< Previous38 ICAO JOURNAL – ISSUE 1 2017XXXXXAircraft and system design and operational procedures still have not totally conquered in-flight icing problems. Flying in icing conditions continues to result in incidents and accidents, with no aircraft type, size, or configuration immune. The availability of advanced simulation tools such as 3D Computational Fluid Dynamics (CFD) can speed up the aircraft certification process, fill important gaps, predict what will eventually be seen in flight testing, and increase safety.Despite considerable advances achieved by 3D CFD simulation of in-flight icing in the past 10 years, the aircraft certification process has failed to adopt them and remains almost at a standstill using technologies from three decades ago. The tools accepted by airworthiness authorities for certifying aircraft for in-flight icing seem to have remained almost frozen in time.CFD tools would permit a more efficient and safer certification methodology for all types of aircraft by reducing the likelihood of ice-induced hazardous events in operation. Aerodynamicists can use CFD to examine and anticipate in-flight icing scenarios and situations that cannot be tested otherwise (with dry and wet icing tunnel testing or flight testing with artificial ice shapes), are too dangerous, or are too difficult to verify by means of natural icing flights. CFD can also play scenarios at the “white sheet” aircraft design phase.ICE PROTECTION CHALLENGEWhen the surface area of an aircraft in a cloud surface strikes supercooled liquid water droplets (the temperature of the droplet is below the freezing point, but the water is in the liquid phase), the droplets form ice – whose shape, location, roughness and dimension can lead to unexpected surface roughness and distortions in the aerodynamic profiles of lifting surfaces, control surfaces, air intakes, fan blades, rotors and propellers. Performance degradation can then occur from a combination of increased drag, reduced stall angle of attack and reduced lift. Higher weight is of secondary significance. Moreover, ice can block engine inlets and internal ducts, distorting flow. If ingested or released, ice may damage engine components, causing power fluctuations, thrust loss, rollback, flameout and loss of transient capability. Asymmetric ice distribution can also cause significant stability and control problems, compounding already reduced aircraft performance. Aerodynamic flow separation (stall), by itself or in combination with other effects, is most often the killer. Some current stall protection systems cannot alert the pilot that the margin between stall warning and actual stall is significantly reduced and perhaps eliminated in icing situations. Also, crew training for stall recovery has been inappropriate for airplanes degraded by ice contamination.In-flight ice accretion can be prevented or removed. It can be prevented by adding energy in the form of heat or by chemically depressing the freezing point. It can be removed after accretion by intermittent heating or mechanical de-icing using pneumatically inflated de-icing boots or other mechanical devices that distort SAFETY38 ICAO JOURNAL – ISSUE 1 2017DR. WAGDI G. “FRED” HABASHIis the Founder of NTI Newmerical. He is Project Leader for CLUMEQ (Consortium Laval-UQAM-McGill and Eastern Quebec) on Supercomputing and is a board member of C3.ca, Canada's coordinating association for supercomputing.JOHN P. DOW, SR.is a well-known in-flight icing certification consultant and designated engineering representative (DER). He was the U.S. Federal Aviation Administration (FAA) senior icing specialist and member of the Ice Protection Harmonization working group.ICING CERTIFICATION: TIME TO CONSIDER 3D CFDXXXXX ICAO JOURNAL – ISSUE 1 2017 39 SAFETYthe leading edge of the airfoil, break the ice-surface bond and fracture the ice, allowing the ice particles to be swept away in the airflow. Unfortunately, the total prevention of ice formation, or its complete removal, is not, and likely will never be, economically feasible because of the large amount of thermal or electrothermal energy required, the problems inherent in mechanical removal, and the weight penalties and residue of freezing-point depressant fluids. Moreover, the controlled amount of anti-icing or de-icing hot air bled from the engines is often needed during climb, especially for smaller airplanes, and may be insufficient during descent, approach and landing because of reduced engine power settings. In practice, therefore, while some areas of the aircraft are anti-iced, other areas can only be de-iced. Large areas are left unprotected. Such unprotected areas must be precisely determined.WEAKNESSES OF ICING TUNNELSThe icing tunnel is one of the most important elements of the currently applied toolbox available to assess in-flight icing parameters during the ice protection system design process. One of the greatest difficulties of icing tunnels is the need for simultaneous scaling of geometric, aerodynamic and droplet characteristics, still a wide-open research area with serious limitations that cast doubt on the quantitative value of the experimental results. Another dilemma is the applicability of data obtained from scaled-down partial geometries due to the small size of tunnels. Some of the limitations of icing tunnels:■■Test only a portion of the aircraft (one wing) and ignore the engine and/or propellers, which cannot be scaled to the tunnel’s dimensions. Propulsive effects are then absent and their effect on the local angle of attack and flow conditions are neglected. Also, the asymmetry of icing between the left and right wings cannot be assessed.■■Test only a portion of the span of the wing, as it cannot fit in the tunnel; the spanwise flows over the real wing are therefore absent.■■Test only a portion of the chord of the wing, also because it cannot fit in the tunnel.■■Test only a limited spectrum of droplet sizes and diameters from the prescribed atmospheric envelope that can be reproduced in the icing tunnel.■■Test for supercooled large droplets (SLDs) using only a small section of the vertical model due to the gravitational settling of the heavy drops before they reach the test section.By the time these approximations are cobbled together, the tested item bears little resemblance to reality.When one observes the way ice shapes are “traced” in an icing tunnel, using a cardboard and a pencil, or how water collection coefficient is measured using blotting paper and a timer, the arbitrary height at which the 2D profile of ice was measured, the non-uniformity of the droplet sizes and water content, and other uncertainties, severe doubts can be cast on published experimental ice shapes. WEAKNESSES OF CALIBRATED CODESThe inaccuracy of icing tunnels then pollutes a large class of simulation codes (which we’ll label “pseudo-CFD” codes). Most pseudo-CFD icing codes in use today are based on 1980’s 2D panel methods for flow and Lagrangian tracking techniques for impingement. How is their “calibration” done? Having identified ice surface roughness as the most important parameter determining ice shapes, developers extract from hundreds of icing tunnel measurements a so-called “heuristic analytical roughness model that will make the pseudo-CFD computer code agree with the icing tunnel.” Such codes are then incorrectly accepted as “validated” because they supposedly yield similar results to the icing tunnel. They are not by any measure “validated,” but are simply “calibrated.” In essence, a calibrated code is one that will regurgitate to the user, through a fancy computational procedure, a result that is preordained. WEAKNESSES OF FLIGHT TESTSThe ultimate step in obtaining certification for flying into known icing is testing the completed prototype of the aircraft in the natural real-world environment. This is the only way to assess, for the first time, how all systems work together. Small modifications to these systems can then be suggested and corrections made, but it is certainly too late for major changes. “ The tools … for certifying aircraft for in-flight icing seem to have remained almost frozen in time.”40 ICAO JOURNAL – ISSUE 1 2017SAFETYThe airplane is flown, guided by meteorologists, to locate in nature a number of points within the mandated certification envelope. It is most desirable to attempt to find the most critical ice accretions – without exposing the pilots to danger. Such campaigns can last as little as a week or as long as several seasons, depending on the time of year and the atmospheric conditions prevailing in the country of test. There currently exists no accepted strategy for precisely determining the critical points within these envelopes, for the particular aircraft tested, and for each of its components exposed to icing. No matter how long a natural icing test can last, the number of test points gathered is generally small and dependent on the whims of the weather. It is also extremely difficult to scientifically ascertain whether the number of points tested are sufficient or not.A MODERN APPROACH TO ICING ASSESSMENTA 3D CFD icing approach is essential to complement the rather weak toolbox available for in-flight icing assessment. First and foremost, it can cover all aircraft speeds and all airworthiness-specified atmospheric envelopes. It requires no scaling of geometry, and analyzes the aircraft and not the airfoil. No other experimental means are available to do that until the final natural icing testing.With 3D CFD icing simulation, all geometric features of the airplane can be taken into account simultaneously, including appendages such as Pitot tubes, angle of attack vanes, antennae, icing probes, etc. It includes propulsive effects of the engine or the propellers. It requires no scaling of atmospheric conditions. It also permits a multi-disciplinary approach (with aerodynamics, stress analysis, etc.), is reproducible, traceable, upgradeable, and continuously decreasing in cost. CFD simulation harmonizes the technologies, permitting icing assessment along with the aerodynamics at the “white sheet” design phase, not only at the certification phase. And it facilitates analyzing a gamut of situations that are difficult or too dangerous to test in nature.Such 3D CFD codes can yield airflow and ice shapes over the entire aircraft, with engines, propellers or rotors running, and yield information that no icing tunnel can deliver. Ice surface roughness no longer need be heuristically deduced from ice tunnels results; it can be predicted analytically in time and in space for any component or the complete aircraft.An emerging technology called Reduced Order Modeling can yield the behavior of the clean aircraft “throughout” the airworthiness envelope, complementing natural icing results and providing additional information for points that could not be located in nature or that would be too dangerous to fly.WHY IS CFD NOT READILY ACCEPTED?Will the aviation community wait for icing tunnels to produce 3D results? If icing tunnel 2D results have proven so unreliable, would 3D ones fare any better? The practical and scientific answer is to modernise the concept of validating 3D icing codes. An analogy is the Space Shuttle; the only way to demonstrate that it works is to launch it. However, before launch, one has to ensure that every separate subcomponent that goes into the Shuttle works well independently, and then works well when all subcomponents are joined into a module. That is exactly how 3D CFD-Icing codes should be validated: one subcomponent at a time (in 2D and 3D for steady, unsteady, laminar, turbulent, transitional flows, smooth walls, rough walls, SLD, ice crystals, verifying solution and mesh convergence, degree of accuracy, etc.), then one module at a time (flow, impingement, ice accretion, heat transfer, performance degradation), progressively tying the modules together into a comprehensive CFD-Icing code. In the final analysis, it is not the intricate details of the shape of ice that matter, but what that ice does to the aircraft. If flight-testing and predictions nearly agree in terms of photographs and observations, and in the measured inflight parameters such as lift, drag, moments, then the 3D CFD-Icing code is validated.The use of CFD-based approaches in support of aircraft icing certification offers enormous advantages: ■■No need for scaling or similitude studies; ■■Exploration of a more complete icing envelope in a risk-free fashion; ■■Synergy between the methods used to design the aircraft and those used to design ice protection systems;■■Elimination of experimental inaccuracies generally associated with icing tunnels (measurement and control of droplet size, relative humidity, ambient temperature, water flow rate, repeatability, start-up times). These advantages translate into significant cost reductions, shortening of the certification process, and improving the safe operation of the air vehicle in service. While we propose the use of 3D CFD simulation as an aid to certification, it certainly should not be the primary tool. There would be two potential dangers to do so: underestimating the adverse consequences of icing and overestimating these consequences. A rational approach should involve a methodical exploration of the icing envelope using CFD, accompanied by “focused” flight tests to verify the analysis and avoid either extreme. It is important for the airworthiness authorities to embrace the new technology and accept or reject it on a rational and predictable basis, by experts. Only then will the OEMs be confident enough to start fully deploying technology that has been available for years. EXPLORE ICAO’S NEW TRAINING SEARCH MECHANISMSelect from more than 130 ICAO-recognized courses to meet your training needs!FILTER multiple training criteria Fast & easy SEARCHFIND scheduled training sessions worldwide REGISTER online!VISIT ICAO’S GLOBAL AVIATION TRAINING SEARCH MECHANISM AT: WWW.ICAO.INT/TRAINING1432Next >