NASA's Aircraft Icing Analysis Program

NASA Technical Memorandum 88791

NASA's Aircraft

Icing Analysis Program

AIRCRAFT ICING N86-31548 (NASA-TM-8879 I) NASA'S ANALYSIS PROGRA_ (NASA) 26 _< CSCL 01C Unclas G3/03 g3986

Robert J. Shaw

Lewis Research Center

Cleveland, Ohio

Prepared for the

International Conference of the Aeronautical Sciences (ICAS)

London, England, September 7-12, 1986

N/ A

NASA'S AIRCRAFT ICING ANALYSIS PROGRAM

pOOR QUALITY

Robert J. Shaw National Aeronautics and Space Administration Lewis Research Center Cleveland, Ohio 44135 of the various computer codes at that time Abstract (1984). This report can be viewed as an update of that document.

An overview of the NASA ongoing efforts to develop an aircraft icing analysis capability is Symbols presented. Discussions are included of the over- all and long term objectives of the program as aircraft drag and lift coefficients CD,CL well as current capabilities and limitations of the various computer codes being developed. Des- airfoil drag, lift, and moment Cd,CI,Cm criptions are given of codes being developed to coefficients analyze two and three dimensional trajectories of water droplets, airfoil ice accretion, aerodynamic airfoil chord C performance degradation of components and complete aircraft configurations, electrothermal deicer, icing cloud liquid water content, glm 3 LWC fluid freezing point depressant antideicer and electro-impulse deicer. The need for bench mark MVD icing cloud median volume droplet and verification data to support the code develop- diameter, um ment is also discussed, and selected results of experimental programs are presented.

electrothermal deicer temperatures, °C T I ,T2 ,T3 Introduction free stream temperature, °C Too The aircraft icing problem has long been t time, sec researched and studied. Reports exist in the literature which trace icing research activities boundary layer velocity ratio UIU e back as early as the late 1920's. Since then research into the aircraft icing problem has been free stream velocity, m/sec Voo an almost ongoing effort at varying levels of intensity by a number of government and private nondimensional coordinates research organizations both in the United States X y C C as well as in many other countries. A common thread which binds these various activities is angle-of-attack that for the most part they have been experimental in nature. The majority of these programs have Overall Objectives been aimed at studying the performance of various ice protection systems, although some limited The long range objective of the NASA aircraft attempts have been made to develop icing analysis icing analysis program is to develop a capability capabilities.

to predict the details of an aircraft icing encounter for both fixed and rotary wing vehicles.

• The current NASA Aircraft Icinq Research Program which was started in 1978( Iy seeks to Figure i suggests four uses which could be made of such an analytical capability. Namely, the take advantage of available computational fluid analysis could be used to predict aircraft per- dynamics capabilities and develop a series of formance and handling qualities changes due to compatible computer codes which will address the icing on unprotected and de-iced components and fundamental icing problems. Once these initial to determine ice protection system performance.

codes are developed, additional codes will then In addition, the analysis could be used to design be developed to broaden and enhance the capabili- ice protection systems for various applications ties. The computer codes will be thoroughly and perhaps even be used to design components evaluated by comparison with appropriate verifi- which are insensitive to icing.

cation data. The development of these computer codes and the acquisition of the required verifi- A majority of the current code development/ cation data bases are two major goals of the NASA verification efforts are directed toward the two- aircraft icing research program.

dimensional airfoil problem. Once the capability to analyze the two-dimensional airfoil has been Once these various computer codes have been developed and the accuracy verified, it will serve developed and adequately validated, they will as the basis for developing a fundamental analy- provide the aircraft designers and certifiers tical capability for treating wings, propellers, with tools which should help to reduce the time helicopter rotors, and complete aircraft con- and, therefore, the cost for the development and figurations.

certification of aircraft with the required ice protection.

Obviously a large number of computer codes are required in order to develop an overall Reference 2 presented the overall icing aircraft icing analysis capability. Figure 2 analysis plan which was developed and the status shows the codes requiredandsome (but byno variety of geometries, from simple two-dimensional airfoils to complete aircraft configurations.

means all) of the interfacesrequiredto develop

Table 1 summarizes the various trajectory codes a unified methodology.

being developed and the geometries which can be handled. As the table suggests, the codes use

Theshaded boxes in FigureP indicatethose

various flowfield analysis methods. Two different

computer codes currentlyunder development and

techniques are used to pass flowfield information

verification. Asthe figure indicates,thereare

to the trajectory routine. The direct solution

many computer codes which are not currentlybeing

approach solves for the flowfield at each point

developed. Whileavailableresources certainly

limit the number of codes which canbedeveloped along each trajectory and the grid generation approach calculates the velocity at the point of concurrently, another, even more significant, interest by interpolating on the two or three

limitation is a lackof knowledge of appropriate

dimensional grid generated by the flowfield

fundamental physics onwhich to base computer

solver.

models.Forexample, the lack of understanding

of basicstructural andfracture properties of

There are advantages and disadvantages to

impactice make it difficult to develop a code

both approaches. The direct solution approach

which models the pneumatic boot, oneof the oldest

can be more expensive computationally, especially of all ice protectionsystems.

as the number of trajectories computed increases.

However, this approach frees the user from having

Thus,beforecomputer codes canbedeveloped

to worry about generating a grid which adequately

to treat variousaspects of the aircraft icing

resolves the flowfield acceleration near the body

problem, a seriesof basicmodeling experiments

surface. In addition, the direct solution approa-

first mustbeconducted in orderthat the key

ches tend to have more difficulty computing the

physicalelements beincorporated in anysubse-

flowfield around a body such as an airfoil which quent mathematical models whichareformulated.

has a leading edge ice accretion. Such calcula-

Thisworkis currentlyunderway in severalareas

tions are required as part of an ice accretion

suchas determining the mechanical propertiesof

impact ice._3). analysis as will be discussed later. The grid

generation approach can handle airfoils with lead- ing edge ice accretions, but the development of

Figure2 also pointsout that an integral

an acceptable grid requires care and skill on the

part of the icing analysismethodology under

part of the analyst.

development is the aerodynamic analysis of both

individual components and even complete aircraft All the trajectory codes indicated in Table 1 configurations. Since these codes have histori- use a predictor-corrector approach to integrate cally been developed for analysis of aerodynamic the droplet equations of motion. Different drop- performance in nonicing environments, their use let drag coefficient expressions are used by the to study icing problems will require that sig- various codes, but these differences are judged nificant modifications and improvements be made.

to not have a significant effect on the calcula- Obviously, the more robust and accurate these tion of droplet trajectories.

codes are, the more robust and accurate the icing analysis methodology will be.

Historically, water droplet trajectory codes have been evaluated by comparisons with available The development of an aircraft icing analysis airfoil collection effi#_ncy data obtained methodology requires that a large number of sup- experimentally by NACA._) In general, the porting benchmark (modeling) and verification codes appear to be in reasonably good agreement experiments be run. The experiments require a with the experimental results which may themselves wide range of ground and flight test facilities.

be subject to some appreciable errors. However, Laboratory tests are required to do some of the this database is limited in scope and in particu- most fundamental experiments while icing wind lar, is confined to low speed studies of airfoil tunnel tests are needed to acquire data such as sections of interest to the aviation community in ice accretion, aero performance, and ice protec- tion system performance for selected components. the 1940 to 1955 time period. Also, no experi- mental data had been available for inlet con- Dry wind tunnel tests are needed to look at the detailed flowfield characteristics about airfoils figurations of interest.

with artificial leading edge ice accretions.

An experimental program has been initiated Flight tests are required, both in simulated and natural icing environments, to determine overall to acquire a comprehensive collection efficiency aircraft performance and handling qualities database for a wide variety of modern airfoil and changes. inlet geomerties which are of interest to the aviation community. The experimental approach, a The following sections of the paper will variation on the dye tracer technique developed by NACA, is discussed in Reference 10.

discuss in detail the particular icing codes being developed as indicated in Figure 2.

The major difference from the NACA technique is that a laser reflectance technique is used to Discussion of Icing Computer Codes determine the local collection efficiency distri- bution curves from the blotter paper strips which Water Droplet Trajectory Analysis have been placed at various locations on the test A knowledge of the water droplet trajectories model. The NACA approach was to dissolve out the about any component provides the first indication dye from small samples of the blotter paper strips of that component's sensitivity to icing. NASA and then perform a colorimetric analysis of each is sponsoring the development of a family of drop- let trajectory codes which can handle a wide resultant water-dye mixture. Figure 3 shows a radius, camber, and maximum thickness on collec- close-up photo of a quarter scale Boeing 737-300 tion efficiency.(12) In addition, correlation inlet model in the NASA Icing Research Tunnel (IRT) with a series of blotter strips attached at expressions have been developed to allow quick estimates to be made of airfoil collection effic- various circumferential locations. The laser set-up used to determine the collection efficiency iency characteristics for geometries other than distribution is shown in Figure 4. A Helium-Neon those analyzed.

laser is used to provide the coherent light source.

The three-dimensional trajectory analysis The amount of light reflected from the blotter can code discussed in References 6 and 7 is being be related to the local amount of dye on the blot- ter. The dye was dissolved in the spray water applied to inlet configurations which have flow- prior to being feed through the spray nozzles. fields which are highly thre_=dimensional in This dye concentration is directly related to nature. The initial results _IJl are highly local collection efficiency. This automated data encouraging and indicate complex inlet geometries reduction procedure allows a much larger quantity can be analyzed for icing sensitivity.

of data to be reduced relative to the labor inten- Airfoil Ice Accretion sive method which had to be employed by NACA researchers.

The NASA airfoil ice accretion analysis code which is called LEWICE -±_:4_n extension of the Figure 5 shows measured collection efficiency work of Lo_ki et al.k J and Ackley and _ata for the NACA 657-015 airfoil for two angles Templeton._ z_) The LEWICE code predicts the of attack (0 ° and 8°7. Also shown are predicted ice growth rate distribution (and thus ice shape) collection efficiency curves. The results are in around the leading edge of the airfoil by locally close agreement for the 0 ° angle-of-attack case solving the quasi-steady enerav balance equation while some differences are noted for the 8 ° case. • (_) The ener first proposed by Messinger. gy balance accounts for the governing heat and mass Initially, drop impingement tests will be transfer processes thought to occur during the icing process. These terms include convection to conducted in the NASA Icing Research Tunnel (IRT) using large number of airfoil and inlet models. the free stream, latent heat release due to freez- Once these tests are completed, follow-on tests ing, sublimation (or evaporation), and aerodynamic will be conducted in a high speed icing wind tun- heating. A more detailed discussion of the LEWICE nel to quantify the effects of flowfield com- code is given in Reference 17.

pressibility on droplet trajectories. Previous The four major elements of the LEWICE code analytical stud_conducted jointly by NASA and the British RAE_11) have indicated that com- are (1) a potential flow panel code to calculate pressibility becomes important only for the small- the airfoil flowfield, (?_ a two-dimensional water est dropsizes. However, this analytical result droplet trajectory code,_ ) (3) an energy bal- must be verified experimentally. ance routine, and (4) integral boundary layer routine to predict heat transfer distributions.

Some current applications of the trajectory The LEWICE code is constructed so that the flow- analysis codes will be mentioned. Figure 6 shows field and droplet trajectory calculations are selected results of three-dimensional trajectory repeated at user specified time intervals as the analysis(8) for the nose region of the NASA ice accretion grows on the airfoil leading edge.

icing research aircraft, a deHavilland Twin Otter.

This updating process attempts to account for the The purpose of this study was to look at the effect of the leading edge ice accretion on the effects of the flowfield on droplet trajectories airfoil flowfield and thus, on the changes in and therefore, on the measurements of icing cloud airfoil collection efficiency characteristics.

liquid water content as measured by various in- struments located on the aircraft. The results The current capabilities of the LEWICE code are shown in the form of concentration factor as are summarized in Figure 7 where ice accretion a function of droplet diameter. Concentration shapes measured on a NACAO012 airfoil in the IRT factor indicates the number of droplets per unit are compared with LEWICE predictions. Two sets volume at the measurement station ratioed to the of comparisons are shown for 5 min rime and glaze number of droplets of that size per unit volume ice accretions. For each calculation, the flow- in the free stream cloud. As the figure indi- field and droplet trajectory analyses were updated cates, for the selected location, the instrument every i min. The agreement is judged to be good for the rime ice case, but the LEWICE code over sees more water droplets per unit volume than actually exist in the cloud especially, for a predicts the glaze ice accretion, although the range of droplet sizes between about 20 and 200 _m general shape appears to be good.

and thus would indicate a liquid water content somewhat higher than the freestream value. This The integral boundary layer code has a simple analysis is being used to help in interpreting treatment of the effects of surface roughness on icing cloud data currently being acquired and to boundary layer transition and heat transfer levels.

However, this model is formulated in terms of assist in future placement of instruments.

equivalent sand grain roughness and the relation- An indepth study of airfoil collection ship to actual ice surface roughness is still a research topic, so currently, "educated guesses" efficiency characteristics has been investigated must be input to the model. Other models for using the two-dimensional trajectory code of wall roughness currently are being considered for Reference 4. Some 30 different airfoil geometries incorporation in the LEWICE code.

were analyzed for a wide variety of velocity and angle-of-attack conditions for a range of water droplet sizes. The results were used to form a Recent high-speed photo-micrographic movies of the icing process( 18] have suggested that computational database to investigate the effects some of the fundamental assumptions made in the of airfoil design parameters such as leading edge ice accretion model may be in error. Efforts judged to be in good agreement with the IRT data with the exception of the over prediction of

will shortlybeginto formulate either changcs to

the existingmodel or analternatemodel in order

C1. Similar over predictions of C1 with to seeif ice shape predictions canbe improved.

ARC2D have been found by othe_Researchers for clean airfoil configurations. _zz) The agreement

Otherresearch begun recentlyin orderto in drag polars is thought to be exceptionally

good. The velocity vector plot shown in the fig-

support the ice accretion modeling efforts

includethe measurement of convective heat

ure indicates that the ARC2D code predicted siz- able zones of separated flow on both upper and

transfercoefficientsfroma smooth andrough

surfaceairfoil bothin lowandhigh turbul,nce lower surfaces, but reattachment was aslo pre-

environments andapplicationof a Navier-Stokes dicted for both surfaces.

solverto predictingheattransfercoefficient

distributionsaround bodies with ice accretions.

While these initial results suggested that the ARC2D code could predict the performance of Aerodynamic Performance Degradation Analysis airfoils with leading edge ice shapes, the data Accurate #re_ct_ns of aerodynamTc perform- acquired in the IRT was judged to be not detailed enough to serve as a validation/verification data

ance degradation of anaircraft dueto icing is

base. It was determined that such a database one of the desired end products of the icing analysis methodology. Currently, the development should include, in addition to the airfoil per- and verification of computer codes to predict formance data (Cl, Cd, Cm), much more detailed surface pressure distributions, velocity and tur- airfoil performance degradation in icing is the primary research effort, but some more empiri- bulence profiles, especially, in the separation- cally based approaches are being developed for reattachment zones, and f_ow visualization data.

In order to acquire this database, an extensive predicting degradations of propellers, helicopter rotors, and complete aircraft. The analysis for experimental program was initiated in the Ohio the airfoil will provide the basis for developing State University low speed wind tunnel. Two summary papers of that effort are given in more exact analyses for components such as wings, References 23 to 24.

tails, propellers, rotors, and eventually for complete aircraft configurations.

The airfoil geometry chosen for the program Currently, two advanced analysis codes are was a 0.53 m (21 in.) chord NACAO012 airfoil. A being evaluated for use in predicting iced airfoil removable leading edge was fabricated which had performance - a Navier-Stokes solver and an inter- an ice shape which approximated an ice accretion active boundary layer code. Code predictions are measured in the IRT. The ice shape and its rela- being compared to data being gathered in the IRT tion to the IRT ice shape are shown in Figure 9.

The ice shape geometry can be described by a con- and in the Ohio State University 0.9 by 1.5 m low-speed wind tunnel. These codes were chosen stant radius section, two rounded edges of speci- fied radius, and two straight line sections.

since they have both demonstrated the ability to treat flows which have regions of separation- Thus, an exact specification could be used to reattachment, a phenomenon which often occurs input the geometry into any flowfield analysis code. The wind tunnel model was instrumented with ice shapes. Previously, more conventional with a surface static pressure every 1 percent airfoil analysis codes were considered, but the codes all were incapable of handling the iced chord in the leading edge region. This large number of pressures gave excellent definition of airfoil problem without major revisions being incorporated. A summary of these investigations the separation-reattachment zones on the surfaces.

is given in Reference 2.

Figure 10 presents the ARC2D predictions The Navier-Stokes code being evaluated is compared to the OSU data. Again, the agreement the ARC2_ code developed the at NASA Ames Research is judged to be quite good for the lower angles- Center.(19) The code has been used to analyze of-attack. However, it was observed that the the flowfields about a large number of airfoil ARC2D code did not converge to a steady state geometries including airfoils with extended spoil- solution for angles of attack greater than about ers which create a large separated zone. The code 6 ° . Currently, flow visualization experiments can be run in either of two modes - the full are being conducted to determine if this is purely Navier-Stokes mode or the so-called thin layer a computational result, or if the experimental Navier-Stokes mode which removes the viscous terms flowfield is not steady at these higher angles- of-attack.

in the axial direction. To date, most calcula- tions have been done with the thin layer Navier- Stokes mode, although, limited comparisons of the To date, velocity and axial turbulence predictions made with both modes failed to reveal intensity profiles have been measured every any significant differences. The ARC2D code uses 2 percent chord for the upper surface separation- the Baldwin-Lomax turbulence model, and the so- reattachment zone. Figure 11 shows two selected profile comparisons - one just downstream of the called GRAPE grid generation,_Q_e developed by _Lu) Sorenson, also of NASA Ames. measured point of separation and the other just upstream of the experimental reattachment point.

Initial ARC2D calculations were made for a The ARC2D predictions are in reasonable agreement with the data; however, the predicted reattachment NACA632-A415 airfoil with a simulated leading occurred more upstream than was observed in the edge glaze ice shape.(21) The ARC2D results experiment. Investigations are underway to are compared in figure 8 to measurements made in improve the code predictions. Areas of investi- the IRT with a general aviation wing section with a NACA632-A415 airfoil section with a wooden gation include the sensitivity to ice shape defi- leading ice shape affixed. The predicted results nition and grid generation as well as to the (assuming no laminar boundary layer growth) are turbulence model employed.

is used to determine the total component or air- A similar evaluation of the capabilities of the interactive boundary layer code was begun craft performance degradation.

more recently. This code was developed by Cebeci The correlation equations are relations and associates under grant to the National Science Foundation(25) to predict clean airfoil behavior developed from an airfoil icing data base which including the higher angles-of-attack and the relate the change in lift, drag, and pitching onset of stall. The code has two major modules - moment to known aerodynamic and icing variables.

a potential flow panel routine to calculate the The first suc_#Rrrelations were developed by inviscid region and an inverse finite difference Gray of NACA._a±j More rece_#ly, Bragg devel- boundary layer routine which is capable of treat- oped a rime ice correlation. _"j Both of these correlations, used data gathered in the NASA _RT.

ing regions of separation-reattachment. The two routines are coupled together to solve further Flemming,(29) acquired a large data base in the airfoil flowfield by iterating back and forth Canadian NRC high speed icing wind tunnel for a series of reduced scale rotor airfoil sections until a converged solution is acquired.

which he used to develop a series of correlation Initial studies have been completed for lower equations.

angle-of-attack results for the "iced" NACAO012 airfoil already discussed with comparisons made Unfortunately, it is not possible to a "of airfoil performance (i.e., Cl, Cd). The pre- priori select one set of correlation equations dictions were in very close agreement with the with assurance that the predictions will yield experimental data. Some oscillations in the final accurate results. Studies made with the above mentioned sets of correlations have determined solutions obtained suggested some additional improvements are required in the numerics. These that the errors can be as large as 100 to improvements are currently being made and the 200 percent for selected conditions. Some of more detailed velocity profile comparisons are this error is undoubtedly due to the rather limi- also being made. Also, comparisons are being ted airfoil icing data base which exists. Also, made for the higher angle-of-attack conditions. questions can be asked about how the icing results from one facility can relate to those from another Currently, it is judged that both the Navier- facility which were calibrated using different Stokes and interactive boundary layer methods techniques. Thus, extreme care must be exercized show great promise for being able to predict iced when using these correlations. Miller(28) gives airfoil performance. It is felt that the two an excellent discussion of typical effects of methods are more complementary than competitive using the different correlation equations to as both can be used for icing analysis. The predict propeller performance.

Navier-Stokes approach, while computationally more expensive (approximately 10 to 20 min per In spite of the above concerns, some reason- solution on the Cray XMP) is thought to be capable able results have been acquired. Figure 12 shows of treating larger, more extreme ice shapes. propeller performance predicted in rime icing conditions when compared to natural icing data Navier-Stokes results, possibly, can be used to help in the more approximate modeling in the gathered by NACA researchers. The propeller effi- interactive boundary layer code. The interactive ciency variation as a function of advance ratio boundary layer code is much faster (approximately measured in flight icing conditions was closely 4 to 30 sec per solution on the Cray XMP) and duplicated by the analytical methodology for this would be very attractive for those problems which particular set of tests. As Reference 26 indi- require many aero performance calculations to be cates, the agreement was not as good for some of made (e.g. the helicopter rotor in forward the other rime icing encounters.

flight).

Korkan, et al., used a similar analytical Besides those areas of improvements currently approach_analyze the helicopter rotor_2_irst in hover_ _°j and then in forward flight._ being investigated for both codes which have been discussed, each will require a model to be devel- The particular rotor configuration analyzed was oped to handle the extreme levels of surface the front rotor of the Boeing Chinook CH47.

roughness which exist for any ice shape. At pre- Unfortunately, no experimental data existed for sent, neither code can handle anything other than comparison with the predictions. All that could be said was the results were reasonable when com- smooth surfaces. Several approaches for treating surface roughness are being evaluated. Regardless pared with previously reported torque rises mea- of the model employed, some means must be found sured in icing for other heliocpter configurations.

for quantifying the nature of the ice shape roughness. Currently, the equivalent sand grain Flemming used the previously discussed air- approach is employed to treat rough surfaces, but foil performance in-icing correlations along with it is not clear that that approach Would work for appropriate rotor performance codes to predict ice accretion roughness. rotor degradation for the S-76, UHIH, and UH-60A helicopters. The predicted results compared to As already indicated, more approximate meth- available experimental data are shown in Fig- ods have been developed for predicting the aero ure 13. Generally the agreement was better for performance degradation of propellers, helicopter the hover comparisons than for the forward flight.

rotors (hover and forward flight), and complete As Figure 13 indicates, the forward flight pre- aircraft. References 26 to 30 review these dictions were generally lower than the experimen- efforts. These various efforts all use the same tal measurements. At present_ it is not known approach - a set of correlation equations is used why this is true, although, as Flemming points to predict airfoil performance degradation due to out,(29) it is not clear how accurate are the icing and, then an appropriate performance code experimental measurements. These results point the BF Goodrich Company, is typical of a flight

out the need for dedicated icing flight programs

design. The model was instrumented with a large for acquiring highquality verification data.

number of thermocouples to measure the transient temperature response at various locations in the

Figure14shows selected results fromthe

model. The test model was run over a wide variety

predictionof overall aircraft performance degrad-

ation in icing usingthe NASA code (30)compared to of dry, wet (above freezing), and icing condi- tions, and a large data base was acquired.

datagathered _2_art of the NASA icing research

Reference 36 gives a description of the test

flight program_ J usingthe TwinOtter4 ) The

airfoil correlationdeveloped by Bragg was program.

used to determine airfoil performance degradation.

Figure 16, taken from Reference 38, shows

Even though this correlationwasdeveloped i.o

one selected result of the comparison of code 2

handle rimeicing conditions,the figure indicates

predictions with experimental measurements. As

it appeared to provide reasonable inputsfor the

the figure shows, thermocouples were located at mixed icing conditions of the icing flight shown.

(I) the abrasion shield surface, (2) the heater,

Thecode,as currentlyconfigured, canpredict

aeroperformance degradations onlyfor lifting and (3) the internal spar of the model. For this comparison, the model predictions were judged to

components (e.g. wings,tails, struts). Thecon-

be in close agreement with the experimental data tributions of othercomponents suchas wheels, for all three cycles of electrothermal deicer

flap' hinges,antennas, etc., mustbeestimated

operation. The model predicted a melting of the andinput. Forthe predictionsshown in Figure14, ice for the second and third heater cycles, and

this contributionwas estimated byforcing the

the experimental results showed the same occur-

code to agree with the dragincrease measured for

rence. While the agreement was not as good for

the completely icedaircraft for the lowestlift

all cases as the one shown here, the one-

coefficientdatapoint (filled symbol in the fig-

dimensional code predictions were found to be

ure). This amount of dragincrease wasnot varied

but keptconstant for all th_ othercode--p-re_Cz generally in good agreement with the data.

Currently, efforts are underway to make similar

tions. Notethe reasonable agreement between

comparisons with the other codes to see if the theoryandexperiment not onlyfor the fully iced, increased complexity in the numerical modeling

but alsofor the wingandwingplustail deiced

can give better agreement with the experimental

data. Additionalcomparisons of the code predic-

results.

tions usingthe variousairfoil correlations

availablewith additionalnaturalicing flight

Ice shedding is a key aspect of electro-

dataareplanned.Also, simple methods for esti-

thermal deicer performance. The one-dimensional

mating the dragincrease dueto nonlifting com-

model of Reference 34 attempts to model this ponents are beingconsidered.

phenomenon by assuming that shedding occurs when the thickness of melted ice reaches a user speci-

Ice Protection Systems

fied amount. A better treatment of shedding may

...... -NASA hassponsored the development of a

be required.

seriesof computer codes which model the electro-

thermal deicer, a system currentlybeingused for

Another ice protection system, the fluid

many helicopterrotor andfixed wingaircraft

freezing point depressant system, has been

applications. Table2 givesthe important charac-

teristics of the six different codes developed to modeled. A simple engineering procedure for pre- dicting the minimum anti-icing flowrates required

dateandhow theydiffer fromoneanother. As

has been developed. (39) The code predictions

the table suggests, the codes varygreatlyin

compared with flowrates measured in the IRT for

complexity andtherefore,in the sizeof computer

a full-scale general aviation wing section model

required. Codes 1 and2 which model the electro-

thermal deicerona one-dimensional, timevarying equipped with this system are shown in Fig-

ure 17. In general, the method is judged to give

basiscanberunon a personal computer.Code 6

reasonable agreement with th_ experimental data.

whichanalyzes the complete two-dimensional geo-

As Reference 39 points out, some of the error can

metry,includingthe variablethickness ice layer

be attributed to the experimental values measured

andthe many layersof the airfoil geometry

for minimum anti-icing flowrates. This deter-

requiresa supercomputer (i.e. a Cray1Sor XMP)

mination based on visual observations was found

to achieve reasonable run times. A unique feature

to sometimes vary significantly for repeat test of all codes is that they treat the melting of conditions and was also sensitive to the inter- the ice through a phase change routiner_AC_s 2 pretations of different observers as to when the through 6 use the wea minimum flow rate was achieved. Given these while code 1 uses an p inherent uncertainties in the experimental approach, the predictive method is felt to be Currently, the code predictions are being acceptably accurate.

compared with electrothermal deicer data from icing wind tunnel and helicopter natural icing The third ice protection system being flight tests. These comparisons are aimed at modeled is the electromagnetic impulse deicer providing a better understanding of how accu- (EIDI). EIDI is a concept which is receiving rately the various codes can predict electro- much attention, both in the United States and thermal deicer performance for various levels of Europe. It shows great promise for being a low modeling complexity.

power consumption, highly efficient deicing sys- tem. NASA has sponsored an intensive research Figure 15 shows the test model used in a effort, the results of which are summarized in recent IRT test. The model was a section of a Reference 40, to acquire an experimental data Bell Helicopter Textron UHIH rotor (0.53 m (21 in.

base to aid in the design of the system for chord)) with a spanwise electrothermal deicer various aircraft applications.

configuration installed. The deicer, provided by ._,,_ .........

POOR QL?L_L!'Pi

Work is also underway to develop the required capability have been developed, many validation/ structural and electrodynamic codes to model the verification experiments are still required. This key features of the EIDI system. The current is especially difficult (and expensive) when it status of the structural dynamic modeling capa- comes to determining the performance in icing of bilities is discussed in Reference 41. The simple complex configurations such as helicopter rotors configuration which is being modeled is shown in and complete aircraft configurations. Analysis Figure 18. The half cylinder is meant to model a of the steady state and dynamic behavior of those wing leading edge and has a single 2.54 cm radius complex configurations in icing may require sig- coil located 18 cm (0.07 in.) beneath the surface.

nificant levels of empiricism to be incorporated Finite element model predictions for surface in the modeling.

acceleration at two locations and the circumferen- tial and longitudinal strains at the coil location Nevertheless, it is felt that once developed are shown in the figure. Agreement in all cases and validated, this icing analysis capability will is judged to be at least reasonable and surpris- be of great benefit to the industry in reducing ingly good in some instances (e.g. peak circum- the fixed and rotary wing development and certi- ferential strain).

fication costs for future aircraft configurations which are required to safely fly into forecast A key input into the structural dynamic icing conditions.

analysis is the force produced by the coil. Both References t'her temporal and spatial characteristics of this force must be known. Until recently, no analyti- cal solutions had been obtained for this force 1. Reinmann, J.J.; Shaw, R.J.; and Olsen, W.A., function, and thus experimental data had to be Jr.: Aircraft Icing Research at NASA. NASA used. The results shown in Figure 18 were TM-82919, 1982.

acquired using a simple force model. However, 2. Shaw, R.J.: Progress Toward the Development recently, a transmission line model obtained from of an Aircraft Icing Analysis Capability.

NASA TM-83562, 1984.

the governing field theory equations has been 3. Scavuzzo, R.J.; Chu, M.L.; and Olsen, W.A.: achieved which appears to yield good predictions of the force for the case of a.coil with its axis Structural Properties of Impact Ices. AIAA perpendicular to a flat plate.(42) Figure 19 Paper 86-0549, Jan. 1986.

4. Bragg, M.B.: Rime Ice Accretion and Its shows a comparison of the force versus time pre- Effect on Airfoil Performance. NASA dicted by this transmission line model with the CR-165599, 1982.

force versus time calculated from the measured 5. Chang, H.P., et al.: Influence of induction fields for a 3.08 cm (2 in.) diameter 0.478 cm (0.188 in.) thick coil rigidly mounted Multidroplet Size Distribution on Icing Collection Efficiency. AIAA Paper 83-0110, next to a fixed 0.0813 cm (0.032 in.) thick 2024 T3 Aluminum disc with a 12.7 cm (5 in.) diameter. Jan. 1983. f 6. Kim, J.J.: Particle Trajectory Computation on It is hoped that the use of this force model will improve theoretical predictions such as those a 3-Dimensional Engine Inlet. NASA CR-175023, 1986.

shown in Figure 18.

7. Norment, H.G.: Calculation of Water Drop While the modeling of the EIDI system shows Trajectories To and About Arbitrary Three- Dimensional Bodies in Potential Airflow.

great promise, the effort is still in its infancy.

NASA CR-3291, 1980.

A better understanding of the ice adhesion prop- 8. Norment, H.G.: Calculations of Water Drop erties and removal mechanisms must be gained before a complete modeling of the system can be Trajectories To and About Arbitrary Three- attempted. Also, the EIDI system modeling for Dimensional Lifting and Nonlifting Bodies in Potential Airflow. NASA CR-3935, 1985.

most applications will probably require three- dimensional structural analysis codes. 9. Gelder, T.F.; Smyers, W.H., Jr.; and VonGlahn, U: Experimental Droplet As already indicated, NASA has sponsored Impingement on Several Two-Dimensional Airfoils with Thickness Ratios of 6 to efforts to measure some of the key physical properties of ice(3) which are required as 16 Percent. NACA TN-3839, 1956.

inputs into any surface deflection type ice i0. Papadakus, M., et al.: An Experimental Method for Measuring Droplet Impingement protection system model. The results are being Efficiency on Two-and-Three-Dimensional used along with a finite element analysis approach Bodies. AIAA Paper 86-0646, Jan. 1986.

to begin a quasi steady modeling of a generic pneumatic boot ice protection system. Companion 11. Gent, R.W.: Calculation of Water Droplet experiments are being planned to measure the ice Trajectories About an Aerofoil in Steady, Two-Dimensional, Compressible Flow. RAE fracture process of the generic boot configuration TR-84060, June 1984.

for comparison with modeling predictions.

12. Bragg, M.B.; and Gregorek, G.M.: An Concluding Remarks Analytical Evaluation of the Icing Properties of Several Low and Medium Speed Airfoils.

AIAA Paper 83-0109, Jan. 1983.

This paper has presented a current status review of the progress being made toward devel- 13. Kim, J.J., et al.: Sand Separate Efficiency Calculation for the JUX T.H. Rake Aircraft oping and validating the computer codes required Inlet. Presented at the 42nd American to assemble an aircraft icing analysis method- ology. While it is felt that much progress has Helicopter Society Forum, June 2-4, 1986.

been made since the initial review presented in Reference 2, much work remains to be done. While many of the codes which are needed to form a core Accretion. J. Aircr, vol. 22, no. 8,

14. Lozowski, E.P.;Stallabrass, J.R.; and

Aug. 1985, pp. 713-718.

Hearty,P.F.: TheIcing of an Unheated

28. Miller, T.L.: Analytical Determination of Non-Rotating Cylinder in Liquid Water Droplet: Propeller Performance Degradation Due to Ice Ice Crystal Clouds. National Research Accretion. NASA CR-175092, 1986.

Council of Canada, LTR-LT-96, Feb. 1979.

29. Flemming, R.J.; and Lednicer, D.A.: High 15. Ackley, S.F.; and Templeton, M.K.: Computer Speed Ice Accretion on Rotorcraft Airfoils.

Modeling of Atmospheric Ice Accretion, CRREL NASA CR-3910, 1985.

Report 79-4, Mar. 1979.

30. Gregorek, G.M., et al.: NASA Twin Otter 16. Messinger, B.L.: Equilibrium Temperature of Flight Test Program - Comparison of Flight an Unheated Icing Surface as a Function of Results with Analytic Theory. SAE Paper No.

Airspeed. J. Aeronaut. Sci., vol. 20, no. i, 850924, Apr. 1985.

Jan. 1953, pp. 29-42.

31. Gray, V.H.: Prediction of AErodynamic 17. MacArthur, C.D.: Numerical Simulation of Airfoil Ice Accretion. AIAA Paper 83-0112, Penalties Caused by Ice Formation on Various Airfoils. NASA TN D-2166, 1964.

Jan. 1983.

32. Ranaudo, R.J., et al.: Performance 18. Olsen, W.A., Jr.; Walker, E.D.; and Sotos, Degradation of a Typical Twin Engine Commuter R.G.: Microscopic High Speed Movies Showing Type Aircraft in Measured Natural Icing tile Droplet Freezing Process of Icing. AIAA Conditions. AIAA Paper 84-0179, Jan. 1984.

Paper 84-0019, Jan. 1984.

33. DeWitt, K.J. and Baliga, G.: Numerical 19. Pulliam, T.H.: Euler and Thin Layer Navier- Simulation of One-Dimensional Heat Transfer Stokes Codes: ARC2D, ARC3D. Computational in Composite Bodies with Phase Change. NASA Fluid Dynamics, University of Tennessee, UTSI Publication No. E02-4005-023-84, 1984, CR-165607, 1982.

Section 15.1.

34. Marano, J.J.: Numerical Simulation of an Electrothermal Deicer Pad. NASA CR-168097, 20. Sorenson, R.L.: A Computer Program to 1983.

Generate Two-Dimensional Grids About Airfoils and Other Shapes by the Use of Poisson's 35. Chou, D.F.K.: Numerical Simulation of Equation. NASA TM-81198, 1980. Two-Dimensional Heat Transfer in Composite 21. Potapczuk, M.G.; and Gerhart, P.M.: Progress Bodies with Application to De-lcing of in the Development of a Navier-Stokes Solver Aircraft Components. NASA CR-168283, 1983.

for Evaluation of Iced Airfoil Performance.

36. Masiulaniec, K.K., et al.: Full Two Dimen- sional Transient Solutions of Electrothermal AIAA Paper 85-0410, Jan. 1985.

22. Mehta, U.; Chang, K.C.; and Cebeci, T.: A Aircraft Blade Deicing. AIAA Paper 85-0413, Comparison of Interactive Boundary Layer and Jan. 1985.

Thin-Layer Navier-Stokes Procedures.

37. Leffel, K.L.: A Numerical and Experimental Numerical and Physical Aspects of Aerodynamic Investigation of Electrothermal Aircraft Flows III, T. Cebeci, ed., Springer-Verlag, Deicing. NASA CR-175024, 1986.

1986.

38. Leffel, K.L., et al.: A Numerical and 23. Bragg, M.B.; and Coirier, W.J.: Detailed Experimental Investigation of Electrothermal Measurements of the Flowfield in the Vicinity Aircraft Deicing. Presented at the American of an Airfoil with Glaze Ice. AIAA Paper Helicopter Society 42nd Forum, June 2-4, 1986.

85-0409, Jan. 1985. 39. Albright, A.E.: Experimental and Analytical 24. Bragg, M.B.; and Coirier, W.J.: Hot Film Investigation of a Freezing Point Depressant Measurements of the Separation Bubble on an Fluid Ice Protection System. NASA CR-174758, 1984.

Airfoil with Glaze Ice. AIAA Paper 86-0484, 40. Zumwalt, G.W., et al.: Analysis and Tests January 1986.

for Design of an Electro-lmpulse De-lcing 25. Cebeci, T., et al.: Airfoils with Separation System. NASA CR-174919, 1985.

and,the Resluting Wakes. Third Symposium on Numerical and Physical Aspects of Aerodynamic 41. Bernhart, W.D.; Gien, P.H.; and Wilson, Flows, California State University, Long B.K.: Structural Dynamics Investigations Related to EIDI Applications. AIAA Paper Beach, CA, 1984, pp. 2-13 to 2-25.

86-0550, Jan. 1986.

26. Korkan, K.D.; Dadone, L.; and Shaw, R.J.: 42. Henderson, R.A.: Theoretical Analysis of the Performance Degradation of Propeller/Rotor Systems Due to Rime Ice Accretion. AIAA Electrical Aspects of the Basic Electro- Paper 82-0286, Jan. 1982. Impulse Problem in Aircraft De-icing 27. Korkan, K.D.; Dadone, L.; and Shaw, R.J.: Applications. Ph.D. Dissertation, Wichita Performance Degradation of Helicopter Rotor State University, 1986.

Systems in Forward Flight Due to Rime Ice TABLE 1. - DROPLET TRAJECTORY ANALYSIS CODE CHARACTERISTICS Geometries Flowfield analysis code used Comments iReference I. Single Element Airfoils Theodorsen Transformation 4 Incompressible Flowfield, Direct Solution 2. Single Element Airfoils Navier-Stokes 5 Compressible Flowfield, Grid Generation 3. Multiple Element Airfoils Panel Method 5 Incompressible Flowfield, Direct Solution 4. 2D, Axisymmetric Inlets Panel Method 5 Incompressible Flowfield, Direct Solution, 2D Trajectories 5. 3D Inlets Transonic, Full Potential Compressible Flowfield Grid Generation 6. Complex Configurations Panel Method Incompressible Flowfield, 7,8 (incl. Complete Aircraft) Direct Solution TABLE 2. - ELECTROTHERMAL DEICER CODES Geometry Ice thickness Number of heaters Phase change Reference I. one Dim#nsional Constant One 33 Approximate 2. One Dimensional Constant One 34 Weak Enthalpy 3. Two Dimensional Constant One Weak Enthalpy (Rectilinear) 4. Two Dimensional Constant Variable Weak Enthalpy (Rectilinear) 5. Two Dimensional Variable One Weak Enthalpy 37 (Rectilinear) 6. Two Dimensional Variable Multiple 36 Weak Enthalpy (Exact)

ICE PROTECTION

SYSTEMDESIGN

AIRCRAFTPERFORMANCE,

HANDLINGQUALITIES

CHANGES

ICE PROTECTION

SYSTEMSPERFORMANCE

COMPONENT DESIGN

CS-86-1714 Figure 1. - Uses of aircraft icing analysis.

FLIGHT, /

E"V%",T_O_A'_

I BODY

I o_o_._ _

I ON ICE

' I

SYSTEM HOT GAS ]

,, J

ICE SHEDDING I VIBRATION BOOT

I

SYSTEM SYSTEM

ME_.A.,OAL ] _.EOMAT,O

NOTE: ICEPHOBICS BY NASA LeRC

CD-85-15526

Figure 2. - Aircraft icing analysis methodology.

OF POOR QU?,J_!_'CI

Figure 3. - Boeing 737-300 inlet n Icing Research Tunnel with blotter strips attached.

OR|GIN_ _::_:'_ '_i_i

OF. POOR QUaLiTY

, Figure 4. - He-Ne laser system to reduce drop impingement blotter samples.

Io0

a = 8o

.8

THEORY

/_ EXPERIMENT

.6

.4

>- C_) Z

.2

C_) L_J

!

z 0

2 4 6 8 i0

SURFACE COORDINATE, ~cm

LJ_

__ 1.0¸

a : 0°

.6'

-4 -2 0 2 4 6

SURFACE COORDINATE, ~cm

Figure 5. - Preliminary collection efficiency results for NACA

6.52-015 airfoil. Voo =82 m/sec; MVD = 15 pm.

"- INSTRUMENT

r,,,"

1.2

O

-- LOCATION ,/_

Z C) t---

1.1

m,, I-- Z (..)

Z O

1.0

(..)

I I0 I00 i000

DROPLET DIAMETER,~pro

Figure 6. - Concentration factor calculations for NASA icing

research aircraft, Voo = 67 mlsec.

CALCULATED COMPARISON EX PERIMENTAL

LWC =1.02 gm/m 3 MVD =12pm Voo =52m/sec Too =-26°C

CA LCU LATED COMPARISON EXPERIMENTAL

LWC =1.20 gm/m 3 MVD =20pm Voo =89m/sec Too =-11°C

' Figure 7. - LEWICE predictions of airfoil ice accretions compared to TRT data. NACA 0012, 0..53m chord, 5 minute ice accretion, 0° angle of attack.

/

J

-_ _. -_'"_-- SEPARATIO N-

REATTACHMENT

ZONES

1.50

,25 --

_T DATA

-D- ARC 2D PREDICTION

I I

-i. 00

-I0. 0 2.5 15

Cl

•ii0

_P

l

-0- IRT DATA I

--El.- ARC 2D I

PRED IC-(_ _

.055

-- TION ,,

o I I

-.5 .5 1.5

C_

Figure 8. - Navier-Stokes analysis

of NACA 632-A415airfoil with

glaze ice (ref. 21).

I

•i0 .20

XIC

\ \

MEASURED IN IRI

SIMULATED

C$-86-1715

Figure 9. - Comparison of measured and simulated ice shapes

NACA 0012airfoil, 0..53m (21 in. ) chord.

.20

THEORY (ARC 2D)

-- O EXPERIMENT

1.0

NO CONVERGED

SOLUTIONS

FOR a >6° --_

\

O0

.i0

.05

.i

E 0

I I

-.i

0 5

a, deg

Figure 10. - Comparison of measured

and predicted airfoil performance

for NACA 0012airfoil with artificial

ice shape.

j._ PREDICTED-_ I

I ,L I _I

y/c

I

, I

[ ....9' ' >'

....I yJc \"\ 1

#

DIVIDING

STREAMLINE

u/u e ,-

---...,._....._

(EX PER IME NTA L)_

/ _

\\ __. U_Ue

/

CS-86-1717

Figure 11. - Comparison of measured and predicted velocity profiles using ARC 2D

code.

1.0

-- /- CLEAR /-CLEAR

i (EXPERIMENTAL) / (THEORETICAL)

I -- / /

.9

>- (J 2: "' .8 m L.I.- I.a--

'" 7

(EXPERIMENTAL) I ICE

,._1

(THEORETICAL)

i.i o .6 13_

.5

I I I I I I I

.4

1.0

i.i 1.2 1.3 1.4 1.5 1.6 1.7

ADVANCE RATIO

Figure 12. - Comparison of measured and predicted

propeller performance in rime icing conditions

(ref. 26).

0 S-76 '81

[] S-76 '82

3O

A UH-IH /_

$7 UH-60A /

2O

-_za

oo

_D

I_ [] / x__LINEOF PERFECT

i0

[] AGREEMENT, o

>1 _/ D I I I

o

(al Hover.

I.LI l-.- r_ 1.1-I \

- /

\\ LINEOF PERFECT

i0

_--_1_ _1 I I

i0 20 30 40

OBSERVED TOROUE RISE, percent

(b) Forward flight.

Figure 13. - Comparisons of measured and predicted

helicopter performance in natural icing (ref. 29).

ORIGIN:t:L P_;_:_::_i _.C

OF POOR QUAISTY

EXPT THEORY

-- -mOB FU LLYICED

•12

---El--- WING DEICED

W + T DEICED

.10

0 CLEAN _-...

___ _.- -

•08

@

o oo

O6

4#o ooo <_@o

.04

I I I I I

•02

.2 .4 .6 .8 1.0

Figure 14. - Comparison of predicted and measured aircraft per-

formance in natural icing conditions (ref. 30).

ELECTROTHERMAL DE-ICER EXPERIMENT IN IRT

6D.-15-16183 Figure 15. - Electrothermal de-icer experiment in I RT.

'I

ICE (~0. 16 cm)

_////,_ T3_ SH,Em

,,_I...,.............,._--.-HH_J HEATER

INSULATIO N--4-1 _T2___ I

EXPERIMENT THEORY

TI (NOTTO SCALE)

TI

T2 T3

#

/

O

r

I_1_1 F-ICE MELTING k_ B

I I I I I

I

0 i00 200 300

0 100 200 300 0 100 200 300

TIME, sec Figure 16. - Comparison of transient temperature profiles measured in IRT with predictions of one dimensional, transient heat conduction code(ref. 38).

m

•28

/

V : 96, 145 KNOTS /

E (_}

a:2.1 °, 5.90 /

r- .m

T =25 OF, 5 OF //

E

•24

.....,.

I.L

0 ///

.20

-- 0 // 0///

_+25%/0 / /"

...I I..i,_

.16

r-,,

i/ J ///

.-..I L.I..

o/6_b///"_25 _0

.12 -- / / i/"

(..)

o/ / .;4

rm l_i_l

0 Jr, O_//'-'./f L LINEOF PERFECT

rv, CL

• 08

- __E_'ME_i .....

>, .,-.I l.--

.04

_._"'o1_ I 1 I I I I

Z ,:IZ

•04 .08 .12 .16 .20 .24

AVERAGED ANTI-ICE SPECIFIC FLUIDFLOW, SFF,

glmincm2

Figure 17.- Comparison ofpredicted and measured

minimum anti-icing flow rates for NACA 2412

(rood)airfoil with fluid freezing point ice protec-

tion system (ref. 3g).

2.54cm RADIUS COIL --x

\ \

z--O.18 cm GAP

\ _ ,,

z"

_- /,_x' o s.o

MEASURED

_._Z o__ 2.5

__ _ 400

,,::E

, / "_ XX___ PREDICTED z"

i.J.J Z "'{::3 "_C3

_- 0

r_

I I °

I I I I I

-400 " z -5.0

CIRCUMFERENTIAL STRAINAT COIL

NORMALACCELERATION AT COIL

m Z

z _ 1.5

rv' Z , I-.-- _

< 400

t--- ._.jr"*' eY Z m .._f/'_ _ _\\\

_ 0

_,,J ¢.v --n (_._

_--- 0

-- -1.5

--1 r_

I I I _/'I ''' I

I I I I I °

:_ -3.

-400 0

200 400 600 800 1000

0 100 200 300 400 500

TIME, psec

LONGITUDINAL STRAINAT COIL

NORMALACCELERATION 11.4 cm FROMCOIL

Figure 18. - Comparison of measured and predicted accelerations and strains for idealized deicer model (ref. 41).

,&

I I_ I I I _o_J__ I i I i

'-rl --°

o_'

o _ _ __

o ! I _:

0_ 1 _

- 91 __

_ I !

2. Government Accession No. 3. Recipient's Catalog No.

1. Report No.

NASA ]M-88791

5. Report Date 4. Title and Subtitle

NASA's Aircraft Icing Analysis Program

6. Performing Organization Code

505-68-li

8. Performing Organization Report No.

7. Author(s)

E-3121

Robert J. Shaw

10. Work Unit No.

9. Performing Organization Name and Address 11. Contract or Grant No.

National Aeronautics and

Space Administration

Lewi_ Research Center

13. Type of Report and Period Covered

Cleveland, Ohio 44]35

12. Sponsoring Agency Name and Address Technical Memorandum

National Aeronautics and

Space Administration 14. Sponsoring Agency Code

Washington, D.C. 20546

15. Supplementary Notes Aeronautical Sciences (ICAS),

Prepared for the International Conference of the

London, England, September 7-12, 1986.

16. Abstract

An overview of the NASA ongoing efforts to develop an aircraft icing analysis

capability is presented, Discussions are included of the overall and long term

objectives of the program as well as current capabilities and limitations of the

various computer codes being developed. Descriptions are given of codes being

developed to analyze two and three dimensional trajectories of water droplets,

airfoil Ice accretion, aerodynamic performance degradation of components and

complete aircraft configurations, electrothermal deicer, fluid freezing point

depressant antldelcer and electro-lmpulse deicer. The need for bench mark and

verification data to support the code development is also discussed and selected

results of experimental programs are presented.

18. Distribution Statement 17. Key Words (Suggested by Author(s))

Unclassified - unlimited

Aircraft icing; Analytical methods;

STAR Category 03

Computer programs

21. No. of pages 22. Price* 19. Security Classif. (of this report) 20. Security Classif. (of this page)

Unclassified Unclassified

*For sale by the National Technical Information Service, Springfield, Virginia 22161