44th American Institute of Aeronautics and Astronautics (AIAA) Aerospace Sciences Meeting and Exhibit in January 2006in Reno, Nevada.

There were a number of common observations in nearly all of the incidents. According to the report, “The engine power loss events occurred in three phases of flight: climb (one event), cruise (17 events) and descent (28 events). The predominance of events during the descent phase of flight can be explained by a combination of two factors. First, engine events related to icing must be colder than the freezing level, and therefore tend to be at higher altitude. Second, compared to climb at high power, during descent the engine throttle is at the idle position, and airflow is at a minimum. The engine’s capability to tolerate ice particles is related to the airflow, and it decreases as density decreases with altitude and decreased power (reduced airflow), where ice particles can constitute a greater portion of the total airflow in the engine. Also, at low power the metal temperature in the compression section is lower than at high power, making it more susceptible to ice accretion.”

Generally the events occurred from ISA+10C to ISA+20C. In fact, most of the events occurred outside the FAR Part 25 Appendix C envelope for engine certification in icing conditions. Aircraft were in the vicinity of convective clouds/thunderstorms, although flight crews reported no flight-radar echoes at the altitude of the event. Precipitation in the form of “rain” was noted on the windscreen, which at first perplexed investigators because the events occurred at altitudes far higher than where supercooled raindrops would exist. No airframe icing was noted. It has since been determined that the “rain on the windscreen” was actually the melting of the high-altitude ice particles.

Events commonly occurred while diverting around a flight-level high reflectivity region associated with an isolated thunderstorm core, as well as in the broad anvil outflow regions from clouds associated with convective storm complexes and tropical storms. Overshooting tops (dome-like protrusions from the top of an anvil cloud) are an indicator that significant convection is occurring and that ice crystal icing may be possible. Downwind from the tops of large areas of convective clouds, which are often signified by the visible anvil shape, is the main risk area for encountering high crystal concentrations.

Satellite data confirm the existence of high concentrations of very small ice crystals in the vicinity of convective weather systems. Convective storms in the tropical latitudes contain much more moisture due to the warmer air in these storms. Why are these more likely to exist in the tropical latitudes? Because warmer air can “hold” much more moisture. In fact, these strong convective systems produce cloud tops that have been observed to burst through the tropopause.

How small are the particles? On the order of roughly 40 microns in diameter, and even in high concentrations, these are not visually detectable even in daytime conditions. Can radar be used to detect these zones? Unfortunately not, for several reasons dealing with the limitations inherent with radar returns. The temperatures at the altitudes of these events are far too cold for supercooled liquid water to exist. With a radar reflectivity of only 5% of average size raindrops, there may be little radar reflectivity at flight altitude above the minimum threshold of the pilot’s onboard weather radar. Secondly, radar returns are highly skewed to the large particles in the distribution, not necessarily where the mass (of ice particles) is concentrated. Today’s pilot is not typically on guard for these conditions because our training has been focused on identifying conventional icing and storm targets that provide strong radar returns.

Note there is a clear distinction between the high concentrations of very small ice crystals that cause this engine problem versus the collection of larger crystals (at lower densities) seen in high-level cirrus, cirrostratus and cirrocumulus clouds. The latter are not hazardous.

According to NASA’s Addy and Veres “Hazardous events such as these will continue to occur unless more-robust means to address the problem are developed.” This will require key technology areas to be developed in order for the engine core icing hazard to be effectively addressed. Existing engine core icing test facilities consist primarily of outdoor, ground-based facilities that are limited in their range of ambient air temperatures and pressures. Very few engine core icing test facilities have the altitude capability for this type of testing.

A small group of engine and airframe manufacturers has formed a consortium called the Ice Crystal Consortium (ICC) to pool resources to address the problem. Research sponsored by the ICC is proprietary and not available to any government agency or any company outside the consortium.

The National Research Council of Canada has capabilities in several areas, to include ice crystal aero-thermodynamic tests, small engine altitude tests and engine ground-level tests over a range of thrusts. Some of the unknowns in this problem are the water content, ice crystal size, air temperature, altitude and horizontal extent conducive to ice crystal icing. Until this can be better defined, we’re really just taking educated guesses for design criteria and can’t make the proper changes to the engine icing certification envelope. In cooperation with NASA, The National Research Council of Canada are developing new instruments needed to help characterize high ice-water content environments.

These will not be easy (or cheap) tasks. Costs of experimental research are high and resources are scarce. New instruments are under development to accurately measure high ice-water content (HIWC) conditions. The ice crystals in HIWC environments have eroded instruments to the point of inoperability, similar to sandblasting. The ice crystals generate the buildup of static electricity, rendering them inoperable. Ice crystals “bouncing” off instruments can cause an artificial concentration. Once these data are collected it will be time-consuming to analyze and ensure their accuracy.

One engineering solution that appears partially effective in preventing engine power loss from ice particle icing focuses on removing the accreted ice from the engine core when it sheds. The intercompressor bleeds, if open during an ice shed, can effectively remove accreted ice upstream of the core. However, during cruise power these bleeds are usually closed for performance and fuel economy reasons. When ice sheds at cruise power with closed bleeds, the ice will pass through the high-power compressor and combustor, posing the threat of surge or flameout.

Until all engines are redesigned and recertified for this yet-to-be-fully-defined problem, it is necessary for pilots to follow interim operational guidance. According to Mason, et al, “The tools available to the pilot of a commercial jet to identify regions of potentially high ice particle concentration are limited. Onboard weather radar does not show significant returns for these events at altitude. It is prudent to assume that flight in the close vicinity of a thunderstorm may lead to high IWC encounters that can lead to engine events. However, it is also clear from the event database that this model is not sufficient as a general avoidance strategy. Incidents have occurred in regions of multiple convective cells with merged anvil regions and where there was no high-reflectivity core at flight altitude.”

Avoidance of high concentrations of very small ice crystals requires effective utilization of the aircraft’s weather radar. The high reflectivity below the aircraft from rain returns associated with these cells may be a good indicator of high ice particle concentrations aloft since that rain would often have formed from falling ice particles. At high altitude, the pilot must tilt the radar down to scan for high reflectivity rain below to determine the existence and position of a convective cell, and gauge the altitude of the high-reflectivity region. However, the height of the cell above this region, if not visible, may only be inferred.

If the top is not visible, it is prudent to conclude that it exceeds the aircraft cruise altitude, and the cell should be avoided by circumnavigation. Also be advised that high-altitude ice crystals may be present for some time after the active convection that produced them has begun to decay, creating a prolonged hazard.

Pilots are advised to avoid reflective regions by more than the typically recommended 20 nm from areas where large convective cells are present. (In engine types that have been identified as at risk while pending modification, a distance of 50 nm from such areas has been recommended). Pilots are advised not to overfly convective cells. Flight upwind of the cell is recommended to avoid the spreading anvil downstream and to limit exposure to high ice particle content conditions. Use of a thrust setting above flight idle during initial descent from high-altitude cruises in the tropics is a prudent precaution.

The seriousness of this issue caused the FAA to issue immediate adoption of an Airworthiness Directive on Boeing 747-800 and 787-800 aircraft without the normal notice and comment period prior to adoption due to recent engine power losses from ice crystal icing. Investigation officials are currently exploring whether it played a role in the fatal crash in northern Mali of a Boeing MD 83 operated for Air Algerie by Spanish operator Swiftair while en route from Ouagadougou, Burkina Faso, to Algiers on July 24, 2014. (See sidebar on next page (52) for further details.)

The ice crystal icing problem must be addressed in training. Ground school training needs to explain the factors involved, the warning signs and the conditions to avoid. This is a technically complex topic and, quite frankly, this author’s experiences listening to well-meaning but not “technically savvy” ground instructors butcher basic aerodynamics leaves me skeptical that the topic can be adequately taught. This is worthy of an industry-wide training product, much like the CFIT training videos, in order to assure proper content explanation and standardized information across the industry.

Would this be worthy of valuable time in a simulator? That is a good question to consider. There are many scenarios worthy of valuable simulator time, and it needs to be balanced with the type of operations you fly. If you fly into the equatorial climates where the ice crystal icing threat seems concentrated, then the incident statistics clearly justify dedicating time to this.

And, by the way, this is a scenario that is quickly life-threatening due to losses of pressurization, thrust, electrics, hydraulics, and being basically down to just the avionics powered by the battery bus, not to mention keeping control of the aircraft and making important decisions on where to point the aircraft for the “safest” landing option. If, on the other hand, you fly a business jet that spends most of its time jetting into mountainous airports such as Aspen-Pitkin (ASE), the unique risks of mountain bowl approaches would dictate focusing your training on those scenarios instead. Clearly much remains to be done in terms of engine modification to prevent these incidents in the future. Until then, enhanced pilot awareness and giving wide berth to suspect areas is the only protection we have.