Icing simulation: A survey of computer models and experimental facilities

/H-c_2--

NASA Tec_cal Memorandum 104366

Icing Simulation: A Survey

of Computer Models and

Experimental Facilities

M.G, Potapczuk and J.J. Reinmann Lewis Research Center Cleveland, Ohio = = N91-23087 " (NASA-TM-I04366) ICING SIMULATION: A SURVEY OF COMPUTER MODELS AND EXPERIMENTAL CSCL 01A unclas FACILITIES (NASA) 29 p 0014440 G_IOZ Prepared for the 68th AGARD Fluid Dynamics Panel Specialists Meeting Toulouse, France, April 29--May 1, 1991 "= °:_-- = _-= " - ....... _ .... : =-7_ =:_ _-_ _ == _" -_- _--_ =:-==_ =_ _ = _ ±_=:= _ _--L = _- _ = == :

I%I A

Icing Simulation: A Survey of Computer Models and Experimental Facilities M.G. Potapczuk JJ. Reinmann NASA Lewis Research Center, Cleveland, Ohio 44135 operation and performance of advanced turboprops, and Summary whether or not ice protectionwill be required; and (8) the FAA has certified only one civilian helicopter for flight This paper is a survey of the current methods for into forecasted icing, which implies a strong need for simulation of the response of an aircraft or aircraft sub- support of helicopter icing.'

system to an icing encounter. The topics discussed Satisfying these needs can require lengthy and include: 1) computer code modeling of aircraft icing and expensive flight test programs if unassisted by icing sim- performance deg/adation, 2) evaluation of experimental ulation methods. Additionally, finding icing conditions facility simulation capabilities, and 3) ice protection sys- over the full certification icing envelope can not be done tem evaluation tests in simulated icing conditions. Cur- ,O within a reasonable time frame. Thus, various methods rent research, which is focussed on upgrading simulation for simulation of icing conditions are an important and fidelity of both experimental and computational meth-.

ods, is discussed. The need for increased understanding necessary part of the design and certification of aircraft and ice protection systems.

of the physical processes governing ice accretion, ice Initial efforts at icing simulation took place in the shedding, and iced airfoil aerodynamics is examined.

late 1920's and early 30's. These activities are described 1. Introduction in References 3-8. World War II precipitated an urgent need for research into icing simulation and ice protection system design. In the United States, this led the National The safe operation of an aircraft under icing condi- Advisory Committee for Aeronautics, NACA, to build tions is a topic of current interest in the aerospace com- the Icing Research Tunnel (IRT) at the Lewis Research munity. A need for development of icing simulation Center in Cleveland, Ohio during the early 1940's. Initial methods has been identified by aircraft manufacturers and certification authorities alike. 1Reinmann, et al.2 activities in the IRT covered a broad range of icing prob- identified several reasons for the current interest in icing; lems. Many of these included the development of ice for- mations on aerodynamic surfaces and the evaluation of '(1) the more efficient high by-pass ratio engines of aerodynamic performance degradation. A complete bib- today and the advanced turboprop engines of tomorrow liography of the NACA research activities during this have limited bleed air for ice protection, so the airframers period is available as a NASA TM. 9 are seeking more efficient systems; (2) airfoil designers Icing simulation activities have increased dramati- do not want their modern, high-performance surfaces cally during the period from the late 1970's to today.

contaminated with ice, so they are intensifying pressure Wind tunnel and flight research has been conducted by to develop ice protection systems that minimize residual many organizations in North America and Europe. In ice and thereby allow the airframer to keep airfoil surface addition, the advent of high speed computer systems has area to the minimum; (3) new military aircraft requiring allowed the development of sophisticated computer sim- severe weather capability are currently under develop- ulations of ice accretion processes and resulting perfor- ment; (4) some existing military aircraft, being used pri- mance degradation. As a result, a new role has been marily for training missions, are experiencing foreign created for wind tunnel and flight research, that is devel- object damage (FOD) due to icing conditions they would not normally encounter in combat; (5) designers of high opment of code validation databases.

performance military aircraft want to avoid burdening Despite the long history of icing research, there still the aircraft with ice protection, so they want to know remains a significant number of unresolved issues in the where and how much ice will build on the aircraft and process of icing simulation. These issues range from the whether the aeroperformance penalties are acceptable; fundamental physics of the icing process to the mecha- (6) designers of future high performance aircraft with nisms underlying the ice removal process. The ability of relaxed static stability need to know how their aircraft the aerospace community to predict the effects of aircraft will perform with contaminated aerodynamic surfaces; icing and the performance of potential ice protection sys- tems will be strengthened by addressing these issues and (7) little is known about the effects of ice accretion on the incorporating anincreased understanding oftheicing setup used to observe ice growth on a cylinder. By illu- process intosimulation efforts. This paper seeks toiden- minating the ice surface with a laser sheet, they con- tifytheissues ofcurrent icing simulation research and to structed a time history of the ice profile (shown in the suggest howcurrent simulation methods might be middle diagram). This sequence of profiles suggested a

improved. three zone heat transfer model that differed significantly

from the Messinger model. Hansman's model attempts to account for the changing conditions on the airfoil surface 2. The Physics of Icing and Ice Protection by tying the transition from the smooth to rough region to the boundary layer transition.

2.1 Ice accretion physics It is hoped that by understanding the ice accretion Icing occurs when an aircraft encounters a cloud process more completely, some of the empiricism containing super-cooled water droplets, which impact present in current icing models may be eliminated. This aerodynamic surfaces and freeze, forming non-aerody- should result in more robust models providing simulation namic shapes on these surfaces. Incoming water droplets of the icing process over a broader range of the icing can vary in size from 2 or 3 microns to over 40 microns.

envelope.

Typically, smaller.droplets tend to follow the airflow over the surface while larger droplets follow more straight- 2.2 Iced airfoil aerodynamics line paths to the surface. Upon impact on the aircraft sur- Ice accretions on the wing leading edge lead to face, the water droplets can either freeze immediately or exist as a water/ice mixture. These two conditions are increases in drag, decreases in lift, changes in the moment distribution, a decrease in the value of C L max, dependent on environmental parameters such as, temper- and a decrease in the stall angle. These effects aredue to ature and cloud liquid water content (LWC), and on air- a change in the pressure distribution on the wing and to craft surface conditions such as, skin temperature and increased viscous losses.

surface roughness.

From the point of view of effects on aerodynamics, The basics of the ice accretion process, as described ice growths have two relevant length scales. Ice growth above, have been known for some time. However, there structures on the scale of the boundary layer height can are details of the process which are still not completely understood. Recent research in these areas has focused be consld-e-re-d_as_oia_hnesselements. Structur_ger than the boundary layer height and on the order of the air- on development of alternatives to the physical model cur- foil thickness influence the aerodynamics in different rently used in ice accretion codes.The current model was ways and are typically referred to as ice shapes or ice proposed by Messinger 10 nearly forty years ago and caps.

includes the following concepts: in rime icing conditions Ice shapes can result in changes in the pressure dis- (i.e. air temperatures well below freezing and low LWC tribution over the airfoil, development of separated flow values), all cloud droplets freeze upon impact with the regions, early transition of the boundary layer, and pre- surface. In glaze icing conditions (i e air temperatures mature stall, ice roughness can result in thickening and close to freezing and high LWC values), only a fraction early transition of the boundary layer, alterations to the of the water will freeze upon impact, and the remainder pressure distribution, and increased drag. Both types of will run back. The close-up photography of Olsen and ice growth can lead to a decrease in maximum lift.

Walker, 11as shown in Figure 1, indicates that some frac- The results of Ingelman-Sundberg, et al.15suggest tion of the water which remains in the liquid state after that ice roughness can result in a decrease in CL maxand impact may not runback along the surface, as is currently that a glaze ice shape can result in an even larger decrease assumed. This water may remain in pools formed by the for single-element airfoils. Their results also indicate that surrounding ice, thus requiring an alteration of the cur- roughness and ice caps produce approximately the same rent model of the ice growth process. Splashing of incoming water droplets 12'13 may occur under certain level of change in C L max for high lift configurations.

Potapezuk and Berkowitz 16 have measured the changes conditions, which could result in less water on the sur- in lift, drag, and pitching moment for a two-dimensional face than indicated by droplet trajectory calculations. It model of a Boeing 737-200 ADV wing section, in both is also suspected that the initial conditions of the icing cruise and high-lift configurations, as ice accumulated on process may have a significant impact on the subsequent the surface. Their results indicated continual increases in ice growth. Factors such as the initial surface roughness, the effects of ice accumulation on drag. The changes to the surface tension at the air-water-airfoil interface, the lift and pitching moment were not evident until the angle water droplet size, and the boundary layer transition of attack was varied from the condition at which the ice location can all influence the final ice shape. 12'13 was accumulated.

Following up on Olsen's work, Hansman, et al. 14 further studied the icing process. Figure 2 shows the test shrunk significantly. First priority for bleed air is given to It is difficult to establish _y clear trends applicable cabin pressurization and air conditioning. To cope with to all icing encounters, however, it is safe to say that the this loss of bleed air, airframers are either eliminating ice degree of aerodynamic degradation due to icing is depen- protection from selected components, or considering dent on airfoil geometry and attitude, ice accretion .time, alternatives to compressor bleed air. Attractive alterna- and icing cloud conditions. As a consequence of this tives are more energy-efficient deicing systems that dependence, it is apparent that simulation of iced airfoil allow some ice buildup before actuating the deicer. Heli- aerodynamics is necessary to thoroughly evaluate the copters, general aviation aircraft, and light transports, all behavior of a given aircraft encountering icing condi- tions. with relatively small payload fractions and low power margins, have always relied on these more efficient ice The changes in airfoil aerodynamics due to icing apply equally to helicopter rotors. Bond et 81.17 has protection systems.

As alternate ice protection methods are incorporated shown that rotor icing can lead to increases in rotor into aircraft designs, there will be a need to simulate their torque on the order of 25 to 50 percent. Additional results capabilities either computationally or in ground testing indicate that ice shedding events can reduce rotor torque facilities such as icing wind tunnels. It therefore will be by 5 to 10 percent temporarily, thus producing loading necessary to fully understand these experimental simula- transients which can increase system vibration. Asym- tion activities.

metric ice shedding can pose a considerable problem due to out-of-balance conditions which may lead to high vibratory loads on the rotor system. 18 3. Analytical Simulation Methods 2.3 Ice Protection System_ Analytical icing simulation has been based on a combi- There are two approaches to aircraft protection from nation of correlations, computer codes, and theoretical icing; anti-icing and de-icing. Anti-icing is the preven- models of the icing process and its consequences. This section will discuss some of the current methods used for tion of ice growth on critical aircraft lifting surfaces, while de-icing is the removal of accumulated ice before modeling ice accretion, aeroperformance degradation, significant degradation of aircraft performance. Anti- and ice protection system performance. For further infor- icing methods provide the greatest safety factor but mation on recent progress in analytidal modeling see the require the greatest amount of energy. De-icing methods report by Shaw, et al.19 on the other hand may provide appropriate safety with lower energy requirements.

3.1 Ice Accretion Computer Models Either approach to ice protection may be imple- Ice accretion codes have been developed by several mented in a number of ways. There are essentially three researchers. 2°-23 Typically, these codes calculate the categories of ice protection techniques: chemical, flowfield surrounding the airfoil, determine the droplet mechanical, and thermal. Chemical methods typically impingement pattern, and calculate the amount and consist of exuding some type of material on the wing sur- shape of ice that grows on the surface. Results of these face that mixes with the ice and depresses the freezing codes are compared to ice shape tracings from tests in point. Mechanical methods utilize various devices that icing wind tunnels. Comparisons to flight data have also break the ice-wing surface bond by imparting a strain or been performed, 24 but are not as common due to the dif- an impulse to the outer structure of the wing. Thermal ficulty in obtaining actual in-flight ice shapes.

methods melt or evaporate the ice by heating the wing Current ice accretion models are based on the con- surface through a variety of methods.

a'ol volume approach, such as that of Messinger. I° In this The conventional approach for protection of com- model, a mass balance and energy balance are performed mercial transport vehicles for the past thirty years has in order to determine the amounts of water that either been anti-icing through the use of hot compressor bleed freeze or runback along the surface to the next control air. Hot air flowing inside the wing raises the temperature volume. This control volume approach is depicted in Fig- of the leading edge above a level which will allow ice to ure 3. The key factors influencing the ice growth in this form on the surface. Usually the temperature of the lead- model are the mass of incoming water from the cloud and ing edge is high enough to evaporate any water on the from the upstream control volume, the convective heat surface. This prevents runback water from refreezing on flux, and the heat flux into the surface at the ice-body the wing surface aft of the heated area. interface.

But more recently, as jet engine manufacturers have The ability to accurately calculate the incoming begun increasing engine by-pass-ratios to achieve higher water from the cloud has been demonstrated quite con- efficiencies, the engine cores have become smaller and vincingly by several comparisons between calculation the amount of hot bleed air available for anti-icing has and experiment. The accuracy of these methods is deter- performance degradation due to icing. Cebeci 2s has con- mined by evaluating the local collection efficiency calcu- pied a two-dimensional interactive boundary layer ('IBL) lation along the surface of the airfoil. The local collection method with the LEWICE ice accretion code in order to efficiency, in a 2D sense, is defined as the ratio of the ver- simulate the entire icing process. Figure 6, taken from tical distance between two particles at the upstream.

Shin et al.,29 shows the results for several icing condi- release point to the distance along the airfoil surface tions. The most encouraging aspect of this calculation is between their impact points. An example of the collec- the ability of the code to determine the drag rise at tem- tion efficiency calculation is shown in Figure 4, where peratures just below freezing. The results for the ice the calculated value along the surface ofa NACA 65-015 shape comparison agree quite well. The differences airfoil is shown to agree quite closely with experimental results. between the code and experiment for the actual drag val- ues suggest that the method requires further refinement.

The convective heat flux is influenced by the devel- opment of the boundary layer on the rough iced airfoil 3.2 Performance De_radatlon Comouter Models surface. As such, the pressure distribution, roughness Determination of an aircraft's response to an icing level, and transition location all play an important roll in encounter requires the evaluation of performance the ice growth process. Currently these effects are changes resulting from a wide variety of ice accretion accounted for by correlations between the heat transfer shapes. This necessitates the use of an extensive series of coefficient and the cloud icing conditions. A typical tests either in ttight or in a wind tunnel. In either case, the example is found in the roughness correlation in LEWICE, 21 the ice accretion code developed by NASA. use of appropriate computational methods could Hansman, et al)4 has indicated an alternate decrease the amount of required testing and thus decrease the cost and time requirements of the certifica- approach which tries to incorporate more of the physics tion process. As such, computer codes currently being of the process, as it is currently understood. His results used for evaluation of clean or un-iced aircraft are being have shown a marked improvement for cases of ice accretion on cylinders. Figure 5 shows a comparison of adapted for use in evaluation of icing performance deg- radation.

calculations using LEWICE with the original ice accre- tion model and with Hansman's 25 updated model. The Until recently, performance changes for iced airfoils updated analysis and experiment agree remarkably well. have been computed using empirical'correlations such as those of Gray, _r°Bragg,3fand Flemming et al.. 32 These This multizone model is undergoing further study and methods work well over a restricted set of environmental refinement, especially regarding surface roughness and its effect on heat transfer and transition location. conditions, but are not adequate for general use by poten- tial aircraft designers. As a result, computer codes have Most computer codes treat the heat flux into the sur- also been developed to evaluate the changes in perfor- face as a specified constant value during the entire simu- mance of airfoils, wings, and rotors due to the presence lation. This constant is normally taken to be zero thus of ice on critical surfaces.

prescribing an insulated boundary. However, heat flux Cebeci 2s has used the IBL code to evaluate the per- into the airfoil surface may play an important role in the development of the ice shape, especially in the initial formance degradation of a NACA 0012 airfoil with a simulated leading edge ice shape. Comparisons of the moments of the ice accretion. This possibility along with IBL results with Bragg's wind tunnel data 33 are shown in the desire to model thermal de-icing systems has led to Figure 7. These results indicate that the IBL method can the development of computational methods for evaluat- ing the heat transfer between the ice and the underlying determine the aerodynamic losses associated with ice airfoil structure. These models will be discussed more growth on an airfoil up to stall. Further work with the IBL method has involved coupling with the LEWICE ice fully in the section on ice protection system simulation.

accretion code, as mentioned previously. An interesting Until recently, these computer models have been result of this work is that the roughness parameters used strictly two dimensional, calculating ice shapes for in the ice accretion calculation were not the same values chord-wise slices along a wing surface. Three dimen- used in the viscous-flow drag calculation. There is some sional ice accretion codes are currently under develop- ment as extensions of the well established 2D methods.

linkage between the role of ice roughness level in heat transfer and in boundary layer development. Further Potapczuk and Bidwel126 have calculated the ice growth on a MS-317 wing section with 30 ° sweep angle. Guf- study is required to determine the influence of roughness on the ice growth process as well as on the iced airfoil fond has calculated ice growth on the tip of a helicopter rotor. 27 These efforts, while promising, require further boundary layer development.

Navier-Stokes calculations, while requiring more development before use as an engineering tool.

computer time, reveal interesting details of the iced air- A recent research activity has been the coupling of the ice accretion code with methods for evaluation of the foil aerodynamics not produced by the IBL method.

electrothermal heater element is shown in Figure 11. The

Potapczuk 34 has used the ARC2D code with a modified

model simulates the thermal behavior of the composite algebraic turbulence model to calculate the flowfield for structure and ice layer. Results indicate that temperature the same iced NACA 0012 geometry that was used in the traces at the ice-skin interface, heater base, and substrate IBL calculations. Results, also shown in Figure 7, !ndi- base agree well with measured values. Current work by cate good agreement with data even beyond the stall con- dition. The structure of the recirculation zone aft of the Wright, et al. 42 in this area is centered on combining this analysis with ice accretion predictions in order to provide ice shape was also examined and compared to the mea- a tool for the evaluation of electrothermal anti-/de-icing surements of Bragg, 35 as shown in Figure 8. These system performance.

results reveal that the reverse flow velocity is not calcu- Development of methods for evaluation of mechan- lated properly. Use of a more appropriate grid or alter- ical ice protection systems is just beginning with some ation of the turbulence model may be required.

initial work on ice structural properties described in Ref- Additionally, results from investigations of Bragg and Khodadoust, 36 and Zaman and Potapczuk 37 suggest that erences 43-45. Computational methods to predict FOD (i.e. shed ice) damage to engine blades are also under there may be significant flow unsteadiness as this iced development. The ability to determine the damage result- airfoil approaches stall.

ing from various sizes and shapes of shed ice will greatly The recircul,_tion zone in the region aft of the ice enhance de-icing system design.

shape results in complicated flow structures which are A mechanical system that has been evaluated com- fundamentally three-dimensional in nature. An effort has putationally is the Pneumatic Impulse Ice Protection begun to calculate the flowfield for a swept wing geom- etry with ice on the leading edge. Kwon and Sankar 38 . (PIJP) system. The PIIP system relies on rapid inflation have used a 3D Navier-Stokes code to evaluate the aero- of pneumatic tubes embedded the wing surface. 46 The dynamics of a finite span wing model with a NACA 0012 pneumatic impulse causes a displacement of the surface which combined with the surface acceleration, cracks, profile and a 30 ° sweep angle. Figure 9 shows particle debonds, and expels the ice. Ramamurthy, et al.47 used a traces obtained from their calculation for an 8 ° angle of attack condition. The traces show the separated flow con- time dependent, compressible flow model for internal duct flow to model this ice protection system. Results to dition that occurs behind the ice shape and increases in date have been encouraging however further work is size from the root to the tip of the wing.Their results for clean and iced conditions agree quite well with the exper- required.

imental results of Khodadoust and Bragg. 39 Figure 10, which shows the comparison of spanwise lift distribution 4. Experimental Simulation Methods for these two conditions, supports the promising results from this simulation effort. The remainder of this paper will discuss approaches to experimental icing simulation, a topic that includes 3.3 Ice Protection System Models both experimental facilities and the testing done in them.

Ice protection systems have also been modeled com- The prevalent approaches are as follows: putationally. Analysis methods for thermal systems are • testing in icing wind tunnels; the most advanced. Both hot air systems and electrother- mal systems have been modeled successfully. There have • testing in engine test cells that can produce super- cooled clouds; been fewer reported results from analysis efforts for mechanical systems. However, some recent activity in • testing in outdoor ground-level spray facilities this area suggests that these systems will also be mod- that can produce supercooled clouds during the efled computationall_, in the future.

winter, AI-Khalil, et al." used a 3D potential flow code, a • testing in environmental chambers that can pro- 3D particle trajectory code, and a control volume energy duce subfreezing air temperatures and super- analysis to evaluate a hot air anti-icing system for an cooled clouds; engine inlet.This method holds the potential for optimi- zation of hot air ice protection systems.

• flight testing behind aircraft spray tankers Electrothermal systems have been modeled using equipped with water tanks and spray booms to finite-difference methods by Keith, et al.. 41 An electro- produce supercooled clouds; and thermal system consists of an array of electric heater ele- • testing with replicated or simulated ice shapes.

ments embedded under the outer skin of the wing surface. The heater elements are activated during an References which contain comprehensive surveys of icing encounter with enough current to melt the ice that all icing facilities in Europe and North America, and may form on a wing surface. A typical 2D model of an briefdescriptions ofthree new icingwindtunnels in sist of supercooled droplets and no ice particles. Fourth, North America arecontained insection 4.1.3.

the spray system must be capable of reproducing nature's Principal applications ofexperimental icing simula- wide range of liquid water contents, droplet size spectra,

tioninclude testing of full-scale and sub-scale aircraft

and droplet median volumetric diameters (MVD). MVD is defined as the diameter where half of the volume of components, developing advanced ice protection sys- tems, conducting basic research on the icing process and water is contained in droplets with diameters smaller (or on the fundamental properties of ice, establishing empir- larger) than this diameter. And fifth, the icing cloud and ical data bases and correlations, and developing and val- air temperature must be repeatable and controllable to idating analytical models or computer codes.

within fz, irly close tolerances. In practice, no icing tunnel fully meets these requirements, but good tunnels do a 4.1 lcine Simulation Facilities reasonable job of approximating them .48 The need to produce both repeatable and predictable Another critical challenge for icing tunnels is the icing test conditions for the evaluation of computer codes accurate measurement of the supercooled cloud proper- and ice protection systems, has led to the development of ties. LWC can be measured to within about +10 percent.

specialized icing test facilities. The following discussion But the accuracy of droplet sizing instruments is not centers on the requirements for and issues surrounding known precisely, and inaccuracies greater than +2 facilities designed to simulate the natural icing environ- microns are probably typical for MVD's from 10 to 40 ment under controlled test conditions.

microns. 49 Finally, methods to detect ice particles in a cloud are primitive, and methods of quantifyin_ the 4.1,1 lcine Wind Tunnels and Test Cells amount of ice in a cloud are not yet available.-" Icing wind tunnels and engine test cells undoubtedly Engine test ceils arc used to evaluate the ice accre- offer the most versatile approaches to icing testing. It tion patterns that may develop on engine inlets, to inves- generally costs much less to test components in an icing tigate the effects of the icing environment on engine wind tunnel than in flight, and conditions can be much operations and performance, and to evaluate the perfor- more closely controlled and repeated. In icing tunnel mance of engine ice protection systems. Engine test testing, productivity is high and the safety risk is very facilities are of two types: sea-level test stands and alti- low. But there definitely is an appropriate role for flight tude test ceils. The former having the advantage of large testing, and that role will be discussed later.

size and the ability to examine crosswind icing, while the A schematic of the closed-loop NASA Icing latter has the advantage of being able to test over a wide Research Tunnel (IRT) is shown in Figure 12. In addition range of Mach numbers, pressure altitudes, and inlet con- to having all the systems of a conventional dry air tunnel, ditions without regard to prevailing weather.

an icing tunnel has two unique systems: a water spray The General Electric icing test facility in Peebles, system that injects water droplets into the airstream to Ohio, is an outdoor engine test stand located downstream create a supercooled cloud and a refrigeration system and of a large, free jet wind tunnel. A schematic of the facil- ity, taken from Reference 51, is shown in Figure 13. The heat exchanger that cools the air to temperatures as low as minus 30 F. The heat exchanger is in the leg just facility has been used to test the icing characteristics of upstream of the spray bars. Closed-loop refrigerated tun- several full-scale engines. The facility is designed to pro- nels can "dial in the weather" any time of the year and are duce supercooled water droplets of 15 to 50 micron therefore very productive. For example, in 1988, the diameter at LWC levels ranging from 0.4 to 3.6 g/m 3.

NASA IRT logged 1330 hours of test time, making it one The facility is also designed to produce these conditions O O of NASA's most productive tunnels.

at temperatures of -20 to 0 C. The temperature condi- Some icing tunnels bring in subfreezing outside air tions are of course dependent on the ambient tempera- to supply the cooling. Since they are restricted to operat- ture.

ing only in the winter, their productivity is lower, and it The lack of altitude pressure capabilities for this is much harder to achieve a systematic and reproducible facility necessitates the development of adjustments to test program in them.

engine variables or to test variables such as LWC, drop The heat exchanger and spray systems in icing wind size, and liquid/air mass flow ratios in order to simulate tunnels introduce unique operational challenges. First, flight conditions. The facility is also limited to humidity the air temperature and velocity profiles must remain conditions prevalent in the atmosphere during the sched- uniform across the test section for the several hours that uled test period. In order to increase the number of test tests usually require, even though as time progresses the opportunities available in this facility a set of scaling heat exchanger surfaces capture and freeze out the water laws would be useful in order to adjust controllable injected by the spray system. Second, the cloud must be parameters, thus allowing simulation of alternate flight uniform over the test section. Third, the cloud must con- conditions.

Altitude test facilities provide the capability to per- another aircraft equipped with cloud instruments must fly through the cloud to measure its conditions. A further form engine icing tests under conditions similar to those experienced in-flighL Examples of altitude icing facili- challenge for tanker spray systems is to provide the copi- ties are the AEDC Engine Icing Test Facility at Tulla- ous supplies of high pressure air required by the nozzles homa, Tennessee, USA; 52 the icing test cells at the to produce the smaller droplets typically found in nature National Gas Turbine Establishment (NGTE) at Pye- (10 to 40 microns MVD). Or conversely, the challenge is stock, UK; 53 and the Centre d'Essais des Propulseurs to develop a nozzle that can produce the smaller droplets (CEPr) at orsay, France. 54 A schematic of the free jet with lower air pressures and flows. Nozzle research for the special needs of the spray tankers is ongoing. 57 If icing test cell at AEDC, taken from Reference 52, is nozzles are developed that make small droplets with shown in Figure 14. This arrangement is typical of the lower air pressures and flow s, they will also be advanta- engine icing test cells.

Testing in altitude test cells generally consists of geous for wind tunnels and engine test cells, because determination of ice accretion patterns on engine compo- lower nozzle air pressures will result in less droplet freeze-out.

nent surfaces, ice protection system evaluation, and engine icing damage assessment. Since the engine mod- Some of the general aviation airplane manufacturers els used in such f&:ilities are generally not full-scale, the use one of their own aircraft as a spray tar_ker, but the two question of scaling plays a predominant role in determin- best known spray tankers in the United States belong to the military: the Air Force's KC 135 spray tanker and the ing test conditions. Typically, all the scaling parameters U.S. Army's Helicopter Icing Spray System (HISS).

necessary for complete similitude cannot be satisfied The Air Force's spray tanker is a KC-135 aircraft during a single test. Hence, there is a need to develop test programs to evaluate the relative importance of scaling equipped with a 2000 gallon water tank and a newsquare parameters in order to relax some of the constraints on spray boom that extends below the aircraft's tail. The test engine icing tests. Ruff 55 and Bartlett -56provide a more aircraft flies 50 to 100 feet behind the boom, and depend- detailed discussion of similitude for engine icing tests. ing upon the distance behind the boom, the cloud size ranges from about 5 to 10 feet square. Most testing is 4.1.2 In-fli_,ht Icine Simulators - Spray Tanker_ done at aircraft speeds from 150 kt to 300 kt indicated.

Although computer simulations and icing tunnel The tanker is used primarily for icing tests of engines, but it is also used for tests of windshields, control surfaces, testing are important steps in an icing program and hold missile/aircraft interface launches, missiles, and the promise for an increased role in the future, flight test- radomes.

ing of the full-scale aircraft is still an essential step in achieving icing certification or qualification. Before The Army's HISS tanker is a Boeing Chinook CH launching into an extensive flight test program in natural 47 D helicopter that carries an 1800 gallon water tank icing, it is sometimes advantageous to precede flights in and a rectangular spray boom that drops down below the natural icing with flights behind a spray tanker. Chinook after the ship is airborne. The test aircraft flies about 180 feet behind the boom, where the cloud size is Icing flight testing behind a spray tanker offers important benefits: furst, it offers a safety advantage approximately 8 feet high by 36 feet wide. Most testing because the pilot can fly the test aircraft out of the cloud is done at aircraft speeds from 80 to 130 kt true. The and terminate the test if he encounters any problems; sec- HISS is used for testing both helicopters and low-speed fixed-wing aircraft. 58 Typical tests for helicopters ond, it can save development time and costs by providing cloud conditions that rarely occur in nature, yet are con- include main rotors, tail rotors, engine inlets, fuselages, ditions in which the aircraft must be certified or qualified stabilators, droop stops, windshields, antennas, external (such as higher LWC's or larger MVD's); and third, it stores, airspeed sensors, production ice detectors, optical serves to uncover major equipment problems that should system sensors (pilot night vision, and target data acqui- be fixed before undertaking an extensive, expensive, and sition), refueling booms, external hoists, and various high risk flight test program in natural icing. mission equipment exposed to the airstrearn.

Producing a supercooled cloud with an in-flight As part of an overall program to assess Army and NASA cloud measurement systems, NASA flew their spray tanker and calibrating that cloud presents all the Twin OUer icing research aircraft behind the HISS 59 in challenges found in an icing tunnel plus some others. For order to acquire main wing ice shape and drag data for example, the outside air humidity profoundly affects the cloud droplet size. Dryer air causes the smaller droplets clouds with large MVD's, a condition that is rare in nat- to evaporate, and since any humidity from zero to 100 ural icing around the Great Lakes area where the Twin Otter is stationed. Figure 15 shows main wing ice shapes percent is possible on any flight, it is difficult or impos- and drag values for two different conditions; 6° first, a sible to achieve predetermined MVD's or LWC's. Thus, before each immersion of the test aircraft in the cloud, flight in natural icing with a 16 micron MVD cloud, and 4.2 lcine Testlne Technioues second a flight behind the HISS with a 35 micron MVD Icing testing can be performed in facilities such as cloud. The HISS cloud, with the larger drop size, pro- those described above and in addition can be done in con- duced a greater extent of ice coverage and a wing section ventional facilities such as dry air wind tunnels. In this drag coefficient about 25 percent higher.

section, several types of test programs will be described 4.1.3 Surveys of lclntt Simulation Facilities in order to illustrate the range of experimental methods used to understand the icing process and to demonstrate Many aircraft icing simulation facilities exist world- the efficacy of ice protection systems.

wide. In 1981, Olsen 61 published a survey of icing simu- lation facilities in North America. Two excellent surveys of these facilities are listed in AGARD References 62, 4,2,1 Ice Accretion Physics Icing tunnels are a good place to study the funda- and 63. AGARD AR-166, surveyed all the facilities in mental processes of ice accretion because they control Europe and North America as of 1981. In 1986, AGARD AR-223 amended the earlier AR-166 to delete those and repeat conditions quite well. As mentioned previ- facilities removed from service and to add new facilities ously, Olsen and Walker 1° observed the icing process in the NASA IRT with close-up movies. These movies developed since 1981.

revealed accretion phenomena for glaze ice quite differ- In February I991, the Society for Automotive Engi- ent from that contained in the currenOy used analytical neers (SAE) sent panel members of the SAE AC-9C model for ice accretion. These differences were dis- Activity AC-9C-90-1 a questionnaire intended to cussed in the analysis section above.

obtained a comprehensive description of all aircraft icing In addition to the close-up movies, Olsen examined facilities in Europe and North America. One goal of th.e the structure of the ice deposits, the effect of ice shape on survey is to determine how facilities are presently cali- droplet catch, ice roughness effects, the effect of initial brated. The activity also has a goal to determine if some surface flow on the resulting ice shape, and the effect of reliable and easily performed calibration methods could be established that would allow all facilities to be com- droplet shedding. Olsen also discussed the use of plastic replicas of the iceshapes, developed during previous pared on a common basis.

testing, 66 to evaluate the local heat transfer coefficient in Since 1986, three new facilities were added in the United States. Fluidyne in Minneapolis, Minnesota, a dry air wind tunnel. This illustrates the inter-relation- ship between testing in an icing facility and in a dry air USA, modified a transonic wind tunnel to include an wind tunnel.The techniques used in Olsen's test program icing spray system. 64 The tunnel uses outside air during illustrate the type of information relating to ice accretion the winter to obtain the subfreezing temperatures. The test section is 22 in. x 22 in., with Mach numbers up to physics which may be obtained in an icing test facility.

Personne 67 examined the effects of roughness on the 0.8. This tunnel offers the potential for obtaining ice shapes and aeroperformance data on full-scale helicopter type of ice developed on cylinders. He found that for low rotor blade sections.

airspeeds (less than 20 m/s) surface roughness increases The BFGoodrich Deicing Systems Group in Union- the droplet collection efficiency. He also suggests that rime ice roughness levels are due to the random nature of town, Ohio, USA, brought on line in 1989, a new icing tunnel. 65 The test section is 22 in. x 44 in., with airspeeds droplet trajectories caused by increased airstream turbu- lence near the surface.

up to 200 mph. This facility also includes a cold room that allows researchers to move icing samples from the test section to the cold room where they can be stored or 4,2.2 Icing Effects on Aeroeerformance studied further.

An important aspect of icing simulation is the exper- imental measurement and analytical prediction of ice The Boeing Mechanical Systems Laboratory in accretions and their effects on wing aerodynamics. The Seattle, Washington, USA, is bringing on line the Boeing combined use of icing wind tunnels and conventional dry Research Aerodynamic/Icing Tunnel (BRArI'). The air tunnels has proven effective in studying the effects of BRAIT will have three test section sizes: 5 ft x 8 ft (150 ice on airfoil aeroperformance. When conventional tun- kt max airspeed), 4 ft x 6 ft (250 kt max airspeed), and nels are employed to study icing effects, the ice_ simu- 2A ft x 4.4 ft (350 kt max airspeed). The spray system lated by attaching a replica of an ice shape to the airfoii's will be removable for dry air testing. The tunnel is sched- leading edge. Ice replicas are obtained by several meth- uled to begin testing in March 1992. Boeing estimates ods as discussed below.

that the use of their new tunnel will save them up to one The best method for obtaining detailed replicas of million dollars for each new airplane they certify for leading edge ice is from silicone rubber molds, which icing.

replicate both the ice shape and its surface texture. 68 Once the female mold is made, epoxy resin is poured into themold toformacasting oftheice. Inasecond method, several angles of attack up to and beyond stall, and for

chordwise cross sections ofthe iceareobtained bymelt-

various flap and slat configurations. The Working Group ingathincutinthe icearound the leading edge, inserting also reported results where they used emery paper to sim- atemplate inthe cut,and drawing anoutline oftheiceon ulate frost (k/c = 1/1300) in the conventional wind tunnel

the template. A third method uses mono or stereographic

tests. Figure I6, taken from Reference 15, shows the photos to obtain ice shapes. effect on CL(alpha) and CL_ax of ice shapes correspond- Ice replicas are typically made from wood, styro- ing to icing in cruise but with lmiling edge flaps extended.

foam, or epoxy resin. Sometimes the airfoil is machined or molded to coordinates that include the shape of the Olsen, 7° in a comprehensive study in the NASA IRT, leading edge ice. Unless the surface texture is obtained systematically varied key icing tunnel and cloud param- from silicone rubber molds, some method of adding sur- eters to obtain a series of ice shapes and resulting drag face roughness is needed. Roughness can be added for coefficients for a NACA 0012, 21 in. chord airfoil (Fig- example, by applying sand grain or by roughing up the ure 17). Recently, Olsen's data was compared with pre- surface with a knurling tool.

dictions made by a version of LEWICE modified to Ice replicas are used in conventional wind tunnels include the interactive boundary layer 0BL) method that for several reason: (1) the icing tunnel in which the ice predicts lift, drag, and pitching moment of the iced air- was formed may not have the necessary force balances or foil. 2-9 The experimental data used in the IBL/LEWICE flow quality to make good aerodynamic measurements; comparisons were from runs made at several air temper- (2) making pressure measurements in a cold, moist cloud atures, while holding cloud conditions and airspeed con- environment doesn't work well; (3) real ice sublimates.

stant. As revealed in Figure 18, air temperature strongly and sometimes rust or dirt circulating in the tunnel affects ice shape and its resultant drag, especially just erodes the ice, so unless the aerodynamic measurements below the freezing point. Figure 6 compares the IBL/ are done quickly, the ice shape and surface roughness LEWICE predictions with the temperature sweeps.

will change during the tests; (4) ice replicas allow the test Potapczuk and Berkowitz 16 tested a 2D, five-ele- to be repeated as many times as necessary to get good ment airfoil, in four different configurations (Figure 19), aeroperformance data; and (5) in the case where the in the NASA IRT. The airfoil was a two-dimensional sec- experimental aeroperformance data base is to be used for tion of the Boeing 737-200 ADV aircraft wing that Boe- code validation, ice replicas machined or molded to pre- ing had used earlier as pan of their ground de/anti-icing cise coordinates on an airfoil will allow the flow codes to fluids evaluation in the NASA IRT. The airfoil was model the exact airfoil shape used in the wind tunnel test.

mounted between two splitter walls, each of which con- As part of their landmark study of icing and its aero- tained a turntable for varying the angle of attack and a dynamic effects, the Swedish-Soviet Working Group on force balance for measuring lift, drag, and pitching moments. Performance characteristics were measured at Flight Safety conducted testing in a Soviet icing tunnel and a Swedish dry air tunnel to study the effects of wing a given angle of attack during the ice accretion process and tail ice on aircraft stability. In one study, they tested and then over a range of angle of attack conditions after several configurations of a 2D, four-element wing sec- the accretion process was complete. All test equipment, tion (I m chord), including advanced high-lift devices.15 including the data acquisition systems, were provided by And in another study, they tested two swept tailplane Boeing.

configurations with [taps. 69 The airfoils were tested in Their results indicate the change in stall mechanism the Special Wind Tunnel T-4 at the Research Institute of that can occur due to the presence of ice on leading edge the Ministry of Civil Aviation, USSR. The tunnel cross surfaces. As seen in Figure 20, the Ct. vs. alpha curve has section was 1.5 x 2.0 m, and its speed range was 10 to 70 a much lower CL ,,ax value and the slope of the curve beyond C Lmaxchanges dramatically. These differences m/s. The tunnel was equ_ped with a spray system and used outside air below 0 C for cooling. In setting the suggest a change from trailing edge stall to leading edge tunnel icing conditions, the Soviets applied approximate stall. Changes of this sort can remain undetected at cruise icing scaling relations to relate model test conditions to conditions, yet could cause severe problems during take- full-scale conditions. Very detailed female silicone rub- off or landing.

ber molds were made of the ice accretions. In Sweden, Flemming32tested several modem helicopter airfoils the molds were attached to airfoils, and epoxy resin was in the National Research Council's icing tunnel in poured into the molds to form detailed replicas of the ice Ottawa, Canada. The chord of the airfoils was about 6 in.

shapes. The airfoils with the ice replicas were then tested Silicon rubber molds were made for many of the ice in the 3.6m diameter FFA conventional dry air wind tun- shapes. The NRC tunnel was chosen because it could be nel in Sweden. These dry air tunnel tests produced a run to about Mach 0,7, which is representative of the comprehensive set of curves of lift, drag, and moment for Mach numbers near the tip of a helicopter rotor. Flem- manufacturers.46, 74-76 Impulse systems have pulse times ming usedthe data todevelop an empirical correlation less than a millisecond and surface accelerations up to for lift, drag, andpitching moment changes caused by 1000 g's, imparting forces strong enough to shatter, deb- "icing. He has used these results in the Sikorsky General- ized Rotor Performance Code (GRP) to predict full-scale ond, and expel the ice. The impulse systems require min- helicopter torque rises and lift loss, and more recelitly imal power (i.e. on the order of the aircraft's landing light has used the correlations, with some modifications, to power) and they have the potential for maintaining ice thicknesses very thin, both before and after actuation.

predict the performance of a sub-scale model rotor that was tested in the NASA IRT.71 These sub-scale model In testing impulse deicers the following parameters rotor tests are described in a recent report by Flem- are measured to characterize deicer performance: (1) ming. 72 maximum size of shed ice particles for a given ice thick- ness and pulse energy; (2) minimum thickness of ice that The aerodynamics of modern swept wing aircraft is can be removed for a given pulse energy; (3) amount, dominated by three dimensional effects. As mentioned texture, and height of residual ice remaining on the sur- earlier, NASA is developing 3D flow codes that can face before and after deicer actuation; (4) energy per unit model the flow over swept, finite length wings with lead- area or per unit span length required for one deicer actu- ing edge ice. To validate the flow codes, NASA has spon- ation; and (5) weight per unit area of deicer coverage. 76 sored a parallel experimental program in a dry air wind tunnel 73 to obtain a comprehensive data base on the In evaluating deicer performance, the systems must aeroperformance of swept, finite length wings with ice be tested under the full range of expected icing condi- replicas on the leading edge. Figure 21 shows a 30 ° tions. Experience has shown that two conditions give impulse deicers the most trouble: near-freezing condi- swept wing model in a dry air wind tunnel. The ice rep- lica can be seen in the edge-on view. The wind tunnel has tions that produce soft, mushy ice with water between the ice and deicer surface; and cold, rime icing conditions the ability to remove the boundary layer through side- wall suction at the wing rooL The model is heavily that cause the ice to adhere strongly to the deicL_:r-sur :-_ face.75, 77 Knowledge of shed ice size would be essential instrumented for surface pressures and is attached to a if a deicer were to be used on a jet engine inlet, because three component force balance in the wind tunnel wall.

Flow diagnostics include laser velocimetry and helium engine fan blades would be damaged if the engine were bubble seeding and tracking. Some typical results from to ingest ice particles greater than a prescribed size as this effort were discussed earlier in section 3.2.

determined by the engine manufacturer.

Power usage is so low for any these impulse systems 4.2.3 Ice Protection Systems Testing that airframers compare them primarily on the basis of weight, complexity, ease of installation, maintainability, Icing tunnels are used to develop and test aircraft ice and aerodynamic penalties caused by the system installa- protection systems, which include pneumatic boot deic- tion and by the surface ice before and after deicer actua- ers, porous leading edge fluid deicers, electrothermal tion.

deicers, electrothermal evaporative and running-wet anti-icers, hot air evaporative anti-icers, pneumatic Figure 22 contains a sequence of photos, from recent low power de-icing tests in the NASA IRT,77 that capture impulse deicers, and electro-mechanical impulse deicers.

These systems apply to wings, tails, rotor blades, propel- an ice shedding event by means of high speed videogra- lers, and engine inlets. phy. Events can be captured at speeds up to 6000 frames When considering new ice protection systems, air- per second on special video tapes. These tapes can be framersare looking for systems that offer some, or all, of examined frame by frame with motion analysis software the following improvements: lower weight; lower power coupled to a micro-computer. This allows the size of the consumption; acceptable aero-penahies that in some largest particles shed during an actuation to be estimated cases may require ice thicknesses not to exceed 0.040 in.; and the ice breakup process in the airstream to be fol- more reliable operation; lower maintenance time and lowed. Special image processing software is being devel- costs; easily retrofitted to existing components; not oped to automate the estimation of particle sizes and dependent on compressor bleed air; and more economi- possibly to obtain size spectrums as well.

cal to manufaca_re.

4,2_3,2 Water Runbaek in Thermal I¢¢ Protection 4.2.3.1 Mechanical !maulse Deicers Thermal deicers or anti-icers are still the most prev- alent types of ice protection systems used on aircraft. If In their search for alternatives to compressor bleed these systems are overwhelmed in severe icing, or if in air and energy-intensive electrothermal anti-icing sys- the case of deicers, their on/off timing sequences are tems, airframers are considering the pneumatic impulse improperly adjusted, water in the form of rivulets can run and electro-mechanical impulse deicer systems now back beyond the heaters and refreeze. Enough runback under early development by the ice protection system water could accumulate over time to cause aerodynamic 4.2.4.1 Sub-scale Model Rotor Testing in Icing Wind performance or stability problems, or ice could shed and runnel be swept downstream to damage aircraft or engine com- Figure 23 shows a sub-scale helicopter model being tested in the IRT. The model consists of a helicopter fuse- ponents. Thus, ice protection manufacturers are begin- ning to develop water runback and refreeze models for lage, four NACA 0012 blades (4.9-in. chord, 6-ft diam.), their ice protection analysis and design codes. For exam- a fully articulated rotor head, and a six-component force ple, see Reference 78 for runback modeling of hot bleed balance housed under the fuselage. 7t'72 Some results from the sub-scale model rotor testing air anti-icers. There does not appear to be extensive experimental icing data for use in validating the runback are shown in Figure 24. This figures shows the torque models, but more work on development of this database rise caused by ice accretion on the rotors versus time in is planned.

icing. The experimental results are compared with an analytical prediction developed by Flemming 72 that includes an ice shedding model. The analytical predic- 4.2.3.3 Heat and Mass Transfer from Wet Surface_ Thermal antl-icers evaporate water that impinges on tion includes empirical airfoil performance-in-icing data that was acquired in previous tests. 32 The comparison the airfoil leading edge. In general, impinging droplets are not evaporatedi immediately upon impact, but rather between analysis and experiment, as shown here, was the water forms a thin f'dm that is heated and evaporated remarkably good for this particular test run. Similar in the process of flowing downstream along the heated agreement was also found between analysis and experi- surface. Heat transfer correlations for air flowing over ment for lift loss versus time in icing.

dry airfoil surfaces are readily available to the manufacl- Another approach to sub-scale model rotor testing turers, but correlations for heat and mass transfer over would involve the use of simulated ice applied tothe wet airfoil surfaces are not adequate. This is an area leading edge of the rotor blades. The rotor blades with where experimental research is still required to support the simulated ice would be tested in a dry wind tunnel.

validation of computer models for evaporative anti-icing The key to successful dry air wind tunnel testing with systems. simulated ice would be having the correct ice shapes, with appropriate roughness, properly located on the 4.2.4 Special Teehnlques for R0torcraft blades. The appropriate simulated ice would be obtained Only one civilian helicopter, the French Super from scale model testing in an icing tunnel, or from pre- Puma, is certified in the United States for flight into fore- dictions with ice accretion codes. It is unlikely that suffi- casted icing conditions. It took approximately ten years ciently accurate ice shapes could be acquired from full- of flight testing in natural icing to receive the FAA's cer- scale helicopter flights in natural icing conditions tification.

because the ice erodes and sublimates substantially dur- The U.S. rotorcraft industry estimates that, if flight ing the time it takes to descend, land, and shut down.

testing in natural icing is the only acceptable means for certification, it would cost about 15 million dollars to 4,2,4.2 Ice Shedding from RoeQr Blades certify a helicopter to the full FAA, Part 25, Appendix C An important problem for propellers and helicopter criteria. This cost is prohibitively high. rotors is shedding of ice from their tips where centrifugal For several years, NASA and the U.S. rotorcraft forces can exceed 1000 g's. When multiple rotor blades industry have been engaged in a joint effort to develop shed ice asymmetrically, the resulting imbalances cause new methods of reducing the cost and time needed to cer- vibrations severe enough to prevent the pilot from read- tify and qualify U.S. rotorcraft for icing. ing his instruments. Another concern is that shed ice par- These methods include (I) computer codes that reli- ticles have considerable energy and can damage aircraft ably predict full-scale rotor performance in icing and (2) structures. For example, in the design of the tilt rotor air- experimental techniques for testing sub-scale model heli- craft, the fuselage was covered with armor plate in those copter rotors in the IRT to acquire data for validating the areas where ice would impact. This armor plate added a codes and to develop a better understanding of the effects severe weight penalty.

of icing on rotor performance.

Rigorous analytical models and supporting experi- The methods derived from these studies will also mental data are needed to predict (1) shed ice events and advance the state-of-the-art for predicting the effects of the size of the shed ice, (2) shed ice trajectories, (3) ice accretion and shedding for the advanced ducted pro- impact energy of the ice particle and (4) structural dam- pellets and other thrusting devices.

age caused by ice impact. A current project at NASA Lewis is attempting to address some of these issues.

4,2,_ Flight Testin2 with Simulated Ice Shaves both to carry the extra weight of hot air ducting and to Icing flight testing with simulated ice shapes on overcome the loss of engine power due to use of more selected lifting surfaces is done as part of the process of engine bleed air.

obtaining icing certification or qualification, or it is done As a recent example for an aircraft powered by high to acquire a data base for use in validating comput_ bypass engines, the Boeing 757 was the first commercial codes that predict overall aircraft performance and stabil- transport to be certified by the FAA without ice protec- ity in icing. tion on the outboard leading edge slats. This need to identify specific components of an aircraft system and If ice replicas are used on lifting surfaces during flight testing as part of the process of obtaining icing cer- selectively remove unnecessary ice protection was driven tification or qualification, then the following important by the reduction in available bleed air. In the process of caution from Reference 15 must be heeded: '...It has acquiring the certification, Boeing first used replicated ice shapes on a sub-scale model of the 757 tested in a dry been found that the shape of the ice deposit on an airfoil air wind tunnel. The stability and control characteristics plays the main role in worsening of the aerodynamic were studied in the wind tunnel tests. Next, simulated ice characteristics. Flight experiments have shown that even a thin layer of ice might have a serious influence. On the shapes were attached to the actual aircraft's outboard other hand, cases'are possible when a thick ice deposit slam, and the airplane was flown in clear air, again to has no significant influence on the flying characteris- study the effect of the ice on performance and stability.

tics...'. Therefore, when simulated ice is used in flight Finally, the airplane was flown in natural icing to com- testing, the selection of ice shapes is critical. Ideally, it plete the icing certification process. Relying on their vast would be desirable if computer codes that predict ice experience in natural icing flight testing, Boeing was able shapes and resulting aeroperformance and stability could to derive empirical and analytical methods to determine be used to judiciously select ice shapes. But until these the most representative ice shapes for the simulation.

codes have been extensively validated, which will take In a recent research program, NASA used replicated many years, certification testing will still depend criti- ice shapes on the horizontal and vertical tails of their cally on flight testing in natural icing, with the aircraft in Twin Otter icing research aircraft to study the effects of its various multi-element wing and tail configurations. tail ice on stability and control (References 79 and 80).

When simulated ice is used, the ice shapes should be First the Twin Otter was flown in natural icing and photos were taken of the ice formations on the tail surfaces.

based on extensive icing flight test experience.

In the certification process, aircraft are ordinarily From these photos, styrofoam shapes were fabricated tested with simulated ice shapes for two prominent rea- and then attached to the tail surfaces (Figure 25). The air- sons. First, in the event of an ice protection failure, the craft was then flown in clear air through a series of airplane will accrete ice that could degrade aircraft per- maneuvers designed to acquire a flight data base for use formance and stability. Thus, the airframer may elect to in determining stability and control derivatives. The use simulated ice to demonstrate flight safety in the event flight data was analyzed by a modified stepwise regres- of ice protection failure. A second reason for using sim- sion algorithm and a maximum likelihood algorithm that ulated ice would be when an airframer deliberately chose yielded estimates of body-axis stability and control not to protect a given component. Then the airframer derivatives related to the short-period, longitudinal motion of the aircraft.

would have to prove that the aircraft could fly safely with ice on the unprotected surface. They could then elect to As mentioned earlier, NASA is developing a com- use simulated ice to demonstrate flight safety. puter flow code to predict performance and stability of As mentioned earlier, newer commercial mansports modem aircraft with given ice shapes on the lifting sur- are powered by advanced turbofan engines with higher faces. Estimated stability and control derivatives and per- formance measurements from the Twin Otter will be bypass ratios and smaller core flows. So airframers are taking a close look at whether ice protection can be used to validate the code at full-scale Reynolds numbers.

safely eliminated from certain aircraft components in In addition to the Twin Otter flight testing with replicated order to conserve on bleed air usage. ice shapes, dry air wind tunnel testing will be conducted Even in the past when commercial jet transports had of a sub-scale model of a modem swept wing aircraft copious supplies of bleed air, airframers examined the with replicated ice on its lifting surfaces. The wind tun- advantages and disadvantages of protecting certain com- nel results will provide code validation data for a modem ponents from ice, especially vertical tails. By making an transport aircrafL After a good data base has been unprotected vertical tail large enough, airframers could acquired from the wind tunnel testing, flight tests with a achieve acceptable tail aerodynamics for expected ice modem swept wing aircraft will be conducted to acquire a validation data base at full-scale Reynolds numbers.

accretions. So airframers compared the amount of fuel required to carry a heavier tail against the fuel required These new results have led Bilanin sl to formulate As noted in Reference 80, 'the successful estimation the scaling laws independently from any mathematical of stability and control derivatives from flight data has model of the icing process. He did this by applying the two important ramifications. First, the values for the Buckingham pi theorem for similitude to the ice accre- derivatives can be compared with values derived from the analytical icing codes and from the IRT. These compari- tion problem. The pi theorem approach showed, for sons allow for an assessment of the confidence that example, that the normalized thickness of the ice accreted on the airfoil is a function of 18 nondimensional should be put in analytical predictions and wind-tunnel groups. Although many of the groups can be satisfied in results as they relate to an aircraft in flight. Second, any scaling test, Mach, Reynolds and Weber numbers flight-derived derivatives can be used judiciously along cannot all be satisfied in the same scaled test. He con- with those provided by analytical predictions and wind cluded that although competing physical effects do not in tunnel tests to upgrade simulator math models to provide general allow a rigorous scalhag methodology, an accept- a realistic set of aerodynamics for pilot-in-the-loop sim- able approximate scaling scheme may still be possible.

ulations of icing scenarios.'

Bilanin's work is continuing under NASA/FAA sponsor- 4.2.6 Icing Scalint, ship.

Since rigorous scaling is not achievable, a different The proposed or desired test matrix for an icing test approach has been taken in experimental icing testing.

usually involves the following variables: airspeed, out- Components are tested in conditions as close as possible side air temperature, altitude, cloud liquid water content, to the desired conditions, or some partial scaling is used.

cloud droplet size distribution or median volume diame- Analytical methods are then used to predict the results of ter, and model size or scale. In a flight test in natural icing the test. The analytical methods are adjusted, if neces- or in an artificial cloud behind an in-flight spray tanker, chances are that the set of variables desired will be unat- sary, to bring the experimental results and predictions into agreement. Then the analytical methods are used to tainable. In a wind tunnel test, certain combinations of variables also will be unattainable. For example, most predict results for the desired conditions. This is essen- tially the same approach used in acquiring a data base for icing wind tunnels have maximum airspew_ far below the speeds of modem transport or military aircraft. And validating any analytical prediction. With intensified due to the practical limits on nozzle turn-down ratios and efforts to develop computer simulations of all key nozzle droplet size ranges, several sets of nozzles would aspects of aircraft icing, the necessity of a prolonged be required to achieve the full FAA Part 25 Appendix C experimental effort to develop and validate scaling rela- tions may be diminished.

operating envelopes over the full speed range of the tun- nel.

If the desired test variables cannot be met, the exper- 5. Concluding Remarks imenter must resort to some form of scaling or simili- From the preceding discussion, it is evident that air- tude. Various scaling objectives can be imagined for any particular icing test, such as: (1) a geometrically similar craft icing simulations consist of three interrelated activ- ice shape; (2) an equivalent drag or lift coefficient; (3) the ities; analytical modeling, ground-based experiments, same water flux distribution around the airfoil leading and flight testing. Analytical methods started as correla- tions of experimental data and have more recently turned edge; (4) the same heat transfer results for a thermal ice protection system; (5) rime icing conditions (i.e., all to computer models based on fLrSt principles. Computer models have been doing a steadily better job of simulat- water must freeze immediately upon impact); and so on.

i

Not all of these objectives can be met simultaneously and ing the icing process and its effects. Current two-dimen- hence the experimenter may have to chose those most sional codes are being usedin industry for the simulation appropriate for the specific test program. of icing effects and for the evaluation of ice protection Scaling laws have always been used, but never rigor- system effectiveness. Three-dimensional methods are ously validated. 32 Flight testing in natural icing clouds currently under development and hold the promise of accurately simulating an icing encounter for a complete will always be a required part of the certification/qualifi- aircraft configuration. Ground testing in icing wind tun- cation process.

nels, icing test cells, and dry air wind tunnels provides Reference 10 gives a good bibliography of the work (1) simulations of conditions not yet capable of being done previously on scaling. Most of these works on scal- modeled correctly, (2) safer and less expensive means of ing rely on an analysis of the ice accretion process described by Messinger.l° As mentioned earlier, more testing than in-flight testing, (3) controlled, repeatable experiments at known icing conditions, and (4) a data- recent studies have revealed that the Messinger model base for computer code validation. The increasing num- does not reflect recent observations regarding the ice ber of these facilities in the industry attests to their accretion process.

References importance in developing effective ice protection mea- sures for all types of aircraft. Finally, flight testing pro- vides a means of determining the effects of icing on 1) Anon., "National Aircraft Icing Technology Plan," complete aircraft configurations and an essential check Federal Coordinator for Meteorological Services and on the fidelity of the analytical and experimental simula- Supporting Research, FCM-P20-1986, U.S. Dept. of tion methods.

Commerce, April 1986.

The futare of icing simulation is tied to this three Reinmann, J3., Shaw, RJ., and Ranaudo, RJ., prong approach. It is important to understand that the 2) advancement of either analytical or experimental simula- "NASA's Program on Icing Research and Technol- tion methods will always require the need of flight test ogy," NASA TM-101989, presented at Symposium data for verification purposes. The acceptance of simula- on Flight in Adverse Environmental Conditions, Gol, tion tools by aircraft designers and certification authori- Norway, May 8-12, 1989.

ties will depend on the icing modelers ability to show the accuracy of simulation methods. Additionally, the inter- 3) Carroll, T.C. and McAvoy, W.H., "Formation of Ice action of modeler, experimentalist, and pilot can lead to on Airplanes," Airway Age, Sept. 1928, pp. 58-59.

a more realistic and therefore more useful simulation tool.

4) B/eeker, W., "Einige Bemerkungen u ber Eisanatz an This paper has highlighted some of the recent devel- Flugzeugen", Meteorologishe Zeitschrift, Sept.

opments in aircraft icing simulation. Several areas 1932, pp. 349-354 (also available as NACA TM No.

requiring further research have been identified. Some of 1027).

these are: 5) Jacobs, E.N., "Airfoil Section Characteristics as • Ice accretion physics, specifically roughness char- Affected by Protuberances", NACA Report No. 446, acterization, heat transfer correlations, splashing, 1932.

runback, surface tension effects, and wetting char- acteristics.

6) Jones, R. and Williams, D.H., "The Effect of Surface • Ice structural properties and ice shedding. Roughness on the Characteristics of the Aerofoils NACA 0012 and RAF 34", British ARC, R&M No.

• Stall mechanisms and post-stall behavior of iced 1708, 1936.

wings. Computational simulation of these phe- nomena.

G ulick, B.G., "Effects of Simulated Ice Formation on 7) • Inclusion of surface roughness effects in aerody- the Aerodynamic Characteristics of an Airfoil", namics codes.

NACA WR L-292, 1938.

• Evaluation of turbulent flow properties for iced 8) Comite d' Etude du Givrage Rapport du Mai 1938, wings and development of appropriate turbulence Bulletin des Services Techniques No. 85, Publica- models tions Scientifiques et Techniques du Ministere de 1' • Three-dimensional ice accretion code develop- Air (also available as NACA TM No. 919).

ment.

9) Anon., "Selected Bibliography of NACA-NASA Air- • Computer code simulation of iced wing and iced craft Icing Publications," NASA TM-81651, NASA aircraft performance. Development of perfor- Lewis Research Center, Cleveland, OH, August mance codes for rotorcraft in icing.

1981.

• Development of computational methods for simu- lO) lating ice protection systems. Messinger, B I.., "Equilibrium Temperature of an Unheated Icing Surface as a Function of Airspeed," • Continued development of experimental methods Journal of the Aeronautical Sciences, Vol. 20, No. 1, for simulating rotorcraft performance in icing.

1953, pp. 29-42.

• Creation and verification of icing scaling laws.

ii) OIsen, W.A. and Walker, E., "Experimental Evi- Certainly the continued development of the icing dence for Modifying the Current Physical Model for simulation methods described in this paper will continue Ice Accretion on Aircraft Surfaces," NASA TM- to present an exciting challenge to the aerospace commu- 87184, May 1986.

nity for some years to come.

Brunet, L., "Conception et Discussion d'un Modele 12) Bilanin, A.J., "Proposed Modifications to Ice Accre- 23) de Formation du Givre sur des Obstacles Varies," tion/Icing Scaling Theory" AIAA Paper 88-0203, Jan. 1988.

ONERA Note Technique 1986-6, Dec. 1986.

13) Hansman, J.R., and Turnock, S.R.,"Investigation of 24) Berkowitz, B.M. and Riley, J.T.,"Analytical Ice Microphysical Factors Which Influence Surface Shape Predictions for Flight in Natural Icing Condi- tions," NASA CR 182234 and DOT/FAA/CT-88/19, Roughness During Glaze Ice Accretion," Fourth Dec. 1988.

International Conference on Atmospheric Icing of Structures, Paris, France, Sep. 1988.

25) Hansman, J.M., Yamaguchi, K., and Kazmierczak, 14) Hansman, J.R., Yamaguchi, K., and Berkowitz, B., M.,"Revision to the Icing Model in LEWICE, AIAA Paper 91-0123, Jan. 1991.

"Modeling of Surface Roughness Effects on Glaze Ice Accretion," AIAA Paper 89-0734, Jan. 1989.

26) Pota_zuk, M.G. and Bidwell, C.S.,"Numreical Simulation of Ice Growth on a MS-317 Swept Wing 15) Ingeiman-Sundberg, M., Trunov, O.K., and Ivaniko, A., "Method_ for Prediction of the Influence of Ice Geometry," AIAA Paper 91-0263, Jan. 1991.

on Aircraft Flying Characteristics," Swedish-Soviet Working Group on Flight Safety, Joint Report No. 27) Guffond, D.P., "Icing and De-Icing Test on a 1/4 Scale Rotor in the ONERA S 1MA Wind Tunnel," JR-l, 1977.

AIAA Paper 86-0480, Jan. 1986.

16) Potapczuk, M.G. and Berkowitz, B.M., "An Experi- Cebeci, T., "Effects of Environmentally Imposed mental Investigation of Multi-Element Airfoil Ice 28) Accretion and Resulting Performance Degradation:' Roughness on Airfoil Performance:' NASA CR- NASA TM-101441, Jan. 1989. 179639, June 1987.

17) Bond, T.H., Flemming, R.J., and Britton, R.K., 29) Shin, J., Berkowitz, B., Chen, H., and Cebeci, T., "Icing Tests of a Sub-Scale Model Main Rotor," Pro- "Prediction of Ice Shapes and Their Effect on Airfoil ceedings of the 46 th Annual American Helicopter Performance" AIAA Paper 91-0264, Jan. 1991.

Society Forum, May 1990, pp. 267-281.

30) Gray, V.H., "prediction of Aerodynamic Penalties 18) Negrette, A., "The Pilot's Responsibility in Icing Caused by Ice Formation on Various Airfoils," NASA TN D-2166, 1964.

Conditions:' Rotor and Wing International, Vol. 25, No. 1, Jan. 1991, pp. 98-99.

31) Bragg, M.B., "Rime Ice Accretion and Its Effect on Airfoil Performance:' Ph.D. Dissertation, The Ohio 19) Shaw, R.J., Potapczuk, M.G., and Bidwell, C.S.

"Predictions of Airfoil Aerodynamic Performance State University, Columbus, OH, 1981.

Degradation Due to Icing;' Numerical and Physical Aspects of Aerodynamic Flows IV, Springer-Verlag, 32) Flemming, R.J., and Lednicer, D.A., "High Speed Ice Accretion on Rotorcraft Airfoils," NASA CR Berlin, 1990, pp. 19-35.

3910, August 1985.

20) Lozowski, E.P. and Oleskiw, M.M., "Computer Modeling of Time-Dependent Rime Icing in the 33) Bragg, M.B. and Spring, S.A., "An Experimental Atmosphere" CRREL Report 83-2, USAF, Jan. Study of the Flow Field about an Airfoil with Glaze 1983.

Ice," AIAA Paper 87-0100, Jan. 1987.

34) Potapczuk, MIG., "Navier-Stokes Analysis of Air- 21) Ruff, G.A. and Berkowitz, B.M., "Users Manual for the NASA Lewis Ice Accretion Prediction Code foils with Leading Edge Ice Accretions" Ph.D. Dis- (LEWICE)," NASA CR-185129, May 1990. sertation, The University of Akron, Akron, OH, May 1989.

22) Cansdale, J.T. and Gent, R.W., "Ice Accretion on Aerofoils in Two-Dimensional Compressible Flow - 35) Bragg, M.B. and Khodadoust, A., "Experimental A Theoretical Model," Royal Aircraft Establishment Measurements in a Large Separation Bubble due to Technical Report 82128, 1983. a simulated Glaze Ice Accretion" AIAA Paper 88- 0116, Jan. 1988.

ing of an Advanced Pneumatic Impulse Ice Protec- 36) Bragg, M.B. and Khodadoust, A., "Effect of Simu- lated Glaze Ice on a Rectangular Wing;' AIAA tion System (PIiP) for Aircraft," AIAA 91-0555, Jan. 1991.

Paper 89-0750, Jan. 1989.

37) Zaman, K.B.M.Q. and Potalx:zuk, M.G., '`TheLow 48) Ide, R.E, "Liquid Water Content and Droplet Size Calibration of the NASA Lewis Icing Research Tun- Frequency Oscillation in the Flow over a nel," NASA TM 102447, AVSCOM TM 8942-014, NACA0012 Airfoil with an Iced Leading Edge;' AIAA-90-0669, January 1990.

NASA TM-102018, June 1989.

49) Oldenburg, LR., and Ide, R.E, "Comparison of Drop 38) Kwon, O.J. and Sankar, L.N., "Numerical Study of Size Distributions from Two Droplet Sizing Sys- the Effects of Icing on Finite Wing Aerodynamics," tems', NASA TM 1102520, AVSCOM TM 9042- AIAA Paper 90-0757, Jan. 1990.

001, March, 1990.

39) Khodadoust, A.and Bragg, M.B., "Measured Aero- 50) Marek, CJ., and Bartlett, C.S., "Stability Relation- dynamic Performance of a Swept Wing with a Sim- dated Ice A6.cretion" AIAA Paper 90-0490, Jan. ship for Water Droplet Crystallization With the 1990.

NASA Lewis Icing Spray," NASA TM 100220, AIAA-88-0289, January 1988.

40) AI-Khalil, K.M., Keith, T.G., DeWitt, K.J., Nath- Keller, R.G., "Measurement and Control of Simu- man, J.K., and Dietrich, D.A., "Thermal Analysis of 51) lated Environmental Icing Conditions in an Outdoor, Engine Inlet Anti-icing Systems;' AIAA Paper 88- 0759, Jan. 1989. Free Jet, Engine Ground Test Facility" Paper No. 7 in AGARD-CP-236, April, 1978.

41) Keith, T.G., DeWitt, K.I., Wright, W.B., and Masiu- laniec, K.C., "Overview of Numerical Codes Devel- Hunt, J.D.,"Engine Icing Measurement Capabilities 52) at the AEDC" Paper No. 6 in AGARD-CP-236, oped for Predicting Electrothermal De-Icing of Aircraft Blades," AIAA Paper 88-0288, Jan. 1988. April, 1978.

53) Swift, R.D., "Icing Test Facilities at the National 42) Wright, W.B., Keith, T.G., and DeWitt KJ., Gas Turbine Establishment," Paper No. 4 in "Numerical Simulation of Icing, Deicing, and Shed- AGARD-CP-236, April, 1978.

ding" AIAA Paper 91-0665, Jan. 1991.

Bongrand, J., "Installations D'essais de Givrage" 43) Chu, M., Scavuzzo, R2., and Olsen, W.A., "Mea- 54) Paper No. 5 in AGARD-CP-236, April, 1978.

surement of Adhesive Shear Strength of Impact Ice in an Icing Wind Tunnel" Proceedings of the 3rd 55) Ruff, G.A., "Analysis and Verification of the Icing International Workshop on the Atmospheric Icing of Scaling Equations, Vol. I;' AEDC-TR-85-30 (AD- Structures, May 1986.

A167976), Nov. 1985.

44) Scavuzzo, RJ., Chu, M., and Lam, ED., "Develop- Bartlett, C.S., "An Analytical Study of Icing Simili- ment of a Composite Technique in the Determina- 56) tion of the Tensile Sa'ength of Impact Ices" tude for Aircraft Engine Testing," AEDC-TR-86-26 Proceedings of the 3rd International Workshop on (AD-A173713), DOT/FAA/L-'T-86-35, Oct. 1986.

the Atmospheric Icing of Structures, May 1986.

Peterson, A.A., and Oldenburg, J.R., "Spray Nozzle 57) 45) Scavuzzo, R.J., Chu, M.L., and Olsen, W.A., "Struc- Investigation for the Improved Helicopter Icing tural Properties of Impact Ices Accreted at Aircraft Spray System (IHISS)," AIAA-90-0666, January 1990.

Surfaces" NASA CR-179580, Jan. 1987.

Belte, D., and Woratschek, R., "Helicopter Icing 46) Martin, C. and Putt, J., "An Advanced Pneumatic 58) Impulse Ice Protection System if'liP) for Aircraft" Spray System (HISS) Evaluation and Improve- ments," USAAEFA Project No. 83-05-3, United AIAA-90-0492, January 1990.

States Army Aviation Engineering Flight Activity, Edwards Air Force Base, Ca 93523-5000, April 47) Ramamurthy, S., Keith, T.G., DeWitt, KJ.Putt, J.C.

1986.

Martin, C.A., and Leffel, KJ.,., "Numerical Model- Flemming, R.J., Bond, T.H., and Britton, R.K., 71) 59) Mikkelsen, K., Juhasz, N., Ranaudo, R., and McK- "Results of a Sub-Scale Model Rotor Icing Test:' night, R., "In-Flight Measurements of Wing Ice NASA TM 103709, AIAA-91-0660, January 1991.

Shapes and Wing Section Drag Increases Caused by Natural Icing Conditions," NASA TM-87307, April Flemming, R.I., Bond, T.H., and Britton, R.K., : 1986. 72) "Model Rotor Icing Tests in the NASA Lewis Icing Research Tunnel," AGARD Specialists Meeting, 60) Belte, D. and Ranaudo, RJ., "Initial Results from Effects of Adverse Weather on Aerodynamics, Tou- the Joint NASA-Lewis/U.S. Army Icing Flight louse, France, 29 Alxil-lMay 1991.

Research Tests," American Helicopter Society 45th Annual Forum Proceedings, May 1989.

73) Bragg, M., Khodadoust, A., Soltani, R., Wells, S., and Kerho, M., "Effect of a Simulated Ice Accretion 61) Olsen, W.A., "Survey of Aircraft Icing SimuLation on the Aerodynamics of a Swept Wing:' AIAA 91- Test Facilities in North America," NASA TM- 0442, January 1991.

81707, Feb. 1981.

Goldberg, J. and Lardiere, B., "Developments in 74) 62) Anon., "Rotbrcraft Icing--Status and Prospects 7 Expulsive Separation Ice Protection Blankets," AGARD Advisory Report No. 166, August 1981.

AIAA Paper 89-0774, Jan. 1989.

Anon., "Rotorcraft Icing--Progress and Potential:' 63) Smith, S.O., and Zieve, RB., 'q'hin Film Eddy Cur- AGARD Advisory Report No. 223, September . 75) 1986. rent Impulse Deicer", AIAA 90-0761, January 1990.

Zumwalt, G.W., Schrag, R.L., Bemhart, W.D., and 64) Idzorek, J.J., "Observations on the Development of 76) Friedberg, R.A., "Electro-Impulse De-Icing Testing A Natural Refrigeration Icing Wind Tunnel," AIAA- Analysis and Design" NASA CR-4175, 1988.

87-0175, January 1987.

Bond, T.H., Shin, J., and Mesander G.A., "Advanced 65) Tenison, G.V., "Development of a New Subsonic 77) Ice Protection Systems Test in the NASA Lewis Icing Wind Tunnel," AIAA Paper No. 89-0773, Jan- Icing Research Tunnel," NASA TM 103757, April uary, 1989.

1991.

Van Fossen, GJ., Semonian, RJ., Olsen, W.A., and 66) Ai-Khalil, K., Keith, T., and Dewitt, K., "Further Shaw, RJ., "Heat Transfer Distributions around 78) Development of an Anti-icing Runback Model:' Nominal Ice Accretion Shapes Formed on a Cylin- AIAA-91-0266, January 1991.

der in the NASA Lewis Icing Research Tunnel," AIAA Paper 84-0017, Jail. 1984.

Ranaudo, R.J., Batterson, J.G., Reehorst, A.L., 79) Bond, T.H., and O'mara, T.M., "Determination of Personne, R, "Effect de la Rugosite sur la Crois- 67) sance du Givre a Faible Vitesse: Resultats Experi- Longitudinal Aerodynamic Derivatives Using Flight mentaux et Modelisation" D.Sc. Thesis, Data from an Icing Research Aircraft:' AIAA Paper 89-0754, Jan. 1989.

L'Universite Blaise Pascal (Clermont-Ferrand I_, June 1988.

Batterson, J.G., and O'Mara, T.M., "Estimation of 80) Longitudinal Stability and Control Derivatives for 68) Reehorst, A.L., and Richter, RG., "New Methods an Icing Research Aircraft from Flight Data:" NASA and Materials for Molding and Casting Ice Forma- TM 4099, 1989.

tions:' NASA "I'M 100126, Sep 1987.

Bilanin, A.J. "Problems in Understanding Aircraft 69) Ingelman-Sundberg, M. and Trunov, O.K., "Wind 81) Tunnel Investigation of the Hazardous Tail Stall due Icing Dynamics:' AIAA Paper 89-0735, Jan. 1989.

to Icing:' Swedish-Soviet Working Group on Flight Safety, Joint Report No. JR-2, 1979.

Olsen, W., Shaw, R., and Newton, J., "Ice Shapes 70) and the Resulting Drag Increase for a NACA 0012 Airfoil", NASA TM 83556, January 1984.

ORIGINAL PAGE IS OF POOR QU/U.JTY Big water drop = on top of ice hill t_ Ice hill below water film ! Stagnation -_-

_ l l!.e

Figure 1.--Close-up grazing-angle still photos of ice formed at below freezing air temperature of-2 °C.

Incorporated new routines in LEWICE that more closely model observed physics __Glas ser s rod

o ,n erl\ I

Tunnel floor Laser sheet Experimental results Multizone LEWICE prediction Figure 2.--Test set-up and results for ice layer growth on a cylinder.

•- 19 m P27_ 1.0 Experimental repeatability band m .9 Analytical prediction ,,n 1. Impinging water Mc iw, T _ .8 .7 B 3. Water flow out of CV MroutSW, 2. Evaporation I_le.iv,. SUrsursur i .5.4"6 Chord: .3302 m 4. Water flow into CV Mrin iw, C Velocity: 80.96 m/sec 5. Ice accumulation leaving the CV Miii, sur "_- m MVD: 20.36 Wn .E .3 Distribution: AEDC 6. Convection qc t_s "i_ m _ 2 Angle of attack: 0 7. Conduction qk AS -J m .1 Mciw, T + MrinlW,SU r + qkAS = o I -15 -12 -9 -6 --3 0 3 6 9 12 15 Meiv, sur + MroutiV, sur +Miii, sur+qc,_S Surface distance, cm Figure 3.--Control volume for ice accretion energy balance.

Figure 4.--Comparison of experimental repeat- ability and analytically predicted local impinge- ment efficiency for a NACA 65-015 airfoil.

Original LEWICE Experirhental results Multi-zone LEWlCE prediction prediction Figure 5.--Comparison of original LEWICE calculation to multi-zone LEWICE calculation for glaze ice accretion on a cylinder.

O0 Experimental Computational .07 -- .05 -- 0 .06 -- _ _ • .04 -- / '.

.02 _ • • • 0 .01 _ 0 0 0 / o/ 1 I I I I I -30 -25 -20 -15 -10 -5 0 Total temperature, "C Figure 6.--LEWlCE/IBL ice shape and drag coefficient predictions compared to experimental results.

.20 -- INTERACTIVE BOUNDARY LAYER (IBL) [] NAVIER-STOKES (NS) 0 0 EXPERIMENT .16 --

1: I

.6t-- 0 _j .12 H- _ .08 .J .2 .04 I I I I I 2 4 6 8 10 0 2 4 6 8 IO ANGLE OF ATrAEK, DEG O, DEG Figure 7.--Comparison of iced airfoil code predictions with experiment.

O Experiment ARC2D X 0.02 .04 ,06 .08 C 1.50 I I i_ I {= .75 ,--- I I_

,t i ,4

IIIIII I ,,,,,,#_ Velocity profiles for 0 ° AOA; XTR = 0 1.50 -- E: .75 -- 0[-14 1 I I I IJ<_ II]111 la Velocity profiles for 0 ° AOA; XTR = 0.05 Figure &--Velocity profiles in the recirculation region aft of a glaze ice horn on a NACA 0(312 airfoil.

1.0 [] Experiment Calculation .8 Voo (z = 8 deg c • .6

-

O ,4 ,=_ ,_1

_

.2 I I I I I ,2 .4 .6 ,8 1.0 Surface oil flow simulation Off body particle traces Span Figure 9.--Particle traces for flow over an iced swept semi- Figure 10.--Spanwise load distribution of an iced wing span wing at 8 ° angle of attack (NACA 0012 profile, 30 ° (NACA 0012 profile, 30 ° sweep angle) at 4° and 8° sweep angle). angle of attack.

\ // // _3 __lLm m m mlmml_ 3 =_3 2 _-_\\\\\\\\\\\'_\\xl _ _,2 Figure 11 .--Two-dimensional finite-difference model of an electrothermai heater. [Ref. 41].

Comer B Flow .... 5000 hp fan 2100 ton var_chron drive cooler control room Balance chamber , ,,---- 300 mph test section 6 x 9 ft f Corner Control A Comer room Shop access door --/ Model O Spray bar / control /

L

room -/ Air lock chamber Figure 12.--Schematic of the NASA Lewis Research Center Icing Research Tunnel (IRT).

_-Wind Tunnel Discharge Thermocouples x (Before Spraybars) \ • Dry bulb temperature • Frost point temperature \ I'-" Cloud Thermocouples \ | • Mixture temperature \ _, • Frost point tempera- \ Spraybars It ture =11.

\ 1.8m !L_I -'--"-r'l n ,,= o,......._

1 I1---_" --_ n

Wind l,Jl 10.07m i Ll /.

tunnel I"11 _. | _. Engine ........ III ,'/ I, "_L]

_ JU ..,, i,_

/ .o .-" /-- Traversing Boom / _/ • Rotating cylinder- LWC / / / / • Cloud velocity profile / L Movable Fixture / • Droplet oil slide / / • Droplet laser histogram L Camera Station • High speed camera • Searchlight • Engine inlet accel films Figure 13._hematic of the General Electric engine icing test facility in Peebles, Ohio, USA. [Ref. 51].

r- Heated total-temperature Free-jet nozzle ] and total-pressure probes i (2.1 m diam) -, I \ I t- Hologram system for water \ l / droplet size determination Air atomizing \ r / water spray nozzle \ \ I / Test article (configuration 2, Fig. 5) _ \ r / /- \ \ I I / _. "-I_°w __Tl_ ....... To facility -I_i ! t_-' - l--I--_,J I I_10 I,_M _ I Flow , •

- .-1 H exhausters

\ / /_L_ -r- -__ _.

_lr 4-- t ' ' / IF _ t -'_'_F-_ / / t I Inlet duct / / L Free-jet nozzle exit _- Test cell (1.8 m diam) -' I L. Free-jet nozzle insert (3.75 m diam) Figure 14.=Schematic of the AEDC engine icing test cell at Tullahoma, Tennessee, USA. [Ref. 52] Point surface analysis within a one foot span wing segment (all points not in same plane) Estimated average ice shape profile over one foot wingspan section xx Thin, ridged, ice accretion extending aft of trans- lucent region on upper and lower wing surfaces [] Translucent ice region [] Discontinuous regions Natural ice ._,_A_ific accretion ial ice _ accretion Inches E I -1 0 1 2

/

_2 240 - Artificial ice / ,--_ 200- "10 x accretion---../ / Ice :onf. I la lb "_ Ic ._c7 / "_-- Natural ice ,_ _ 120 - ZRoughness 1/1300 of the chord ""--Baseline (no ice)

V

I ] I ] I I I ...... [ ] l I 0 -10 -2 -5 0 5 t0 15 20 0 2 4 6 8 10 a, deg Angle of attack (degrees referenced to floor) Figure 15.---Stereographic analysis of natural and artificial Figure 16.--The effect on CL(,-, ) and CLmax of ice (HISS) ice accretions on NASA DH-6 wing section and shapes from the icing wind tunnel at the Research resulting measured drag coefficients. [Ref. 60] Institute of the Ministry of Civil Aviation, USSR (Cruise condition with flaps extended). [Ref. 15] 0 AIRSPEED, 209 KM/HR;LWC, 1.3g/M3; TIME, 8 MIN TOTAL TEMPERATURE -26 OC -20 Oc -18 Oc -15 °C -12 OC -80C -5 °C -2 °C -I °C 0 °C [] AIRSPEED, 358 KM/HR;LWC, 1.05g/M3; TIME,6.2 MIN TOTAL TEMPERATURE -26 OC -17 °C -12 Oc -8 °C -2 °C 00C Figure 17.--The effect of total temperature on ice shape development. (LWC x V x time) = const.

NACA 0012 airfoil at 4° angle of attack.

(a) CRUISE WING.

0 V = 209 KM/HR, LWC = 1.3 g/M T, T = 8 HIN [] V = 338 KH/HR, LWC = 1.05 g/M 3, T = 6.2 MIN .08 (b) FLAPS 1, _J m .06 -- [] [] .04 (c) FLAPS S.

_r-CLEAN .02 G -20 -10 0 -30 (d) FLAPS 15.

TOTAL TERPERATUREo°C Figure 18.--The effect of total temperature on drag. (LWC x Figure lg.--Boeing 737-200 ADV wing section models tested in the NASA IRT.

V x time) = const. NACA 0012 airfoil at 4° angle of attack.

Run Duration, Temperature, Velocity, LWC, MVD, number rain °F ft/s g/m 3 pm 17 10 26 159 O.92 14.4 2.00 .5 CLEAN D CLEAN o ICEDA-ROST o I CED/FROST ,,._'_x 1.75 v IC[D/NO FROST • II 1.50 v ICED/RO FROST //" r_ 1.25 ,j_ 1,00 ,_P .75 .50 .25 _ ._l .. L - t _ i J _5 0_5 0 5 10 15 20 -5 ALPHA ALPttA Figure 20.--Changes in lift and drag resulting from glaze ice growth on the Boeing 737-200 ADV wing model in the flaps 1 configuration.

ORIGINAL PAGE BLACK AND WHITE P HOTOGi_APH | Figure 21 .--Swept, finite wing with simulated ice on the leading edge installed in the University of Illinois subsonic wind tunnel.

Figure 22.--High speed video of ice shedding event during activation of a low power de-icing system.

2?

ORIGINAL PAGE BLACK AND WHITE PHOTOGRAPH Prediction 6O t a Test A_" Ice sheds e" /I \\\ /1\\ o 4O 2O 8O 20 40 60 C-89-13351 Icing time, sec Figure 23.--Powered force model operating during simulated Figure 24.---Comparison of calculated and measured icing encounter in NASA Lewis Icing Research Tunnel, torque dse of PFM rotor during simulated icing encounter in the NASA IRT.

Figure 25.--Simulated ice shape attached to tail of NASA DHC-6 for evaluation of changes to stability and control characteristics.

N/L /X

Report Documentation Page

Space Administration 1. Report No. I 2. Government Accession No. 3. Recipienrs Catalog No.

NASA TM - 104366

I

5. Report Date 4. Title and Subtitle Icing Simulation: A Survey of Computer Models and Experimental Facilities 6. Performing Organization Code 7. Author(s) 8. Performing Organization Report No.

E -6164 M.G. Potapczuk and J.J. Reinmann 10. Work Unit No.

505-68-10 9. Performing Organization Name and Address 11. Contract or Grant No.

National Aeronautics and Space Administration Lewis Research Center Cleveland, Ohio 44135 - 3191 13. Type of Report and Period Covered Technical Memorandum 12. Sponsoring Agency Name and Address National Aeronautics and Space Administration !14. Sponsoring Agency Code Washington, D.C. 20546 - 0001 15. Supplementary Notes Prepared for the 68th AGARD Fluid Dynamics Panel Specialists Meeting, Toulouse, France, April 29--May 1, 1991.

Responsible person, M.G. Potapczuk, (216) 433-3919.

16. Abstract This paper is a survey of the current methods for simulation of the response of an aircraft subsystem to an icing encounter. The topics discussed include: 1) computer code modeling of aircraft icing and performance degradation, 2) evaluation of experimental facility simulation capabilities, and 3) ice protection system evaluation tests in simu- lated icing conditions. Current research, which is focussed on upgrading simulation fidelity of both experimental and computational methods, is discussed. The need for increased understanding of the physical processes governing ice accretion, ice shedding, and iced airfoil aerodynamics is examined.

18. Distribution Statement 17. Key Words (Suggested by Author(s)) Unclassified - Unlimited Icing Aircraft hazards Subject Category 02 Aeronautics 22. Price* 19. Secudty Classif. (of the report) 20. Security Classif. (of this page) 21. No. of pages Unclassified Unclassified 28 A03 NASA FORM 1626 OCT 86 *For sale by the National Technical Information Service, Springfield, Virginia 22161